Content uploaded by A. K. M. Abdullah Al-Amin
Author content
All content in this area was uploaded by A. K. M. Abdullah Al-Amin on Jun 20, 2024
Content may be subject to copyright.
An economic assessment of autonomous
equipment for field crops
A thesis submitted in partial fulfilment of the requirements of Harper Adams University for
the degree of Doctor of Philosophy.
By
A. K. M. Abdullah Al-Amin
MS in Agricultural Economics (Production Economics) (Bangladesh Agricultural
University)
B.Sc.Ag.Econ.(Hons.) (Bangladesh Agricultural University)
August 2023
Department of Food, Land and Agribusiness Management
Harper Adams University, Edgmond, Newport, Shropshire, TF10 8NB, UK
i
Table of contents
Table of contents
i-ii
List of tables
iii
List of figures
iv
List of publications
v
Contributions of authors
vi-vii
Research outcomes dissemination
viii
Abbreviations
ix
Declaration
x
Dedication
xi
Acknowledgements
xii-xiv
Abstract
xv
Chapter 1: General introduction
1-18
1.1
Introduction
1-2
1.2
Autonomous machines for arable field crops
2-10
1.3
Autonomous machines for sustainable intensification solutions
10-12
1.4
The research problem
12-14
1.5
Research objectives
14
1.6
Research hypotheses
14
1.7
Theoretical grounds
14-15
1.8
Research approach
15-17
1.9
Outline of the thesis
18
Chapter 2: State of the art
19-45
2.1
Introduction
19-20
2.2
Economics of field crop robotics and autonomous systems (RAS)
20-28
2.3
Field size and shape (AND/OR) autonomous machines: Whole
field sole cropping (Objective 1)
28-34
2.4
Automating mixed cropping
34
2.5
Strip cropping (AND/OR) autonomous machines: Objective 2
34-40
2.6
Regenerative agriculture (AND/OR) autonomous machines:
Objective 3
40-45
Chapter 3: Economics of field size and shape for autonomous crop
machines
46-72
3.1
Introduction
47-49
3.2
Methods
50-55
3.2.1
Field time and efficiency estimation subject to field size and shape
50-52
3.2.2
Modelling the economics of field size and shape
53-54
3.2.3
Case study and data sources
54-55
3.3
Results
55-67
3.3.1
Field efficiency and times: rectangular fields
55-57
3.3.2
Field efficiency and times: non-rectangular fields
57-59
3.3.3
Economics of rectangular fields
60-63
3.3.4
Sensitivity tests: rectangular fields
63-64
3.3.5
Economics of non-rectangular fields
64-67
3.3.6
Sensitivity tests: non-rectangular fields
68
3.4
Discussion
68-71
ii
3.5
Conclusions
71-72
Chapter 4: Economics of strip cropping with autonomous machines
73-96
4.1
Introduction
74-76
4.2
Materials and methods
76-84
4.2.1
Approach and data
76-80
4.2.2
Base economic model
80-83
4.2.3
Modelling sensitivity scenarios
83-84
4.3
Results
84-93
4.3.1
Baseline results
84-86
4.3.2
Equipment investment costs
86-90
4.3.3
Allocation of farm expenses
90-92
4.3.4
Impacts of soybean/corn price ratios and increasing demand for
human supervision
92-93
4.4
Discussion
93-96
4.5
Conclusions
96
Chapter 5: Economics of autonomous machines for regenerative
agriculture
97-111
5.1
Introduction
98-101
5.2
Materials and methods
101-106
5.2.1
Case study and data
101-102
5.2.2
Base modelling
102-105
5.2.3
Sensitivity scenarios
105-106
5.3
Results
106-109
5.3.1
Baseline results
106-108
5.3.2
Sensitivity results
108-109
5.4
Discussion
109-110
5.5
Conclusion
111
Chapter 6: General discussion and conclusions
112-120
6.1
General discussion
112-114
6.2
Limitations and future research
115-117
6.3
Conclusions
117-120
References
121-164
Appendices
165-220
iii
List of tables
Title
Page No.
Table 1.1: Example initiatives of autonomous prototype machines for
arable field crops.
4-6
Table 1.2: Example initiatives of autonomous commercial machines for
arable field crops.
6-9
Table 2.1: State of the arts of automated crop robotics.
21
Table 2.2: State of the art of autonomous crop robotics.
22-24
Table 3.1: Equipment times of the machinery sets for rectangular fields of
1 ha and 10 ha.
57
Table 3.2: Equipment times of the machinery sets for non-rectangular
fields of 1 ha and 10 ha.
59
Table 3.3: HFH-LP outcomes on the economics of technology choice
subject to different sized rectangular fields.
61
Table 3.4: HFH-LP outcomes on the economics of technology choice
subject to different sized non-rectangular fields.
65-66
Table 4.1: Comparative labour requirements and profitability of whole field
sole cropping and strip cropping practices under conventional and
autonomous machine (crop robot) scenarios in the Corn Belt of central
Indiana, US.
85
Table 4.2: Conventional larger machine inventory and costs for whole field
sole cropping in US$.
88
Table 4.3: Conventional smaller machine inventory and costs for strip
cropping in US$.
89
Table 4.4: Hardware and software needed to retrofit for autonomous
system.
90
Table 5.1: Optimization models outcomes for five-year winter wheat-
winter barley-nectar flower mix-winter wheat-spring bean rotations in the
UK arable farm.
108
iv
List of figures
Title
Page No.
Figure 1.1: Structure of the thesis and chapters overview.
18
Figure 2.1: Costs of production of wheat for conventional (triangles) and
autonomous equipment (circles) subject to farm sizes.
33
Figure 2.2: Five principles of regenerative agriculture.
41
Figure 3.1: Typical field path for rectangular fields considered in the study
based on the HFH demonstration project experience.
51
Figure 3.2: Typical field path for non-rectangular (i.e., right-angled
triangular) fields considered in the study based on the HFH demonstration
project experience.
52
Figure 3.3: Estimated (weighted average) whole farm field efficiency of
HFH equipment (i.e., 28 kW conventional equipment with human operator
and autonomous machine), large conventional and small conventional
machines with human operators in different sized rectangular fields.
56
Figure 3.4: Estimated (weighted average) whole farm field efficiency of
HFH equipment (i.e., 28 kW conventional equipment with human operator
and autonomous machine), large conventional and small conventional
machines with human operators in different sized non-rectangular fields.
58
Figure 3.5: Wheat unit cost of production in euro per ton for farms with
rectangular fields of different sized farms. The labels on the data points for
1 ha and 10 ha fields are the size of the tractor used and the number of
equipment sets. The curves without labels are the baseline analysis which
was done without field size and shape modelling.
63
Figure 3.6: Wheat unit cost of production in euro per ton for farms with non-
rectangular fields of different sized farms. The labels on the data points for
1 ha and 10 ha fields are the size of the tractor used and the number of
equipment sets. The curves without labels are the baseline analysis which
was done without field size and shape modelling.
67
Figure 4.1: Corn-soybean strip cropping field layout planted in six, 0.76 m
row strips based on Ward et al. (2016).
82
Figure 4.2: Comparative returns and expenses of whole field sole cropping
and strip cropping practices.
91
Figure 4.3: Cost elements as percentage of total costs.
92
Figure 5.1: Five-year rotational layout of regenerative strip cropping to
maximize edge effects.
105
Figure 6.1: New Hands Free Farm demonstration research of
autonomous strip cropping.
114
v
List of publications
The PhD thesis is based on the published/submitted papers given below:
Paper
Published
Paper 1
Al-Amin, A.K.M. Abdullah, Lowenberg DeBoer, J., Franklin, K. and Behrendt,
K. (2023) ‘Economics of field size and shape for autonomous crop machines',
Precision Agriculture, p. 0123456789. Available at:
https://doi.org/10.1007/s11119-023-10016-w (Accessed: 12 July 2023)
Note: The preliminary versions of the first paper were published as
proceedings papers as:
Al-Amin, A.K.M. Abdullah, Lowenberg DeBoer, J., Franklin, K. and Behrendt,
K. (2022) ‘Economics of Field Size for Autonomous Crop Machines.’, in A
paper from the Proceedings of the 15th International Conference on Precision
Agriculture Minneapolis, Minnesota, United States. pp. 1–15. Available at:
https://ageconsearch.umn.edu/record/322755?ln=en (Accessed: 12 July
2023).
Al-Amin, A.K.M. Abdullah, Lowenberg-DeBoer, J., Franklin, K. and Behrendt,
K. (2021) ‘Economic Implications of Field Size for Autonomous Arable Crop
Equipment.’ In: K. Behrendt and D. Paparas (2021) Proceedings of the 4th
Symposium on Agri-Tech Economics for Sustainable Futures. Global Institute
for Agri-Tech Economics, Food, Land & Agribusiness Management
Department, Harper Adams University. pp. 25–44. Available at:
https://ageconsearch.umn.edu/record/316595?ln=en (Accessed: 12 July
2023).
Paper 2
Al-Amin, A. K. M. Abdullah, Lowenberg‑DeBoer, J., Erickson, B. J., Evans, J.
T., Langemeier, M. R., Franklin, K., & Behrendt, K. (2024) ‘Economics of strip
cropping with autonomous machines’, Agronomy Journal, 1–18. Available at:
https://doi.org/10.1002/agj2.21536 (Accessed: 12 February 2024)
Paper 3
Al-Amin, A.K.M. Abdullah, Dickin, E., Monaghan, J.M., Franklin, K. and
Lowenberg‑DeBoer, J. (2023) 'Economics of autonomous machines for
regenerative agriculture'. Proceedings at the 14th European Conference on
Precision Agriculture, 2-6 July 2023. Bologna, Italy. Precision Agriculture’ 23.
Wageningen Academic Publishers. pp. 749-755. Available at:
https://www.wageningenacademic.com/doi/10.3920/978-90-8686-947-
3_94?fbclid=IwAR19C5uYtaNb96lWo5mdXwz6dx96kBAf2_YD9sdwcc8u7S1-
Tjx_EiAzOBA (Accessed: 18 July 2023).
vi
Contributions of authors
The contributions of authors on publications included in this PhD thesis are as follows:
Paper
Authors
Contributions
Paper 1
A. K. M. Abdullah Al-Amin
Conceptualization, Data curation, Initial and
Formal analysis, Investigation, Methodology,
Validation, Visualization, Writing - original draft,
Writing - review & editing
James Lowenberg-
DeBoer
Conceptualization, Data curation, Funding
acquisition, Investigation, Methodology, Project
administration, Software, Supervision,
Validation, Writing - review & editing
Kit Franklin
Methodology, Project administration,
Supervision, Writing - review & editing
Karl Behrendt
Conceptualization, Funding acquisition,
Investigation, Project administration,
Supervision, Writing - review & editing
Paper 2
A. K. M. Abdullah Al-Amin
Conceptualization, Data curation, Initial and
Formal analysis, Investigation, Methodology,
Validation, Visualization, Writing - original draft,
Writing - review & editing
James Lowenberg-
DeBoer
Conceptualization, Data curation, Funding
acquisition, Investigation, Methodology, Project
administration, Software, Supervision,
Validation, Writing - review & editing
Bruce J. Erickson
Conceptualization, Data curation, Investigation,
Methodology, Project administration,
Supervision, Writing - review & editing
John T. Evans
Conceptualization, Investigation, Methodology,
Resources, Supervision, Writing - review &
editing
Michael R. Langemeier
Conceptualization, Investigation, Methodology,
Writing - review & editing
Kit Franklin
Conceptualization, Methodology, Project
administration, Supervision, Writing - review &
editing
Karl Behrendt
Conceptualization, Funding acquisition,
Investigation, Methodology, Project
administration, Supervision, Writing - review &
editing
Paper 3
A. K. M. Abdullah Al-Amin
Conceptualization, Data curation, Initial and
Formal analysis, Investigation, Methodology,
Validation, Visualization, Writing - original draft,
Writing - review & editing
E. Dickin
Conceptualization, Methodology, Writing -
review & editing
J.M. Monaghan
Conceptualization, Methodology, Writing -
review & editing
vii
Kit Franklin
Methodology, Project administration,
Supervision, Writing - review & editing
James Lowenberg-
DeBoer
Conceptualization, Data curation, Funding
acquisition, Investigation, Methodology, Project
administration, Software, Supervision,
Validation, Writing - review & editing
viii
Research outcomes dissemination
Objective
Seminar Presentation
Objective 1
• Presented at the 4th Symposium on Agri-Tech Economics for
Sustainable Futures organized by organized by Global Institute for
Agri-Tech Economics, Food, Land and Agribusiness Management
Department, Harper Adams University, Newport, Shropshire,
TF108NB, UK on 20-21 September 2021. Available at:
https://www.youtube.com/watch?v=PGzGKVMnIlI&t=148s
• Presented at the DIGICROP2022 conference organized by the
German Cluster of Excellence “PhenoRob – Robotics and
Phenotyping for Sustainable Crop Production” at the University of
Bonn and the AI Institute for Next Generation Food Systems.
Available at: https://digicrop.de/a-al-amin-economic-implications-
of-field-size-for-autonomous-arable-crop-
equipment/?fbclid=IwAR0pdsW6WKJy2lKLDYK59eL3Y9INhyQ57
0FlChqwiPVori14u8H_D8MJaig
• Presented at the Annual Postgraduate Student Colloquium 2021
at Harper Adams University, UK. Available at:
https://www.youtube.com/watch?v=RUCtickJWiY&t=29s
• Presented at the 15th International Conference on Precision
Agriculture organized by International Society of Precision
Agriculture (ISPA). Available at: https://www.ispag.org/icpa
(Presented in-person by Professor James Lowenberg-DeBoer on
behalf of Al-Amin)
Objective 2
• Presented at the 5th Online Symposium on Agri-Tech Economics
for Sustainable Futures, organized by Global Institute for Agri-
Tech Economics, Food, Land and Agribusiness Management
Department, Harper Adams University, Newport, Shropshire,
TF108NB, UK on 19-20 September 2022. Available at:
https://www.youtube.com/watch?v=xNmSUkKn6Ww&t=425s
Objective 3
• Presented at the 14th European Conference on Precision
Agriculture (ECPA) 2023. Available at: https://www.ecpa2023.it/
• Presented at the 6th symposium on Agri-Tech economics for
sustainable futures to be held on 18-19 September 2023 at Harper
Adams University, UK. Will be available at:
https://www.agritechecon.co.uk/
ix
Abbreviations
AI
Artificial Intelligence
AES
Agri-Environment Schemes
BREXIT
British Exit from the European Union
BPS
Basic Payment Scheme
CSS
Countryside Stewardship Scheme
ELMS
Environmental Land Management Scheme
GAMS
General Algebraic Modelling System
GBP
Great British Pound
GHG
Greenhouse Gas
GNSS
Global Navigation Satellite Systems
GPS
Global Positioning System
HFH
Hands Free Hectare
HFF
Hands Free Farm
HFH-LP
Hands Free Hectare Linear Programming
ICT
Information and Communication Technology
LP
Linear Programming
PA
Precision Agriculture
PAR
Photosynthetically Active Radiation
PPFD
Photosynthetic Photon Flux Density
RAS
Robotics and Autonomous Systems
ROLMRT
Return to Operator Labour Management and Risk Taking
SI
Sustainable Intensification
DSS
Decision Support System
MIDAS
Model of an Integrated Dryland Agricultural System
UK
United Kingdom
US
United States
x
Declaration
The author hereby confirms that the research is originally based on the experience of the
Hands Free Hectare (HFH) and Hands Free Farm (HFF) demonstration project at the
Harper Adams University in the United Kingdom. The author himself wrote the report, with
a few definitions adopted from different sources and references cited properly. The report
has not been partially and fully submitted to achieve institutional qualification or award to
any university.
A. K. M. Abdullah Al-Amin
August 2023
xi
Dedication
The PhD thesis is dedicated to the departed soul of my beloved father (late) who was a
close friend of mine. I would like to extend my profound gratitude to him as he launched
the voyage to educate me going against his socio-economic constraints that was totally
unfavourable for him to start with. His lifelong sacrifices pushed me ahead.
My father is still alive in the great thoughts and good deeds of mine. Have a great time in
‘haven’ dear. We will meet again.
xii
Acknowledgements
First and foremost, the author would like to express his deepest sense of gratitude and
indebtedness to Professor James Lowenberg-DeBoer. As an international student, at the
very early stage of PhD research it would be difficult for the author to move ahead without
the erudite guidelines, scholastic criticisms, and kind encouragement of Professor
Lowenberg-DeBoer. The author would like to extend his gratitude to Professor James
(also known as Professor Jess to the colleagues and scientific world) for creating such a
wonderful and comfortable working environment, where the author enjoyed unrestricted
access and idea sharing facilities when necessary. Throughout the PhD journey,
Professor James acted not only as a professor, but as a mentor and a real friend.
Professor James navigates research idea conceptualization, analyses and corrections of
original manuscripts published/submitted based on PhD research objectives. Apart from
PhD research, the author also learnt regarding student supervising modes and norms that
are expected to help ahead as the author himself is working as a staff at Bangladesh
Agricultural University as an Assistant Professor.
In the same fashion, the author would like to offer his sincere gratitude and respect to
Professor Karl Behrendt for his constructive advice, erudite guidance, encouragement,
and kind cooperation throughout the PhD research work. Professor Karl always helps in
fine tuning the research ideas and original manuscripts. The unrestricted access and idea
sharing scope helps the author heading in the right direction. Especially the outcomes of
the PhD research were dissemination and recommended by Professor Karl in social
media and to the scientific community that motivated the author in a good manner.
The author is also grateful to the Principal Investor of the Hands Free Hectare and Hands
Free Farm (HFH&HFF) and Senior Engagement Fellow of Engineering at Harper Adams
University, Kit Franklin for his constructive comments and suggestions in the PhD
research and original manuscripts. Acknowledgement also goes to the HFH&HFF teams,
especially to Mike Gutteridge for his constructive guidelines.
