Available via license: CC BY 4.0
Content may be subject to copyright.
Citation: Terranova, P.; Dean, M.E.;
Lucci, C.; Piantini, S.; Allen, T.J.;
Savino, G.; Gabler, H.C. Applicability
Assessment of Active Safety Systems
for Motorcycles Using
Population-Based Crash Data:
Cross-Country Comparison among
Australia, Italy, and USA.
Sustainability 2022,14, 7563.
https://doi.org/10.3390/su14137563
Academic Editor: El ˙
zbieta Macioszek
Received: 20 April 2022
Accepted: 10 June 2022
Published: 21 June 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
sustainability
Article
Applicability Assessment of Active Safety Systems for
Motorcycles Using Population-Based Crash Data:
Cross-Country Comparison among Australia, Italy, and USA
Paolo Terranova 1,2 ,* , Morgan E. Dean 1, Cosimo Lucci 2, Simone Piantini 2, Trevor J. Allen 3,
Giovanni Savino 2,3 and Hampton C. Gabler 1, 2, †
1Department of Biomedical Engineering and Mechanics, Virginia Tech, Blacksburg, VA 24061, USA;
morgandean@vt.edu (M.E.D.)
2
Department of Industrial Engineering—DIEF, University of Florence, Via di S. Marta, 3, 50139 Florence, Italy;
cosimo.lucci@unifi.it (C.L.); simone.piantini@unifi.it (S.P.); giovanni.savino@unifi.it (G.S.)
3Accident Research Centre, Monash University, Clayton, VIC 3800, Australia; trevor.allen@monash.edu
*Correspondence: pterranova@vt.edu; Tel.: +1-540-824-8570
† This author has passed away.
Abstract:
The role of powered two-wheeler (PTW) transport from the perspective of a more sustain-
able mobility system is undermined by the associated high injury risk due to crashes. Motorcycle-
based active safety systems promise to avoid or mitigate many of these crashes suffered by PTW
riders. Despite this, most systems are still only in the prototype phase and understanding which
systems have the greatest chance of reducing crashes is an important step in prioritizing their de-
velopment. Earlier studies have examined the applicability of these systems to individual crash
configurations, e.g., rear-end vs. intersection crashes. However, there may be large regional differ-
ences in the distribution of PTW crash configurations, motorcycle types, and road systems, and hence
in the priority for the development of systems. The study objective is to compare the applicability of
five active safety systems for PTWs in Australia, Italy, and the US using real-world crash data from
each region. The analysis found stark differences in the expected applicability of the systems across
the three regions. ABS generally resulted in the most applicable system, with estimated applicability
in 45–60% of all crashes. In contrast, in 20–30% of the crashes in each country, none of the safety
systems analyzed were found to be applicable. This has important implications for manufacturers
and researchers, but also for regulators, which may demand country-specific minimum performance
requirements for PTW active safety countermeasures.
Keywords:
motorcycle; powered two-wheelers (PTWs); safety systems; antilock braking (ABS);
autonomous emergency braking (MAEB); collision warning; curve warning; curve assist
1. Introduction
Motorcycles and mopeds—also called powered two-wheelers (PTWs)—play an in-
creasingly important role in personal mobility in several countries. In the state of Victoria,
Australia, from 2007 to 2012, the number of registered motorcycles increased by 30.6%
(from 122,825 to 160,390), while passenger vehicle registrations increased by only 9.2%
over the same period [
1
,
2
]. Similarly, there have been large increases in PTW numbers in
Sweden (+50% from 2000 to 2013) [
3
] and Italy (+44% from 2004 to 2016) [
4
]. However, this
large increase in registered PTWs increases the effect of the dangerous nature of these types
of vehicles [5].
Despite their many advantages (e.g., low cost, minimum space occupation, and fuel
efficiency) the risk of PTW riders and passengers suffering serious and fatal injuries is
much greater than that of passenger car occupants: in Australia, fatality and serious injury
rates per kilometer are 29 and 37 times higher for a PTW rider, respectively, than for other
Sustainability 2022,14, 7563. https://doi.org/10.3390/su14137563 https://www.mdpi.com/journal/sustainability
Sustainability 2022,14, 7563 2 of 20
vehicle types [
6
,
7
]. Furthermore, the comparison between the number of PTW riders killed
and the percentage of motorcycles in the fleet highlights the dangerous nature of PTWs:
in the state of Victoria, Australia, motorcyclists represented 12.0% of people killed in road
crashes in 2014, despite PTWs comprising only 3.9% of the registered vehicle fleet [
8
,
9
].
Similar results were found in the United States and Italy, where the accident statistics show
that PTW riders or passengers represent, respectively, 11.8% and 25.3% of people killed
on the roads, although motorcycles represent only 2.8% to 13.1% of the total [
10
,
11
]. New
developments in road safety are required to reverse this trend by reducing PTW rider risk.
The increasing prevalence and effectiveness of active safety systems, e.g., electronic
stability control [
12
], has greatly improved the safety of passenger cars. However, many
factors, such as the unstable nature of PTWs and the increased influence of external forces,
make the development of active safety systems for PTWs more challenging. The European
Transport Safety Council (ETSC) has noted that, while design and construction improve-
ments in cars have contributed to reductions in deaths, this has not been the case for
motorcycles [
13
]. This lack of development could explain why, while road death numbers
are generally declining, the deaths among motorcyclists have increased over the last decade
in the European Union [
14
]. Many studies have proven the potential effectiveness of PTW
active safety systems in avoiding PTW crashes or reducing their resulting injuries [
15
–
18
],
but, at the same time, the improvements and the implementations of these systems have
become increasingly difficult, both from a technical and economic point of view. To address
these challenges, in recent years it has become increasingly important to perform applica-
bility assessment studies. They make it possible to evaluate whether a safety system would
apply to a crash based on crash features such as the crash configuration: “The active safety
system X would have been applied in the accident scenario Y”. Therefore, applicability
studies represent the starting point of efficient effectiveness studies, determining the sys-
tems on which it would be most beneficial to concentrate the available economic resources
in terms of safety system research (effectiveness analysis that consider the benefits follow-
ing the systems’ activation—e.g., [
19
,
20
]) and development (market penetration facilities
and physical limits improvements).
Previous research has estimated the applicability of PTW active safety systems as a
function of crash scenarios in Australia [
21
]. However, how these results can be generalized
to other regions of the world is unknown. Several factors may result in regional differences
in countermeasure applicability, such as the primary use of PTWs, the composition of the
PTW fleet, the road system, and the traffic conditions. Each of these factors may directly
impact the applicability and priority of the proposed active safety systems. There is a need
for a multi-region applicability study to identify needs, opportunities, and priorities for
PTW active safety systems development worldwide.
This study aims to assess and compare the applicability of five promising PTW active
safety systems [
22
] in the Australian, Italian, and US PTW fleets, using real-world crash data
from each region. The systems considered are: antilock braking systems (ABS), motorcycle
autonomous emergency braking (MAEB), collision warning, curve warning, and curve
assist. The new proposed approach aims to determine which systems have the greatest
chance of reducing crashes and how differences between countries that influence system
applicability could lead to region-specific priorities for developing systems.
2. Materials and Methods
Country-specific applicability for each system, alone and in combination, was esti-
mated using three real-world crash databases containing crash data from Australia, Italy,
and the US. In addition, using the approach developed in an earlier Australian study [
21
], it
was possible to compare the results with those obtained from the analysis of the Australian
Road Crash Information System (RCIS) database. The new proposed approach is one of the
first methods to allow an applicability comparison among datasets originally characterized
using different standards.
Sustainability 2022,14, 7563 3 of 20
2.1. Settings: USA, Prato (IT), and Victoria (AUS)
Data for this study were taken from three different regions: the state of Victoria
(Australia), the Municipality of Prato (Italy), and the United States of America (USA). Some
characteristics of the three regions are shown in Table 1.
Table 1.
Comparison among the three regions considered: US, Victoria, and Prato, plus the addition
of the Italian data. In the upper section of the table, general data about the country and the fleets
are reported (N
◦
of vehicles and N
◦
of PTWs in the region, N
◦
of all crash typologies and N
◦
of
PTW crashes, fatalities in all road crashes, and fatalities in PTW crashes). In the central and lower
sections are reported the ratios and proportions used for a first comparison between the countries.
The Victorian” percentage of PTW crashes w.r.t. (with respect to) all crashes” has no value due to
the lack of data regarding all the crashes that occurred, i.e., only crashes with fatal and hospitalized
injured riders were counted in national reports.
Victoria (2014) Prato (2018) USA (2018)
Population 5,800,000 190,000 326,895,465
N◦vehicles 4,483,098 154,557 297,042,658
N◦PTWs 174,336 18,080 8,305,171
All crashes 5098 964 6,734,000
PTW crashes 958 294 82,124
Killed in all crashes 249 9 35,560
Killed in PTW crashes 30 0 4181
Ratio vehicles/population 0.77 0.81 0.91
% PTW in the fleets 3.9% 11.7% 2.8%
(PTW per 1000 vehicles) (38.9) (116.9) (27.9)
Mortality rate (killed in all crashes for 100,000 inhabitants)
4.29 4.73 10.87
% PTW crashes (w.r.t. all crashes) - 30.5% 1.2%
% Killed in PTW crashes (w.r.t. all fatal crashes) 12.0% 0.0% 11.8%
As of December 2014, the state of Victoria had an estimated population of 5.8 mil-
lion [
23
] and presented 4,483,098 registered vehicles (773 vehicles per 1000 persons) [
8
].
