Wu Z, Zeng X, Wang L.

A New Traffic Conflict Measure for Electric Bicycles at Intersections

ZHIZHOU WU, Ph.D1. Safety and Security in Traffic

E-mail: wuzhizhou@tongji.edu.cn Original Scientific Paper
XIN ZENG, Master Degree Candidate1 Submitted: 1 Apr. 2019
E-mail: zengxin13579@qq.com Accepted: 27 Nov. 2019
LING WANG, Ph.D.1
(Corresponding author)
E-mail: wang_ling@tongji.edu.cn
1 The Key Laboratory of Road and Traffic Engineering,
Ministry of Education, Tongji University
Cao’an Road #4800, Jiading District, Shanghai, 201804,
China

A NEW TRAFFIC CONFLICT MEASURE

FOR ELECTRIC BICYCLES AT INTERSECTIONS

ABSTRACT 1. INTRODUCTION
As electric bicycles (e-bikes) are becoming popular in In China, electric bicycles (e-bikes) have be­
China, concerns have been raised about their safety con­
come a popular travel mode for citizens. They are
ditions. A traffic conflict technique is commonly used in
traffic safety analysis, and there are many conflict mea­
as convenient and flexible as conventional bicycles,
sures designed for cars. However, e-bikes have high flex­ but can reach much higher speeds (30 km/h or even
ibility to change speed and trajectories, which is different higher) [1]. According to the China Traffic Manage­
from cars, so the conflict measures defined for e-bikes ment Bureau, the number of e-bikes was 250 mil­
need to be independently explored. Based on e-bike driv­ lion in 2017. Meanwhile, from 2013 to 2017, e-bike
ing characteristics, this paper proposes a new measure, related crashes have resulted in about 56,200 inju­
the Integrated Conflict Intensity (ICI), for traffic conflicts ries and 8,431 fatalities in China. The traffic safety
involving e-bikes at intersections. It measures the degree of e-bikes cannot be ignored.
of dangerousness of a conflict process, with consider­
In order to assess the safety of e-bikes when
ation of both conflict risk and conflict severity. Time to
passing through intersections, there is need to pro­
collision is used to measure the conflict risk. Relative ki­
netic energy is used to measure the conflict severity. ICI pose e-bike traffic conflict technique which has
can be calculated based on video analysis. The method been commonly used for cars. Traffic conflict was
of determining ICI thresholds for three conflict levels defined for the first time by Perkins & Harris [2]
(serious, less serious, and slight) and two conflict types as a surrogate safety measure for crashes at inter­
(conflicts between two e-bikes, and conflicts between an sections. Baker [3] described traffic conflict as the
e-bike and a car) is put forward based on the question­ situation in which a driver tries to avoid a poten­
naires about safety perception of e-bike riders, which is tial accident or a situation of danger through the
regarded as the criterion of e-bike safety conditions at application of an evasive manoeuver (braking, lane
intersections. The video recording and a questionnaire
change, or acceleration).
survey about conflicts involving e-bikes at intersections
have been conducted, and the unified thresholds applica­
After decades of development, the traffic conflict
ble to different intersections have been determined. It is technique is becoming more mature and sophisticat­
verified that ICI and its thresholds meet the criterion of ed. However, since cars account for the majority of
e-bike safety conditions. This work is expected to be used traffic structure in the cities, most of the research
in the selection of intersections for safety improvement of focuses on conflict measures for cars. In terms of
e-bike traffic. e-bikes, the driving characteristics are different
from cars, so it is inappropriate to directly apply the
KEY WORDS methods for cars to e-bikes. It is necessary to form
electric bicycle; traffic conflict; conflict measure; time to an improved new safety measure for e-bikes based
collision; kinetic energy; threshold; on a traffic conflict technique.

Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320 309
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

Due to the lack of physical protection, provid­ To assess the conflict levels, the criteria of TTC
ed by cars, e-bike riders are directly exposed to should be determined. A TTC-threshold of 1 s was
safety risks at intersections, and they take direct originally formulated by Hayward [9] in order to
damage from an accident. Therefore, in terms of distinguish between the so-called “near-misses” and
e-bike-related conflicts, both risk and severity are safe driving situations. Hyden & Linderholm [10]
important factors. In this study, the conflict risk and proposed a comparable threshold of 1.5 s. Higher
the conflict severity are considered together in the thresholds have been put forward by other research­
proposed measure for e-bikes. ers to suit different situations. For example, Hirst &
The remainder of this paper is organized as fol­ Graham [11] regarded a TTC measure of 4 s as the
lows. Section 2 reviews the previous studies of traf­ safe-critical value considering the driver’s percep­
fic conflict measures. Section 3 discusses the driv­ tion. These studies make TTC easy to use in traffic
conflict analysis.
ing characteristics of e-bikes. Section 4 describes
Another popular indicator is Post-Encroachment
the methodology. Section 5 applies and verifies the
Time (PET), initially introduced by Allen et al. [12].
proposed measure. Finally, Section 6 summarizes
PET is defined as the time between the moment when
the findings of this study and provides suggestions
the first road user leaves the path of the second road
for future research. user and the moment when the second user reaches
the path of the first user (i.e. PET indicates the extent
2. LITERATURE REVIEW to which they miss each other). PET is a period of
Current studies on e-bike safety are mainly time that has already elapsed in reality, and there is
about the speed or crash characteristics based on the only one PET for a single conflict process.
sample data analysis. For example, Lin et al. [4] got Besides, some improved indicators based on
the operating speed and its distribution of e-bikes in time are put forward to extend usability, such as TA
China, Schepers et al. [5] compared crash severity (Time-To-Accident) proposed by Hyden [13], TET
of e-bikes with classic bicycles in the Netherlands, (Time Exposed Time-to-Collision) and TIT (Time
Gorenflo et al. [6] revealed participants’ safety con­ Integrated Time-to-Collision) proposed by Minder-
cerns about the speed of e-bikes in Canada, and houd & Bovy [14], RTTC (Relative Time to Colli­
sion) proposed by Chen et al. [15]. But in general,
Hertach et al. [7] analysed crash causes of e-bikes in
TTC and PET are the most commonly used measures
Switzerland. Traffic conflicts related to e-bikes are
for the conflict analysis at intersections.
rarely studied, and the only few studies are about
2) Measures based on distance
conflict types and regions at intersections [8], with­
This kind of measures considers the distance re­
out referring to safety measures.
lated to conflict, such as the remaining distance to
As for current traffic conflict measures, although
potential point of collision introduced by Allen et al.
they are not specially designed for e-bikes, the basic
[12]. Its implication and effect are similar to the mea­
methods and ideas can be of advantage. In general, sures based on time, so it is not commonly used and
the conflict measures can be divided into five cate­ discussed.
gories: based on time, based on distance, based on 3) Measures based on speed
speed, based on kinetic energy, and based on com­ TTC and PET reflect the possibility of crashes,
bined indicators. but sometimes the crash severity (once it happens) is
1) Measures based on time taken into consideration rather than the crash prob­
The most widely used conflict measure based ability. Kloeden et al. [16] found that the vehicle
on time is Time-To-Collision (TTC). The origi­ speed when a crash occurs, significantly contributes
nal definition of TTC proposed by Hayward [9] is to the severity of that crash. Shelby [17] regarded
the time required for two vehicles to crash if they the change in vehicle speed because of collision as a
continue moving at their current speeds and in the measure oftraffic conflict severity. These studies take
same direction. Obviously, TTC is a “crash-esti­ speed into consideration for its high correlation with
mated” time that has not actually happened yet, fatal accidents.
and for every moment there is a corresponding 4) Measures based on kinetic energy
TTC as long as two vehicles are in conflict. So, It is commonly thought that at the same speed,
there may be numerous TTCs in a complete con­ heavier vehicles may contribute to higher crash se­
flict process. verity. Kinetic energy combines mass with speed,