The author feels proud to express his profound respect to Harper Adams University in the
UK and Elizabeth Creak Charitable Trust for funding the PhD study. The author would
also like to express his gratitude to the devoted researchers and academics at Harper
Adams University, worth mentioning the name of Senior Lecturer, E. Dickin and Professor
J. M. Monaghan. Both of their constructive criticisms help in ex-ante economic modelling.
Sincere gratitude also goes to the first - and second - years examiners of the author’s PhD
research, Professor, Richard Godwin and Senior Lecturer, Sven Peets for their very
xiii
cordial and erudite comments and suggestions. The author is also grateful to the Head of
Food, Land and Agribusiness Management Department, Rebecca Payne for her very kind
co-operation and logistic support. Acknowledgement also goes to Postgraduate Research
Student Advisor, Emma Hancox; Director of Postgraduate Research, Martin Hare; and
Associate Pro Vice-Chancellor, Dawn Arnold for timely co-operations. Sincere gratitude
goes to the Food, Land and Agribusiness Management Department, Bamford Library,
Service desk, Postgraduate office, and finally to all members of Harper Adams University.
The author would like to thank all his colleagues and everybody who contributed to his
research journey. Dr. Mohammed Rashed Chowdhury, Harper Alumni deserves thanks for
ensuring cordial help at Harper during initial settlement of Harper dormitory. Thanks to
Senior Lecturer, Karim Farag, and his family for helping the author in family settlement
when author’s wife visited Lilleshall, Shropshire, UK for a short period of time. Hello
Harper colleagues and well-wishers, sorry! not to mention all of your names here, but you
all are in my memory. I am really grateful to Iqbal Hussain and his family based on
Telford, Shropshire, UK for their kind family support, worth mentioning delicious
Bangladeshi cuisine.
The co-authors and mentors from Purdue University, West Lafayette, Indiana, US also
deserve cordial gratitude. During author’s affiliation as a Visiting Scholar at the Agronomy
department of Purdue University, the author exchanged research ideas to conceptualize
the second PhD research objective and complete the ex-ante analysis. Professor Bruce
Erickson of Agronomy education Distance and Outreach Director, John T Evans of
Department of Ag and Biological Engineering and Professor Michael R Langemeier of
Department of Agricultural Economics and Associate Director, Center for Commercial
Agriculture, deserve thanks for valuable comments and suggestions.
The author is also indebted to colleagues of Agronomy and Farm Power and Machinery
departments of Bangladesh Agricultural University, worth mentioning the name of
Professor Md. Harun Or Rashid, Professor Md. Rostom Ali and Professor Md. Hamidul
Islam for their valuable suggestions in agronomic and technical affairs. Sincere gratitude
to Computer Science and Engineering graduate Mr. Mazharul Evan and Engineering of
Data-intensive Intelligent Software Systems graduate Md Raisul Kibria for your valuable
suggestions.
The authors’ father, Abdur Rouf is always an inspiration for the author. The friendly
behaviour of the author’s late father always went beyond his fatherly attitude. Mr. Rouf
was really a best and Almighty Allah gifted friend of mine. The author still recalls the
memory and pursues the moral education of his father. The advice of the author’s father:
xiv
“Try to educate yourself and achieve the highest degrees of the world and use the
acquired knowledge for the welfare of the universe” is a lifetime learning for the author.
The author would like to take the opportunity to express his love and gratefulness to his
beloved wife Tahmina Akhter, Assistant Superintendent of Bangladesh Police. Without her
constant support, the author would not be able to complete the PhD degree. Staying away
from family for PhD degree was really a great challenge for the author, where Mrs.
Tahmina holds the hands of the author with her loveable words and motivation. The
author's relationship with his wife is a hub of peace and encouragement.
Finally, acknowledgement goes to the author’s family members for their spiritual support,
especially to his beloved mother Minuara Begum, brother A. K. M. Maniruzzaman, sister
Rubina Sultana, brother-in-law Md. Jashim Uddin, nephews Rhythm and Mahib, and
nieces Redita and Manha. Acknowledgement also goes to authors in in-law’s family.
Last but not least, gratitude goes to the Almighty Allah. Finally, the authors would like to
offer his profound respect to all the farmers all over the world for ensuring food for all and
motivating the author through their struggles with risk and uncertainty.
The Author
A. K. M. Abdullah Al-Amin
xv
Abstract
Research suggests autonomous machines in open field arable farming can enhance
biodiversity conservation and ecosystem services restoration. It is hypothesized that
autonomous equipment could be a profitable alternative to conventional machines with
human operators irrespective of field size and shape or cropping systems. However, lack
of agronomic, economic and technical data has constrained economic assessment.
Noting this, this study evaluated the economics of field size and shape, and mixed
cropping with autonomous machines using the Hands Free Hectare and Hands Free Farm
(HFH&HFF) demonstration experience of Harper Adams University, UK. Using the Hands
Free Hectare Linear Programming (HFH-LP) optimization model results indicated that
autonomous machines in British farming decreased wheat production cost by €15/ton to
€29/ton for small rectangular fields and €24/ton to €46/ton for small non-rectangular fields.
Sensitivity scenarios of increasing wage rates and labour scarcity shows that autonomous
farms adapted easily and profitably to changing scenarios, whilst conventional
mechanized farms struggled. The ex-ante economic analysis of corn-soybean strip
cropping in the North American Corn Belt of Indiana found that per annum return to
operator labour, management and risk-taking (ROLMRT) was $568.19/ha and $162.58/ha
higher for autonomous strip cropping as compared to whole field sole cropping and
conventional strip cropping. Conventional strip cropping was only feasible with a
substantial amount of labour availability. The ex-ante economic analyses of wheat - barley
- flower mix - spring bean regenerative strip cropping practices show that for Great Britain
autonomous regenerative strip cropping ROLMRT was £57,760 and £25,596 higher
compared to whole field sole cropping and conventional regenerative strip cropping
practices. The profitability of autonomous machines in small fields irrespective of field size
and shape, strip cropping systems and regenerative practices imply that autonomous
machines could offer a win-win farming solution that help achieve the production and
environmental goals of arable farming.
1
General introduction
Chapter 1
General introduction
“Autonomous field management represents the next evolutionary step in agricultural
technology. ... The integration of multifaceted objectives into a common decision-making
process poses a great challenge to human farmers and their capacities. Liberated from
labour constraints, autonomous systems have the potential to align decisions with the
complex requirements of multiple − even contradicting – goals more easily, and to execute
them accordingly without exhaustion.”
Gackstetter et al. (2023): Agricultural Systems, 206, p.103607.
1.1 Introduction
Robotics and autonomous systems (RAS) for arable open-field crop farm operations are
being introduced worldwide to reduce social, economic and environmental costs of
farming (Duckett et al., 2018; Rose and Chilvers, 2018; Lowenberg-DeBoer et al., 2021a;
Pearson et al., 2022). The robots for livestock rearing (e.g., performing operations like
milking, feeding, barn cleaning and silage handling) have developed more rapidly as this
technology is similar to industrial robots and requires less mobility and decision-making
capacity due to the structured environment. The pioneers among crop robots were the
greenhouse robots operated on rails in a controlled environment (Lowenberg-DeBoer et
al., 2020; Daum, 2021). However, the farming environment of arable open-field crops is
more complex and unstructured. Arable crop farm operations depend on the risk and
uncertainty of weather, soil attributes, travelling between fields, rolls and slopes, and other
farm level challenges, and legal and policy constraints (Bechar and Vigneault, 2016;
Grieve et al., 2019; Fountas et al., 2020; Shockley et al., 2021; Kubota, 2023).
Open-field crop robots are currently used for agricultural tasks (i.e., land preparation,
transplanting, seeding, plant protection, weed control and harvesting) and supporting
tasks (i.e., guidance, navigation, mapping, and localization) to maximize production, and
environmental and food safety (Bechar and Vigneault, 2016; Bellon-Maurel and Huyghe,
2017; Davies, 2022; Finger et al., 2019). The development of mechatronic technology,
information and communication Technology (ICT), increasing agricultural labour scarcity
and higher demand for food and nutritional security has pushed arable crop farming
towards crop robotics (Duckett et al., 2018; Lowenberg-DeBoer et al., 2020). Arable farm
2
General introduction
machines having some autonomy and mobility in operations are simultaneously termed
‘robots’, ‘field crop robots’, ‘automated machines’, and ‘autonomous machines’.
The term ‘robots’ refers to the equipment with decision making capacity through the use of
artificial intelligence (AI) (Kyriakopoulos and Loizou, 2006; Lowenberg-DeBoer et al.,
2021a). In this study ‘field crop robot’ indicates “a mobile, autonomous, decision making,
mechatronic device that accomplishes crop production tasks (e.g., soil preparation,
seeding, transplanting, weeding, pest control and harvesting) under human supervision,
but without direct human labour”. The term ‘automated machines’ refers to the partially
robotized mechatronic technology that accomplish arable field operations such as
seeding, weeding, and harvesting, but with mobility assured by a human operator
(Lowenberg-DeBoer et al., 2020). In this study ‘autonomous machines’ (or ‘autonomous
crop robots’ or ‘autonomous crop machines’) are a subset of field crop robots which have
autonomy in arable field operations using predetermined field paths and itinerary often
with relatively little decision-making capacity.
Autonomous machines are precision agriculture (PA) technology because they have the
potential of cost effectively increasing the precision of input applications and to collect
very detailed data on agricultural production. Autonomous machines considered in this
study are modelled on the arable field crop machines of the Hands Free Hectare (HFH)
(https://www.handsfree.farm/) project demonstrated at the Harper Adams University
(HAU) in the UK (Hands Free Hectare (HFH), 2021). The HFH autonomous machines are
also labelled as ‘swarm robots’ (or ‘swarm robotics’) because multiple units of smaller
equipment were used to accomplish arable farm operations that would typically be done
by a larger conventional machine with a human operator (Lowenberg-DeBoer et al.,
2021a).
1.2 Autonomous machines for arable field crops
Worldwide autonomous machines for arable field crops are in an early wave of the
development and commercialization processes (Shockley et al., 2021; Lowenberg-
DeBoer, 2022b). The market for autonomous machines is expanding with a robust growth
in large and medium scale farming contexts, such as Australia, the UK, US and European
Union. Research in the US and UK found that small and medium sized farms could
purchase low cost small autonomous machines as such farms are potentially profitable
with autonomous farming (Shockley, Dillon and Shearer, 2019; Lowenberg-DeBoer et al.,
2021a). Autonomous machines could be the future of arable farming (Hekkert, 2021).
Recent market report shows that autonomous machines manufacturers are investing
substantially in research and development of equipment where industry players expect
3
General introduction
that the market for autonomous machines would be a US$ 150 billion industry by the near
future in 2031. The report also pointed out that apart from large and medium scale
farming, smallholders farming could be the thriving market for autonomous tractors and
harvesters (Claver, 2021).
The interest in autonomous machines is increasing in smallholder’s contexts because of
the challenges of agricultural labour scarcity, aging of farmers, reluctance of young people
to choose agriculture as a career and their desire for off-farm employment opportunities in
the city (Feike et al., 2012; Yanmar, 2017; Tofael, 2019; Devanesan, 2020; Al-Amin and
Lowenberg-DeBoer, 2021; World Bank, 2021a). Recent market reports identified that the
autonomous equipment market for smallholders is swelling dramatically (Devanesan,
2020; Xinhuanet, 2020; Business Wire, 2021; PR Newswire, 2022). Countries like China
and Bangladesh are taking initiatives for rural revitalization strategies, national
digitalization vision and smart farming strategy that include automation (Business Wire,
2021; Bangladesh Delta Plan, 2018; Al-Amin and Lowenberg-DeBoer, 2021; Globe
Newswire, 2022; Al-Amin, Lowenberg-DeBoer and Mandal, 2023). Among smallholders of
Asia, more proactive autonomous on-field trails have been demonstrated by universities,
research institutes, and agribusinesses in Japan (Farm Equipment, 2021; Ministry of
Agriculture Forestry and Fisheries (MFF), 2022; Nature, 2022; Yanmar, 2022), China
(New China, 2018; Aguilar, 2021; Qin, 2021), Philippine (Bautista et al., 2021) and
Thailand (Precision Farming Dealer, 2022c).
Apart from developing new machines, retrofitting conventional machines for autonomy has
received growing attention worldwide (Karsten, 2019a; Koerhuis, 2021a; Azevedo, 2022;
Lowenberg-DeBoer, 2022b; Torres, 2022). Agribusiness innovators have been marketing
retrofit kits for autonomy (Andrews, 2020; Advanced Navigation, 2022; Claver, 2022a;
Future Farming, 2022; Precision Farming Dealer, 2022a; Sveaverken, 2022).
Autonomous machines as a service model (i.e., custom hire service) has also been
initiated by service companies (Claver, 2020a, Claver, 2022d; Wilde, 2020). Academics
and researchers hypothesized that part of the future market may be captured by the
service model, like Uber or other modes of custom hire services (Lowenberg-DeBoer et
al., 2020; Al-Amin and Lowenberg-DeBoer, 2021; Daum, 2021; Al-Amin, Lowenberg-
DeBoer and Hasneen, 2022; Al-Amin et al., 2023).
The initial development of autonomous machines for arable field crops were prototypes
that were demonstrated by universities and research institutes on parking lots and
playgrounds. A few on-field trials were carried out for specific crop(s) and/or operation(s)
4
General introduction
(Lowenberg-DeBoer et al., 2020; Lowenberg-DeBoer, 2022b). In the last few decades,
commercial manufacturers worldwide, ranging from commercial giants to start-ups have
initiated development of autonomous machines to revolutionize arable crop farming. Many
companies have developed autonomous prototypes and a smaller number have moved
towards commercialization. It is often not exactly clear which are prototypes or
commercial because limited information is publicly available about technology scaling up.
This study classified the example initiatives as prototype or commercial. Commercial
initiatives are considered those which have public news available about commercialization
or the company itself announces this as commercial. Example initiatives of autonomous
prototype (Table 1.1) and commercial (Table 1.2) machines mainly for arable field crops
are given as follows.
Table 1.1: Example initiatives of autonomous prototype machines for arable field crops.
Year
Autonomous
milestones
Primary
operation
considered
Enterprise
Company or
organization
Country
Reference
2011
Autonomous
operating
systems
Till, puddle,
transplant
and harvest
Rice
National
Agriculture
Research
Organization
(NARO)
Japan
Nagasaka
et al. (2011)
2017
Autonomous
machines
Whole farm
operations
(Plant to
harvest)
Wheat,
oilseed
rape,
barley,
beans and
grass ley
Harper
Adams
University
(HAU)
UK
Hands Free
Hectare
(HFH),
(2021)
2018
Driverless
tractors
Plough,
rake, seed,
fertilizer
and mulch
Cotton
Lovol, and
South China
Agricultural
University
(SCAU)
China
New China
(2018)
2019
Autonomous
ground
vehicle
Plant,
spray,
fertilizer,
crop health
monitor
and cover
crop seed
Wheat,
soybeans,
corn and
sorghum
Easton
Robotics
US
Groeneveld
(2021a) and
Easton
Robotics
(2023)
2019
Autonomous
weeders
Weed
Vegetables
FarmWise
and Roush
US
Claver
(2019) and
FarmWise
(2023)
2020
Autonomous
rice-
transplanter
Transplant
Rice
Kubota
Tractor
Corporation
Japan
Kubota
(2023)
5
General introduction
Table 1.1: Example initiatives of autonomous prototype machines for arable field crops
(Continued).
2020
Yanmar’s
agro-bot
Monitor
crops,
detect and
treat
diseases,
soil
sample,
and spray
Vineyard
and
spinach
Yanmar
R&D
Europe
(YRE) and
Florence
University
Italy
Claver
(2020b)
2020
Fendt Xaver
robots
Plant,
protect,
weed
control
and
fertilizer
Grains
AGCO
US
Fendt
(2020) and
Fendt
(2023)
2021
Autonomous
implement
carriers
Seed,
plant,
weed and
grass cut
Cereals,
vegetables
and
vineyard
3D Radar
AS and
Norwegian
University
of Science
and
Technology
Norway
AutoAgri
(2023)
2021
Uncrewed
agricultural
machinery
Transplant
and
harvest
Wheat and
rice
AIForce
Tec
China
Qin (2021)
2021
Autonomous
Hand
Tractor
Till
Rice
University
of Santo
Tomas
Philippine
Bautista et
al. (2021)
2021
Autonomous
tractor
swarm
Till
Open field
Yanmar
and
Hokkaido
University
Japan
Yanmar
(2021)
2021
Autonomous
asparagus
harvester
Harvest
Asparagus
University
of Waikato
and
Robotics
Plus
New
Zealand
Groeneveld
(2021b)
2022
Autonomous
electric
planter
Plant
Corn and
soybean
Salin 247
US
Salin 247
(2022)
2022
Autonomous
diesel-
electric field
robots
Cultivate,
plough,
sow, mow,
ted and
rake
Forage
Krone /
Lemken
Germany
Krone
(2022)
2022
Horsch
autonomous
robot tractor
Plant
Open field
Horsch
Germany
TractorLab
(2022) and
Azevedo
(2023a)
2022
Autonomous
weeding
robot
Weed
Carrot and
onion
Ulf
Nordbeck
Sweden
Koerhuis
(2022b)
and Ekobot
(2022)
6
General introduction
Table 1.1: Example initiatives of autonomous prototype machines for arable field crops
(Continued).
2022
Autonomous
weeding
robot
Weed
Carrot
and
onion
Odd.Bot
The
Netherlands
Koerhuis
(2022b)
and
Odd.Bot
(2023)
2023
Autonomous
Flex-Ro
Robot
Phenotype
data
collect
Cereals
University
of
Nebraska
(UNL)
and
Farmobile
US
Asscheman
(2023) and
University
of
Nebraska -
Lincoln
(UNL)
(2023)
Source: Author's own compilation.
Table 1.2: Example initiatives of autonomous commercial machines for arable field crops.
Year
Autonomous
milestones
Primary
operation
considered
Enterprise
Company or
organization
Country
Reference
2011
Naio
Technologies
Hoe, weed,
furrow,
seed,
transport
and
harvest
Vegetables
and
vineyards
Naio
Technologies
French
Naio
Technologi
es (2023)
2019
ROBOTTI
autonomous
robot
Seed,
weed and
spray
Cereals,
oilseeds,
vegetables
and grass
for seeds
Agrointelli
Denmark
Agrointelli
(2019)
2019
Ecorobotix
Weed
Row crops,
vegetable
crops and
grassland
Ecorobotix
SA
Switzerland
EcoRobotix
2023)
2019
GUSS
autonomous
herbicide
sprayer
Spray
Orchard
GUSS
(Global
Unmanned
Spray
System)
US
Agromillora
(2023) and
GUSS
(2023)
2019
Greenbot and
X-pert
Plough,
mow, sow
and
fertilizer
Vegetables,
horticultural
crops,
orchard and
golf courses
Precision
Makers
The
Netherlands
Van
Hattum
(2019) and
Precision
Makers
(2023)
2019
Pixelfarming
Robot One
Plant, hoe
and crop
protect
Cereals,
vegetables
and flower
Pixelfarming
Robotics
The
Netherlands
Pixelfarmin
g Robotics
(2019)
7
General introduction
Table 1.2: Example initiatives of autonomous commercial machines for arable field crops
(Continued).