The operator must be at least 18 years of age to hold a motorcycle learner’s permit [
9
].
Likewise, as of December 2018, the Municipality of Prato had a population of 190,000 with
154,557 registered vehicles, i.e., 813 vehicles per 1000 persons [
24
]. To hold a moped or
motorcycle license, the operator must be at least 14 years old or 18 years old, respectively. In
2018, the United States had a population of 327 million, with a fleet of 297 million vehicles
(909 vehicles per 1000 persons) [
25
]. Each state has its own rules, but generally, a PTW
license can be obtained from 16 years of age in the US.
One major difference among the three regions is the higher prevalence of PTWs present
in Prato (and Italy (Appendix ATable A1)) compared to the US and Victoria (Table 1, middle
section), although the ratio between the number of registered vehicles and the population
is similar for all three regions (Victoria = 0.77; Prato = 0.81; US = 0.91). In 2018, 11.7% of
Prato’s fleet of vehicles were PTWs (18,080), and the percentages for the wider contexts of
Tuscany (the region that contains the Municipality of Prato) and Italy were similar: 15.8%
and 13.1%, respectively [
10
,
24
]. In contrast, in the state of Victoria (2011) and the US (2018),
only 3.9% and 2.8%, respectively, of registered vehicles were PTWs [
8
,
26
]. These differences
in PTW prevalence and use are likely to be greater than estimated based on registered
vehicle numbers, since for countries such as the US and Australia, motorcycles are more
often not the user’s primary motor vehicle for transport. For example, in Victoria, Australia,
it is estimated that less than 1% of vehicles actually in traffic are motorcycles [27].
The lower part of Table 1also highlights the high risk of PTW riders and passengers
suffering fatal crashes worldwide, with fatally injured PTW riders/passengers in the United
States (from 2016 to 2018) representing 11.8% (4181) of the total crash fatalities, despite
PTW crashes representing only 1.2% (82,124) of the total crashes [
25
]. The same dangerous
nature of PTWs is also found in Italy: even if, in 2018, only 14.2% (24,550) of total crashes
were PTW crashes, the PTW riders or passengers killed in crashes represented 25.3% (844)
Sustainability 2022,14, 7563 4 of 20
of the people killed in all crashes—see also Appendix ATable A1 [
10
]. Since the Australian
database contains only crashes with seriously injured people [
28
], to avoid possible bias,
the relative occurrence was not calculated (“% PTW crashes (w.r.t. all crashes)” of Table 1).
The statistical validity of PTWs’ higher risk of being involved in a crash and the riders’
higher risk of being killed was assessed using a two-proportion Z-test with p< 0.01. The
null hypothesis “there is no difference being involved in a crash between PTWs and other
vehicles” and “there is no difference being killed during a crash between PTWs and other
vehicle drivers” were both rejected. The greater risk of PTWs being involved in a crash
and of riders being killed was found to be statistically significant (p< 0.01) for all three
regions. It was not possible to reject the null hypothesis for the Prato region, as there were
no people killed in a PTW crash in 2018.
2.2. Data Sources: CRSS, Prato-X, and MICIMS Database
The active safety systems applicability assessment was carried out based on real-world
crashes contained in the Australian Managing Increasing Challenges in Motorcycle Safety
(MICIMS) database from Victoria, the new Prato-X database from Italy, and the US’ Crash
Reporting Sampling System (CRSS) database. Table 2lists the main characteristics of the
three datasets.
Table 2.
Comparison of the main characteristics of the three databases analyzed in the study: Prato-X,
CRSS, and MICIMS. The data presented are: extended database name; number of crashes extracted
from each dataset (for US, sampling weights were applied); crash location; crash period; road types;
injury types; and a special characteristic of the MICIMS dataset. The last row reports the original
dataset categorization method.
MICIMS Prato-X CRSS
Extended name
Managing Increasing
Challenges in
Motorcycle Safety
–Crash Reporting
Sampling System
Location Victoria, Australia Municipality of Prato United States of
America
Considered crashes
(with sampling weights)
235 294 6088
(265,361)
Period 01/2012–08/2014 2018 2016–2018
Type of roads Urban and rural Only urban Only urban
Included crashes Hospital admission
(non-fatal injury) Police reported Police reported
Particular
characteristic
Only crashes where
injured riders were
admitted to one of the
hospitals within the
study area
– –
The Australian dataset, MICIMS, contained 235 PTW crashes from 2012 to 2014 that
occurred on urban and rural roads in Victoria. MICIMS includes only crashes where
motorcycle riders were non-fatally injured and admitted to one of the hospitals within
the study area. A description of the characteristics of the MICIMS database can be found
in previous studies [
29
,
30
]. The Prato-X database comprises data provided by the Prato
Police for the 294 PTW crashes that occurred in the urban area of the Municipality of
Prato during 2018. For each crash, the police collected information regarding the crash
circumstances, the environment, the vehicles, and the people involved. The US Crash
Report Sampling System (CRSS) is a nationally representative sample of all police-reported
crashes in the US. Case weight values were assigned to each sample, following the national
estimates procedure [
25
], to estimate the national incidence of each crash type. The CRSS
comprised 6088 PTW cases from the years 2016 to 2018. To better compare with Prato-X
Sustainability 2022,14, 7563 5 of 20
(which contains exclusively urban crashes), only PTW crashes that occurred on urban roads
were extracted from the CRSS dataset. An estimated total of 265,361 US PTW crashes was
obtained after applying the CRSS sampling weights. The results obtained from the analysis
of these three datasets were also compared with the applicability results obtained from the
analysis of the crashes that occurred in the state of Victoria between 2001 and 2011 and
were contained in the Road Crash Information System (RCIS) [21].
2.3. Crash Classification
Each database codes crashes according to its own national standards and rules. This
section describes the method used to express all the crashes with the same crash configura-
tion variables.
MICIMS codes crash configurations according to the Australian Definition for Clas-
sifying Accidents (DCA) chart [
31
]. The Australian DCA chart (Appendix AFigure A1)
is divided into 10 categories, comprising 81 total crash configurations, each described
by a pictogram containing the trajectory of the vehicles involved in the crashes without
distinguishing between types of vehicles. Following Savino et al., it was necessary to extend
the crash configurations from 81 to 152, because the original Australian DCA code does
not specify the type of vehicle to which the trajectories correspond [
21
]. The expansion
was achieved by introducing a new variable, “Type”, that specifies all the possible combi-
nations of the position of motorcycles and other vehicles. Figure 1shows an example of
this extension process. Therefore, each of the MICIMS crashes was re-coded in one of the
152 possible extended DCA crash configurations.
Sustainability 2022, 14, 7563 5 of 22
Table 2. Comparison of the main characteristics of the three databases analyzed in the study: Prato-
X, CRSS, and MICIMS. The data presented are: extended database name; number of crashes ex-
tracted from each dataset (for US, sampling weights were applied); crash location; crash period;
road types; injury types; and a special characteristic of the MICIMS dataset. The last row reports the
original dataset categorization method.
MICIMS Prato-X CRSS
Extended name Managing Increasing Chal-
lenges in Motorcycle Safety -- Crash Reporting
Sampling System
Location Victoria, Australia
Municipality of
Prato
United States of
America
Considered crashes
(with sampling weights) 235 294
6088
(265,361)
Period 01/2012–08/2014 2018 2016–2018
Type of roads Urban and rural Only urban Only urban
Included crashes Hospital admission
(non-fatal injury) Police reported Police reported
Particular characteris-
tic
Only crashes where injured
riders were admitted to one
of the hospitals within the
study area
-- --
2.3. Crash Classification
Each database codes crashes according to its own national standards and rules. This
section describes the method used to express all the crashes with the same crash configu-
ration variables.
MICIMS codes crash configurations according to the Australian Definition for Clas-
sifying Accidents (DCA) chart [31]. The Australian DCA chart (Appendix A Figure A1) is
divided into 10 categories, comprising 81 total crash configurations, each described by a
pictogram containing the trajectory of the vehicles involved in the crashes without distin-
guishing between types of vehicles. Following Savino et al., it was necessary to extend the
crash configurations from 81 to 152, because the original Australian DCA code does not
specify the type of vehicle to which the trajectories correspond [21]. The expansion was
achieved by introducing a new variable, “Type”, that specifies all the possible combina-
tions of the position of motorcycles and other vehicles. Figure 1 shows an example of this
extension process. Therefore, each of the MICIMS crashes was re-coded in one of the 152
possible extended DCA crash configurations.
