Robo-Taxi Service Fleet Sizing: Assessing the Impact of User Trust and Willingness to Use

Abstract

The first commercial fleets of Robo-Taxis will be on the road soon. Today important efforts are made to anticipate future Robo-Taxi services. Fleet size is one of the key parameters considered in the planning phase of service design and configuration. Based on multi-agent approaches, the fleet size can be explored using dynamic demand response simulations. Time and cost are the most common variables considered in such simulation approaches. However, personal taste can affect the demand and consequently the required fleet size. In this paper, we explore the impact of user trust and willingness to use on Robo-Taxi fleet size. This investigation is carried out by simulating the transportation system of the Rouen Normandie metropolitan area in France using MATSim, a multi-agent activity-based simulator. A local survey is made in order to explore the variation of user trust and their willingness to use future Robo-Taxis according to the sociodemographic attributes. After applying this survey data in the simulation, the obtained results reveal the significant importance of traveler trust and willingness to use variations on Robo-Taxi use and the required fleet size.

Full text

Robo-Taxi Service Fleet Sizing: Assessing the Impact of User Trust and Willingness to Use R. Vosooghi1,2, J. Kamel1, J. Puchiner1,2, V. Leblond1, M. Jankovic2
1SystemX Instute of Research and Technology
2Industrial Engineering Research Department (LGI), CentraleSupeléc, Université Paris-Saclay
Problem Statement
Robo-Taxi Services Simulaon
Research Gap
 Neglecng user taste variaon
Contribuons
 Integrang user heterogeneity in terms of trust and willingness to use future Robo-Taxi services
 Local survey made in order to explore user taste variaon
Categorized Scoring Funcon
The scoring funcon as the core of co-evoluonary algorithm has been modied and categorized by travelers’
socio-professional categories in order to:
 Integrate the dierent behavior of travelers according to their personal aributes,
 Dierenate the similar daily acvity paern of groups of individuals according to their main daily acvity tour.
User Taste Variaon Factors
A recent survey made in France that addressed 457 persons with dierent individual aributes and current modes
shows that user trust varies according to age and gender. In general:
 Men are more likely to use Robo-Taxis than women,
 Younger persons are more likely to use Robo-Taxis in comparison to older ones.
It is assumed that the constant ulity of mode Robo-Taxi varies according to those aributes. This is given by using
variable user trust factor:
These variaons are derived from the analysis of the local survey and the distribuon graph based on it (Figure 1).
According to this analysis, the Robo-Taxi willingness to use is strongly correlated to the household income.
Therefore, we assumed that the percepon of in-vehicle and waing mes varies with income:
Figure 1: Distribuon graph of user tendency factors.
Methodology
Preparing Data
Synthec Populaon Generaon
Synthec populaon of the case study region has been generated applying tness‐based synthesize with
mullevel controls (FBS-MC) approach developped by the authors. The process is simple: while a set of
households is drawn randomly from the sample, a mullevel controller measures the tness of marginal synthec
and real data by zone and by aributes of interest.
Acvity Chain and Locaon Analysis
Once the synthec populaon is generated, the next step is to allocate acvity chains to each individual. This has
been done using the frequency of each acvity chain in the transport survey according to socio-professional
aributes. By analyzing the transport survey of the case study area, we found that the acvity chains are
signicantly correlated to those aributes (Figure 3).
Figure 2: Synthec populaon with allocated located daily acvity chain generaon framework.
Figure 3: Top ten acvity chains of Rouen Normandie metropolitan populaon and the frequency of socio-
professional categories of individuals.
Figure4: Acvity start me models esmated from regional transport survey (EMD Rouen 2017).
'
, , , , , , , , , ,
()
trav cat ut m cat trav m cat ivt ivt wt wt dist m cat trav co m cat pl m cat
S C t t d
      