2019
XAG R150
Unmanned
Ground
Vehicle
Spray,
weed,
crop
monitor
and on-
farm
transport
Orchard and
vegetables
XAG
China
XAG (2023)
2020
Autonomous
electric robot
Seed
and
weed
Rapeseed
and
vegetables
(e.g., sugar
beets, onions,
spinach, and
salad)
FarmDroid
UK
FarmDroid
(2020)
2020
Monarch MK-
V tractor
Till,
weed
and
spray
Vineyard
Monarch
US
Monarch
(2023)
2020
Sitia
autonomous
robot
Till,
spray
and hoe
Gardens, tree
crops, and
vineyards
Sitia
France
Sitia (2023a)
2020
Harvest
CROO
Harvest
Strawberry
Harvest
CROO
Robotics
US
Harvest CROO
(2020) and
Koerhuis
(2020)
2021
H2L
autonomous
robotics
Crop
protect
Flower (tulip)
H2L
Robotics
The
Netherland
s
H2L Robotics
(2021)
2021
E-tract
Weed,
spray
and trail
Vegetable,
vineyard and
flowers
Elatec
France
Koerhuis
(2021b) and
Elatec (2023)
2021
EarthAutomat
ions Dood
Plough,
spray,
top,
shoot
remove,
disease
detect
and pest
detect
Cereals,
vegetables
and vineyard
EarthAutom
ations
Italy
EarthAutomati
ons Dood
(2023) and
Future
Farming
(2023)
2021
Pek
Automative
Slopehelper
Till,
fertilizer,
prune
and
spray
Orchard
Pek
Automative
Slovenia
Pek
Automative
(2021)
2022
Autonomous
larger tractor
Plough
Commodity
crops
John Deere
US
John Deere
(2022)
2022
CNH
autonomous
solutions
Till,
spray
and
harvest
Commodity
crops
CNH
Industrial
Italian-
American
multination
al
corporation
CNH
Industrial
(2023)
8
General introduction
Table 1.2: Example initiatives of autonomous commercial machines for arable field crops
(Continued).
2022
La Chevre
Weed
Vegetables
Nexus
Robotics
Canada
Agtecger
(2022) and
Nexus
Robotics
(2022)
2023
Raven
autonomous
solutions
Till, spray
and
harvest
Cereals
and
vegetables
Raven
Industries
US
Bedord
(2022) and
Raven (2023)
2022
AgBots
Soil and
seedbed
prepare,
seed,
spray, roll
and mow
Cereals,
cover crop,
and grass
AgXeed
The
Netherlands
AgXeed
(2023)
2022
Swarm Farm
Robotics
Crop
protect,
Mow and
slash
Grain,
cotton and
grass
Swarm
Farm
Robotics
Australia
Groeneveld
(2023a) and
SwarmFarm
Robotics
(2023)
2022
AIGRO UP
Weed
and mow
Orchared
AIGRO
The
Netherlands
AIGRO
(2023)
2022
Amos Power
A3/A4
Till, inter-
seed,
mow and
spray
Cereals
and
vineyards
Amos
power
US
Precision
Farming
Dealer
(2022b) and
Amos (2023)
2022
Directed
Machines
Land Care
Robot
Light
plough,
spray,
mow and
trim
Grass and
orchard
Directed
Machines
US
Bloch (2022)
and Directed
Machines
(2023)
2022
Exxact
robotics
Till and
spray
Vineyard,
cereals and
vegetables
Exxact
Robotics
France
Exxact
Robotics
(2023)
2022
Robotics
Plus
Unmanned
Ground
Vehicle
Spray,
weed
control,
mulch,
mow and
crop
analyse
Orchard
Robotics
Plus
New
Zealand
Power and
Motion
(2022) and
Robotics
Plus (2023)
2022
VitiBot
Bakus
Plough,
weed,
mow and
spray
Vineyard
SAME
Deutz
Fahr (SDF
Group)
French
Vitibot (2023)
2023
Korechi
RoamIO
Cultivate,
seed,
weed,
mow, soil
sample
and data-
log
Cereal,
vineyard,
golf course,
Korechi
Canada
Korechi
(2023)
9
General introduction
Table 1.2: Example initiatives of autonomous commercial machines for arable field crops
(Continued).
2023
Smart
Machine
Oxin
Mow,
mulch, trim
and spray
Vineyard
The Smart
Machine
Company
New
Zealand
Koerhuis
(2023) and
Oxin (2023)
2023
Trabotyx
autonomous
robot
Weed
Carrot
Trabotyx
The
Netherlands
Trabotyx
(2023)
2023
Tevel Flying
Harvest
Robots
Pick, thin,
and prune
Orchard
Tevel
Aerobotics
Technologies
Israel
Agtecher
(2023) and
Tevel Tech
(2023)
2023
Solix Hunter
and Sprayer
Monitor,
map,
protect and
spray
Soybean,
corn,
sugarcane
and cotton
Solinftec
Brazil
Azevedo
(2023b)
and
Solinftec
(2023)
Source: Author's own compilation.
Although autonomous machines are increasingly used for the production of grain-oilseed
(Shockley, Dillon and Shearer, 2019; Lowenberg-DeBoer et al., 2021a), forage,
vegetables, fruits and tree nursery (Sitia, 2020; Edwards, 2021; Koerhuis, 2021b; H2L
Robotics, 2023; Sitia, 2023b), the present study concentrated on autonomous grain-
oilseed farms because of data availability from Hands Free Hectare and Hands Free Farm
(HFH & HFF) and the worldwide market implications of autonomous grain-oilseed
production. The grain-oilseed farms, especially for medium and large-scale farming
contexts are already mechanized, so the transition towards autonomous farming should
be relatively easy (Gackstetter et al., 2023) compared to the fruit and vegetable farms
which still depend heavily on manual labour. The engineering challenge for grain-oilseed
farms is primarily making already mechanized systems autonomous.
The HFH & HFF also demonstrated autonomous grass ley cutting that could have
implications for forage harvesting. Prototypes such as Krone and Lemken developed
autonomous combined powers for forage production including operations such as
cultivating, ploughing, sowing, mowing, tedding and raking (Claver, 2022c; Krone, 2022).
There are few commercial autonomous initiatives for grass ley production as detailed in
Table 1.2.
The HFH & HFF at Harper Adams University, UK, were the world’s first whole farm
commercial autonomous grain-oilseed farming public demonstration (including planting,
spraying and harvesting). HFH was initiated in 2016 with the first harvest in 2017. Major
agricultural machinery companies have had autonomous machine development programs
10
General introduction
for many years and may have completed full cropping cycles with autonomous machines,
but their results are proprietary. HFH was a simplified farming system with one hectare of
a single crop. The HFH focus was on cost effective retrofitting of conventional farm
machines for autonomy using modified open-source drone software. HFF scaled up
autonomous farming to 35 hectares with several crops, using commercial auto-guidance
systems. At the initial stage, HFH was concentrated on whole field sole cropping
(Lowenberg-DeBoer et al., 2021a). In 2023, the HFF has extended the focus with
demonstration trials of strip cropping to show the relevance of autonomous machines for
agroecological and regenerative farming (Franklin, 2022; Harper Adams University (HAU),
2023). Most autonomous initiatives worldwide other than HFH & HFF for grain-oilseed
farming are highly concentrated on specific field operations (e.g., seeding, weeding and
spraying) rather than whole farm production operations.
1.3 Autonomous machines for sustainable intensification solutions
Autonomous machines are the potential successors of large conventional machines with
human operators that could lead to a paradigm shift of arable farming (Goense, 2005;
Shockley, Dillon and Shearer, 2019, Shockley et al., 2021; Revell, Powell and Welsh,
2020). It is hypothesized that autonomous machines have the possibility to revolutionize
PA and facilitate the ‘Fourth Agricultural Revolution’ which is also labelled as ‘Agriculture
4.0’ (Klerkx and Rose, 2020). A number of potential benefits are hypothesized for
autonomous arable farming that could promote sustainable intensification solutions
(Duckett et al., 2018; Daum, 2021).
Autonomous machines could help to solve the problem of agricultural labour shortage and
thereby help to feed the growing population of the world (Kolodny and Brigham, 2018).
The list of economic benefits of autonomous machines usually start with labour saving
because worldwide agricultural labour is scarce and agricultural real wage rate is
increasing over time (Lowenberg-DeBoer et al., 2021a; Lowenberg-DeBoer, 2022b). The
availability of agricultural labour is one of the prime challenges for medium and large-scale
arable farming (OECD, 2020; Charlton and Castillo, 2021; World Bank, 2021a; The
Environment Food and Rural Affairs Committee, 2022) owing to the economic and political
reasons (e.g., BREXIT, new immigration policies for COVID-19 pandemic) (Shockley et
al., 2021; Sandford and Hanrahan, 2022; The Migration Observatory, 2022). Smallholders
around the world also face labour scarcity in agriculture (World Bank, 2021b) due to socio-
economic reasons (Devanesan, 2020; Al-Amin and Lowenberg-DeBoer, 2021; Yanmar,
2021).
11
General introduction
Small autonomous machines have numerous benefits beyond labour saving potential
such as efficiency, reliability, accuracy, economies of size, lower machinery investment
costs, higher field work rates, timeliness of operations, working 24/7, increasing labour
and land productivity, and profit maximization (Shockley, Dillon and Shearer, 2019,
Shockley et al., 2021; Farm Equipment, 2021; Lowenberg-DeBoer et al., 2021a).
Autonomous machines have the potential of reducing off-target application with localized
on-the-go application of pesticides, herbicides and fertilizer, plant specific husbandry and
collection of on-field data supporting farm management decision making (Duckett et al.,
2018; Daum, 2021; Lowenberg-DeBoer, 2022b).
Apart from facilitating production goals (i.e., least cost of production and profit
maximization) (Shockley, Dillon and Shearer, 2019; Lowenberg-DeBoer et al., 2021a)
autonomous machines have the potential to support environmental goals of farming
(Ditzler and Driessen, 2022; Pearson et al., 2022; Gackstetter et al., 2023). In the longer
run, the biggest impact of autonomous machines may not only be confined to technical
and economic (i.e., techno-economic) feasibility, rather extend to environmental
sustainability. Small autonomous machines are expected to reduce environmental
footprints of agriculture through reducing soil compaction and carbon footprint (Chamen et
al., 2015; Asseng and Asche, 2019; Karsten, 2019b; McPhee et al., 2020; Revell, Powell
and Welsh, 2020; Keller and Or, 2022; AutoAgri, 2023).
Autonomous machines are expected to help restore in-field biodiversity that has been
reduced through whole field sole cropping with larger conventional machines with human
operators (Blackmore, Have and Fountas, 2001; Robinson and Sutherland, 2002; Duckett
et al., 2018; Santos and Kienzle, 2020; Lowenberg-DeBoer et al., 2021a). Autonomous
machines are hypothesized to be capable of farming small, irregularly shape fields that
will reduce land consolidation pressure, promote hedges, wetlands and in-field trees
(Lowenberg-DeBoer et al., 2021a). Agricultural intensification solutions are suggested
through addressing spatial and temporal (i.e., spatio-temporal) heterogeneity with
autonomous mixed cropping systems (Slaughter, Giles and Downey, 2008; Ward, Roe
and Batte, 2016; Tanveer et al., 2017; van Oort et al., 2020; Ditzler and Driessen, 2022;
Donat et al., 2022) that could reduce synthetic input use, pest and diseases infestation,
improve soil health, ecosystem services, and soil carbon and nitrogen.
Although many of the early-stage autonomous machines are powered by fossil fuels (i.e.,
typical diesel-powered combustion engines) (e.g., HFH & HFF autonomous machines), an
increasing number of autonomous machines are powered by alternative renewable
electricity from solar, wind, methane and hydrogen, etc. (FarmDroid, 2020; Hekkert, 2020;
12
General introduction
Fuel Cells Works, 2021; Vale, 2021; Claver, 2022b; Groeneveld, 2022; Hein, 2022;
Karsten, 2022). For example, battery-based autonomous electric machines are suggested
to reduce greenhouse gas emissions and increase driveline efficiency. Research in the
context of Swedish agriculture using systems analysis, economic analysis and life cycle
assessment found that autonomous electric tractors reduced energy use, per annum
costs, soil compaction and greenhouse gas emissions (Lagnelöv, 2023).
Autonomous machines could facilitate sustainable intensification by integrating
multifaceted goals in a common decision-making process. Integrating those goals is often
too complex with human operated conventional machines. The diverse goals of
individuals (i.e., increasing productivity and/or profit maximization) and society as a whole
(i.e., environmental sustainability) could be jointly optimized with use of autonomous
machines. For instance, autonomous machines could facilitate the net zero agricultural
goal through facilitating agroecological and regenerative farm management. These
cropping systems will help to achieve simultaneously food and nutrition security, and
environmental sustainability (DEFRA, 2020; Davies, 2022; Pearson et al., 2022).
As part of the agricultural intensification solution of autonomous machines, the present
study considered labour saving potential, opportunity costs of capital investment, higher
field work rates, and timeliness of operations. Moreover, the advantage of mixed cropping
systems (i.e., agroecological strip cropping and regenerative agriculture) with autonomous
machines was also considered to reconcile the production and environmental goals of
arable open-field crop farming. Other anticipated benefits mentioned above were not
included in the present study due to lack of data.
1.4 The research problem
Research on autonomous machines shows that autonomous machines could solve the
real-world problems of arable crop farming as part of sustainable intensification solutions
(Lowenberg-DeBoer, 2022b). The technical potential of autonomous machines are well
accepted through worldwide prototypes and commercial on-farm demonstrations
(Shamshiri et al., 2018; Fountas et al., 2020; Hands Free Hectare (HFH), 2021).
Economic research has been focusing on guiding wide scale adoption through farm
profitability assessment with autonomous farming systems. Prior to 2018, the economic
research on autonomous operations mostly concentrated on horticultural crops
(Lowenberg-DeBoer et al., 2020). Recently, some studies focused on the economics of
autonomy for arable cereal farming. For instance, most recent autonomous farming
research focused on the profitability of this precision agriculture technology on whole field
sole cropping commodity crops production considering the context of the UK (Lowenberg-
13
General introduction
DeBoer et al., 2020, Lowenberg-DeBoer et al., 2021a) and the US (Shockley, Dillon and
Shearer, 2019). Some studies also examined the implications of regulation on economics
of autonomous farming (Shockley et al., 2021; Maritan et al., 2023). The study of
Lowenberg-DeBoer et al. (2021a) hypothesized that autonomous machines could facilitate
biodiversity by superseding the ‘‘get big or get out’’ rule of thumb of conventional
mechanization through facilitating farm operations in small, irregularly shaped fields
farmed with whole field sole cropping systems. But they were unable to assess the
hypothesis owing to the lack of data on field times (h/ha) and field efficiency (%).
Apart from whole field sole cropping, heterogeneous within field mixed cropping systems
are also envisaged with autonomous machines (Slaughter, Giles and Downey, 2008; van
Oort et al., 2020; Juventia et al., 2022; Ward, Roe and Batte, 2016). Research also
hypothesized that autonomous machines would facilitate the regenerative agriculture
practice which could help to achieve the net zero target in addition to the production goals
of arable cereal farm (Davies, 2022; Pearson et al., 2022; Manshanden et al., 2023).
Delving into the state of the knowledge it is clear that research on autonomous machines
in whole field sole cropping system is still unable to address the implications of field size
and shape (Lowenberg-DeBoer et al., 2021a). Understanding the economics of field size
and shape for autonomous machines is crucial because over the last few decades,
conventional machines with human operators have been largely motivated by the rule of
thumb of conventional machines (i.e., ‘’get big or get out’’) to achieve labour productivity in
arable farming. Beyond unlocking the economics of autonomous machines for whole field
sole cropping system subject to field size and shape, this study attempted to address the
economics of autonomous machines for mixed cropping and regenerative agriculture.
Mixed cropping systems and regenerative agriculture are suggested with autonomous
machines to simultaneously achieve both the production goals of productivity and
profitability, and environmental goal of limiting environmental footprints of arable crop
farming (Duckett et al., 2018; Daum, 2021; Pearson et al., 2022; Davies, 2022). In this
study, strip cropping system is considered to represent mixed cropping and regenerative
agriculture practice because strip cropping is the simplest mixed cropping system. It is
feasible with conventional mechanization with human operators, but requires more labour
than conventional whole field production (Ward, Roe and Batte, 2016; Exner et al., 1999;
van Apeldoorn et al., 2020; Alarcón-Segura et al., 2022). The mixed cropping and
regenerative strip cropping practices may be less profitable for conventional mechanized
farms operated with human operators, while autonomous machines may the change the
cost calculus. This study assumed that profitability assessment will guide the
autonomous machines adoption because farm economics is one of the prime drivers for
14
General introduction
technology adoption and scaling up (Lowenberg-DeBoer et al., 2021a; Tey and Brindal,
2022). The context of the of the UK and the US was considered to achieve the following
objectives because of economic, agronomic, and technical data availability. The case
study contexts are described in detail in the respective objective sections.
1.5 Research objectives
The overall objective of this study was to assess how autonomous machines could
maximize the profitability of arable field crop production compared to farming with
conventional machines with human operators both in whole field sole cropping and mixed
cropping systems considering agroecological and regenerative agriculture. The specific
objectives were to:
(i) Assess how field size and shape impact the profitability of autonomous crop
machines (detailed in Chapter 3);
(ii) Estimate the profitability of strip cropping with autonomous machines (detailed in
Chapter 4); and
(iii) Determine the profitability of autonomous machines for regenerative agriculture
(detailed in Chapter 5).
1.6 Research hypotheses
The following hypotheses were examined to achieve the specific objectives of the study:
(i) Autonomous crop machines make it possible to farm small, non-rectangular fields
profitably, thereby preserving field biodiversity and other environmental benefits;
(ii) Autonomous machines make strip cropping profitable, thereby allowing farmers to
gain additional agroecological benefits; and
(iii) Autonomous machines make regenerative strip cropping profitable, thereby
supporting the agricultural transition plan to improve soil health, biodiversity and
achieve carbon net zero target.