Prato-X codes crashes using the national rules released by the Istituto Nazionale di
Statistica (ISTAT), which employs short text descriptive phrases. Using the sanitized po-
lice crash information, each Prato-X crash was manually coded using the extended DCA
code, after first adjusting the codes for driving on the right side of the road. This new chart
(Appendix A Figure A2), called the Italian DCA, also contained 152 crash configurations,
which specify the vehicle positions and types.
Figure 1. Example of the extension process applied to DCA scenario n° 110 (left). It was divided into
four different crash scenarios (right) using the variable “Type”, which specifies all possible crash
combinations, i.e., if vehicle 1 and 2 are PTWs or other vehicles and all the possible combination of
crashes between them.
Figure 1.
Example of the extension process applied to DCA scenario n
◦
110 (left). It was divided into
four different crash scenarios (right) using the variable “Type”, which specifies all possible crash
combinations, i.e., if vehicle 1 and 2 are PTWs or other vehicles and all the possible combination of
crashes between them.
Prato-X codes crashes using the national rules released by the Istituto Nazionale di
Statistica (ISTAT), which employs short text descriptive phrases. Using the sanitized police
crash information, each Prato-X crash was manually coded using the extended DCA code,
after first adjusting the codes for driving on the right side of the road. This new chart
(Appendix AFigure A2), called the Italian DCA, also contained 152 crash configurations,
which specify the vehicle positions and types.
CRSS codes crash configurations based on an Accident Type classification scheme
(Appendix AFigure A3), which contains 64 crash configurations divided by typology [
11
].
The main difference between the Accident Type and the DCA tables is that CRSS’ Accident
Type table is based largely on the type of impact, while the DCA table is mainly based on
the intentions and maneuvers of the riders. To make comparison with the other datasets
possible, the Accident Type configuration chart used in CRSS was translated to the Italian
DCA configuration chart, as they both use right-hand driving. To better specify the cor-
respondence, additional CRSS variables [
11
] describing riders’ intentions and maneuvers
were included as needed (Appendix ATable A2 shows the correspondence between the
charts). Figure 2shows an example of the translation between Accident Type scenario n
◦
83
and DCA n
◦
113, as well as the CRSS variable “Initial Contact Point”, which was used to
make a distinction between DCA n◦113 type C or D.
Sustainability 2022,14, 7563 6 of 20
Sustainability 2022, 14, 7563 6 of 22
CRSS codes crash configurations based on an Accident Type classification scheme
(Appendix A Figure A3), which contains 64 crash configurations divided by typology [11].
The main difference between the Accident Type and the DCA tables is that CRSS’ Acci-
dent Type table is based largely on the type of impact, while the DCA table is mainly
based on the intentions and maneuvers of the riders. To make comparison with the other
datasets possible, the Accident Type configuration chart used in CRSS was translated to
the Italian DCA configuration chart, as they both use right-hand driving. To better specify
the correspondence, additional CRSS variables [11] describing riders’ intentions and ma-
neuvers were included as needed (Appendix A Table A2 shows the correspondence be-
tween the charts). Figure 2 shows an example of the translation between Accident Type
scenario n°83 and DCA n°113, as well as the CRSS variable “Initial Contact Point”, which
was used to make a distinction between DCA n° 113 type C or D.
In this case, Accident Type scenario 83 was found to be equivalent to:
- DCA scenario n°113 type C when the CRSS variable “Initial Contact Point” had a
value between 1 and 11 for the PTW and between 7 and 11 for the other vehicle.
- DCA scenario n°113 type D when the CRSS variable “Initial Contact Point” had a
value between 1 and 5 for the PTW and between 1 and 11 for the other vehicle.
Figure 2. Example of the translation process between AccType scenario n°83 and DCA n°113. The
picture on the right shows the CRSS variable “Initial Contact Point”, used to make a distinction
between the two possible correspondences of Accident Type scenario n° 83 with DCA scenario n°
113 Type C or Type D.
Figure 3 depicts the method proposed to express all the crashes with the same varia-
bles. MICIMS crashes were re-coded using the extended DCA chart containing 152 picto-
grams and obtained with the addition of the variable “Type”. Prato-X data were re-coded
using the same extended DCA chart, after adjustment for right-hand driving. Regarding
CRSS crashes, firstly, the exact correspondence between the Accident Type and the DCA
scenarios was found. Then, the crashes were re-coded using the DCA chart with right-
hand driving. However, Accident Type scenarios 1, 3, 6, 8, 34 to 41, and 54 to 61 did not
correspond to any DCA scenario and were analyzed separately.
Figure 2.
Example of the translation process between AccType scenario n
◦
83 and DCA n
◦
113. The
picture on the right shows the CRSS variable “Initial Contact Point”, used to make a distinction
between the two possible correspondences of Accident Type scenario n
◦
83 with DCA scenario n
◦
113
Type C or Type D.
In this case, Accident Type scenario 83 was found to be equivalent to:
−
DCA scenario n
◦
113 type C when the CRSS variable “Initial Contact Point” had a value
between 1 and 11 for the PTW and between 7 and 11 for the other vehicle.
−
DCA scenario n
◦
113 type D when the CRSS variable “Initial Contact Point” had a value
between 1 and 5 for the PTW and between 1 and 11 for the other vehicle.
Figure 3depicts the method proposed to express all the crashes with the same variables.
MICIMS crashes were re-coded using the extended DCA chart containing 152 pictograms
and obtained with the addition of the variable “Type”. Prato-X data were re-coded using
the same extended DCA chart, after adjustment for right-hand driving. Regarding CRSS
crashes, firstly, the exact correspondence between the Accident Type and the DCA scenarios
was found. Then, the crashes were re-coded using the DCA chart with right-hand driving.
However, Accident Type scenarios 1, 3, 6, 8, 34 to 41, and 54 to 61 did not correspond to
any DCA scenario and were analyzed separately.
Sustainability 2022, 14, 7563 7 of 22
Figure 3. Different re-coding methods used for the three datasets. MICIMS crashes, originally cata-
logued into 82 scenarios, were re-coded into 152 scenarios. Prato-X and CRSS, using different pro-
cedures, were both re-coded using the extended DCA adjusted for right-hand driving (Italian DCA).
2.4. Determining Device Applicability
To make comparison possible also with the results obtained for the Australian RCIS
crash database, the evaluation of safety system applicability was carried out using the
same method as Savino et al. [21], who developed a scoring scheme for each safety system
ranging from 1 (“system would have definitely not applied to crashes belonging to a given sce-
nario”) to 4 (“system would definitely have applied”). In this study, the authors revised the
definition of category 4 to improve clarity (“system would have applied if the technology was
activated”) to emphasize the potential relevance of the systems. A score value of 2 was
used in cases where the application was controversial, while a score value of 3 indicated
“Would probably have applied but before some technical challenge need to be solved”. The scheme
illustrated in Figure 4, regenerated from Table 1 in Savino et al. (2019), was based on a
detailed taxonomy of the functionality, purpose, and applicability of the five safety sys-
tems [21]: for each score, specific and detailed rules were developed based on the current
understanding of the device’s functionality.
The antilock braking system (ABS), already implemented in high-level PTWs, is use-
ful for reducing braking distance and preventing wheel lock during intense braking or
conditions of low adherence [32,33]. Motorcycle autonomous emergency braking (MAEB)
performs a braking maneuver when the rider has no time to brake. AEB has been imple-
mented in passenger cars but has not yet been implemented in PTWs, as it is still in the
prototype phase [19,20]. Collision avoidance technologies for PTWs are systems designed
to scan the environment and surroundings of the vehicle, issuing a warning to the rider
when a risk of a collision has been computed. A similar system was recently implemented
in production bikes (e.g., 2020 Ducati Multistrada) [34–36]. Curve warning is a prototype
system that monitors the real-time state of the PTW and issues a warning to the rider when
it identifies curve hazards [37,38]. Lastly, the prototype curve assistance system checks
the vehicle dynamics in real time and makes some adjustments (i.e., engine torque/brake)
when a loss of control is detected [17,39].
Two of the authors of [21] independently categorized each of the 152 extended DCA
crash scenarios into one of the four relevance categories using the rules reported in Figure
4 (adjusted from Table 1 of [21]). A third researcher independently assessed the
Figure 3.
Different re-coding methods used for the three datasets. MICIMS crashes, originally
catalogued into 82 scenarios, were re-coded into 152 scenarios. Prato-X and CRSS, using different pro-
cedures, were both re-coded using the extended DCA adjusted for right-hand driving (Italian DCA).
Sustainability 2022,14, 7563 7 of 20
2.4. Determining Device Applicability
To make comparison possible also with the results obtained for the Australian RCIS
crash database, the evaluation of safety system applicability was carried out using the
same method as Savino et al. [
21
], who developed a scoring scheme for each safety system
ranging from 1 (“system would have definitely not applied to crashes belonging to a given scenario”)
to 4 (“system would definitely have applied”). In this study, the authors revised the definition
of category 4 to improve clarity (“system would have applied if the technology was activated”)
to emphasize the potential relevance of the systems. A score value of 2 was used in cases
where the application was controversial, while a score value of 3 indicated “Would probably
have applied but before some technical challenge need to be solved”. The scheme illustrated in
Figure 4, regenerated from Table 1in Savino et al. (2019), was based on a detailed taxonomy
of the functionality, purpose, and applicability of the five safety systems [
21
]: for each
score, specific and detailed rules were developed based on the current understanding of
the device’s functionality.