310 Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

and it indicates the potential energy of a moving of e-bikes is 15-25 km/h [22], much higher than that
vehicle, which may be released in a crash. There­ of conventional bicycles, even close to cars on some
fore, recent studies evaluate crash severity in the city roads. Due to the lack of necessary protection
viewpoint of kinetic energy. Chen et al. [18] de­ as provided by cars and proper speed limit like con­
fined kinetic energy of conflict as the sum of the ventional bicycles, e-bike riders withstand higher
kinetic energy of conflict entities. Sobhani et al. safety risks.
[19] considered the change of kinetic energy using Acceleration characteristics. E-bike accelera­
kinetic equations in physics. It is believed that ki­ tion is rapid relative to that of conventional bicy­
netic energy measures crash injury severity well. cles without electric motors. It is not challenging
5) Measures based on combined indicators for e-bikes to accelerate from 0 to 20 km/h in 4 s
Moreover, some studies combine two or more [23]. Fast speed change makes it difficult to predict
indicators together, making the analysis more e-bike behaviour.
comprehensive. For example, Fazekas et al. [20] Trajectory characteristics. The turning of
defined DRAC (Deceleration Rate to Avoid a e-bikes is flexible, resulting in changeable driving
Crash) which is a combination of speed difference trajectories. If disturbed frequently, the trajectories
with distance. Alhajyaseen [21] proposed CI (Con­ of e-bikes would be fluctuant.
flict Index) which is a combination of PET with the These driving characteristics distinguish the con­
change of kinetic energy. This idea is worth taking flict characteristics of e-bikes from those of the cars.
in because it considers both the crash probability Car drivers are inclined to brake rather than to turn
and the crash severity.
the steering wheel when facing an emergency, because
Briefly, there are numerous traffic conflict mea­
cars are relatively bulky and the direction adjustment
sures which can be divided into five categories.
within a short time is not easy. However, e-bikes can
These measures refer to crash probability and crash
be flexibly controlled, so the riders tend to change tra­
severity. However, they are not specially designed
jectories as well as speed in traffic conflicts.
for e-bikes, so a new measure should be established
As a result, in the conflicts involving e-bikes,
based on the e-bike driving characteristics.
the conflict points are changeable as the trajecto­
ries change. That is, during a conflict process, the
3. E-BIKE DRIVING position of the expected crash point is no longer
CHARACTERISTICS fixed, but changeable with time. Similarly, TTC is
Learning more about the driving characteristics time-related in a conflict process. However, there
of e-bikes helps to develop the traffic conflict mea­ is only one value for PET during a conflict process
sure. According to the studies based on solid field according to Section 2 of this paper. Therefore, the
observations, the main driving characteristics of concept of PET is not applicable in e-bike conflicts,
e-bikes in China are concluded as follows: while TTC is still suitable as it describes a state.
Speed characteristics. Due to the absence of Furthermore, the change of TTC over time can be
strict implementation of e-bike product standards, analysed. In a typical conflict process, there should
many e-bikes have a maximum design speed of over be only one local minimum TTC which represents
30 km/h [1]. Besides, the average operating speed the absolute minimum TTC (Figure 1a) [9]. While in

a) Typical TTC-time diagram with one b) TTC-time diagram with several

local minimum TTC value local minimum TTC values involving e-bikes
Figure 1 - TTC-time diagram ofa conflict process

Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320 311
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

the conflict involving e-bikes, it is likely that there As described in Section 3 of this paper, TTC, in­
is more than one local minimum TTC according to stead of PET, is used as the crash-estimated time.
field observation of this study which collected the The smaller the TTC, the higher the conflict risk.
TTC of e-bike related conflicts from videos taken However, CPR is supposed to increase as conflict
from intersections in China (Figure 1b). It indicates risk grows, so TTC is not directly used.
that because of the driving flexibility and uncertain­ The form “1/TTC” was considered once. How­
ty of e-bikes, the conflict resolution is probably not ever, when TTC tends to zero, its value tends to in­
a one-time process, but a dynamic process with mu­ finity, missing the meaning of the measuring possi­
tual feedback and continuous adjustment, usually bility. Thus, it is not appropriate here.
taking more time. Instead, CPR is defined as follows:
CPR = eTTC (1)
4. METHODOLOGY
The value of CPR ranges from 0 to 1, because
This study focuses on developing a measure ap­ TTC>0 indicates eTTC>1 and 0 < e^ # 1. When
plicable to conflicts involving e-bikes at intersec­
tions. The main idea includes three aspects: (1) con­ TTC is large, CPR is close to 0, suggesting low con­
sidering both conflict risk and conflict severity; (2) flict risk. CPR close to 1 means high conflict risk,
assessing the entire conflict process in discrete time and CPR = 1 indicates a crash as TTC = 0. So this
steps, and “the worst moment” representing this form describes the conflict risk reasonably.
conflict process; (3) considering conflicts between Since CPR is connected with TTC, it is a time-re­
two entities: one is an e-bike, one is another e-bike lated variable like TTC. For every moment in a con­
or other kind of vehicle (usually a car). flict process, the value of CPR can be calculated.
The first step in traffic conflict analysis is to In addition to CPR, CPH is proposed to measure
identify the traffic conflict. The identification of the severity of a possible crash (which is called con­
traffic conflicts is based on the observed evasive flict severity here, since the crash has not actually
actions between two conflict entities, such as brak­ happened) from the perspective of kinetic energy.
ing, swerving, and deceleration [1]. Swerving is fre­ As mentioned in Section 2 of this paper, some
quently used to eliminate danger for e-bikes while research formed physical equations to depict pos­
deceleration is used for cars. sible kinetic changes of vehicles after crash. How­
In order to analyse a conflict process, the start ever, this method is based on the analysis of scenes
and the end time of the process should be deter­ that have not actually happened, accompanied with
mined. Considering the mixed traffic conditions in many unverifiable assumptions. In this study, the
China, the TTC of an observed conflict is not high, conflict severity is considered based on reality.
normally varying between 0.5 and 2.6 s [1]. Here, a For convenience, the two conflict entities are
TTC value of 2.6 s is adopted as the threshold of a marked as A and B. For a moment during the con­
conflict process. That is, if TTC drops to less than flict process, A and B run at speeds va and vb, re­
2.6 s, the conflict process starts at that time; if TTC spectively. Now, connect with a line between A and
rises to higher than 2.6 s, the conflict process ends B, and establish a rectangular coordinate system in
at that time. which X-axis is parallel to the line (Figure 2). Mark
the angle between va (or vb) and the positive direc­
tion of X-axis as a (or ft). The value of a (or ft) rang­
4.1 Integrated Conflict Intensity (ICI)
es from 0 to 2 n.
This study defines Integrated Conflict Intensity
(ICI) as the measure for traffic conflicts involving vb vby
y
e-bikes. In order to make the derivation clearer,
V va
Conflict Potential Risk (CPR), Conflict Potential ay