1.7 Theoretical grounds
Based on microeconomic theory and opinion of farm management experts, the choice of
cropping systems (e.g., whole field sole cropping and/or within field heterogeneous crop
mixes such as strip cropping and/or regenerative agriculture) and farm mechanization
levels (e.g., whole farm conventional mechanization with human operators and/or
autonomous machines) should maximize utility (Henderson and Quandt, 1958; Boehlje
and Eidman, 1984; Lowenberg-DeBoer, 2022b). However, utility maximization
encompasses numerous factors such as profit, leisure time, risk, capital, resource
constraints and transaction costs.
15
General introduction
Maximizing profit is the starting point to analyse farm management decisions in the short
run. The cropping systems and farm mechanization levels should at least cover the costs
of production. The economic payoffs would motivate wide-scale adoption (Lowenberg-
DeBoer, 2022b). In farm mechanization levels and crop choices, economic benefits are
considered as the prime driver (Lowenberg-DeBoer et al., 2021a; Tey and Brindal, 2022).
Consequently, the theoretical grounds of the research would be consistent with typical
neoclassical microeconomic farm theory (Shockley, Dillon and Shearer, 2019). The
objective function of the research was to maximize gross margin (i.e., return over variable
costs) subject to primary farm resource constraints. The net return to operator labour,
management and risk-taking (ROLMRT) was examined to address the impacts of
overhead costs in mechanized farming (Lowenberg-DeBoer et al., 2021a).
1.8 Research approach
Farmers and farm management specialists traditionally make less complex farm
management decisions using budgeting (Hazell and Norton, 1986). A review study
conducted in 2018, found that most production economics studies on automation (i.e.,
automated, and autonomous machines) to that point mostly used partial budgeting
methods, where only some specific costs and returns associated with automation were
changed while crop rotations, field operation timing, and other aspects of production were
unchanged (Lowenberg-DeBoer et al., 2020). Although budgeting can account for a whole
farming system, it is feasible only for very simple farming systems. With scenario-based
budgeting, in complex cropping and farming systems, the analyst quickly gets lost in the
alternatives and options for enterprise rotations, plant and harvest time, labour hiring, etc.
Rarely budgeting could find the optimal or most profitable plan (Boehlje and Eidman,
1984). Deterministic linear programming (LP) is computationally easier compared to
tedious and burdensome budgeting (Hazell and Norton, 1986; Boehlje and Eidman,
1984). The LP does not require substantial additional data but automates optimization that
best allocate farm resources (Boehlje and Eidman, 1984).
Apart from deterministic LP models, there are various options for whole farm planning that
could capture more complex interactive effects such as integer mathematical
programming, non-linear programming and/or simulation studies. Simulation can capture
more of the biological and physical details of farming, but it was not used in this study
because it fails to capture the key human tendency to optimize. Also, interpretation of
simulation results can be challenging because it involves comparisons of many options
and scenarios.
16
General introduction
The choice of optimization model depends on the trade-offs between model complexity
and the credibility of results given limited data. In an ex-ante analysis, data is usually very
limited. Often the parameters must be estimated by extrapolating from experimental
results and expert opinion. Constructing a complex optimization model based on this
limited data is often not credible. In contrast, LP requires only slightly more data than
budgeting but can provide insights at the farming system level.
To achieve the objectives of this study, whole farm deterministic LP model was used as
the simplest analytic tool for a farming system level analysis. The LP model utilizes a set
of “optimizing rules” to identify the most profitable plan to quickly sort through thousands
of potential crop rotations, technologies, and plant and harvest timing options (Hazell and
Norton, 1986; Boehlje and Eidman, 1984). Through shadow prices LP capture important
interactions between resource availability, constraints, and choice of activities. This study
considered LP model to maximize gross margin subject to the binding constraints of land,
human labour, and equipment time.
In keeping with the concept of using the simplest model that captures farming system
changes, the choice of machinery sets (i.e., tractor, implements, combine) was done
manually by comparing solutions with specific machinery assumptions. Because
machines within a machinery set must be compatible, choosing machinery sets within the
algorithm would require integer programming. This integer programming approach was
used by Shockley et al. (2019, 2021). In this case integer programming would add to the
complexity of the model, without adding substantially to the insights.
The whole farm deterministic LP model used in this study was adopted from the Hands
Free Hectare-Linear Programming (HFH-LP) model of Lowenberg-DeBoer et al. (2021a).
The HFH-LP model follows the ‘Steady State’ concept, which refers that the solutions
could be repeated annually over time. This steady state HFH-LP model was originally
based on the Purdue Crop/Livestock Linear Program (PC/LP) model for Midwestern
farmers (Dobbins et al., 1994). The PC/LP model was later adapted for use in various
countries of the world (Fontanilla-Díaz et al., 2021; Lowenberg-DeBoer et al., 2021a).
The whole farm deterministic HFH-LP model used in this study can be expressed as
follows following Boehlje and Eidman (1984):
The objective function:
Subject to:
17
General introduction
where, π is the gross margin, is the level of jth production activities, is the gross
margin per unit over fixed farm resources () for the jth production activities, is the
amount of ith resource required per unit of jth activities, is the amount of available ith
resource.
Following Lowenberg-DeBoer et al. (2021a) the primary constraints considered in this
study were:
(i) Land: This study assumed that the sum of land in productive activities is less
than or equal to the arable crop land available. For example, if in each rotation
the crops are q, the used land for a unit of a rotation is the fractional unit 1/q of
each crop. Taking for example, one hectare of a wheat-oilseed rape (OSR)
rotation is equal to half a hectare of wheat and half a hectare of OSR.
(ii) Human labour: This study assumed that the sum of the labour needed in each
month for each crop in the rotation multiplied by the fractional unit (1/q) of each
crop in each rotation. Here in this study the sum of the human labour required
must be less than the labour available from the operators, permanent farm
labour and temporary farm labour on the number of good field days.
(iii) Equipment time: The equipment time is that the sum of equipment time per
crop in each month on good field days, weighted by the rotation fraction (i.e.
1/q), must be less than or equal to the amount of equipment time available.
(iv) Cashflow: Sum of the variable costs for each crop in a rotation in each month
multiplied by the rotation fraction (1/q) must be less than or equal to the
working capital available. This study considered that the cashflow is not
binding.
The optimization model used in this study was coded in the General Algebraic Modelling
Systems (GAMS) software (GAMS Development Corporation, 2020). Although the R
software has also been used for optimization modelling of PC/LP type models (Griffin et
al., 2023), this study used the GAMS software because it is a standard mathematical
optimization algorithm used around the world and the HFH-LP was already available in
the form of GAMS code (Lowenberg-DeBoer et al., 2021a).
18
General introduction
1.9 Outline of the thesis
The outlines of the PhD thesis are represented in Figure 1.1. Chapter 1 explains the
general difference between 'robots', 'field crop robot', 'automated machines', and
’autonomous machines'. Subsequently, it provides examples of autonomous initiatives
worldwide with the implications for agricultural intensification solutions. The knowledge
gaps and rationale of the study are identified in 'The Research Problem' section. The
research objectives, research hypotheses, theoretical background and research approach
are explained briefly to give a general overview of the research. Chapter 2 shows the
state of the art and limitations of the existing research which is linked with research
objectives and hypotheses. Chapter 3 represents the outcomes of first research
hypothesis regarding how field size and shape impact the economics of autonomous
machines for grain-oilseed farms. Chapter 4 estimates the ex-ante economic scenarios of
the economics of strip cropping with autonomous machines. Chapter 5 analyses the
economics of regenerative agriculture with the ex-ante scenarios. Finally, Chapter 6
contains the general discussion and conclusions with the limitations of the study.
Worldwide implications of the research and future research directions are also suggested.
Figure 1.1: Structure of the thesis and chapters overview.
Chapter 1
General introduction
Chapter 2
State of the art
Chapter 3
Economics of field size and shape for
autonomous crop machines
Chapter 4
Economics of strip cropping with
autonomous machines
Chapter 5
Economics of autonomous machines for
regenerative agriculture
Chapter 6
General discussion and conclusions
19
State of the art
Chapter 2
State of the art
" … your task is to build an argument, not a library."
Rudestam and Newton (1992): Surviving your dissertation. Fourth Edition. p. 49.
2.1 Introduction
The review of literature as presented in this chapter explored the existing state of the art
of the economics of field crop robotics and autonomous systems (RAS) and associated
literature. The objective is to contribute to the scientific knowledge. The review mainly
concentrated on three research objectives regarding the economics of field size and
shape for whole field sole cropping system, and mixed cropping (i.e., within field
heterogeneous agroecological strip cropping systems and regenerative agriculture) with
different levels of mechanized farms. Mechanization levels here refer conventional
machines operated with human operators and autonomous machines (i.e., HFH retrofitted
autonomous machines). This chapter identified the limitations of the existing production
economics studies on autonomous machines. The chapter also proposed simulation
methodology to compare arable open field farming with conventional mechanization and
autonomous machines.
Simulation methods other an econometric approach is suggested because the
autonomous machines for field crops considered in this study are not yet widely marketed
and adopted for the context of large, medium and small-scale farming. The technologies
are in the pipelines and on the verge of commercialization processes (Shockley et al.,
2021). The HFH autonomous machines are prototypes that were demonstrated in the
context of the UK. The ex-post scenarios evaluation using econometric analysis was not
feasible here. To understand the economic potential of whole field sole cropping subject to
field size and shape and autonomous farming beyond whole field sole cropping simulation
methodology (here Linear Programming (LP)) is the right choice because it goes beyond
the biological and physical relationships to incorporate human motivation and drive to
seek better solutions. Consequently, LP analysis was suggested to fill the research gaps
that usually incorporates the basic elements of human decision making and overcome the
limitations of partial budgeting. In partial budgeting only crop and enterprise specific
20
State of the art
changes in costs and revenues are considered with all other things remaining the same
assumption.
This research anticipated that the profitability analysis of autonomous machines using
optimization LP model irrespective of field size and shape for whole field sole cropping
and farming beyond whole field sole cropping would facilitate the game changing
autonomous machines to achieve both production goals of productivity and profitability
and environmental goals of agroecological and regenerative farming to limit environmental
footprints of agriculture. The implications of this research for agri-tech economists,
engineers, agronomists, environmentalists, agribusinesses innovators, and policy makers
and planners are also pointed out in this chapter.
2.2 Economics of field crop robotics and autonomous systems (RAS)
The state of the knowledge of the economics of field crop robotics and autonomous
systems (RAS) reveals that the RAS used in arable open-field crop operations are viewed
in two perspectives: Firstly, automated machines (or automated crop robots) (i.e., partially
robotized mechatronic technology that accomplish arable field operations such as
seeding, weeding, and harvesting, but with mobility assured by a human operator).
Secondly, autonomous machines (or autonomous crop robots) (i.e., are a subset of field
crop robots which have autonomy in arable field operations using predetermined field
paths and itinerary with relatively little decision-making capacity) (Lowenberg-DeBoer et
al., 2020).
The most updated review of literature as of 2018 addressed the economics of automated
operations of one or two horticultural crops (detailed in Table 2.1). The economic studies
on automated machines for field crops are primarily focused on the cost saving potentials
of one or two field operations in production of horticultural crops. The production
economics literature on automated machines to date does not cover the broader
implications of the technology (e.g., the implications of machinery performance subject to
field size and shape and biodiversity conservation). The existing studies mostly
concentrated on horticultural crops, ignoring the whole farm systems analysis of
commodity crops production (Table 2.1) (Lowenberg-DeBoer et al., 2020). However, this
study did not delve in into the economics of automated machines because this is out of
the scope of this study. The HFH demonstration experience at Harper Adams University in
the UK represents autonomous machines, also known as autonomous crop robots or
swarm robots or swarm robotics. Considering the research objectives, the review of
literature in the subsequent sections concentrated the focus on the economics of
21
State of the art
autonomous machines to identify the knowledge gaps and to contribute to the state of the
art.
Table 2.1: State of the arts of automated crop robotics.
Authors and
year
Country
Economic
tools used
Goal
Arable
operation
considered
Machinery
performance
Crops
Tillett (1993)
UK
Partial
budget
Labour
cost saving
Harvest
No
Tomato
and fruit
Arndt et al.
(1997)
US
Partial
budget
Recovery
rate of
breakeven
harvest
Harvest
Yes: Harvest
rate (28%
and 15%)
Asparagus
Tsuga
(2000)
Japan
Partial
budget
Cost
saving
Transplant
No
Cabbages
and
lettuces
Ruhm
(2004)
Germany
Partial
budget
Recovery
rate of
breakeven
harvest
Harvest
and grade
No
Asparagus
Clary et al.
(2007)
US
Partial
budget
Breakeven
harvest
recovery
rate
Harvest
Yes: Harvest
rate (70%
and 80%)
Asparagus
Cembali et
al. (2008)
US
Partial
budget
Breakeven
harvest
recovery
rate
Harvest
Yes: Spear
collection
rate (85%)
Asparagus
Fennimore
et al. (2014)
US
Partial
budget
Max. net
return
Weed and
Thin
No
Leafy
vegetables
Mazzetto
and
Calcante
(2011)
Italy
Partial
budget
Cost
saving
Transplant
No
Vineyard
Pérez-Ruíz
et al. (2014)
US
Partial
budget
Cost
saving
Weed
control
No
Tomato
Zhang,
Pothula and
Lu (2016)
US
Partial
budget
Max. net
return
Harvest
Yes:
Commented
Apples
Source: Lowenberg-DeBoer et al. (2020) and author’s compilation.
The research on the economics of autonomous machines primarily focused on the cost
saving potential (Gaus et al., 2017; Goense, 2005; Pedersen et al., 2006; Pedersen et al.,
2017). A very few production economics research pointed out the significance of
machinery performance in terms of field efficiency and equipment times (Lowenberg-
DeBoer et al., 2021a; Revell, Powell and Welsh, 2020; Sørensen, Madsen and Jacobsen,
2005). From 1990 to 2018, a total of eight studies investigated economics of autonomous
machines in arable open-field farms (Lowenberg-DeBoer et al., 2020). In 2018 onwards,
nine studies focused on the economics of autonomous machines (Table 2.2). The
production economics research on autonomous machines did not cover how machinery
22
State of the art
performance subject to field size and shape in the whole field sole cropping system
impacts the economics of autonomous machines. A detailed overview of existing
literature, limitations and contribution of the present study are provided as follows:
Using partial budgeting approach, Edan, Benady and Miles (1992) assessed the
potentiality of automation for melon harvesting, where in sensitivity analysis they
considered autonomy. They found that if the manual harvesting operation is less than
US$494/ha, then autonomous operation is economically viable for a 202.4 ha harvesting
operation. They showed that breakeven investment lies within the range of US$ 50,000 to
US$250,000. Goense (2005) investigated the economics of autonomous equipment and
examined how autonomous implement size affects mechanization cost in row crop
cultivation of different sized farms. The study showed that row crop cultivation with
autonomous technology is an attractive alternative to manually operated machinery, if the
navigation is cost effective and large areas are covered.
Table 2.2: State of the art of autonomous crop robotics.
Authors
and year
Country
Economic
tools used
Goal
Arable
operation
considered
Machinery
performance
Crops
Edan,
Benady and
Miles (1992)
Israel and
US
Partial
budget
Cost
saving
Harvest
No
Melon
Sørensen,
Madsen and
Jacobsen
(2005)
Denmark
Scenario
planning
Cost
saving
Weed
Yes:
Weeding
efficiency
(80%)
Whole
farm
Pedersen et
al. (2006)
Denmark
Partial
budget
Cost
saving
Scout
/Weed
No
Sugar
beets
and
cereals
Pedersen,
Fountas and
Blackmore
(2008)
Denmark
/Greece
/UK/US
Partial
budget
Cost
saving
Scout
/Weed
No
Sugar
beet
McCorkle et
al. (2016)
US
Financial
simulation
Cost
saving
Prune
and Thin
No
Vineyard
Pedersen et
al. (2017)
Denmark
Partial
budget
Max.
gross
margin
Seed
No
Sugar
beet
Gaus
et al. (2017)
Germany
Partial
budget
Cost
saving
Weed
No
Cereals
Shockley and
Dillon
(2018) and
Shockley,
Dillon and
Shearer
(2019)
US
Linear
Programmi
ng (LP)
(Whole
farm)
Max. net
return
All
production
operations
No
Maize and
soybean
De Witte
(2019)
Germany
Partial
budgeting
Cost
saving
Harvest and
till
No
Grain
crops
23
State of the art
Table 2.2: State of the art of autonomous crop robotics (Continued).
Lowenberg
-DeBoer et
al. (2019)
UK
Hands
Free
Hectare
(HFH)-LP
model
(Whole
farm)
Max.
net
return
All
production
operations
Yes:
Field
efficiency
(70%) for
all
operations
and
equipment
sets
Wheat,
oilseed
rape
and barley
Revell,
Powell and
Welsh
(2020)
Australia
Discounte
d Cash
Flow
(DCF)
Analysis
Cost
saving
Spray
Yes:
Considere
d field
time
(h/ha)
Cotton,
wheat
and
chickpea
Lowenberg
-DeBoer,
Pope and
Roberts
(2020),
UK
HFH-LP
model
(Whole
farm)
Max.
net
return
Spray
No details
are
provided
Wheat,
barley,
oilseed
rape,
beans,
and
linseed
Lowenberg
-DeBoer et
al. (2021a)
UK
HFH-LP
model
(Whole
farm)
Max.
net
return
All
production
operations
Yes: Field
efficiency
(70%) for
all
operations
machines
Wheat,
oilseed
rape,
and barley
Lowenberg
-DeBoer et
al. (2021b)
Worldwide
,
especially
China,
Brazil, UK,
US,
Australia,
Belgium,
Netherlan
ds,
Canada,
and New
Zealand.
Discussio
n
Policy
lesson
with
discuss
ion of
max.
net
return
All
production
operations
No
specific
analysis,
but
inclusive
of field
efficiency
in UK
case
study
UK
and US
case
studies of
maize,
soybean,
wheat,
oilseed
rape
and barley
Shockley et
al. (2021)
US and
UK
Linear
Programm
ing (LP)
(Whole
farm)
Max.
net
return
and
policy
lesson
All
production
operations
Yes: Field
efficiency
(70%)
for all
operations
and
equipment
set
Corn and
soybeans
for the US
and
Wheat,
oil seed
rape
and barley
for the UK
24
State of the art
Table 2.2: State of the art of autonomous crop robotics (Continued).