Sustainability 2022, 14, 7563 8 of 22
classification agreement. The three researchers were academics with at least 10 years of
experience in the development and assessment of vehicle safety systems. This method was
characterized by: (a) a large set of configurations used for the definition of the crash sce-
narios; (b) a high degree of detail in the definition of the taxonomy; and (c) a limited num-
ber of researchers involved in the evaluation process.
As reported in the Appendix A of Savino et al. (2019), the quadratic weighted kappa
was used [40,41] as a measure of the inter-rater reliability statistic. The authors found that
the quadratic weighted kappa was statistically significant and therefore the level of agree-
ment was substantial.
Regarding the Accident Type scenarios with no correspondence to the DCA scenar-
ios (n° 1, 3, 6, 8, 34 to 41, and 54 to 61), two authors of the present work (P.T. and M.D.)
separately assigned the relevant category to the scenarios for each safety system based on
their functionality (Figure 4). After carrying out the assessment, two experts on road safety
(G.S. and H.C.G.) assessed the classifications to make a final decision for the scenarios
where there was no agreement.
Figure 4. Scoring scheme of relevance. Using this flowchart, each safety system was evaluated in
each scenario, obtaining a score from 1 to 4. Regenerated from Savino et al. (2019) [21].
Having assigned a relevance score to each scenario (Appendix A Table A3), the ap-
plicability of the safety systems to each real crash contained in the three datasets was de-
termined. Each safety system was considered independently and in combination.
3. Results
This research aimed to perform an applicability assessment of five PTW safety sys-
tems using real-world crash data from three different regional databases. Tables 3 and 4
show the percentage of crashes to which each safety system is relevant, alone and in com-
bination.
Figure 4.
Scoring scheme of relevance. Using this flowchart, each safety system was evaluated in
each scenario, obtaining a score from 1 to 4. Regenerated from Savino et al. (2019) [21].
The antilock braking system (ABS), already implemented in high-level PTWs, is use-
ful for reducing braking distance and preventing wheel lock during intense braking or
conditions of low adherence [
32
,
33
]. Motorcycle autonomous emergency braking (MAEB)
performs a braking maneuver when the rider has no time to brake. AEB has been imple-
mented in passenger cars but has not yet been implemented in PTWs, as it is still in the
prototype phase [
19
,
20
]. Collision avoidance technologies for PTWs are systems designed
to scan the environment and surroundings of the vehicle, issuing a warning to the rider
when a risk of a collision has been computed. A similar system was recently implemented
in production bikes (e.g., 2020 Ducati Multistrada) [
34
–
36
]. Curve warning is a prototype
system that monitors the real-time state of the PTW and issues a warning to the rider when
it identifies curve hazards [
37
,
38
]. Lastly, the prototype curve assistance system checks
the vehicle dynamics in real time and makes some adjustments (i.e., engine torque/brake)
when a loss of control is detected [17,39].
Two of the authors of [
21
] independently categorized each of the 152 extended DCA
crash scenarios into one of the four relevance categories using the rules reported in Figure 4
(adjusted from Table 1of [
21
]). A third researcher independently assessed the classification
Sustainability 2022,14, 7563 8 of 20
agreement. The three researchers were academics with at least 10 years of experience in the
development and assessment of vehicle safety systems. This method was characterized
by: (a) a large set of configurations used for the definition of the crash scenarios; (b) a high
degree of detail in the definition of the taxonomy; and (c) a limited number of researchers
involved in the evaluation process.
As reported in the Appendix Aof Savino et al. (2019), the quadratic weighted kappa
was used [
40
,
41
] as a measure of the inter-rater reliability statistic. The authors found
that the quadratic weighted kappa was statistically significant and therefore the level of
agreement was substantial.
Regarding the Accident Type scenarios with no correspondence to the DCA scenarios
(n
◦
1, 3, 6, 8, 34 to 41, and 54 to 61), two authors of the present work (P.T. and M.D.)
separately assigned the relevant category to the scenarios for each safety system based on
their functionality (Figure 4). After carrying out the assessment, two experts on road safety
(G.S. and H.C.G.) assessed the classifications to make a final decision for the scenarios
where there was no agreement.
Having assigned a relevance score to each scenario (Appendix ATable A3), the
applicability of the safety systems to each real crash contained in the three datasets was
determined. Each safety system was considered independently and in combination.
3. Results
This research aimed to perform an applicability assessment of five PTW safety systems
using real-world crash data from three different regional databases. Tables 3and 4show
the percentage of crashes to which each safety system is relevant, alone and in combination.
Table 3.
Applicability comparison between countries with percentages of crashes for each category
and safety system.
Category 1 Category 2 Category 3 Category 4
System would definitely not
have applied
System would possibly have
applied (the applicability is
controversial)
System would probably have
applied (technical challenge
still needs to be solved)
System would have applied if
the technology was activated
MICIMS
Prato-X CRSS
MICIMS
Prato-X CRSS
MICIMS
Prato-X CRSS
MICIMS
Prato-X CRSS
ABS 6.4% 7.7% 15.8% 40.3% 29.1% 28.0% 4.7% 3.2% 10.6% 48.7% 60.0% 45.6%
MAEB 44.1% 33.0% 32.7% 22.5% 28.1% 44.3% 26.3% 31.2% 8.8% 7.2% 7.7% 14.2%
Collision warning 36.9% 31.2% 30.1% 9.7% 7.0% 11.0% 30.5% 29.8% 33.1% 22.9% 31.9% 25.9%
Curve warning 66.1% 88.1% 84.2% 9.3% 7.4% 5.7% 0.0% 0.0% 0.0% 24.6% 4.6% 10.1%
Curve assist 54.2% 60.4% 70.4% 16.5% 32.6% 11.4% 4.2% 2.5% 7.9% 25.0% 4.6% 10.4%
Table 4. Applicability comparison of safety systems alone or in combination.
Category 4 Category 3 + 4
MICIMS Prato-X CRSS MICIMS Prato-X CRSS
Safety Systems Alone
ABS alone (A) 25.8% 28.1% 19.7% 0.8% 1.7% 0.8%
MAEB alone (B) – – – – – –
CollWarn alone (C) – – – 0.4% – 1.1%
CurveWarn alone (D) – – – – – –
CurveAssist alone (E) 0.4% – 0.3% 0.4% – 0.5%
Two Systems in Combination
A + C 15.7% 24.2% 11.6% 15.3% 20.4% 27.2%
D + E 24.6% 4.6% 10.1% 24.6% 4.6% 10.1%
C + E – – – 0.4% 0.4% 2.5%
Other double combination – – – – – –
Three Systems in Combination
A + B + C 7.2% 7.7% 14.2% 33.5% 38.9% 23.0%
A + C + E – – – 3.8% 2.1% 5.2%
Other triple combination – – – – – –
None apply 26.3% 35.4% 44.0% 20.8% 31.9% 29.7%
Sustainability 2022,14, 7563 9 of 20
3.1. Independent Systems Analysis
ABS was potentially relevant (category 2 + 3 + 4) for 93.7% of crashes contained in
MICIMS, 92.3% in Prato-X, and 84.2% in CRSS. MAEB, on the other hand, “would have
applied if the technology was activated” (category 4) for 7.2% of MICIMS crashes, 7.7% of Prato-
X, and 14.2% of CRSS. Collision warning systems were considered potentially relevant
(category 2 + 3 + 4) for 63.1% of crashes in MICIMS, 68.8% in Prato-X, and 69.9% in CRSS.
On the contrary, curve warning was considered definitely not relevant (category 1) for
66.1% of MICIMS crashes, 88.1% of Prato-X, and 84.2% of CRSS. Similarly, curve assist was
definitely not relevant for 54.2% of MICIMS, 60.4% of Prato-X, and 70.4% of CRSS crashes.
The safety system with the highest percentage of crashes classified in category 4 was
ABS (MICIMS = 48.7%, Prato-X = 60.0%, CRSS = 45.6%), followed by collision warning
(MICIMS = 22.9%, Prato-X = 31.9%, CRSS = 25.9%). Categories 2 and 3 were found to be
particularly important for MAEB, as they included 48.8% of MICIMS, 59.3% of Prato-X,
and 53.1% of CRSS crashes.
To assess if the differences in the percentages between countries were statistically
significant, a chi-square test with p< 0.05 was performed. Except for “curve warning”
category 3 (because of its null result), the null hypothesis “there is not a statistical difference
between the percentages of applicability for the safety system in the countries” was rejected.
3.2. Combined Systems Analysis
For each dataset, Table 4presents the percentage of crashes that were sensitive to a
single system at a time (i.e., crashes that were not sensitive to more than one system) or
to a set of multiple systems (i.e., crashes in which more than one system may have been
applied). Focusing on the category 4 rating (high relevance), ABS alone obtained maximum
applicability compared to the other systems: the percentage of crashes sensitive to only
ABS (category 4) was between 19.7% and 28.1%. When considering ABS and collision
warning in combination (A + C), i.e., both systems may have been applied, the percentage
of sensitive crashes was between 11.6% and 24.2%.