Harm (CPH), and their combination Conflict Po­

tential Risk and Harm (CPRH) are proposed with b
a
different meanings.
A Vax vx B
CPR is proposed to measure the possibility of
------------- ►
a conflict converting into a crash (which is called x
conflict risk here). Figure 2 - Diagram for analysis ofvehicle speeds

312 Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

According to a (or ^), va (or vb) can be decom­ In fact, indicator CPRH is well adapted for con­
posed into vax and vay (or vbx and vby) along the di­ flicts involving e-bikes. E-bikes run flexibly, often
rections of X-axis and Y-axis. There is no conflict leading to low TTCs. Although CPR is large, it does
between vay and vby because they are parallel to each not mean the situation is critical. In most cases, even
other. The conflict exists between vax and vbx. if there is an accident, the severity is slight, such as
Here, Av is defined as the “relative speed” be­ lateral scratching. So CPRH will not be too large as
tween A and B. The term “relative speed” implies CPH is small, which is expected in this study.
the speed difference along the direction of A and B, CPRH is a variable related to time since CPR or
i.e. the difference between vax and vbx. It is calculat­ CPH are. It is a measure for the conflict state at a
ed as follows: certain moment.
Dv = |v«t - v&t| = |va cos a - Vb cos b\ (2) Now, an entire conflict process can be studied.
In order to assess a conflict process, the moments
Although the decomposition analysis of va and
vb above (Figure 2) is based on the ground frame, rel­
in the process are selected with a fixed time step
ative speed Av is defined from the perspective of A (0.1 s in this study), and the corresponding CPRHs
(or B). In A’s view, B is coming at a speed of Av, and are calculated for these moments. Suppose there are
in B’s view, A is coming at a speed of Av. n moments selected, and the CPRH at moment i is
Then, AKa and AKb are defined as the “relative expressed as:
kinetic energy”:
(
CPRHi = Oaimnb eT-iC , i = 1, . f,n (6)
DKa = 1-ma(Dvh2, Kb = 1 mb (Dvh2 (3)
where Avi is Av at moment i, and TTCi is TTC at
where ma and mb are the mass of A and B.
moment i.
In the perspective of A, B is coming towards
A with the relative kinetic energy of AKb. And in The largest one among these CPRHs represents
the perspective of B, A is coming towards B with the worst moment of this conflict, which is critical for
the relative kinetic energy of AKa. For A, the larger measuring the safety. It is defined as the final mea­
AKb (i.e. larger mb and Av), the more harm A would sure of a conflict involving e-bikes, called ICI here:
suffer once the crash happens; similarly for B. So
AKa and AKb can be used to measure the conflict
ICI = max" CPRHi}n= 1 = JmOmb • max ( ^^vCC | (7)
severity.
To consider AKa and AKb together, their geomet­ where max (( dvih2
ric mean can be simply taken as the final definition (
eTTCi
r
2i=1
means the maximum val-

of CPH. Here, the coefficient “1/2” in AKa and AKb

is omitted for a simple form.
ue of all numbers in the set ((Dv C |
:i = 1,....n .