Maritan et
al. (2022)
UK
HFH-LP
model
(Whole
farm)
Max. net
return
All
production
operations
Yes: Field
time (hr/ha)
and
Field
efficiency
(70%) for all
operations
and
equipment
sets
Wheat,
oilseed
rape
and barley
Lowenberg-
DeBoer
(2022b)
World
wide
Qualitative
Economics
of adoption
Miscellaneous
Performance
discussion
Miscellane
ous
Source: Adopted from Lowenberg-DeBoer et al. (2020) and authors own compilation.
Pedersen et al. (2006) compared economic feasibility of autonomous robotic systems in
three different agricultural applications. The findings revealed that agricultural robotic
operations were economically feasible compared to the conventional operating systems.
In robotic weeding on sugar beet, micro spraying reduces herbicide application by 90%
and total costs of robotic and conventional weeding were per annum €260.4/ha and
€296.6/ha. It means that autonomous weeding reveals €36.20 cost advantage than
conventional one. Likewise, robotic crop scouting ensured per annum cost savings of
€3.80/ha. The study pointed out several benefits such as, weed mapping, working hour’s
advantage and improved efficiency in modern production. The robotic grass cutting had a
cost saving advantage of more than €300/ha per annum. Pedersen, Fountas and
Blackmore (2008) analysed economic feasibility of robotic weed scouting and robotic
weeding for the US, UK, Greece, and Denmark. They pointed out that robotic weeding
had a cost advantage for all of the countries studied except Greece. They found that
autonomous operations are comparatively flexible and reduce labour expenses and had
advantage of extended working hours.
Considering early seedling and re-seedling of sugar beet, Pedersen et al. (2017)
quantified the economic perspectives of agricultural robots. They compared gross margins
of new seeding systems (i.e., early seeding and reseeding) and conventional cultivation
practice. The study mentioned that robotic operations lead to minimum overlaps, and it is
possible to ensure economies of scale in small and medium sized farms. Among the three
scenarios (conventional practice, early seeding, and re-seeding) considered, early
seeding was the most profitable system. Even though they assumed a yield increase of
2.5%, the system is expected to offer cost advantage due to the use of robots that leads
to labour savings. However, the expected increase of yield in re-seeding will be 5%, but
the system will require conventional seeding due to its dual seeding operations. Results
showed that in early seeding, the gross margin will increase by 7.7% and in re-seeding
25
State of the art
there is a possibility to increase gross margin of 6.5%. The study of Gaus et al.
(2017) using partial budgeting techniques investigated economics of autonomous swarm
robots for weeding in wheat. The study commented on the future product prices
and robot’s requirement for field operation. Results showed that swarm robots could be a
possible alternative for crops, especially for crops with high costs intercultural operations.
The production economics studies mentioned above used partial budgeting, whereas a
very few studies considered methodological rigour in economic assessment to overcome
the limitations of partial budgeting (Lowenberg-DeBoer et al., 2019, Lowenberg-DeBoer et
al., 2020, Lowenberg-DeBoer et al., 2021b; McCorkle et al., 2016; Shockley, Dillon and
Shearer, 2019; Shockley and Dillon, 2018; Sørensen, Madsen and Jacobsen, 2005).
In Denmark, using scenario planning Sørensen, Madsen and Jacobsen (2005)
investigated the potentiality of organic crops robotic weeding. They found that the benefits
of robotics weeding were highly sensitive to weed intensity and initial equipment price.
Results showed that for robotic weeding, farmers paid up to €40,000, but they are still in a
better off position than manual weeding. The study mentioned efficiency is the critical
prerequisite for improved profitability and assumed 80% weeding efficiency for sugar beet
and maize. McCorkle et al. (2016) investigated the economics of robotic technology in the
production of wine grapes using a financial simulation model. They showed how
substituting manual labour with robotic equipment affects vineyards of different sizes.
Shockley and Dillon (2018) examined the economic feasibility of autonomous field
machinery compared to conventional manned machinery using the whole farm planning
model in corn and soybean production. Results showed that net returns were greater
when the farm was operated with autonomous machinery. With the anticipated benefits of
10% reduction in input costs and 7% increase in yields, the net return increased
significantly up to 19%. Findings of the sensitivity analysis showed that autonomous
machinery had the potentiality to ensure greater profitability for different sizes of farm,
especially for small sized farms.
De Witte (2019) pointed out that small autonomous equipment will be less capital
intensive and hypothesized that in addition to labour cost saving potential, small
autonomous machinery will positively influence profitability with yield increase and other
resource savings in arable farming. Shockley, Dillon and Shearer, (2019) compared the
economic feasibility of conventional and autonomous machinery to produce grain crops in
the United States for a given farm size of 850 hectares. Results showed that autonomous
machinery was profitable over conventional machinery when the intelligent control
establishment was cost effective. They also found that relatively small autonomous
machinery was likely to have economic advantage for various farm sizes, especially for
26
State of the art
small farms. Lowenberg-DeBoer et al. (2019) examined the economic impacts of
autonomous equipment subject to farm size in the using Hands Free Hectare (HFH)
demonstration experience. Although economic analyses of autonomous crop robotics
throughout the world are constrained due to lack of data on economic parameters, the key
strength of the HFH on-farm demonstration was that it provided first-hand experience with
autonomous whole farm production operations. Using HFH demonstration experience
they showed that crop production with swarm robots was economically feasible, where
small and medium sized farms had cost advantage, and production costs of the United
Kingdom were internationally competitive. Revell, Powell and Welsh (2020) examined the
economic feasibility of autonomous tractors used in spraying operations for producing
cotton in irrigated and dryland including cotton, wheat and chickpeas. They found that
adoption of autonomous equipment was economically feasible.
Lowenberg-DeBoer, Pope and Roberts (2020) used HFH-LP model to investigate
economics of arable autonomous technology for biopesticide application in break crop,
namely, oilseed rape, beans, and linseed. They found that application of low cost
biopesticide is feasible with both conventional and autonomous technology, but
autonomous equipment still demands more human labour in field operation compared to
conventional herbicide treatments. Lowenberg-DeBoer et al. (2021a) identified economic
implications of autonomous equipment. The study showed technical and economic
feasibility of autonomous equipment and found that medium sized farms had a cost
advantage with autonomous technology. They also commented on the economic
potentiality of autonomous technology in irregular shaped arable fields and restoration of
in-field biodiversity. In another study, considering the context of the United Kingdom,
Lowenberg-DeBoer et al. (2021b) suggested that economic and social implications of
autonomous equipment adoption will be affected by the rules of autonomous equipment
use. Using the context of the United States, Shockley et al. (2021) pointed out that
profitability of autonomous equipment was sensitive to the rules of automation for arable
farming. They found that small farms gain more through using autonomous machinery in
arable farm operation. Maritan et al. (2022) investigated economically optimum farmer
supervision time for open-field autonomous machines. The study found that for field crop
production economically optimum supervision time lies between 13% to 85% depending
on the reliability of the machine and type of supervision (i.e., on-site or remote).
Lowenberg-DeBoer (2022b) pointed out the economics of digital technology adoption
worldwide, where autonomous machines adoption and implications for large, medium and
small-scale economies are vividly described.
27
State of the art
The state of the art of economics of autonomous machines (i.e., autonomous crop robots)
reveals that research on autonomous machines economics highly concentrated the focus
on costs saving potential. In economic analysis, a very few studies encompassed or
commented on the implications of farm size (Edan, Benady and Miles, 1992; Gaus et al.,
2017; Goense, 2005; Lowenberg-DeBoer et al., 2019, Lowenberg-DeBoer et al., 2021a;
McCorkle et al., 2016; Pedersen et al., 2017; Shockley, Dillon and Shearer, 2019;
Shockley and Dillon, 2018). Nevertheless, apart from economic parameters, how
machinery performance subject to field size and shape impact the farm economics of
autonomous arable open-field farming are still unexplored. On the contrary, in
conventional mechanized farms, farm size and shape received substantial attention to
increase labour productivity and economies of size. Relatively larger rectangular fields are
preferred which support the ‘get big or get out’ rule of thumb of conventional
mechanization (Robinson and Sutherland, 2002; Lowenberg-DeBoer et al., 2021a).
The economic and technical data limitations of autonomous farming are the prime reason
for such a research gap (Lowenberg-DeBoer et al., 2020). The technological development
and research of autonomous machines are well advanced (Shamshiri et al., 2018;
Fountas et al., 2020). Academic, researchers and agribusiness innovators envisioned that
autonomous machines will be able to reconcile techno-economic and environmental goals
(Duckett et al., 2018; Daum, 2021; Pearson et al., 2022; AutoAgri, 2023). Up-to-date
production economics research are based on autonomous whole field sole cropping cost
economies, whilst machinery performances subject to field geometry are yet to be
explored (Shockley, Dillon and Shearer, 2019; Lowenberg-DeBoer et al., 2021a; Shockley
et al., 2021; Maritan et al., 2023). Similarly, open-field autonomous arable crop farming
beyond whole field sole cropping (i.e., mixed cropping to address spatial and temporal
heterogeneity) need investigation to guide win-wing farming synergies. The multifaceted
benefits to reconcile both production goals (i.e., productivity and/or profitability) and the
goals of the society as a whole (i.e., limiting environmental footprint of agriculture) are yet
to be answered.
To address the production economics research gaps on autonomous machines and to
navigate the game changing technology innovation and adoption, the present study
examined the economics of field size and shape for autonomous machines in the whole
field sole cropping system (Objective 1). In addition, the study extended the research
focus beyond autonomous whole field sole cropping economics to reconcile production
goals and environmental goals through evaluating the economics of autonomous
agroecological strip cropping systems (Objective 2) and the economics of autonomous
regenerative agriculture (Objective 3). The following sections dealt with objective specific
28
State of the art
state of the knowledge, limitations and the contribution of the present study to scientific
knowledge:
2.3 Field size and shape (AND/OR) autonomous machines: Whole field sole
cropping (Objective 1)
Field size and shape received substantial attention in the field of geography (Davis, 1926;
Miller, 1953; Boyce and Clark, 1964; White and Renner, 1957) and more importantly in
the last decades in agricultural sciences (Batte and Ehsani, 2006; Griffel et al., 2018;
Janulevičius et al., 2019; Larson et al., 2016; Zandonadi et al., 2013). Research in
agricultural sciences, considered field size and/or shape to examine machinery
performances (Amiama, Bueno and Álvarez, 2008; Gónzalez, Marey and Álvarez, 2007;
Oksanen, 2013; Spekken and Bruin, 2013), input application overlap (Luck, Zandonadi
and Shearer, 2011; Jernigan, 2012; Zandonadi et al., 2013), and agricultural production
economics literature to investigate profitability of precision agriculture technology,
especially on Global Navigation Satellite Systems (GNSS) guidance and related
technologies such as boom control (Batte and Ehsani, 2006; Larson et al., 2016; Shockley
et al., 2012).
In arable field operations, field size and shape received significant attention. Studies
showed that conventional agricultural mechanization always favoured large sized
rectangular fields and most of the land consolidation studies around the world in the last
decades have been motivated by the desire for larger fields (Kienzle, Ashburner and
Sims, 2013; Lindsay et al., 2013; Robinson and Sutherland, 2002; Van den Berg et al.,
2007). Likewise, machinery performances are always sensitive to field sizes and shapes
(Keicher and Seufert, 2000; Spekken and de Bruin, 2013; Janulevičius et al., 2019).
Majority of the research on machinery performance subject to field size and shape mainly
concentrated on two domains of field operations: (i) Numerous studies focused on the
path planning to minimize non-productive time in agricultural field operations (Oksanen,
2013; Spekken and de Bruin, 2013), and (ii) research generally highlighted machinery
performance, especially time efficiency during agricultural operations (Anigacz, 2015;
Ebadian et al., 2018; Fedrizzi et al., 2019; Griffel et al., 2018; Janulevičius et al., 2019).
In southern Finland, Oksanen (2013) aimed to find a computationally faster method of
examining the relationship between field shape and operational efficiency. They compared
their findings of a path planning algorithm with a set of real plots. Likewise, considering
field shape, Oksanen and Visala (2007) developed a coverage path planning algorithm
which is applicable to any kind of agricultural equipment. Spekken and de Bruin (2013)
focused on route optimization with a reference to different field sizes to reduce non-
29
State of the art
productive time in field operations. Janulevičius et al. (2019) provided a method for
estimating time efficiency of farm tractors during tillage operation in fields of different
sizes. In Bangladesh, Islam, Kabir and Hossain (2017) investigated existing plot size and
shape to understand the effects on operational efficiency of mechanical walk behind type
rice transplanter. Gonzalez, Alvarez and Crecente (2004) considered plot size and shape
to evaluate land distribution in Spain and presented an index considering plot size and
shape factor. Similarly, Gónzalez, Marey and Álvarez (2007) examined effects of plot
shape and size on effective field capacity of machinery operation in potato farming. In
Spain, Amiama, Bueno and Álvarez (2008) considered field shape and proposed two new
shape indices to investigate the effects of field shapes on the effective field capacity of
self-propelled forage harvester. The study of Koniuszy et al. (2017) investigated power
performance of farm tractors in tillage operation subject to different field sizes.
Agricultural scientists considered field size and/or shape to minimize input application
overlap in PA literature. For instance, Jernigan (2012) considered field shape to examine
the relationship between diversified fields and planter overlap in Tennessee, US. Luck,
Zandonadi and Shearer (2011) in Kentucky, US, examined the effects of field size and
shape on overlap errors of automatic section control and manual application. In Central
Kentucky, US, Luck et al. (2010a) investigated pesticide and nutrient savings based on
three different irregularly shaped grain fields. In another study, Luck et al. (2010b)
compared effectiveness of automatic section control with manual section control to
investigate pesticide application overlap in fields of different shapes and sizes in
Kentucky, US. Zandonadi et al. (2011) developed a computational method based on field
shape to calculate overlap errors of machinery and concluded that off-target spray
application area varied depending on shape and size of field boundary. Likewise,
Zandonadi et al. (2013) evaluated field shape descriptors to calculate off-target application
area.
However, most of the existing research on the effects of field size and/or shape on arable
field operations are mainly concentrated on technical aspects of machinery management.
A very few production economics studies addressed field size and shape issues in
economic feasibility assessment of PA technology. In Tennessee, US, Larson et al. (2016)
examined effects of field size and shape on profitability of chemical application with PA
equipment. They concluded that field size and shape significantly affect profitability of
precision spraying using automatic section control. Batte and Ehsani (2006) compared
economic benefits of farmer-owned precision sprayers with a traditional non-precision
system in three differentiated field shapes (i.e., a rectangle, parallelogram, and trapezoid).
They analyzed a set of hypothetical farm fields each of which was 40.47 ha sized with and
30
State of the art
without the inclusion of grass waterways through the fields at 450 and 600 angles. In
Kentucky, US, Shockley et al. (2012) investigated impacts of field size and shape on
automatic section control profitability. The study was conducted for planting and
spraying operations in four fields and within the fields there were a spectrum of shape,
size, and obstacles. Smith et al. (2013) considered on-farm field parameters (i.e., field
size and shape) to evaluate the profitability of precision spraying technologies in
Colorado, Kansas, and Nebraska, US. They found profitability was sensitive to size and
shape of the irregular fields. Although the existing studies investigated profitability of PA
technologies considering field size and shape, most of the studies were based on partial
budgeting methods and concentrated on one or two crop operations (i.e., weeding and/or
harvesting). The review of the existing literature reveals that the economics of PA
technologies with a reference to field sizes and shapes focused on the input savings
potentials in economic assessment. Nevertheless, the economic implications of machinery
performance subject to field sizes and shapes considering whole farm operations from
planting to harvesting (i.e., systems analysis) were unexplored.
Autonomous machines have the potential to revolutionise PA (Lowenberg-DeBoer et al.,
2020, Lowenberg-DeBoer et al., 2021a). Even the studies on the economics of automated
machines considered different sized farms (Tsuga, 2000; Ruhm, 2004; Mazzetto and
Calcante, 2011) and commented on the operational efficiency in field operations (Clary et
al., 2007; Cembali et al., 2008). Nonetheless, the economic implications of field sizes and
shapes with the lens of machinery performance are still unexplored. For example, in
Japan, Tsuga (2000) showed that automated transplanters can economically compete
with human labour with a minimum area covered over 8.2 ha. Ruhm (2004) evaluated
economics of harvesting, grading and cultivation of asparagus in Germany. The study
showed that automated asparagus grading technology would be cost-effective if the area
is more than 13 ha and the optimum size of the field for automated technology is 29 ha.
They pointed out that future effort should concentrate on efficient production with
minimum costs. In Italy, Mazzetto and Calcante (2011) developed an innovative system
for completely automated transplant operation of vine cutting. They considered farm size
and tested the developed method in various field topographic conditions. They found that
the automated system reduced the requirements of labour and increased the transplanting
rate by 15% compared to the conventional system. The cost curve estimation revealed
that automated transplanter had lowest cost potentiality with annual area transplanted
over 23 ha. Zhang, Pothula and Lu (2016) conducted an economic assessment of a self-
propelled harvesting and automated in-field sorting machine systems in the US apple
industry. The study mentioned that farm size played an important role in cost savings and
automated machines increased harvest efficiency. Cembali et al. (2008) determined the
31
State of the art
efficiency level for profitable automation asparagus harvesting and compared it with
manual harvesting methods. The study assumed the spear collection and collateral
damage efficiency as the primary trial was unable to demonstrate exact efficiency. They
concluded that the efficiency of the spear collection rate should be 85% with 5% collateral
damage for profitable selective mechanical harvesting compared to manual methods.
These studies focused on farm size but did not analyse the impact of the field size within
the farm.
Similarly, the economic feasibility assessment of autonomous machines incorporated farm
size in arable field crops and fruit production. For example, Edan, Benady and Miles
(1992) found that if manual harvesting operation was less than US$494/ha, then
autonomous operation was economically viable for 202.4 ha harvesting operation. In the
US, McCorkle et al. (2016) showed how substituting manual labour with robotic equipment
affects different sized vineyards. Pedersen et al. (2017) mentioned that robotic operations
lead to minimum overlaps, and it was possible to ensure economies of scale in small and
medium sized fields. Goense (2005) showed that row crop cultivation with autonomous
technology was an attractive alternative to manually operated machinery, if large areas
were covered. The above production economics studies were unable to disclose the
economic implications of field sizes and shapes, in addition, these studies lacked systems
analysis.