Table 4also shows the percentage of crashes where no safety system was found
applicable (“None Apply”). None of the five selected safety systems obtained a score of
4 for 26.3% of crashes in MICIMS, 35.4% in Prato-X, and 44.0% in CRSS. Furthermore,
considering category 3, the crash percentages were reduced by 4–6% in the case of Prato-X
and MICIMS, and by almost 15% in the case of the CRSS dataset.
4. Discussion
The main challenge of the comparison was that, initially, crashes were coded differently
from country to country: MICIMS crashes used the DCA scenarios with left-hand traffic,
Prato’s crashes were cataloged using the ISTAT standard, and the CRSS database used the
Accident Type classification. Despite this, our thorough re-coding methodology yielded
comprehensive results showing the variability of the applicability of the five safety systems
for the three countries considered.
4.1. ABS
ABS showed the largest independent applicability: its category 4 “Would have applied if
the technology was activated” contained 48.7% of MICIMS crashes, 60.0% of Prato-X crashes,
and 45.6% of CRSS crashes. Savino et al., using the same method but considering both
urban and rural Australian crashes in the RCIS dataset, obtained an ABS applicability of
40.6% [
21
]. Rizzi et al., using different methods, obtained for Sweden an ABS relevance of
38% for all crashes and 48% for severe or fatal crashes [16].
Difference between Countries—ABS Category 4
ABS maximum relevance for Prato-X (60.0%) was quite different from both the litera-
ture and the other datasets analyzed in this paper: more than 10 percentage points greater
than MICIMS and CRSS, and 20 percentage points greater than the RCIS database studied
Sustainability 2022,14, 7563 10 of 20
by Savino et al. [
21
]. The reduced number of crashes contained in Prato-X, compared
to CRSS and RCIS, may have influenced this result. However, this does not explain the
10-percentage-point difference between the Prato-X and MICIMS databases, which contain
a comparable number of crashes: 294 and 235, respectively. In this case, the difference in the
results was likely influenced by differences in the features of the two countries’ transporta-
tion systems, such as the different road systems and the high number of mopeds in Prato’s
fleet. MICIMS contains crashes that occurred within 150 km of the city of Melbourne,
the capital of the state of Victoria and the second-most populous city in Australia [
30
].
Urban areas of the city contain large arterial roads, extra-urban roads, highways, and
freeways, while non-urban areas are higher-speed zones with access to recreational routes
with curved roads. Motorcycles also tend to be larger with relatively large engine capacities.
Prato-X, on the other hand, contains crashes that occurred in the Municipality of Prato, a
typical small/medium-sized Italian city, characterized by narrow streets, no highways, and
high traffic on the roads. Due to these factors, 74% of Prato’s crashes occurred from contact
with other vehicles or with pedestrians/bikes (only 4.6% of Prato-X crashes were detected
while the PTW was cornering or in high-speed zones): these are scenarios where braking
can be useful, and therefore, where ABS scored higher. On the contrary, 38.6% of MICIMS
crashes occurred while cornering or in high-speed zones, scenarios where ABS has lower
applicability. In addition, the high presence of mopeds—PTWs with an engine capacity
equal to or less than 50 cc—in Prato’s fleet could have influenced the high applicability of
ABS: they represented 36% of the PTWs involved in Prato’s crashes and, after an analysis
of the Prato Police data, none were found to be equipped with ABS. The lack of vehicles
equipped with this system could have increased its applicability: if it had been more
widespread, it is possible that many crashes would have been mitigated or avoided.
The greater diffusion of ABS in the future will definitely bring the ABS category 4 of
Prato-X to a value closer to that of other countries. This improvement could be achieved by
reducing the cost of ABS, facilitating its implementation on mopeds, and by increasing the
presence of high-level PTW types.
4.2. MAEB and Collision Warning
MAEB obtained the maximum applicability (category 4) in 7.2% of MICIMS, 7.7%
of Prato-X, 14.2% of CRSS, and 5.7% of RCIS [
21
] crashes. Meanwhile, almost half of the
crashes contained in the datasets belonged to controversial categories where the MAEB
applicability was uncertain (category 2 + 3): 48.8% for MICIMS, 59.3% for Prato-X, 53.1%
for CRSS crashes, and 41.6% for the RCIS dataset analyzed by Savino et. al. [
21
]. These
results are aligned with studies that performed detailed crash reconstruction [
20
]. Typical
scenarios where MAEB obtained a rating of 2 or 3 were crashes where the PTW was
initially partially adjacent to the car (e.g., sideswipe crashes): at present, instruments to
detect obstacles located laterally/frontally are reaching a high technology readiness level
(TRL) and approaching a market release [
36
,
42
,
43
]. Such systems, once validated in field
operational tests with motorcycles, could shift many of these crashes into category 4 of the
MAEB. Such developments could also influence category 3 of collision warning, shifting
many of these crashes into category 4.
Difference between Countries—MAEB Category 3
Substantial differences between regions were found for MAEB category 3: 26.3% for
MICIMS, 31.2% for Prato-X, 8.8% for CRSS, and 17.3% for RCIS crashes [
21
]. While the
MICIMS result is discussed later in the paper, the difference between Prato-X and CRSS
was likely influenced by the variations in the number of sideswipe crashes between the
datasets (MAEB has score 3 in sideswipe scenarios): making up 10.2% in Prato-X but only
2.9% of the total crashes in CRSS. These different percentages reflect the different traffic
conditions in the two countries: in Italy, due to the layout of the roads and the typical traffic
conditions, it is usual for PTWs to overtake other vehicles sideways when they are in a line.
This maneuver is the reason for the high number of sideswipes in Prato-X crashes: cars in a
Sustainability 2022,14, 7563 11 of 20
row often turn to enter into secondary roads or private properties and hit PTWs intent on
overtaking. This fact could have influenced the MAEB category 3 results for the Italian city.
On the contrary, the high percentage of crashes contained in MAEB category 3 of
MICIMS was caused not by transportation features, but by one particular characteristic of
this database: it contains only crashes where the rider/passenger was injured and admitted
to one of the hospitals within the study area (Table 2). As a result, in MICIMS, there is
a prevalence of scenarios where the rider has a high probability of being injured. For
example, scenarios DCA 121 type C and 113 type C (where the other vehicle (Vehicle 1)
cuts in front of the PTW (Vehicle 2)) have a high probability of the rider being injured, and
in fact represented 16.5% of the total crashes, while in the Prato-X and CRSS datasets, they
represented only 6.7% and 0.6% of the total, respectively.
4.3. ABS Plus MAEB Plus Collision Warning
ABS and collision warning are non-automatic safety systems, as they require the rider
to take action for any safety benefits to be achieved. For this reason, riders may benefit
more from the combined use of ABS and/or collision warning with MAEB. The advantages
of these systems can add up in many situations, such as in a case where a rider may not
have braked due to not recognizing the risk of a crash. In a situation like this, a collision
warning system could trigger braking intervention by the rider, increasing the effectiveness
of ABS. Additionally, MAEB could perform braking for riders if they do not brake despite
the collision warning. When the DCA vehicle movement was classified as category 4, the
combination of ABS and collision warning (A + C) covered a large number of crashes
(Table 4): 15.7% in MICIMS, 24.2% in Prato-X, 11.6% in CRSS, and 17.4% in RCIS [
21
]. Good
coverage was also provided by the combination of ABS, collision warning, and MAEB
(A + B + C): 7.2% in MICIMS, 7.7% in Prato-X, 14.2% in CRSS, and 5.7% in RCIS [21].
4.4. Curve Warning and Curve Assist
When considering only category 4 scenarios, both curve warning and curve assist
applied to only 4.6% of Prato-X crashes. This result differs from those of Savino et al. [
21
]
and Biral et al. [
44
], where the systems were each found to be applicable to approximately
16% of crashes. For the CRSS dataset, the percentage was almost double (CW = 10.1%;
CA = 10.4%), and it was even higher for MICIMS (CW = 24.6%; CA = 25.0%).
Differences between Countries—Curve Warning and Curve Assist Category 4
The higher MICIMS percentage was obtained because 58 out of 235 crashes (24.6%)
were DCA scenarios involving a vehicle being run off the road on a curve, (scenarios
between 180 and 184, “Off path on curve”; Appendix AFigure A1). In contrast, due to
the narrow urban streets and the low speed of the PTWs (comprising mostly mopeds),
Prato-X crashes that occurred while navigating a turn represented only 4.6% of the total
crashes (scenarios 180 to 184). It is interesting to note that the two countries that exhibited
the highest applicability for curve warning and curve assist systems, the US and Australia,
are the countries in this study known for recreational PTW use. This may indicate a link
between recreational PTW use and the loss of control at higher speeds while navigating
turns, which promotes the implementation of these systems especially when motorcycles
are used recreationally.