ICI is used to assess a conflict based on the en­

CPH = maia (Dv h2 $ mb (Dvh2 = mmmnb (Dvh2 (4)
tire process instead of a single moment. It is deter­
Clearly, a larger CPH indicates a higher de­ mined by the CPRH of the worst moment. A large
gree of conflict severity. Similar to CPR, CPH is a ICI indicates a conflict that not only lasts long, but
time-related variable as Av changes over time. also has a high conflict risk and high degree of se­
In order to consider CPR and CPH together, verity at one time.
CPRH is defined as follows: In the expression of ICI, TTCi is the indicator
CPRH = CPR $ CPH = ^"a^CD'h (5) related to conflict risk, while ma, mb and Avi are the
indicators related to conflict severity. So the term
CPRH indicates the integrated meaning of CPR “intensity” in ICI refers to both conflict risk and
and CPH. If the possibility of a potential crash is conflict severity, together indicating the degree of
high, but its severity is slight, CPRH will not be dangerousness of the conflict.
large. Likewise, if a potential crash is estimated to In one word, ICI is an integrated measure for the
be severe, but its possibility is negligible, CPRH degree of dangerousness of a conflict, and it is ap­
will not be large either. A rather large CPRH indi­ plicable to traffic conflicts that involve e-bikes.
cates both high conflict risk and high degree of con­ The procedure
summarized of calculating
in Figure 3. ICI for a conflict is
flict severity.

Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320 313
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

To be as practical as possible here, this study

determines the thresholds for the three conflict
levels based on the field survey data. One idea is
to collect crash data at various intersections, and
divide ICI data into three groups as conflict lev­
els according to the severity classification of crash
data. This idea is theoretically feasible, but the use
of crash data is problematic for its unavailabili­
ty. Besides, it takes a great deal of time to collect
enough samples.
Instead, it is considered to survey subjective
safety perceptions of e-bike riders at intersections,
and use these data as the criterion to determine the
conflict level thresholds. In this study, two types
of conflicts are considered: conflicts between two
e-bikes, and conflicts between an e-bike and a car.
The questionnaire is designed as shown in
Table 1. The questionnaire refers to four choice
questions as shown in Table 1, and these questions
Figure 3 — Flowchart ofICI calculation refer to crashes between two e-bikes (Questions 1,
2) and crashes between an e-bike and a car (Ques­
tions 3, 4), as well as conflict risk (Questions 1, 3)
4.2 Threshold determination and conflict severity (Questions 2, 4).
Now, a value of ICI can be calculated, but the For each question in the questionnaire, three
value itself has no direct practical meaning. From options a, b, c are assigned three values 3, 2, 1,
the value, it is not known to which level the conflict respectively, indicating the levels of conflict risk
or conflict severity. Thus, for each question in each
belongs, whether the conflict is serious or not, and
completed questionnaire, there is a corresponding
to what extent the conflict is serious.
value. In consideration of combining the conflict
Imitating the severity levels of traffic events pro­ risk and the conflict severity, the two values for
posed by Hyden [24] and Svensson [25], this study each conflict type are multiplied (i.e. multiply the
divides traffic conflicts into three levels from the values of Questions 1 and 2 as the result for con­
perspective of safety: serious conflicts, less serious flicts between two e-bikes, and multiply the values
conflicts, and slight conflicts. of Questions 3 and 4 as the result for conflicts be­
Distinguishing different conflict levels is of great tween an e-bike and a car). Obviously, each multi­
significance since there is no other similar definition plied value is in the set of {9, 6, 4, 3, 2, 1}.
of conflict levels related to e-bikes. However, it is The multiplied value is used to represent the
not easy to determine the criterion. For one thing, conflict level. In this study, {9} is regarded as se­
the determination is rather subjective because dif­ rious conflicts, {6, 4} as less serious conflicts, and
ferent people may hold different opinions about the {3, 2, 1} as slight conflicts. According to this cri­
seriousness of the same conflict. For another, criteria terion, the values derived from the questionnaire
data at one intersection can be divided into three
for other measures have no reference value because
parts, representing three conflict levels. So the pro­
systems are different. For example, many people
portions of serious conflicts, less serious conflicts
regard TTC < 1 s as a serious conflict, but others
and slight conflicts at one intersection can be ob­
believe it is 1.5 s or some other numbers. Even if tained.
there is a universally accepted criterion, it cannot be Besides, the ICI data of the observed conflicts
applied here because the factors considered are dif­ at the intersection can be calculated. After being
ferent. Therefore, it is necessary to know about the sorted from large to small, these ICI data can be
traffic safety conditions of e-bikes at intersections divided according to the proportions of three con­
as the criterion. flict levels. Then, at the two division positions, two