To date, only very few studies focused on systems analysis in their economic assessment
and mentioned the significance of farm sizes. However, economic implications of field
sizes and shapes subject to the performance of machineries’ were overlooked
(Lowenberg-DeBoer et al., 2019, Lowenberg-DeBoer et al., 2021a; Shockley, Dillon and
Shearer, 2019; Shockley and Dillon, 2018). Shockley and Dillon (2018) examined the
economic feasibility of autonomous field machinery to produce corn and soybean in the
US. They concluded that farm size should be considered into market size determination.
Likewise, Shockley, Dillon and Shearer, (2019) found that relatively small autonomous
machines are likely to have economic advantages for medium and small-scale farms.
Shockley et al. (2021) examined how regulation will impact the commercial viability of the
use of autonomous equipment in the US. They mentioned that smaller farms had the
advantage to gain more from farming with autonomous equipment. Lowenberg-DeBoer et
al. (2019) went beyond the study of Shockley, Dillon and Shearer (2019), they assessed
the economic feasibility of swarm robots incorporating seeding to harvesting operations
based on field data. They found small and medium sized farms with swarm robotic
operations had cost advantage. Similarly, using systems analysis, Lowenberg-DeBoer et
al. (2021a) identified economic implications of autonomous equipment for grain-oil-seed
32
State of the art
farms in the UK. They found that medium sized farms had a cost advantage with
autonomous technology. They also commented on the economic potentiality of
autonomous technology in irregular shaped arable fields. In their analysis, they assumed
all farms had 70% field efficiency for all operations and equipment sets, but did not reflect
the economic implications of field efficiency differences subject to field sizes and shapes.
Sørensen, Madsen and Jacobsen (2005) mentioned efficiency is a critical prerequisite for
improved profitability and assumed 80% robotic weeding efficiency. However, these
studies assumed constant field efficiency for different farm sizes and operations. They
overlooked the crucial question about economic implications of field sizes and shapes on
the use of autonomous crop robotics.
On the contrary, the ecological management studies, especially studies conducted in the
US, UK, Canada, and European Union considered field sizes with utmost importance to
promote environmental schemes (Clough, Kirchweger and Kantelhardt, 2020; Europe,
2008; Fahrig et al., 2015; González-Estébanez et al., 2011; Stanners and Bourdeau,
1995). For instance, Fahrig et al. (2015) considering Canadian context, found that field
size had a strong relationship with biodiversity. Results showed that higher biodiversity
exists in small arable crop fields. They suggested for ensuring biodiversity conservation,
field size reduction should be considered. Likewise, in the context of eastern Ontario,
Canada, Flick, Feagan and Fahrig (2012) examined effects of the structure of landscape
on the diversity of butterfly species. The results showed there was a positive relationship
between declining patch size and richness of butterfly species. Lindsay et al. (2013)
investigated the relationship between structure of farmland and bird species composition,
diversity and richness in six watersheds in the Midwest, US. They found avian richness
decreased with the increase of field size. In the context of Great Britain, Robinson and
Sutherland (2002) found increased use of machinery promoted the expansion of field size
that resulted in 50% removal of the stock of hedgerows. In northwest Spain, González-
Estébanez et al. (2011) found that butterfly diversity is higher in smaller fields. They
mentioned landscape attributes are important for biodiversity conservation. Gaba et al.
(2010) examined the richness and diversity of weed species in France. The study found
increased richness and diversity of weed in small fields. They suggested that fields having
more crop edges could shelter numerous species of weed. Clough, Kirchweger and
Kantelhardt (2020) pointed out that in European landscapes, biodiversity declined with the
increase in field size. They suggested that ecological and economic trade-offs should be
addressed in policy and research, where field size could be the mediator to mitigate the
trade-offs.
33
State of the art
The state of the art reveals that small fields are advantageous for environmental
management. On the contrary, the performance of conventional agricultural
mechanization has an inverse relationship with small fields. Nonetheless, it has yet to be
demonstrated how and to what extent autonomous machines performance is sensitive to
field sizes and shapes, and what would be the economic implications of field
size and shape on autonomous machines. Although the most up to date study conducted
by Lowenberg-DeBoer et al. (2021a) estimated wheat costs of production subject to farm
sizes (Figure. 2.1). They hypothesized that autonomous swarm robots will minimize the
pressure to "get big or get out", indicating that small farms are economical with
autonomous machines. However, they did not address the implications of field size and
shape. Consequently, the economic implications of field sizes and shapes remain in
question.
Figure 2.1: Costs of production of wheat for conventional (triangles) and autonomous
equipment (circles) subject to farm sizes. Source: Lowenberg-DeBoer et al. (2021a).
To shed light on the research gap, the present study hypothesized that autonomous crop
machines would make it possible to farm small, non-rectangular fields profitably, thereby
preserving field biodiversity and other environmental benefits. The study took advantage
of systems analysis considering autonomous arable farm operations from drilling to
harvesting. Using HFH demonstration experience at Harper Adams University in the UK,
the study investigated the economics of field size and shape on autonomous grain-oilseed
production (Objective 1). The findings of the study have implications for the development
and improvement of autonomous machines and facilitate the decision-making process of
34
State of the art
the farmers and agribusiness adopters, environmentalists, and policy makers and
planners.
2.4 Automating mixed cropping
Apart from the cost economies (i.e., economies of size) of autonomous machines for
whole field sole cropping system (Shockley, Dillon and Shearer, 2019; Lowenberg-DeBoer
et al., 2021a), the mixed cropping farm management potentials of autonomous machines
are in planning (Daum, 2021; Davies, 2022; Pearson et al., 2022) and demonstration
stage (Ditzler and Driessen, 2022; Harper Adams University (HAU), 2023). Autonomous
machines are expected to reconcile the multifaceted goals of arable farming such as
production goals of productivity and profitability and environmental goals of sustainability
(Gackstetter et al., 2023).
Research suggests several mixed cropping systems with the advent of autonomous
machines, such as strip cropping (Ward, Roe and Batte 2016), pixel cropping (Ditzler and
Driessen, 2022) and patch cropping (Grahmann et al., 2021; Donat et al., 2022).
However, technical challenges of farm management constraints more complex mixed
cropping due to different plant heights and growth patterns (Ditzler and Driessen, 2022).
Strip cropping (refers to a farming practice of simultaneously growing two or more crops in
adjacent strips, where the strips are wide enough for independent cultivation, whilst
narrow enough for facilitating crop interaction) is considered as the simplest and most
technically feasible mixed cropping systems even with conventional mechanization (Exner
et al., 1999; van Apeldoorn et al., 2020; Alarcón-Segura et al., 2022).
2.5 Strip cropping (AND/OR) autonomous machines: Objective 2
Strip cropping is considered as part of sustainable intensification solution because strip
cropping has the potential to address within field spatial and temporal (i.e., spatio-
temporal) heterogeneity, while increasing production and reducing synthetic inputs use
(Cruse and Gilley, 2008; Du et al., 2019; Juventia et al., 2022). Agroecology (FAO, 2019)
has been suggested to bring a new paradigm in arable crop farming through redesigning
spatio-temporal heterogeneity. The agroecological farming systems has the potential to
reconcile production and environmental sustainability while substituting external inputs
use through optimizing the ecological processes (Lacombe, Couix and Hazard, 2018;
Boeraeve et al., 2020). Under the umbrella of agroecological farming, strip cropping is
advocated with existing machinery to increase productivity and resource-use-efficiency
(Munz et al., 2014a; Song, 2020; Juventia et al., 2022; Bejo, 2023; Chongtham, 2023).
The agronomic (West and Griffith, 1992; Agyare et al., 2006; Munz et al., 2014a) and
35
State of the art
ecological (AlarcónSegura et al., 2022) benefits of strip cropping are well documented in
research throughout the world.
Research in the large-scale farming context showed the agronomic benefits of strip
cropping. For instance, Borghi et al. (2012) using field experiments in Brazil, investigated
the effects of different row spacing on maize and forage intercropping. The study found
that narrow-row spacing maize yields were higher compared to wide-row spacing at the
same plant density. Field experiments in Argentina by Verdelli, Acciaresi and
Leguizam´on (2012) found that strip cropping corn yield in three seasons increased 13
to16% in the border rows, whilst soybean yield decreased 2 to11% compared to whole
field sole cropping (i.e., monocultures). The study found no significant difference in centre
rows yield in strip cropping. The study pointed out that yield increase of corn in border
rows was highly associated with the radiation interception and crop growth rates
advantage of taller corn plants and the opposite happened for subordinate soybean that
leads to yield penalty. West and Griffith (1992) conducted maize and soybean strip
cropping trials from 1986 to 1990 in the Corn Belt of Indiana, US. The study found that the
strip cropping system increased outside corn rows yield on average by 25.8% and
decreased outside soybean yield by 26.6% compared to unstripped cropping system. The
review study conducted by Francis et al. (1986) pointed out that in the Eastern and
Midwest US, narrower strips corn had yield advantage of 10 to 40% and soybean yield
reduction was 10% to 30% over sole cropping systems owing to the light water and
nutrient competition between taller corn and smaller soybean plants. The study also
mentioned that in wider strips the corn yield increase and soybean yield decrease were
less than sole cropping. Ghaffarzadeh, Préchac and Cruse (1994) evaluated the yield
response of corn-soybean-oat-legume strip intercropping in two experiments conducted in
1989 and 1990 in Iowa, US. Results showed that outside corn rows produced significantly
higher yield, but competition of water caused yield loss. Rainfall and water adequacy
affected soybean yield. The study suggested strip intercropping as a suitable alternative to
current monocropping practices. Cruse and Gilley (2008) in Iowa, US found that corn yield
was 10% to 30% higher in the edge rows whilst soybean yield decreased 5% to 10%
compared to the strip in centres.
Experiments of medium scale farming context in Germany and small scale context in
China by Munz, Claupein and Graeff-Hönninger (2014b) showed that strip widths have
significant impact on crop yield. The study found that on average maize yield increase in
border rows for 18 to four rows by 3% to 12% in Germany and 5% to 24% in China. Yang
et al. (2014) used the experience of maize and soybean relay strip intercropping
experiments in China and found that planting geometrics had yield effects. The study
36
State of the art
pointed out that spatial pattern differences have implications for soybean owing to the light
environment. Results showed that total yield in strip intercropping systems was higher
than that of sole cropping systems. Yang et al. (2015) found that maize yield increased
with bandwidth reduction and plant spacing had significant impacts on yield. The yield of
relay strip intercropping was higher compared to sole cropping maize and soybean
farming. The optimum bandwidth and narrow-row spacing of maize were 200 and 40 cm.
The study suggested appropriate reduction of narrow rows maize plant spacing and
increased distance of maize-soybean rows for higher yield. Research in China by Iqbal et
al. (2019) suggested appropriate planting geometry for yield increase, nutrition acquisition,
and mechanical operations in maize-soybean strip intercropping systems. They suggested
increasing distance between soybean and maize rows and decreasing distance of maize
rows. Qin et al. (2013) using experimental trails in arid land of China found that maize
based intercropping systems such as maize-pea and maize-wheat had significant yield
advantages compared to sole cropping systems. The study also found land equivalent
ratios of 1.2 to 1.5 (the benefit greater than 1 indicates intercropping benefits). The study
also advocated that intercropping systems incorporating a legume such as pea has the
capacity to increase crop productivity, reduce soil respiration and decrease carbon
emission. Jun bo et al. (2018) in China found that increase of plant density resulted in
higher maize and soybean yield and the land equivalent ratios as of 2.0. The study
mentioned that the outer rows of maize and soybean were expanded enough to facilitate
light use and equipment work efficiency. The simulation study of van Oort et al. (2020)
using Chinese case study of wheat and maize relay strip intercropping found that wider
strip decreased intercropping benefits. The study suggested optimum strip width less than
1 m. Liu et al. (2022) using three years maize and soybean experiments in southwest
China found that soybean strip width had substantial effects on leaf photosynthetically
active radiation (PAR) compared to maize strip width.
Agronomic studies also found the effects of strip orientation in strip cropping systems. The
maize-soybean-oat strip cropping study of Jurik and Van (2004) on four farms of Iowa,
US, found that outside edge rows of corn in north-south direction received 2% to 38%
higher daily photosynthetic photon flux density (PPFD) (i.e., the number of photons per
unit time on a unit surface) compared to inner rows in strip systems and the outside
soybean row far away from corn received 36% to 140% greater PPFD. Cruse and Gilley
(2008) found that in east-west oriented strip cropping systems, south border rows corn
yield increased substantially compared to north borders. They also found that strips
oriented in the north-south favoured corn yield on both side edges. Liu et al. (2022) based
on three years of maize and soybean strip cropping experiments on North-south and
West-east strip orientation in southwest China found more photosynthetically active
37
State of the art
radiation (PAR) (i.e., solar radiation that photosynthetic organisms capable to use in
photosynthesis where the solar radiation lies between 400 to 700 nanometres)
interception by soybean plants compared to maize plants while strip orientation angled
increase from 00 to 900. Iragavarapu and Randall (1995) based on the experiment in
southern Minnesota, US found that yield of corn increased by 3% when the corn and
soybean strip cropping was oriented in East-west rows and 13% in North-south rows. On
the contrary, soybean yield decreased by 10% for East-west rows and 7% for North-south
rows. Iragavarapu and Randall (1996) showed that strip orientation has substantial effects
on yield based on the experimental trials of southern Minnesota, US. The layout followed
south side soybean, northside wheat and east-west side-oriented corn to maximize light
interception and minimize shading. The four-year (1991 to 1994) average yield found that
outside rows corn yield was 12% higher in east-west rows and 25% higher in north-south
rows compared to non-border rows. In the case of soybean, the study found 13% yield
penalty for east-west and 12% for north-south rows.
The study of Cruse and Gilley (2008) considering the North American context of Iowa
pointed out that total application of pesticide and fertilizer was less in strip cropping
systems compared to whole field sole cropping. The review synthesizes of Iqbal et al.
(2019) found that intercropping helps in higher resource capture due to the advantage of
capturing spatial and temporal dimensions. The inclusion of legumes in intercropping
systems served as a strategy to save nitrogen owing to the biological nitrogen fixation
process. They also pointed out that cereal-legume intercropping systems improve water
use efficiency and soil fertility. The findings of meta-analysis showed that intercropping
reduced anthropogenic inputs (i.e., less fertilizer N is required) compared to the sole
cropping system (Xu et al., 2020). Based on the southwestern Chinese context, the study
of Du et al. (2019) found that legume-nonlegume intercropping such as maize and
soybean intercropping systems reduce N input through biological N fixation. The study of
Głowacka et al. (2018) using a field experiment in south Poland from 2008 to 2010 found
that strip cropping is an effective strategy to improve maize biofortification (i.e., the
process of improving food nutritional quality). The study found that strip cropping
significantly increased Magnesium (Mg) and Calcium (Ca) accumulation in maize biomass
(i.e., renewable organic matter) and grain.
Strip cropping systems also have the potential to maximize temporal variability and reduce
the negative border effects. Small grains same in height such as oat and wheat could be
considered to take the advantage of edge effects. Small grains typically sown a few
months before the typical taller plant maize and subordinate plant soybean. This cropping
system competes less for light. When the small grains (e.g., oat and wheat) reach maturity
38
State of the art
the taller plant (maize) ensures wind shelter that reduces grain lodging (Iragavarapu and
Randall, 1996; Cruse and Gilley, 2008). The findings of meta analysis by Xu et al. (2020)
showed that intercropping increased temporal variability by sowing or harvesting one crop
earlier than others.
Apart from agronomic benefits, research shows ecological benefits of strip intercropping
(Qin et al., 2013; Tajmiri et al., 2017b). For instance, based on a field experiment in China
Ju et al. (2019) pointed out that strip intercropping increases crop biodiversity and could
be used for successful conservation and biological control tools. The study of Alarcón
Segura et al. (2022) found that strip intercropping systems increased biodiversity and
biological pest control in conventionally mechanized farms with larger 27-36m strips in
German farms. Cong et al. (2015) conducted wheat, maize and faba bean strip cropping
experiments from 2008 to 2011 in Northwest China. The study found that root biomass in
intercrops was 23% higher compared to whole field sole cropping. Results also found soil
carbon (C) sequestration, Nitrogen (N) fixation. The study pointed out that strip
intercropping systems have aboveground productivity owing to species complementary
and belowground productivity due to C sequestration and biological N fixation. The review
study by Kremen and Miles (2012) pointed out that strip intercropping has less disease
spread due to the spacing and enterprise diversity. The crop roots interaction in strip
intercropping is less than other types of row intercropping. The review of Hiddink
Termorshuizen and van Bruggen (2010) found that strip cropping and other mixed
cropping systems reduced diseases in 74.5% of cases compared to whole field sole
cropping. The study of Raseduzzaman and Jensen (2017) based on the metal analysis
pointed out that cereals and grain legumes intercropping has the potential to promote
biodiversity and sustainable intensification with higher yield stability. Using field
experiments in south China, Liang et al. (2016) found that intercropping substantially
lower disease infestation. The net rate of photosynthesis and leaf chlorophyll content were
increased for rice in the edge rows of water spinach. Ning et al. (2017) found that in south
China, rice and water spinach intercropping reduced diseases and pest infestation in rice.
Chen et al. (2017) in their experiments in China found that maize and soybean relay strip
intercropping increased land productivity and reduced environmental pollution.
The study of Cruse and Gilley (2008) considering the North American Corn Belt context of
Iowa pointed out that due to the advantages of easily defined planting positions, no-till or
ridge tillage and contour planting strip cropping system limit soil erosion. Moreover,
inclusion of small grain/forage strips in the strip cropping system act as an efficient
vegetation filter for sediment removal during water runoff. Strip cropping is a management
system to control soil drift and improve water storage (Bravo and Silenzi, 2002).
39
State of the art
Mohammadi et al. (2021) considered Iranian context and found that strip intercropping is
an effective strategy to encourage pest predators. Study by Iqbal et al. (2019) in China
showed that cereal-legume intercropping systems reduce weed infestation, soil erosion
and improve water use efficiency and soil fertility. The study of Brennan (2013) in
California, US, showed that lettuce and alyssum strip cropping is an effective strategy for
biological aphids control. Using 2014 and 2015 cropping seasons of Iran, Tajmiri et al.
(2017a) found that canola and alfalfa strip cropping increased pest predator species
diversity. Another study of Tajmiri et al. (2017b) in Iran pointed out that potato and alfalfa
strip intercropping could be an effective strategy to reduce pest density. Based on
experimental outcomes from 2009 to 2011 in China, Qin et al. (2013) pointed out that
adoption of strip intercropping is an effective strategy to reduce soil respiration and lower
carbon emission.