4.5. Remaining Crashes
After considering the coverage of category 3 and 4 scenarios, for either an independent
system or a combination of systems, it was possible to determine the percentage of crashes
where no safety system was considered applicable. The raw “None Apply” value in the
right-side of Table 4represents the number of crash scenarios where all five safety systems
obtained scores lower than 3. The crashes in this group for Prato-X and MICIMS largely
comprised loss-of-control scenarios (“Off path on straight” and “Off path on curve”, DCA
columns 8 and 9; Appendix AFigure A1). For CRSS, nearly half of the crashes in this group
Sustainability 2022,14, 7563 12 of 20
consisted of PTWs that were rear-ended, run-off-road cases (not due to traction loss), and
cases where the PTW was sideswiped.
The large percentages of cases for which none of these systems would apply emphasize
the need to develop new solutions, especially in cases where the PTW is sideswiped or
experiences a loss of control.
5. Limitations
One limitation of this study was that the systems’ applicability was based only on
the crash configuration, which was a typology of data available in all the datasets (even if
originally coded differently) but that did not provide specific information about the crashes.
This influenced the applicability of the non-automatic safety systems—i.e., that require the
rider to take action to achieve any safety benefits—such as ABS or collision warning. A
more confident assessment of applicability would be possible using information regarding
the riders’ activation of the brakes or avoidance maneuver (for collision warning), and
the locking of the wheels (for ABS); however, this type of data was not available for CRSS
and Prato-X. This could have led to an overestimation of the applicability of these two
systems. Future studies that analyze riders’ actions (e.g., naturalistic data) are encouraged
to validate these results.
An additional limitation was that we did not consider rider behavior factors (ability to
react, medical or other impairment such as fatigue or alcohol), environmental conditions
(i.e., weather and lighting), the physical limitations of the systems, and the riders’ ability to
disable the system, which are variables that could influence a system’s applicability. Future
effectiveness studies should consider these factors, as well as the presence of passenger
vehicle technologies that might prevent crashes involving a motorcyclist [
45
], given that a
relatively high proportion of these crashes involve errors by other drivers [29].
Lastly, even if it is proven that crashes within these countries have different con-
tributing factors and that therefore the benefits of these technologies would not be evenly
distributed, it is possible that, even within a single country, different areas could have
different tendencies. Other than incorporating new crash datasets, future research should
statistically relate the benefits of the safety systems to the tendencies of the crashes’ con-
tributing factors.
6. Conclusions
This research was performed to assess and compare the applicability of five up-
and-coming active safety systems in three different countries. Furthermore, the new
proposed approach shows that there is space and need for PTW-based active safety systems,
highlighting which system has the greatest chance of reducing crashes.
Aiming to achieve maximum coverage with the minimum number of systems, ABS
and collision warning have the greatest chance of reducing crashes, as they were found to
be applicable to half and one-quarter of the crashes of each database, respectively. Although
these results are consistent with the literature, the applicability obtained by using a scenario-
based approach could be improved in future studies that consider more in-depth crash
datasets. Curve warning and curve assist are the only systems considered useful during
the navigation of a curve. Even if the results for these systems varied widely between
countries, their percentages were the same within individual countries. As Savino et al.
determined from the RCIS dataset, these results suggest that all riders who could benefit
from curve assist would also benefit from curve warning. Considering three systems at a
time, riders may benefit from the combined use of ABS and collision warning with MAEB
in one-third of crashes (category 3 + 4).
Furthermore, this study proved that the device relevance for each of these systems de-
pends largely on the geographical location of interest. The countries analyzed in this paper
differ according to their roadway structures (narrow roads versus wider highways); rules
of the road (left-hand versus right-hand traffic); how PTWs are generally used (recreational
use versus practical use, such as for commuting); and PTW interactions with passenger
Sustainability 2022,14, 7563 13 of 20
vehicles (US fleets contain larger vehicles and more pickup trucks, while the Italian fleet
contains a consistent percentage of PTWs, including mopeds). Despite these differences,
the proposed applicability assessment of PTW active safety systems presented in this study
shows the need to prioritize the development of such systems. This has important impli-
cations for both researchers and manufacturers seeking to prioritize the development of
active safety countermeasures for a particular PTW fleet. In addition, because active safety
countermeasures may differ by motorcycle type, these regional differences suggest that
regulators may need to consider country-specific minimum performance standards.
Author Contributions:
Conceptualization, H.C.G. and G.S.; methodology, H.C.G., G.S., P.T. and
M.E.D.; software and data analysis, P.T., S.P., M.E.D., T.J.A. and C.L.; formal analysis and investigation,
G.S., M.E.D. and P.T.; resources, G.S., H.C.G., S.P. and T.J.A.; data curation and maintenance, T.J.A.
and S.P.; writing—original draft preparation, P.T., M.E.D. and G.S.; writing—review and editing, all;
supervision, G.S. and H.C.G.; funding acquisition, T.J.A., H.C.G. and G.S. All authors have read and
agreed to the published version of the manuscript.
Funding:
The MICIMS study was funded by the Australian Research Council (LP110100057), Vi-
cRoads, the Transport Accident Commission ofVictoria, and the Victorian Government Department of
Justice, with generous support from Victoria Police, the Victorian Automobile Chamber of Commerce,
and Ambulance Victoria.
Institutional Review Board Statement: Not applicable.
Data Availability Statement:
Restrictions apply to the availability of these data. Prato-X data are
available to the MOVING research group members, and MICIMS data to the MICIMS research
team members.
Acknowledgments:
The authors would like to thank the Municipality Police of Prato, Italy, for
providing the processed data. Thanks also to the MOVING research group of the University of
Florence, the Centre for Injury Biomechanics of Virginia Tech, and the MICIMS research team. In
addition, the authors would like to thank Miguel A. Perez and Luke E. Riexinger for their help and
support in the last part of the research. Special thanks to Hampton C. Gabler for the fruitful and
important supervision and contribution.
Conflicts of Interest: The authors declare no conflict of interest.
Appendix A
Table A1.
National crash statistics of Australia and Italy [
10
,
46
]. The Australian “percentage of PTW
crashes w.r.t. (with respect to) all crashes” has no value due to the lack of data regarding all the
crashes that occurred, i.e., only crashes with fatal and hospitalized injured riders were counted in
national reports.
Australia (2014) Italy (2018)
Population 23,490,700 60,480,000
N◦vehicles 17,633,493 51,682,370
N◦PTWs 780,174 6,780,733
All crashes 34,091 172,553
PTW crashes 7734 24,550
Killed in all crashes 1156 3334
Killed in PTW crashes 192 844
Ratio vehicles/population 0.75 0.85
% PTW in the fleets 4.4% 13.1%
(PTW per 1000 vehicles) (44.2) (131.2)
Mortality rate (killed in all crashes for 100,000 inhabitants) 4.92 5.51
% PTW crashes (w.r.t. all crashes) - 14.2%
% Killed in PTW crashes (w.r.t. all fatal crashes) 16.6% 25.3%
Sustainability 2022,14, 7563 14 of 20
Table A2.
Part of the translation of the Accident Type scenarios into DCA scenarios. In many cases,
additional CRSS variables were used to better characterize the scenarios and express the correspondence.
DCA + Type (ITA and AUS) AccType (USA) Use of Additional Variables
for Translation
100/101/102/103/104/105/167
13 No
106/107/108/109/169 15/16 No
110A 86 No
110B 89 No
110C 88 No
110D 87 No
112A/B 81 No
112C 80 Yes
112D 80 Yes
113A/B 82 No
113C 83 Yes
113D 83 Yes
114A 74 + 75 Yes
114B 74 + 75 Yes
114C 74 + 75 Yes
114D 74 + 75 Yes
115/117AB/CD 84 + 85 Yes
115/117CD/AB 84 + 85 Yes
116A 78 Yes
116B 78 Yes
116C 79 Yes
116D 79 Yes
118ABCD 84 + 85 Yes
119 84 + 85 + 90 + 91 Yes
120 50 + 51 Yes
121A/B 68 No
121C 69 Yes
121D 69 Yes
129 52 + 53 + 62 + 63 + 66 + 67 Yes
130/131/132A 20 + 24 + 28 Yes
130B 21 + 25 + 29 Yes
131B 22 + 26 + 30 Yes
132B 23 + 27 + 31 Yes
133A 44 Yes
133B 44 Yes
133C 45 Yes
133D 45 Yes
134A 46 No
135A 47 No
136A 70 Yes
136B 71 Yes
137A 72 No
137B 73 No
139 32 + 33 + 42 + 43 + 48 + 49 No
140A 76 Yes
140B 76 Yes
140C 77 Yes
140D 77 Yes
141 11 + 12 Yes
144/146 92 No
145A 21 + 25 + 29 Yes
145B 20 + 24 + 28 Yes
149 98 + 99 Yes
150A 50 + 51 + 64 Yes
150B 50 + 51 + 65 Yes
159 98 + 99 Yes
Sustainability 2022,14, 7563 15 of 20
Table A2. Cont.
DCA + Type (ITA and AUS) AccType (USA) Use of Additional Variables
for Translation
160/161/162A 11 No
163 12 Yes
164 12 Yes
165 12 Yes
166 12 Yes
171 7 Yes
173 2 Yes
175 14 No
179 04 + 05 + 09 + 10 Yes
180/182/184 02 + 07 Yes
181/183 02 + 07 Yes
189 04 + 05 + 09 + 10 Yes
198 98 Yes
199 99 Yes
Table A3.