314 Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

Table 1 - Questionnaire about e-bike riders ’ safety perceptions at intersections

No. Questions and options

Based on this passing experience, do you think that there is a possibility of a crash between two e-bikes at this
intersection?
1 a. very likely
b. likely
c. not very likely
Based on this passing experience, if there is a crash between two e-bikes at this intersection, what do you think the
crash would be like?
2 a. a severe crash, causing serious injury or even death
b. a less severe crash, causing persons to fall or be bruised and vehicles to be deformed
c. a minor crash, causing persons to be slightly scratched and vehicles to be slightly damaged at most
Based on this passing experience, do you think that there is a possibility of crash between an e-bike and a car at this
intersection?
3 a. very likely
b. likely
c. not very likely
Based on this passing experience, if there is a crash between an e-bike and a car at this intersection, what do you think
the crash would be like?
4 a. a severe crash, causing serious injury or even death
b. a less severe crash, causing persons to fall or get bruised and vehicles to be deformed
c. a minor crash, causing persons to be slightly scratched and vehicles to be slightly damaged at most

thresholds can be determined: one is the minimum In order to collect ICI data, videos were recorded
value of serious conflicts (also the maximum value at the three intersections during rush hours (04:30­
of less serious conflicts), and one is the minimum 06:30 p.m.). The movement of vehicles in the videos
value of the less serious conflicts (also the maxi­ was analysed by the software George 2.1 (Figure 4).
mum value of slight conflicts). Based on the videos, conflicts were identified and
In short, traffic conflicts involving e-bikes are di­ extracted by certain means, i.e. recording the first
vided into three levels, and the safety perceptions of appeared conflict in every two minutes, which could
e-bike riders are regarded as the criterion for thresh­ avoid the subjectivity of artificial choice, so the
old determination. conflict samples could represent the overall situa­
tion. Two types of conflicts (conflicts between two
5. APPLICATION AND VERIFICATION e-bikes, and conflicts between an e-bike and a car)
were recorded and analysed separately.
In order to get the specific values of thresholds Fifty conflicts were recorded for every type, ev­
and to verify the proposed measure, field surveys ery intersection, and the corresponding ICIs were
were conducted at intersections, and ICI data were calculated with the method described previously. In
obtained and analysed as well as the questionnaire the calculation, the mass of vehicles (including rid­
data. ers or drivers) was estimated according to the size,
the time step was 0.1 s, and other parameters were
5.1 ICI data calculated as described. The International System
Three four-arm signalized intersections (Table 2) of Units of Measurement was adopted as the units
were selected for investigation in Shanghai, China. of these parameters, i.e. TTCi [s], ma and mb [kg],
They vary from each other in signal phases, scales, Avt [m/s].
locations, etc.; thus the safety conditions for e-bikes The values of those ICIs are shown as box­
are very likely to be different, which is necessary for plots in Figure 5 (a few extremely large values are
testing the applicability of the proposed measure. omitted). For conflicts between two e-bikes, most
Table 2 - Information ofthe investigated intersections

ID Name Number of Phases Scale Location

A Jianhe-Xianxiaxi 2 small, subarterial-subarterial city subcentre
B Changji-Moyu 3 medium, subarterial-subarterial suburb
C Siping-Dalian 4 large, arterial-arterial city centre

Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320 315
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

Figure 4 — Movement ofvehicles analysed by George 2.1

Intersection Intersection
a) Conflicts between two e-bikes b) Conflicts between an e-bike and a car
Figure 5 — ICI boxplot of two conflict types