The state of the knowledge of strip cropping shows agronomic and ecological
(agroecological) benefits. Strip cropping has been practiced based on agronomic and/or
environmental considerations in human intensive farming and/or conventional mechanized
systems (Cruse and Gilley, 2008; Qin et al., 2013; Brooker et al., 2015; van Oort et al.,
2020; Rahman et al., 2021). Mixed cropping is often evident in manual agriculture as
compared to whole field sole cropping (Francis et al., 1986; Brooker et al., 2015). In
conventional mechanized systems operated with human operators strip cropping is
envisaged and practiced to maximize productivity and ecological benefits (Munz et al.,
2014c; Wang et al., 2015; AlarcónSegura et al., 2022). However, questions about
economics of strip cropping are yet to be answered as strip cropping economics is
constrained by the substantial added labour requirements. Research in the large-scale
farming context of the Midwest, US found that higher labour requirements and associated
fixed costs in conventional mechanized systems offset the profitability of maize and
soybean strip cropping (Ward, Roe and Batte, 2016; West and Griffith, 1992). The
economics of strip cropping is even constrained for smallholder’s context in China by
labour scarcity (Feike et al., 2012; Munz et al., 2014c).
Researchers have hypothesized that economically feasible agricultural intensification with
the strip cropping system could be possible with technological innovation. For example,
Exner et al. (1999) in their strip intercropping study based on Iowa, US farms pointed out
that precision management is demanded for strip cropping. The study of Lesoing and
Francis (1999) in their study of corn-soybean and grain sorghum-soybean strip cropping in
eastern Nebraska, US mentioned that new planting equipment could be more effective for
cereal-legume strip intercropping. van Oort et al. (2020) pointed out that future machinery
and crop growth models will enable farmers to achieve the benefits of intercropping. The
study also suggested that intercropping and autonomous swarm robotics co-evolution will
40
State of the art
help to achieve maximum benefits of intercropping and labour productivity. The maize and
soybean strip cropping study by Ward, Roe and Batte (2016) in the Corn Belt of the US
found that strip cropping profitability is constrained by higher labour requirements and
associated fixed costs related to conventional machines. Their study hypothesized that
innovative technology such as small supervised autonomous machines (i.e., robots) could
change the cost calculus of strip cropping.
The economic benefits of strip cropping as found in literature was measured using partial
indicators such as Land Equivalent Ratio (LER), Gross Margin Ratio (GMR), Monetary
Equivalent Ratio (MER) and/or harvested yields (Francis et al., 1986; Smith and Carter,
1998; Lesoing and Francis, 1999; Yu et al., 2015; van Oort et al., 2020; Rahman et al.,
2021). A very few studies used partial budgeting (West and Griffith, 1992; Exner et al.,
1999; Ward, Roe and Batte, 2016; Kermah et al., 2017). The most up-to-date economic
analysis of strip cropping using partial budgeting was conducted by Ward, Roe and Batte
(2016) considering the context of the Corn Belt of the US. However, the study was unable
to test their robot hypothesis due to a lack of autonomous whole farm operations
experience and data.
The existing strip cropping economics studies to date concentrated on partial assessment
of economic scenarios instead of whole farm systems analysis capturing planting to
harvesting operations. To help close this research gap, the study assessed the profitability
of maize and soybean strip cropping with autonomous machines on a central Indiana, US
farm. Maize and soybean farm of the Corn Belt of the US was considered because the
study evaluated the hypothesis of Ward, Roe and Batte (2016), where agronomic benefits
of yield increase for corn and penalty for soybean related to edge effects are available.
2.6 Regenerative agriculture (AND/OR) autonomous machines: Objective 3
The scientific definition of regenerative agriculture is not yet clear (Schreefel et al., 2020).
The existing definitions are based on processes (i.e., incorporating cover crops, livestock
and tillage reduction or elimination), outcomes (i.e., improvement of soil health, carbon
sequestration and biodiversity enhancement) and/or combination of both (Newton et al.,
2020; Manshanden et al., 2023). In this study, regenerative agriculture is considered with
strip cropping practices that diversify crop production within the same field in strips to
minimize soil disturbance, improve resource use efficiency of the farm through reducing
synthetic chemical input use, and boost soil health, biodiversity, and farm productivity.
The combination of processes and outcomes based definition assumed in this study
considered five soil health principles because soil health is the entry point to achieve the
multiple objectives of arable farming, such as production and nature conservation
41
State of the art
(LaCanne and Lundgren, 2018; Schreefel et al., 2020; Schreefel et al., 2022b). The soil
health principles include minimising soil disturbance, maximizing crop diversity, keeping
soil covered, maintaining living roots year round and including livestock components
(Jaworski, Dicks and Leake, 2023; Manshanden et al., 2023) as shown in Figure 2.2.
Figure 2.2: Five principles of regenerative agriculture.
Source: Lower Blackwood Catchment. Adopted from: Cool Farm Tool. Available at:
https://coolfarmtool.org/2020/12/regenerative-agriculture-and-climate-change/ (Accessed: 21 December
2022).
Strip cropping systems are the simplest regenerative mixed cropping system. It is
technically feasible even on conventional mechanized farms (Exner et al., 1999; van
Apeldoorn et al., 2020; Alarcón-Segura et al., 2022). This cropping system could also be
considered as agroecological farming because strip cropping is an innovative
agroecological practice to produce more (i.e., owing to edge effects) (Ward, Roe and
Batte, 2016) with less external resources (FAO, 2019). Although the definitions of
agroecology and regenerative agriculture differ, field biodiversity is a common element
(FAO, 2019; Tittonell et al., 2022). One of the key differences between agroecology and
regenerative agriculture is that ‘agroecology’ refers 10 elements (FAO, 2019), which
include 'political' or 'activist' elements as well as production aspects, whereas regenerative
agriculture is increasingly supported by commercial and large-scale farming (Tittonell et
al., 2022; Manshanden et al., 2023). At recent times, regenerative agriculture has been
promoted by civil society, NGOs, media and multinational food companies considering
agronomic and ecological grounds (Gosnell, Gill and Voyer, 2019; Giller et al., 2021;
Umantseva, 2022).
42
State of the art
However, the production economics of regenerative agriculture shows mixed results
(WBCSD, 2023; Schreefel et al., 2022a; Constantin et al., 2022). Profitability of
regenerative practices are constrained by higher labour requirements and farm
management challenges which limit the adoption and scaling up (Pearson 2007;
Keshavarz and Sharafi, 2023). In this context, regenerative agriculture is envisaged with
autonomous machines to reduce labour needs and enable intensive management
(Davies, 2022; The Pack News, 2022). Research hypothesized that robotics and
autonomous systems (RAS) could provide emerging opportunities that will help to achieve
net zero agriculture targets with regenerative agriculture (Pearson et al., 2022).
The state of the knowledge of regenerative practices reveal that regenerative agriculture
is a prominent alternative which could transform food production and ecosystem
restoration degraded by industrial monocultures (Gordon, Davila and Riedy, 2023). For
instance, considering the US farming systems, the study of Day and Cramer (2022)
pointed out the significance of regenerative agricultural transformation through linking
policy, process and education. Gosnell, Gill and Voyer (2019) mentioned that regenerative
agriculture is a climate smart mitigation and adaptation measure supported by
technological innovation, policy, education, and outreach. The study of Gremmen (2022)
suggested scientific and technology driven as well as nature-based regenerative
agriculture solutions to meet the demand for food of the increasing population. In the US,
LaCanne and Lundgren (2018) found that corn pest abundance was more than 10-fold
lower in regenerative multispecies cover crop systems compared to conventional systems.
Regenerative agriculture directly conserves and restores soil health, increases biodiversity
and ecosystem services, and sequesters atmospheric carbon (CO2). Similarly, part of the
co-benefits of regenerative agriculture is that it helps in producing healthy and nutritious
food (White, 2020).
Using an Australian case study, Bartley et al. (2023) found that regenerative grazing (i.e.,
rotational grazing with rest included strategically) improved vegetation, and soil and land
condition, but it took longer, a period of at least three to five years and a maximum fifteen
to twenty years, to capture the benefits. The study found that regenerative grazing
increased total nitrogen and soil organic carbon compared to control sites that did not
follow regenerative grazing. Eckberg and Rosenzweig (2020) pointed out that
regenerative agriculture is a farmer-led movement that adopts nature-based principles to
restore soil health, biodiversity, and farm economics because cereal grain intensive food
systems degrade natural resources. The study of Rehberger, West and Spillane, (2023)
pointed out that regenerative agriculture increases soil organic carbon, soil health and
biodiversity.
43
State of the art
Using participatory monitoring and evaluation, the study of Soto, de Vente and Cuéllar
(2021) in Spain, found that regenerative agriculture is a promising approach to restore
degraded agroecosystems. Rhodes (2017) mentioned that regenerative agriculture not
only increases soil organic carbon but also builds new soil that helps to improve soil
health and structure, increase soil fertility and crop yield, facilitate water retention and
aquifer recharge. McLennon et al. (2021) mentioned that regenerative agriculture helps to
reduce external inputs dependency and restore and maintain natural systems. Hellwinckel
and Ugarte (2011) argued that a regenerative agricultural transition is necessary to avoid
locking into a system that depletes the soil and fossil fuels.
Although the food production and nature conservation potentials of regenerative
agriculture is well known, the economics of regenerative agriculture has mixed literature
depending on the specific regenerative practices considered (Bennett, 2021; Boston
Consulting Group, 2023). A review study on the economics of regenerative agriculture in
Western Australia conducted by Bennett (2021), found that the profitability of
regenerative agriculture is lower compared to conventional agriculture. However, the
regenerative agriculture production economics is sensitive to enterprise type. The study
also pointed out that a loss of income is a significant barrier to scaling up regenerative
practices’ adoption. The study of Ogilvy et al. (2018) using financial data from sixteen
regenerative agriculture grazing farms found that before interest and tax, the earnings
from regenerative agriculture were more profitable compared to conventional grazing
systems. However, this study was contested by Francis (2019) who found that using the
same data, conventional systems grazing sheep achieved a return on assets of 4.22%,
while regenerative agriculture practitioners’ return on assets was only 1.66%. The
difference was mainly because the study of Ogilvy et al. (2018) only considered enterprise
or animal level analysis, not whole farm analysis considering all the factor costs and
assets invested as used in the study of Francis (2019).
A survey study based on North American context found that in the short run for the first
three to five years, farmers have to face loss of profitability, but in the long run the profit
increased with regenerative practices (WBCSD, 2023). The study of LaCanne and
Lundgren (2018) in the US, examined the profitability of regenerative corn and
conventional corn production. The study found that regenerative corn yield was 29% lower
compared to conventional corn production. However, profit was 78% higher for
regenerative practice which was due to the reduced inputs costs and higher output price
as there was positive association between organic matter and profit, not corn yield. The
survey study of Taylor and Dobbs (1988) in South Dakota, US found that regenerative
44
State of the art
agriculture was more profitable compared to conventional farming due to the lower input
costs for regenerative practice and improved prices for the regenerative agriculture
products.
In the Dutch dairy case, the study of Schreefel et al. (2022a) using an ex-ante modelling
approach found improved soil function at the expense of farm profitability. Another study
of Schreefel et al. (2022b) considered arable, dairy and mixed farming system in the
Netherlands. The study found that regenerative practices improve environmental
performances at the expense of farm profitability on an average by 50% across all case
study farms. Constantin et al. (2022) did a review of literature of comparative
environmentally friendly farming systems and found that regenerative agriculture is not an
economically feasible farming system, while suggesting large-scale practice to achieve
farm profitability.
The study of Pearson (2007) pointed out that regenerative systems required higher
labour. The study suggested technological innovations to solve the problem. In the Iranian
context, the study of Keshavarz and Sharafi (2023) found that climate smart regenerative
agriculture is a plausible solution to restore agroecosystems, but scaling up was a
significant challenge. Along with other socio-economic and institutional attributes, the
study suggested technological changes to help wide scale adoption. Precision
technologies are suggested to facilitate regenerative agriculture (Green Biz, 2020; Listen
Field, 2021; Futures Centre, 2023; Manshanden et al., 2023). For instance, the study of
Pearson et al. (2022) hypothesized that autonomous systems could help regenerative
agriculture. McLennon et al. (2021) suggested digital agriculture and sustainable
agricultural management using agricultural technologies with artificial intelligence (AI) for
regenerative agriculture. ‘Robot one’ a cutting-edge agricultural robot is expected to help
regenerative farming practices (Pixel Farming Robotics, 2023). The Kibb autonomous
initiative in Sweden is designed to promote regenerative farming (The Pack News, 2022).
However, the economics of regenerative agriculture with precision agriculture technology
is yet to be explored. In this study, autonomous machines are considered as precision
agriculture technology because they have the potential to cost effectively increase the
precision of input applications and to collect very detailed data on agricultural production.
This economic analysis will guide the regenerative agriculture practices, development of
autonomous prototypes and commercialization of autonomous technology and
regenerative farming.
The context of the UK was considered because Great Britain is one of the most nature
depleted countries which need ecosystem services regeneration. Moreover, the HFH &
45
State of the art
HFF was the first whole farm commercial operation conducted at Harper Adams
University in the UK (Hands Free Hectare (HFH), 2021). Another underlying reason is
associated with the government vision of net zero agriculture. The British government has
set an ambitious plan to achieve a net zero target by 2050 (Bank of Scotland, 2021;
RASE, 2021). The study of Davies (2022) hypothesized that autonomous machines could
facilitate regenerative agriculture that will help decarbonize cereal production in the UK. In
British agriculture and policy, regenerative agriculture has received growing attention as
part of a soil management strategy (Jaworski, Dicks and Leake, 2023).
Recent research and on-farm regenerative practices experience motivate regenerative
farming in Great Britain. Using simulation methods, Jordon et al. (2022a) estimated that
arable farming in Great Britain could mitigate 16-27% of agricultural emissions without
losing crop yield through the adoption of regenerative agriculture. The analysis found that
cover cropping, reduced intensity of tillage and incorporation of grass-based ley rotations
are effective regenerative practices that increase soil organic carbon. The study
mentioned that even though Great Britain farming systems have adoption constraints in
the existing farming systems, the adoption of regenerative agriculture could contribute to
the net zero target. The Bayesian meta-analysis of Jordon et al. (2022b) considering no or
reduced tillage, cover crops and ley-arable rotations in Great Britain found that
regenerative practices increased soil organic carbon compared to conventional practice.
However, these typical regenerative rotations have to yield increasing benefit, while the
study suggested future work could think of win-win farming with regenerative practice.
Considering the context of the UK, Ken Hill Farms and Estate at Snettisham, Norfolk,
found that farming using regenerative agriculture principles has the potential to save fixed
costs (Abram, 2020). They also pointed out that regenerative agriculture is an effective
strategy when farming practices include intercropping in the farms to reduce substantial
amounts of fertilizer and chemical inputs. Based on farmer’s experience it is hypothesized
that technology is the key to regenerative farming transition (Abram, 2021). With these
backdrops, delving into the profitability of regenerative strip cropping practices under both
conventional mechanized farming with human operators and autonomous system will help
Great Britain to link with the transitional vision related to productivity and environmental
sustainability that is “public money for public goods” subsidization policy.
46
Economics of field size and shape for autonomous machines
Chapter 3
Economics of field size and shape for
autonomous crop machines
“Arable crop production with autonomous equipment is technically and economically
feasible, allowing medium size farms to approach minimum per unit production cost
levels. The ability to achieve minimum production costs at relatively modest farm size
means that the pressure to “get big or get out” will diminish. … The ability of autonomous
equipment to achieve minimum production costs even on small, irregularly shaped fields
will improve environmental performance of crop agriculture by reducing pressure to
remove hedges, fell infield trees and enlarge fields.”
Lowenberg-DeBoer et al. (2021a): ‘Precision Agriculture, 22, pp. 1992–2006.
This chapter is published in the 'Precision Agriculture' journal as:
Al-Amin, A.K.M. Abdullah, Lowenberg DeBoer, J., Franklin, K. and Behrendt, K. (2023)
‘Economics of field size and shape for autonomous crop machines', Precision Agric 24,
1738–1765. Available at: https://doi.org/10.1007/s11119-023-10016-w (Accessed: 12 July
2023)
Rights and permissions:
Open Access: This article is licensed under a Creative Commons Attribution 4.0
International License, which permits use, sharing, adaptation, distribution, and
reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and
indicate if changes were made.
To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Note: This chapter writing style is adopted and formatted as submitted to the ‘Precision Agriculture’ Journal.
47
Economics of field size and shape for autonomous machines
3.1 Introduction
Field size and shape have substantial consequences for environmental management
(Bacaro et al., 2015; Clough, Kirchweger and Kantelhardt, 2020; Konvicka, Benes and
Polakova, 2016; Marja et al., 2019), technical (Fedrizzi et al., 2019; Griffel et al., 2018;
Griffel et al., 2020; Islam, Kabir and Hossain, 2017; Janulevičius et al., 2019; Luck,
Zandonadi and Shearer, 2011) and economic feasibility (Batte and Ehsani, 2006;
Carslaw, 1930; Larson et al., 2016; Miller, Rodewald and McElroy, 1981; Sturrock, Cathie
and Payne, 1977). To facilitate conventional agricultural mechanization, comparatively
large rectangular fields are needed and most of the land consolidation around the world in
the last decades have been motivated by the desire for larger fields (Kienzle, Ashburner
and Sims, 2013; Van den Berg et al., 2007). Field size and shape has been a key factor in
determining international crop competitiveness. Since the advent of motorized
mechanization countries with relatively large, roughly rectangular fields have had a major
economic advantage (e.g., US, Canada, Australia, Brazil, Argentina). In the UK, field size
has increased through removing hedgerows and in field trees to allow use of larger
machinery and ensure economies of size (MacDonald and Johnson, 2000; Pollard,
Hooper and Moore, 1968; Robinson and Sutherland, 2002). On the contrary, small fields
are often neglected and considered as non-economic. For instance, in the US many small
irregular-shaped fields were abandoned in the 20th Century. The European Union and
Switzerland retained small fields in production with subsidies (Lowenberg-DeBoer et al.,
2021a; OECD, 2017).