          
()
22
Age Sex
ut




1
ivt
Income
wt Income




0.85
0.9
0.95
1
1.05
1.1
1.15
Sex
Female
Male
0.7
0.8
0.9
1
1.1
1.2
1.3
0 20000 40000 60000 80000 100000 120000
Income (€)
0.7
0.8
0.9
1
1.1
1.2
1.3
0 20 40 60 80 100
Age (Years)
Simulation Setup
Framework
 MATSim: Mul-agent acvity-based Simulaon
 30 hours daily travel and acvity paern of whole populaon with queue-based trac simulaon
 New scoring funcon and subpopulaon factories
Baseline Scenario : Rouen Normandie Metropolitan Area (France)
 Around 490,000 inhabitants
 Fine-grained daily acvity chains (929) based on recent transport survey analysis (EMD2017)
 Calibrated by actual modal shares
Scenarios
 Various eet sizes (1k-10k)
 Simulaons with and without considering user trust and willingness to use
 No ride-sharing
(a) (b) (c)
Simulation Results
Modal Shares
 Robo-Taxi modal share increases proporonally to the eet size
 Modal shis toward Robo-Taxi come mainly from public transport, car and walk
 The use of public transport decreases signicantly relave to other modes
 Mode shares with and without considering user taste variaon are almost similar
 However, aer considering taste variaon, Robo-Taxi users are dierent in terms of sociodemographic
aributes, socio-professional proles, average trip distances and acvity chains (trip paerns)
Detailed Analysis
Hourly Robo-Taxi In-Service Rate
 User taste variaons lead to dierent temporal Robo-Taxi usage paerns.
 Maximum eet usage occurs when the eet size is between 2k and 3k.
 For the eet size of 1k, passenger wait mes are high, and therefore Robo-Taxi
usage is lower.
Total duration of in-service drive over total duration of all tasks
010 20 30 40 50 60 70 80 90 100
w/ considering user trust and willingness to use
10000 0 0 1 1 3 13 31 70 87 87 67 64 67 65 63 65 68 68 64 51 35 15 6 3
9000 0 0 1 1 3 14 34 71 84 81 68 68 69 72 68 69 75 79 71 54 39 15 6 3
8000 0 0 1 1 4 16 39 82 97 99 81 69 71 69 67 72 80 76 71 59 44 16 7 3
7000 0 0 1 2 4 17 43 91 100 100 86 78 81 82 75 82 90 90 83 70 49 18 8 4
6000 0 0 1 2 6 18 48 96 100 100 87 82 86 85 78 80 95 99 90 72 54 22 9 4
5000 0 0 1 2 6 24 53 99 100 100 95 85 86 84 77 87 94 99 90 76 58 24 11 5
4000 0 1 1 3 7 29 66 100 100 100 95 86 91 87 78 92 100 100 96 78 57 27 12 7
3000 0 1 2 3 9 32 78 100 100 100 96 93 91 96 89 93 100 99 98 89 70 30 15 7
2000 0 1 2 5 12 40 85 100 100 100 100 100 97 94 93 92 100 100 96 93 76 35 16 9
1000 0 2 2 6 16 49 92 100 100 100 100 99 89 79 78 85 100 97 96 79 65 30 18 13
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
w/o considering user trust and willingness to use
10000 0 0 1 1 3 13 30 68 83 85 72 65 66 68 61 65 69 70 64 49 32 12 6 3
9000 0 0 1 1 3 14 34 70 87 85 68 67 71 70 66 69 71 73 68 55 40 16 6 3
8000 0 0 1 1 4 15 35 79 93 90 72 68 70 70 66 72 75 76 70 56 40 17 7 4
7000 0 0 1 1 4 18 42 83 95 94 75 72 74 75 73 76 82 80 70 64 46 18 7 3
6000 0 0 1 1 5 20 48 93 100 100 86 84 82 84 77 88 95 96 79 65 50 19 9 3
5000 0 1 1 3 6 22 54 99 100 100 89 80 85 81 79 86 88 98 98 81 56 23 10 5
4000 0 0 2 3 7 27 65 100 100 100 92 86 83 85 77 88 96 99 87 75 58 26 13 5
3000 0 0 2 4 8 30 73 100 100 100 91 87 83 74 76 89 100 100 91 72 60 30 14 6
2000 0 1 2 4 10 40 88 100 100 100 96 98 92 88 81 86 100 100 97 78 71 34 14 8
1000 0 1 4 6 17 54 91 100 100 100 96 87 93 93 73 80 95 97 92 60 54 27 20 12
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Fleet size
Hour of day
Fleet size
Hour of day
Robo-Taxi Service and User Related Relave Changes
 The indicators vary for all eet sizes with unlike rao but with similar trends.
 The dierence on average passenger waing me is posive and less than 1.1
minute for all scenarios.
 Service demand and average vehicle with passenger on-board driving mileages
have two major changes in the eet sizes of 3k and 7k vehicles.
Average Robo-Taxi Daily, Peak Hour and O-Peak In-Service Rates
 The peak hours are assumed to be 8-10 a.m. and 5-7 p.m.
 Average daily in-service rate changes aer introducing user tendency variaon are signicant for the eet sizes of 3k and 7k vehicles.
 For the eet sizes of less than 5k, the average morning peak hour in-service rate remains unchanged due to the excessive demand.
 In o-peak hours an important alteraon for the eet size of 3k vehicles is observed.
Conclusion & Outlook
By introducing user trust and willingness to use, signicant changes occur for two eet sizes:
1) The eet required to meet the maximum demand (with all vehicles in-service at least for one hour),