DCA scenarios that obtained a categorization value of 3 (system would probably have
applied) or 4 (system would have applied) for the safety systems analyzed. DCA scenarios not
included in this table did not obtain a relevance level of 3 or 4 for any safety system.
Code (DCA) ABS MAEB CollWarn CurveWarn CurveAss
100–105 4 – 4 – –
106 – – 3 – 3
107 4 – 3 – –
108 4 3 4 – –
110 4 3 (PTW into OV) 4 – –
111 4 (1 PTW), 3 (2 PTW into 1 OV) 3 (1 PTW into 2 OV) 3 – 3 (2 PTW)
112 4 (1 PTW) – 3 – 3 (2 PTW)
113 3 (1 PTW, 1 OV into 2 PTW),
4 (2 PTW into 1 OV) 3 (2 PTW into 1 OV) 3 – 3 (1 PTW)
114 3 (2 PTW) – 3 – 3
115 3 – 3 – 3
116 3 (1 PTW), 4 (2 PTW) 3 (2 PTW into 1 OV) 3 – 3 (1 PTW)
117–118 3 – 3 – 3
120 4 – 4 – –
121 3 (1 PTW), 4 (2 PTW) 3 (2 PTW into 1 OV) 3 – 3 (1 PTW)
122 4 (1 PTW), 3 (2 PTW) – 3 – 3 (2 PTW)
123–125 3 – 3 – 3
130–132 4 (1 PTW) 4 (1 PTW) 4 (1 PTW) – –
133 4 (1 PTW) – – – –
134–135 3 (1 PTW) – 3 (1 PTW) – 3 (1 PTW)
136–137 4 (2 PTW) 3 (2 PTW) 3 (2 PTW) – 3 (1 PTW)
140 4 – 3 – 3 (1 PTW)
141 – – – – 3
142–143 4 (2 PTW) 3 (2 PTW) 4 (2 PTW) – –
145 4 (2 PTW) 4 (2 PTW) 4 (2 PTW) – –
147 4
3 (1 PTW into 2 OV, 2 PTW)
4 – –
148 4
3 (1 PTW into 2 OV, 2 PTW)
3 – –
150 4 – 3 – –
152 – – 3 (2 PTW) – –
153 3 (1 PTW), 4 (2 PTW) – 3 (1 PTW) – –
154 – – 4 (1 PTW) – –
160–162 4 (1 PTW) 4 (1 PTW) 4 (1 PTW) – –
163 4 – 4 – –
164 4 3 4 – –
165 4 4 4 – –
166–167 4 – 4 – –
175 4 – 4 – –
180–184 – – – 4 4
189 – – – – 4
192 4 4 4 – –
193 4 – 4 – –
Sustainability 2022,14, 7563 16 of 20
Sustainability 2022, 14, 7563 18 of 22
Figure A1. Definitions for Classifying Accidents (DCA) chart: Australian DCA chart (left-hand traf-
fic) used for MICIMS categorization.
Figure A1.
Definitions for Classifying Accidents (DCA) chart: Australian DCA chart (left-hand traffic)
used for MICIMS categorization.
Sustainability 2022,14, 7563 17 of 20
Sustainability 2022, 14, 7563 19 of 22
Figure A2. Italian DCA chart (right-hand drive) used for Prato-X and CRSS categorization.
Figure A2. Italian DCA chart (right-hand drive) used for Prato-X and CRSS categorization.
Sustainability 2022,14, 7563 18 of 20
Sustainability 2022, 14, 7563 20 of 22
Figure A3. CRSS US “Accident Type” chart, used in the original CRSS categorization.
References
1. Australian Bureau of Statistics. Survey of Motor Vehicle Use, Australia 2008. Months Ended 31 October 2007. Available online:
http://www.abs.gov.au/AUSSTATS/abs@.nsf/Lookup/9208.0Main+Features112 (accessed on 19 April 2022).
2. Australian Bureau of Statistics. Motor Vehicle Census, Australia. 2013. Available online:
https://www.abs.gov.au/AUSSTATS/abs@.nsf/Lookup/9309.0Main+Features131%20Jan%202013?OpenDocument= (accessed
on 19 April 2022).
3. Swedish Transport Administration. Increased Safety for Motorcycle and Moped Riders. Joint Strategy Version 2.0 for the Years 2010–
2020; Swedish Transport Administration: Borlänge, Sweden, 2020.
Figure A3. CRSS US “Accident Type” chart, used in the original CRSS categorization.
References
1.
Australian Bureau of Statistics. Survey of Motor Vehicle Use, Australia 2008. Months Ended 31 October 2007. Available online:
http://www.abs.gov.au/AUSSTATS/abs@.nsf/Lookup/9208.0Main+Features112 (accessed on 19 April 2022).
2.
Australian Bureau of Statistics. Motor Vehicle Census, Australia. 2013. Available online: https://www.abs.gov.au/AUSSTATS/
abs@.nsf/Lookup/9309.0Main+Features131%20Jan%202013?OpenDocument= (accessed on 19 April 2022).
3.
Swedish Transport Administration. Increased Safety for Motorcycle and Moped Riders. Joint Strategy Version 2.0 for the Years 2010–2020;
Swedish Transport Administration: Borlänge, Sweden, 2020.
Sustainability 2022,14, 7563 19 of 20
4.
The Motorcycle Industry in Europe, “Brexit—ACEM—The Motorcycle Industry in Europe”. 2018. Available online: https:
//www.acem.eu/policy- areas/brexit (accessed on 19 April 2022).
5. World Health Organization. Global Status Report on Road Safety; World Health Organization: Geneva, Switzerland, 2018.
6.
Department of Infrastructure, Transport, Regional Development and Local Government: Road deaths Australia 2008 statistical
summary. Road Safety Report No. 4. Canberra: The Department. 2009. Available online: https://www.google.com/url?sa=t&
rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwiD3dz36qj4AhWdpXIEHcQbBcUQFnoECAsQAQ&url=https%3A%2F%
2Fwww.bitre.gov.au%2Fsites%2Fdefault%2Ffiles%2FRDA_Summary_2008.pdf&usg=AOvVaw1W5PxGiGHROPA-LkOzp_go
(accessed on 19 April 2022).
7.
Henley, G.; Harrison, J. Serious Injury due to Land Transport Accidents—Serious Injury Due to Land Transport Accidents.
2008. Available online: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwj_r_vJ6qj4
AhWOnHIEHaG2D0wQFnoECAoQAQ&url=https%3A%2F%2Fwww.aihw.gov.au%2Fgetmedia%2F7c390016-ebd1-4ca6-9813-
254e85d99441%2Finjcat-129-11033.pdf.aspx%3Finline%3Dtrue&usg=AOvVaw1PO3_GpRo4TJEpGtxMOxlv (accessed on 19 April
2022).
8.
Australian Bureau of Statistics. Survey of Motor Vehicle Use, Australia. 2014. Available online: https://www.abs.gov.au/
AUSSTATS/abs@.nsf/Lookup/9309.0Main+Features131%20Jan%202014?OpenDocument (accessed on 19 April 2022).
9. Vic Roads. Motorcycle Licence and Learner Permit. In Melbourne, Australia—State Government of Victoria; 2014. Available online:
https://www.vicroads.vic.gov.au/licences/licence-and-permit-types/motorcycle-licence-and-learner-permit (accessed on 19
April 2022).
10.
Istituto Nazionale di Statistica. Incidenti Stradali. 2019. Available online: https://www.istat.it/it/files//2019/07/Incidenti_
stradali_2018_aggiornamento_Ottobre2019.pdf (accessed on 19 April 2022).
11.
National Highway Traffic Safety Administration. Crash Report Sampling System (CRSS) Analytical User’s Manual 2020. Available
online: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/813236 (accessed on 19 April 2022).
12.
Lie, A.K.A.; Tingvall, C.; Krafft, M. The effectiveness of ESP (Electronic Stability Program) in reducing real life accidents. Traffic
Inj. Prev. 2004,5, 37–41. [CrossRef] [PubMed]
13.
European Transport Safety Council. Reducing Motorcyclist Deaths in Europe. 2007. Available online: https://www.google.
com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwiCjpnT8aj4AhVRoHIEHao-A3MQFnoECAkQAQ&url=
https%3A%2F%2Farchive.etsc.eu%2Fdocuments%2Fcopy_of_copy_of_copy_of_PIN%2520Flash%25207%2520Reducing%2520
motorcyclist%2520deaths%2520in%2520Europe.pdf&usg=AOvVaw0dHDj8C7KZRDl0fUXitRN2 (accessed on 19 April 2022).
14.
Yannis, G.; Antoniou, C.; Evgenikos, P.; Papantoniou, P.; Kirk, A. Characteristics and Causes of Power Two Wheeler Accidents in
Europe. Procedia—Soc. Behav. Sci. 2012,48, 1535–1544. [CrossRef]
15.