Nevertheless, under the umbrella of landscape management, small fields are promoted by
researchers. Research in Canada and the US found higher biodiversity in smaller fields
(Fahrig et al., 2015; Flick, Feagan and Fahrig, 2012; Lindsay et al., 2013). Likewise,
studies in the UK and the European Union also showed that small fields and more
fragmented landscapes have higher biodiversity (Firbank et al., 2008; Gaba et al., 2010;
González-Estébanez et al., 2011). Using the context of the agricultural low lands of
England, Firbank et al. (2008) pointed out that the pressure on biodiversity may be
reduced through minimizing habitat loss in agricultural fields. The German case study
found that East Germany's large-scale agriculture reduced biodiversity while small-scale
agriculture of West Germany had higher biodiversity (Batáry et al., 2017). As the
environmental benefits of small fields are well documented in research, it would be
interesting to explore the economics of small fields to better identify the win-win scenarios
for small fields. Consequently, this study hypothesized that autonomous crop machines
would make it possible to farm small, non-rectangular fields profitably, thereby preserving
field biodiversity and other environmental benefits.
48
Economics of field size and shape for autonomous machines
Autonomous crop machines in this study refer to the mechatronic devices which have
autonomy in operation usually through a predetermined field path. More specifically, the
autonomous machines are mobile, having decision making capability, and accomplish
arable farm operations (i.e., drilling, seeding, spraying fertilizer, fungicide and herbicide,
and harvesting) under the supervision of humans, but without the involvement of direct
human labour and operator (Lowenberg-DeBoer et al., 2020). Autonomous machines are
precision agriculture technology because they have the potential to cost effectively
increase the precision of input applications and to collect very detailed data on agricultural
production. The autonomous machines, demonstrated by the HFH project used swarm
robotics concepts in which multiple smaller robots are used to accomplish farm work
usually done by larger conventional machines with human operators. The autonomous
swarm robotics of the HFH project are developed by retrofitting conventional diesel
operated machines (Hands Free Hectare (HFH), 2021).
Autonomous machines are considered as a game changing technology that could
revolutionize precision agriculture (PA) and facilitate the 'fourth agricultural revolution'
often labelled ‘Agriculture 4.0’ (Daum, 2021; Klerkx and Rose, 2020; Lowenberg-DeBoer
et al., 2021a). Owing to population and economic growth, agricultural labour scarcity,
technological advancement, increasing requirements of operational efficiency and
productivity, and mitigating environmental footprint, autonomous machines are suggested
as a sustainable intensification solution (Duckett et al., 2018; Guevara, Michałek and
Cheein, 2020; Santos and Kienzle, 2020). Robotic systems for intensive livestock and for
protected environments have been commercialized more rapidly than for arable cropping.
Research on autonomous arable crop machines has mostly concentrated on the technical
feasibility, not economics (Fountas et al., 2020; Shamshiri et al., 2018). Understanding the
economic implications of autonomous machines is key to their long-term
adoption. Economic feasibility plays a crucial role in attracting investment, guiding
adoption decisions, and further understanding of environmental and social benefits
(Grieve et al., 2019; Lowenberg-DeBoer et al., 2020).
Most production economic studies on autonomous machines prior to 2019 focused on
horticultural crops and rarely on cereals using prototype testing and experimental data
(Edan, Benady and Miles, 1992; Gaus et al., 2017; McCorkle et al., 2016; Pedersen et al.,
2017, Pedersen, Fountas and Blackmore, 2008, Pedersen et al., 2006; Sørensen,
Madsen and Jacobsen, 2005). Lack of information on economic parameters and
machinery specifications has been a bottleneck in economic feasibility assessment
because autonomous machines are at an early stage of the development and
commercialization processes (Lowenberg-DeBoer et al., 2021a; Shockley et al., 2021).
49
Economics of field size and shape for autonomous machines
Most of the earlier economic studies used partial budgeting where only the changes in
cost and revenue linked to automation of a single field operation were analysed omitting
the economic consequences of farming systems changes (Lowenberg-DeBoer et al.,
2020). To date, four studies have considered systems analysis of autonomous machines
(Al-Amin et al., 2021; Lowenberg-DeBoer et al., 2021a; Shockley, Dillon and Shearer,
2019; Sørensen, Madsen and Jacobsen, 2005).
Using a Linear Programming (LP) model with data from prototypes at the University of
Kentucky, US, Shockley, Dillon and Shearer, (2019) showed that relatively small
autonomous machines are likely to have economic advantages for medium and small
farms. The most comprehensive study so far was reported by Lowenberg-DeBoer et al.
(2021a). They assessed the economic feasibility of autonomous machines from seeding
to harvesting operations using on-farm demonstration data and estimated equipment
times based on methodology from the agricultural engineering textbook of Witney (1988).
The study assumed 70% field efficiency from drilling to harvesting operations for both
autonomous machines and conventional equipment sets with human operators. They
showed that autonomous machines are technically and economically feasible for medium
and small sized farms. The study concluded that autonomous machines diminished the
pressure of “get big or get out”. The study hypothesized that in the context of the UK,
autonomous machines would be economically feasible in small fields. Nonetheless, the
study was unable to test the hypothesis because of field efficiency estimates by field size
and shape were not available.
To help fill this knowledge gap, the objective of the study is to assess the economics of
field size and shape for autonomous machines. Using the experience of the HFH
demonstration project, the study developed algorithms to estimate equipment times (h/ha)
and field efficiency (%) for different sized rectangular and non-rectangular fields.
Historically, in the UK rectangular fields were considered as the most efficient, whereas
non-rectangular fields were substantially less efficient to farm (Carslaw, 1930; Sturrock,
Cathie and Payne, 1977). Triangular fields were among the least efficient field shape
because of the numerous short rounds. To analyse the economic scenarios, the study
adopted and re-estimated the Hands Free Hectare-Linear Programming (HFH-LP) model
(Lowenberg-DeBoer et al., 2021a) by incorporating equipment times and field efficiency
parameters estimated with field size and shape algorithms. The HFH-LP model replicates
farm management and machinery selection decisions. It helps researchers understand
choices that farmers would make if they had the alternative of using autonomous
machine.
50
Economics of field size and shape for autonomous machines
3.2 Methods
3.2.1 Field time and efficiency estimation subject to field size and shape
To date the production economics studies on autonomous machines did not consider field
size and shape because of lack of data (Lowenberg-DeBoer et al., 2021a; Shockley,
Dillon and Shearer, 2019; Sørensen, Madsen and Jacobsen, 2005). Over time, the
performance of arable field machinery has received growing attention for farm
management and the ability to model field times has accelerated through the development
of the technology and modelling approaches (Bochtis et al., 2010; Sørensen, 2003;
Sørensen and Nielsen, 2005). Nonetheless, existing studies on arable crop machinery
performance lack information of equipment times (h/ha) and field efficiency (%) subject to
field size and shape.
Even though logistics software is well developed in trucking and other transportation
sectors (Software Advice, 2021), there is no readily available commercial software in the
UK to estimate equipment times and field efficiency encompassing field and machine
heterogeneity. In the farm equipment path planning research literature, field times were
sometimes generated as a by-product (Hameed, 2014; Jensen et al., 2012; Oksanen and
Visala, 2007; Spekken and de Bruin, 2013). The agri-tech economic studies often rely on
the general estimates of agricultural engineering textbooks like Hunt (2001) and Witney
(1988). In conventional mechanization and PA literature, few studies estimated field
efficiency, but prior studies treated the headlands of the field as non-productive areas,
excluded overlap percentage, amalgamated productive field times (i.e., field passes,
headlands turning, and headlands passes) and non-productive field times (i.e., replenish
inputs, refuelling, and blockages), and ignored the headland turning patterns.
Studies suggested that future research should separately calculate the headlands turning
time, and stoppages time because productive times and non-productive times play a
significant role in field efficiency estimation. Keeping these points in consideration, the
study developed field time approximation algorithms by field size and shape for 28 kW,
112 kW and 221 kW conventional equipment sets with human operators, and for the HFH
sized 28 kW autonomous equipment set. The combine harvesters were assumed to have
head widths of 2 m, 4.5 m and 7.5 m respectively. Using the experience of the HFH
demonstration project, the algorithms addressed the research gaps identified from the
prior studies. The study estimated field efficiency as the ratio of theoretical field time
based on machine design specifications like the estimates of theoretical field time to its
actual field productivity as follows:
… … (1)
51
Economics of field size and shape for autonomous machines
where, is the field efficiency, TT is the theoretical field time, is the total observed
time in the interior field and passes, is the total headland round time, and total
stoppage time “within” in the field.
Based on user input of equipment and field measurements, the first step was to calculate
field area, number of headlands rounds and other values that were used repeatedly
throughout the algorithm. Secondly, headland area and field times were calculated.
Afterwards, observed times in the interior field and passes were estimated. Fourthly, the
algorithms estimated non-productive times. Fifthly, total field operation times were
calculated. The theoretical field times were estimated based on the machine design
specifications. For details of the estimation processes of the algorithms see the technical
note in Appendix A (i): Supplementary Text (i.e.., STEXTT Supplementary Text, which
includes Main Text of the Technical Note).
The algorithms were calibrated for 1 ha, 10 ha, 20 ha, 50 ha, 75 ha, and 100 ha
rectangular fields considering the typical farm field sizes of the UK that were assumed to
follow the field path of Figure 3.1. To illustrate the impact of field size on technical
efficiency, estimates were made for rectangular fields with the length ten times the width
of the field, up to one kilometre length. Rectangular field algorithms are detailed in the
algorithm’s spreadsheet in Appendix A (ii): Algorithms Spreadsheets (i.e., SM1
Rectangular Field Algorithms).
Figure 3.1: Typical field path for rectangular fields considered in the study based on the
HFH demonstration project experience.
Interior Field
Headland
52
Economics of field size and shape for autonomous machines
Similarly, non-rectangular fields algorithms were tested for 1 ha, 10 ha, 20 ha, and 25 ha
sized right-angled triangular fields assuming the height equalling twice the base up to a
height of one kilometre. The equipment sets were assumed to follow the typical field path
given in Figure 3.2. The non-rectangular fields algorithms were estimated with the same
equipment sets (for details of the right-angled triangular field algorithms see spreadsheet
in Appendix A (ii): Algorithms Spreadsheets (i.e., SM2 Non-Rectangular Field Algorithms
(i.e., Right-Angled Triangular Field)).
Figure 3.2: Typical field path for non-rectangular (i.e., right-angled triangular) fields
considered in the study based on the HFH demonstration project experience.
The study assumed that the equipment enters the field from the lower left corner and
completes the headlands first for all field operations (i.e., drilling, spraying, and
harvesting). Afterwards, the machine makes a “flat turn” to start the interior passes.
Subsequently, follows the “flat turn” to complete the interior headland turns. Finally, the
study assumed that the equipment ends on the entry side of the fields as shown in Figure.
3.1 and Figure. 3.2.
Interior Field
Headland
53
Economics of field size and shape for autonomous machines
3.2.2 Modelling the economics of field size and shape
To understand the whole farm effects of field size and shape with different types of farm
equipment, the study adopted and re-estimated the Hands Free Hectare - Linear
Programming (HFH-LP) model. The HFH-LP model is a decision-making tool which
assesses the economics of autonomous machines compared to conventional equipment
sets with human operators. Consistent with typical neoclassical microeconomic farm
theory, the objective function of the HFH-LP model was to maximize gross margin (i.e.,
return over variable costs) subject to primary farm resource constraints in the short-run. In
the subsequent stages, using the outcome of the HFH-LP model, the study examined net
return to operator labour, management and risk taking (ROLMRT) and evaluated the
wheat cost of production to explore the cost economies (i.e., economies of size) (Debertin,
2012; Duffy, 2009; Hallam, 2017; Miller, Rodewald and McElroy, 1981). The HFH-LP
model is a one-year “steady state” model for arable grain-oil-seed farm, where the model
assumed a monthly time step from January to December. It is steady state in the sense
that it is assumed that solutions would be repeated annually long term. The concept of
“steady state” was carried over from the Orinoquia model (Fontanilla-Díaz et al., 2021)
which used the same software. Following Boehlje and Eidman (1984), the HFH-LP
deterministic economic model can be expressed as:
The objective function:
Subject to:
where, π is the gross margin, is the level of jth production activities, is the gross
margin per unit over fix farm resources () for the jth production activities, is the
amount of ith resource required per unit of jth activities, is the amount of available ith
resource.
The HFH-LP model encompassed limiting constraints i.e., land, human labour, equipment
times (i.e., tractor use time for drilling and spraying, and combine use time for harvesting),
working capital and cashflow. The equipment scenarios encompassed four farm sizes: 66
ha, 159 ha, 284 ha and 500 ha farms, but did not model field size or shape. This study re-
estimated the labour use, tractor use and combine use times for larger fields (10 ha) or
54
Economics of field size and shape for autonomous machines
smaller fields (1 ha), that were either rectangular or non-rectangular (i.e., right-angled
triangular). The assumptions regarding variable costs, crop yields, and land use were
same as Lowenberg-DeBoer et al. (2021a). The crop variable costs were the same across
scenarios, but machinery costs differed. Details of the linear programming (LP)
coefficients including machinery investment and operating costs are available from the
supplementary materials of Lowenberg-DeBoer et al. (2021a). The 10 ha field size was
selected for the large fields, because the field efficiency algorithm estimates showed that
over 10 ha, field efficiency does not vary much by field size. A 1 ha field size was selected
to represent small fields, because relatively few fields in the UK are smaller than 1 ha. The
rectangular shape was selected as the shape usually considered most efficient for
mechanized farming, and the triangular as the field shape that is among the least efficient
(Carslaw, 1930).
The time window is crucial because agricultural operations are sensitive to weather
conditions and crop activities. In literature the probability of good field days is considered
as primary mechanism to model risk-aversion. The PC/LP model used good field days
available in the 17th worst year out of 20 (McCarl et al., 1977) that is 85% of the time.
Following Agro Business Consultants (2018) the study assumed that number of good field
days available was in 4 years out of 5 years that is 80% of times. Similar to the original
HFH-LP model, the conventional machines assumed that field operations of drilling,
spraying and harvesting were conducted during daytime that is on an average 10 h/day.
The autonomous machines assumed that tractor for drilling and spraying was operated for
22 h/day (2 h for repair, maintenance, and refuelling) while autonomous combine operated
for 10 h/day limited for night dew. The LP models of the study were coded using the
General Algebraic Modelling System (GAMS) (https://www.gams.com/). Details of other
associated assumptions and the programming code is available at Appendix C (GAMS
code used) or at the supplementary materials of Lowenberg-DeBoer et al. (2021a).
3.2.3 Case study and data sources
Because the Hands Free Hectare (HFH) was a demonstration project, it was difficult to
separate on-field stops and down time while the engineers tinkered from those stoppages
that would have occurred in normal field operations. Consequently, the model parameters
were based on published machine specifications and farm budget information, and guided
by the qualitative experience of the HFH project demonstrated at Harper Adams
University, Newport, Shropshire, UK (Hands Free Hectare (HFH), 2021). The Lowenberg-
DeBoer et al. (2021a) HFH-LP model represented the arable grain-oil-seed farm in the
West Midlands of the UK, this study re-estimated field times to reflect the range of field
sizes and shapes often found in Britain. To calibrate the HFH-LP model, the study used
55
Economics of field size and shape for autonomous machines
parameters from different sources. The information about commodity produced and the
costs estimates were from the Agricultural Budgeting and Costing Book (Agro Business
Consultants, 2018) and the Nix Pocketbook (Redman, 2018). To facilitate comparability
with the Lowenberg-DeBoer et al. (2021a) results, 2018 input and output price levels were
retained. Prices were converted following daily average exchange rate of 2018 from Great
British Pounds (GBP) to Euro (€) of €1.1305 (Bank of England, 2018). Details of the
machine inventory, costs of machines, hardware and software, crop rotations and key
baseline assumptions are available at Lowenberg-DeBoer et al. (2021a). Field operation
timing was adopted from Finch, Samuel and Lane (2014) and Outsider’s Guide (1999).
Equipment timeliness (i.e., HFH 28 kW conventional equipment set with human operator
and autonomous machine, 112 kW and 221 kW conventional equipment sets with human
operators) were estimated through the developed algorithms, where the equipment and
field specifications were collected from HFH demonstration experience
(https://www.handsfree.farm/) (Hands Free Hectare (HFH), 2021), conventional machine
specifications from John Deere (https://www.deere.co.uk/en/index.html) (John Deere,
2022), Arslan et al. (2014) and Lowenberg-DeBoer et al. (2021a). For more details of the
technical parameters used and data sources see Appendix A (ii): Algorithms
Spreadsheets (i.e., SM1 Rectangular Field Algorithms and SM2 Non-Rectangular Field
Algorithms (i.e., Right-Angled Triangular Field)).
3.3 Results
3.3.1 Field efficiency and times: rectangular fields
The study evaluated the technical feasibility of the HFH 28 kW conventional equipment
with human operator and autonomous machines, and 112 kW and 221 kW conventional
equipment sets with human operators for all field operations including direct drilling, five
spray applications and harvesting operation. The spray application included pre-drill burn
down, two nitrogen top dressing and fungicide applications, late season fungicide and pre-
harvest desiccant. The human and equipment times were re-estimated subject to field
size and shape scenarios. Results show that average whole farm field efficiency for 112
kW and 221 kW equipment sets differed substantially between 1 ha and 10 ha rectangular
fields, whereas for rectangular fields a given equipment set the field efficiency was almost
the same for 10 ha to 100 ha fields (Figure 3.3). The whole farm field efficiency of HFH
equipment sets was relatively high irrespective of different sized rectangular fields, but
efficiency for 112 kW and 221 kW conventional equipment sets with human operators
dropped for small 1 ha fields. Beyond 10 ha, the field efficiency for a given equipment set
was similar for all rectangular field sizes (i.e., 20 ha, 50 ha, 75 ha, and 100 ha).
56
Economics of field size and shape for autonomous machines
Figure 3.3: Estimated (weighted average) whole farm field efficiency of HFH equipment
(i.e., 28 kW conventional equipment with human operator and autonomous machine),
large conventional and small conventional machines with human operators in different
sized rectangular fields.
Operation specific equipment times (h/ha) and field efficiency (%) results of the
rectangular fields show that equipment times for drilling and harvesting operations were
longer for small 1 ha fields operated with equipment of all sizes and types, but field sizes
had least impact for the HFH equipment sets (Table 3.1). The higher time for small 1 ha
fields was largely due to the fact that the full width of the larger equipment could not be
used effectively in the smaller fields.