2) The eet required to meet approximately the maximum usage.
The results indicate, that demographic structure as well as preference variaons need to be taken into account in
Robo-Taxi eet sizing.
Future work will:
 Extend the current framework towards Robo-Taxi ride-sharing services.
 Take into consideraon vehicle-related aspects such as scapacity and range.
 Mul-Agent Transport Simulaon, MATSim. hp://matsim.org/
 Kamel, J., Vosooghi, R., and J. Puchinger. Synthec Populaon Generator for
France. hps://github.com/josephkamel/SPGF2.
 Al Maghraoui, O. Designing for Urban Mobility - Modeling the Traveler
Experience. CentraleSupéléc, Université Paris Saclay, 2019.
 Kamel, J., R. Vosooghi, J. Puchinger, F. Ksonni, and G. Sirin. Exploring the
Impact of User Preferences on Shared Autonomous Vehicle Modal Split : A
Mul-Agent Simulaon Approach. 21st EURO Working Group on
Transportaon Meeng, EWGT 2018, 2018.
This research work has been carried out in the framework of IRT SystemX, Paris-Saclay, France, and therefore
granted with public funds within the scope of the French Program “Invesssements d’Avenir”. The authors
would like to thank Groupe Renault for parally nancing this work and Métropole Rouen Normandie for
providing the data.
0.7
1.1 1.0
0.8
0.0
0.7
1.0
0.7
0.1
0.3
0.7
1.1 1.0
0.8
0.0
0.7
1.0
0.7
0.1
0.3
1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Robo-Taxi fleet size (number of vehicules)
Average passenger waiting time differences (min)
Robo-Taxi fleet size (number of vehicules)
1.6%
3.9%
6.8%
4.2% 3.5%
2.4%
5.3%
1.4% 1.6% 1.4%
Service demand changes (%)
3.4%
1.4%
4.4%
2.4%
1.6% 0.9%
3.5%
0.7%
-1.4%
-0.6%
1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Average vehicle on-board driving mileage changes (%)
Synthetic Population Generation
Transport
Survey
Public Use
Microdata
Sample
Activity Chain Analysis
Activity Location Analysis
Synthetic Population with
Allocated Located Daily
Activity Chain (Plan)
Facility Data
Categorized by Socio-Professional Groups
Missed attribute
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
RF
12.9%
RF
6.7%
RF
5.8%
RF
4.2%
RF
2.7%
RF
2.5%
RF
2.1%
RF
1.86%
RF
1.84%
RF
1.48%
Home
|
Study
|
Home
Home
|
Work
|
Home
Home
|
Shopping
|
Home
Home
|
Leisure/Visit
|
Home
Home
|
Study
|
Home
|
Leisure/Visit
|
Home
Home
|
Study
|
Home
|
Study
|
Home
Home
|
Personal Errands
|
Home
Home
|
Work
|
Home
|
Work
|
Home
Home
|
Shopping
|
Shopping
|
Home
Home
|
Shopping
|
Home
|
Leisure/Visit
|
Home
Relative frequency (%)
Activity chains and its relative frequency (RF %)
Employed Unemployed Students Under 14 years Retired Homemakers
3% 4%
7%
3%
1%
2%
9%
4%
1%
1%
1 000
2 000
3 000
4 000
5 000
6 000
7 000
8 000
9 000
10
000
Comparision of Average In-Service Rates Before and After the Introduction of User Trust and Willingness to Use for
Different Robo-Taxi Fleet Size
Average daily in-service rate
0% 0%
0%
0%
0%
1%
7%
6%
-3%
3%
1000
2000
3000
4000
5000
6000
7000
8000
9000
10 000
Average morning peak hour
in-service rate
3%
0%
2%
5%
0%
5%
14%
3%
7%
-1%
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Average evening peak hour
in-service rate
5% 7%
12%
3%
2%
2%
8%
4%
0%
1%
1000
2000
3000
4000
5000
6000
7000
8000
9000
10 000
Comparision of Off-Peak Hours In-Service Rate Before and After the
Introduction of User Trust and Willingness to Use for Diffrent Fleet Size
Average off-peak hours in-service rate
Robo-Taxi
Operation
(characteristics)
Fleet
Specification
Relocation
Strategies
Assignment
and Routing
Dynamic Fleet
Management
Sharing
Strategies
…
Robo-Taxi
Users
(demand)
Travel Behavior
Changes
Mode Share
Changes
Traffic State
Changes
Air Quality
Changes
Land Use
Changes
…
Transportation
System
(impact)
55.1
55.1
54.9
54.8
54.4
54.4
54.1
53.9
53.8
53.7
53.4
53.4
53.3
53.0
53.1
53.0
52.9
52.8
52.6
52.6
3.1
3.2 5.0
5.2
6.8
7.0
8.2
8.6
9.3
9.6
10.3
10.6
11.1
11.6
11.8
12.1
12.5
12.7
13.2
13.4
10.2
10.0
8.9
8.9
7.9
7.9
7.1
7.0
6.4
6.4
5.9
5.9
5.4
5.4
5.1
5.2
4.7
4.8
4.5
4.5
30.4
30.4
30.0
29.9
29.8
29.6
29.6
29.4
29.5
29.2
29.4
29.1
29.2
29.0
29.0
28.8
28.9
28.6
28.7
28.5
1.2
1.2
1.1
1.1
1.1
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
w/o w/ w/o w/ w/o w/ w/o w/ w/o w/ w/o w/ w/o w/ w/o w/ w/o w/ w/o w/
1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Modal split (%)
Robo-Taxi fleet size (number of vehicules)
Bike
Walk
Public Transport
Robo-Taxi
Private Car