Sporner, A.; Kramlich, T. Motorcycle braking and its influence on severity of injury. In Gesamtverband d. Dt. Versicherungswirtschaft
e. V., GDV, Inst. f. Fahrzeugsicherheit; 2001; pp. 1–7. Available online: https://www-nrd.nhtsa.dot.gov/pdf/esv/esv17/Proceed/
00168.pdf (accessed on 19 April 2022).
16.
Rizzi, M.; Strandroth, J.; Tingvall, C. The effectiveness of antilock brake systems on motorcycles in reducing real-life crashes and
injuries. Traffic Inj. Prev. 2009,10, 479–487. [CrossRef]
17.
de Filippi, P.; Tanelli, M.; Corno, M.; Savaresi, S.M.; Santucci, M.D. Electronic stability control for powered two-wheelers. IEEE
Trans. Control Syst. Technol. 2014,22, 265–272. [CrossRef]
18.
Matschl, G.; Mörbe, M.; Gröger, C. Motorcycle Stability Control—MSC The next step into safety solutions for motorcycles. In
Proceedings of the 10th International Motorcycle Conference, Köln, Germany, 29–30 September 2014; pp. 128–154.
19.
Savino, G.; Mackenzie, J.; Allen, T.; Baldock, M.; Brown, J.; Fitzharris, M. A robust estimation of the effects of motorcycle
autonomous emergency braking (MAEB) based on in-depth crashes in Australia. Traffic Inj. Prev. 2016,17, 66–72. [CrossRef]
20.
Lucci, C.; Allen, T.; Pierini, M.; Savino, G. Motorcycle Autonomous Emergency Braking (MAEB) employed as enhanced braking:
Estimating the potential for injury reduction using real-world crash modeling. Traffic Inj. Prev. 2021, 1–7. [CrossRef]
21.
Savino, G.; Pierini, M.; Fitzharris, M. Motorcycle active safety systems: Assessment of the function and applicability using a
population-based crash data set. Traffic Inj. Prev. 2019,20, 406–412. [CrossRef]
22.
Gil, G.; Savino, G.; Piantini, S.; Baldanzini, N.; Happee, R.; Pierini, M. Are automatic systems the future of motorcycle safety?
A novel methodology to prioritize potential safety solutions based on their projected effectiveness. Traffic Inj. Prev.
2017
,
18, 877–885. [CrossRef]
23.
Australian Bureau of Statistics. Australian Demographic Statistic. 2014. Available online: https://www.abs.gov.au/AUSSTATS/
abs@.nsf/Lookup/3101.0Main+Features1Jun%202014 (accessed on 19 April 2022).
24.
Istituto Nazionale di Statistica. Incidenti Stradali in Toscana. 2020. Available online: https://www.istat.it/it/files//2019/11
/Focus-Toscana-2018.pdf (accessed on 19 April 2022).
25.
National Highway Traffic Safety Administration. Traffic Safety Facts 2018 Data: Summary of Motor Vehicle Crashes. November
2020; pp. 1–10. Available online: https://www.nhtsa.gov/risky-driving (accessed on 19 April 2022).
26.
Teoh, E.R. Motorcycles registered in the United States, 2002–2021. Insur. Inst. Highw. Saf.
2021
. Available online: https:
//www.iihs.org/api/datastoredocument/bibliography/2225 (accessed on 19 April 2022).
27.
Allen, T.; McClure, R.; Newstead, S.V.; Lenné, M.G.; Hillard, P.; Symmons, M.; Day, L. Exposure factors of Victoria’s active
motorcycle fleet related to serious injury crash risk. Traffic Inj. Prev. 2016,17, 870–877. [CrossRef]
Sustainability 2022,14, 7563 20 of 20
28.
Vic Roads. 2014 Analysis of the Victorian Road Toll Fatalities and Serious Injuries. 2015. Available online: https://www.google.it/
url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwjApqS2tqn4AhXQTd8KHXq_B9kQFnoECA0QAw&url=https%
3A%2F%2Fwww.vicroads.vic.gov.au%2F-%2Fmedia%2Ffiles%2Fdocuments%2Fsafety-and-road-rules%2Farchived%2F201
4_analysis_of_the_victorian_road_toll__fatalities_and_serious_injuries.ashx&usg=AOvVaw1QWDcJRCJ8RvzVmeoR9GFc
(accessed on 19 April 2022).
29.
Allen, T.; Newstead, S.; Lenné, M.G.; McClure, R.; Hillard, P.; Symmons, M.; Day, L. Contributing factors to motorcycle injury
crashes in Victoria, Australia. Transp. Res. Part F Traffic Psychol. Behav. 2017,45, 157–168. [CrossRef]
30.
Day, L.; Lenné, M.G.; Symmons, M.; Hillard, P.; Newstead, S.; Allen, T.; McClure, R. Population based case-control study of
serious non-fatal motorcycle crashes. BMC Public Health 2013,13, 1–6. [CrossRef]
31.
Vic Roads. Crashstats—User Guide. 2013. Available online: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=
web&cd=&ved=2ahUKEwjbpqbLtqn4AhUOnXIEHclVCXIQFnoECAUQAQ&url=https%3A%2F%2Fdata.vicroads.vic.gov.
au%2Fmetadata%2Fcrashstats_user_guide_and_appendices.pdf&usg=AOvVaw2f9Xa_hyXHKfyieFk8uLDa (accessed on 19
April 2022).
32.
Rizzi, M.; Strandroth, J.; Holst, J.; Tingvall, C. Does the improved stability offered by motorcycle antilock brakes (ABS) make
sliding crashes less common? In-depth analysis of fatal crashes involving motorcycles fitted with abs. Traffic Inj. Prev.
2016
,
17, 625–632. [CrossRef] [PubMed]
33.
Teoh, E.R. Motorcycle antilock braking systems and fatal crash rates: Updated results. Traffic Inj. Prev.
2022
,23, 1–5. [CrossRef]
[PubMed]
34.
Silla, A.; Leden, L.; Rämä, P.; Scholliers, J.; Van Noort, M.; Morris, A.; Hancox, G.; Bell, D. A headway to improve PTW rider
safety within the EU through three types of ITS. Eur. Transp. Res. Rev. 2018,10, 1–12. [CrossRef]
35.
Fang, C.Y.; Hsu, W.H.; Ma, C.W.; Chen, S.W. A vision-based safety driver assistance system for motorcycles on a smartphone.
In Proceedings of the 17th International IEEE Conference on Intelligent Transportation Systems (ITSC), Qingdao, China, 8–11
October 2014; pp. 328–333. [CrossRef]
36.
Gil, G.; Savino, G.; Piantini, S.; Pierini, M. Motorcycle that see: Multifocal stereo vision sensor for advanced safety systems in
tilting vehicles. Sensors 2018,18, 295. [CrossRef]
37.
Huth, V.; Gelau, C. Predicting the acceptance of advanced rider assistance systems. Accid. Anal. Prev.
2013
,50, 51–58. [CrossRef]
[PubMed]
38.
Biral, F.; Bosetti, P.; Lot, R. Experimental evaluation of a system for assisting motorcyclists to safely ride road bends. Eur. Transp.
Res. Rev. 2014,6, 411–423. [CrossRef]
39.
Lich, T.; Block, W.; Prashanth, S.; Heiler, B. Motorcycle Stability Control—The Next Generation of Motorcycle Safety and Riding
Dynamics. SAE Int. J. Engines 2016,9, 491–498.
40.
Cohen, J. Weighted kappa: Nominal scale agreement provision for scaled disagreement or partial credit. Psychol. Bull.
1968
,
70, 213. [CrossRef]
41. Cohen, J. A Coefficient of Agreement for Nominal Scales. Educ. Psychol. Meas. 1960,20, 37–46. [CrossRef]
42.
Gil, G.; Savino, G.; Piantini, S.; Pierini, M. Is stereo vision a suitable remote sensing approach for motorcycle safety? An analysis
of LIDAR, RADAR, and machine vision technologies subjected to the dynamics of a tilting vehicle. In Proceedings of the 7th
Transport Research Arena TRA, Vienna, Austria, 16–19 April 2017; Volume 12, pp. 1–10.
43.
Raj, T.; Hashim, F.H.; Huddin, A.B.; Ibrahim, M.F.; Hussain, A. A survey on LiDAR scanning mechanisms. Electronics
2020
,9, 741.
[CrossRef]
44.
Biral, F.; da Lio, M.; Lot, R.; Sartori, R. An intelligent curve warning system for powered two wheel vehicles. Eur. Transp. Res. Rev.
2010,2, 147–156. [CrossRef]
45.
Dean, M.E.; Haus, S.H.; Sherony, R.; Gabler, H.C. Potential Crash Benefits of Motorcycle-Detecting Automatic Emergency Braking
Systems. In Proceedings of the IRCOBI Conference, online, 8–10 September 2021; pp. 39–50.
46.
Australian Bureau of Statistics “Road Trauma Australia 2014-Statistical Summary”. 2015. Available online: https://www.nrspp.
org.au/resources/bitre-road-deaths-australia- 2014-statistical-summary/ (accessed on 19 April 2022).