(98.7%) ICI data vary from 0 to 5,000; for conflicts A larger ICI-index indicates a more serious
between an e-bike and a car, most (98.7%) ICI data conflict. ICI-index=0 means no conflict, while
vary from 0 to 20,000. Non-typical ICI values are ICI-index=1 means an extremely serious conflict.
some extremely large values, which indicates the ICI-index is more convenient to imply the serious­
conflict is so serious that it almost causes a crash. ness of a conflict, and it is used for the later analysis.
To make the interpretation of ICI more intui­
tive, ICI data were converted into an index called 5.2 Questionnaire data

*
ICI-index, which ranges in a fixed interval [0, 1]. For
conflicts between two e-bikes, the conversion is: To know the safety conditions of intersections
as reference, questionnaire surveys were conducted
ICI-index 1= -C0, 0 #ICI # 5000 (8) at the three intersections, that is, for e-bike riders
1, ICI 2 5000 about their subjective safety perceptions of the spe­
For conflicts between an e-bike and a car, the cific intersection they passed through.
conversion is: For the sake of safety, the questionnaires were not
filled in the field but through Internet. The research
T^T , [ cm, 0 # ICI # 20000
ICI-index2 = 20000 , team handed out papers, and the questionnaire link
1, ICI 2 20000 (9) was on the papers. This work was implemented at

316 Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

Table 3 - Proportions of three conflict levels based on questionnaire data

Conflicts between two e-bikes Conflicts between an e-bike and a car

Conflict levels
A B C A B C
Serious conflicts 0.061 0.024 0.063 0.091 0.095 0.375
Less serious conflicts 0.394 0.524 0.583 0.636 0.667 0.354
Slight conflicts 0.545 0.452 0.354 0.273 0.238 0.271

the three intersections respectively, on the same day the minimum value of less serious conflicts. Figure 6
of the week and at the same hours when the videos compares the conflict level ranges of different inter­
were recorded. In total, 250 papers were handed out sections for each conflict type.
at each intersection, and 33, 42, 48 valid question­ The thresholds for serious conflicts derived from
naires were collected for Intersections A, B, C, re­ different intersections are close to each other, and so
spectively. The collection rate was not high maybe are the thresholds for less serious conflicts. To quan­
because many riders just forgot the papers. But the tify the differences, the Coefficient of Variation is
samples were basically sufficient like some oth­ calculated in Table 5.
er studies (e.g. the study by Lowry et al. [26]: 92 In the four cases, the Coefficient of Variation of
participants, and the study by Wang et al. [27]: 72 the threshold is no more than 0.08. It indicates that
participants). the threshold deviation does not exceed 8%. There­
The proportions of three conflict levels at each fore, although the three intersections are different
intersection based on the questionnaire data are in many aspects, their ICI-index thresholds are sim­
shown in Table 3. ilar to each other. In order to finally determine the
thresholds, the values in Table 6 are taken as the uni­
5.3 Thresholds and verification fied ICI-index thresholds.
With the proportions derived from questionnaire The invariance of thresholds derived from dif­
data, the ICI data were divided, and the thresholds ferent intersections indicates that the ICI-index
determined at the division positions. Table 4 shows the thresholds are independent of the intersections. It is
two thresholds derived from each intersection: one verified that ICI and its thresholds are applicable for
is the minimum value of serious conflicts, and one is different scales of four-arm signalized intersections.
Table 4 - ICI-index thresholds for conflict levels derivedfrom each intersection

Conflicts between two e-bikes Conflicts between an e-bike and a car

Conflict levels
A B C A B C
Serious conflicts 0.64 0.64 0.69 0.37 0.38 0.34
Less serious conflicts 0.17 0.19 0.18 0.06 0.06 0.07

a) Conflicts between two e-bikes b) Conflicts between an e-bike and a car

□ Serious conflicts □ Less serious conflicts I Slight conflicts

Figure 6 - ICI-index ranges ofthree conflict levels for two conflict types

Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320 317
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

Table 5 - Threshold statistics of the three intersections

Conflicts between two e-bikes Conflicts between an e-bike and a car

Statistics Threshold for serious Threshold for less Threshold for serious Threshold for less
conflicts serious conflicts conflicts serious conflicts
Mean 0.657 0.180 0.363 0.0633
Standard deviation 0.0236 0.00816 0.0170 0.00471
Coefficient of Variation 0.0359 0.0454 0.0468 0.0744

Table 6 - Unified ICI-index thresholds and ranges for conflict levels

Conflicts between two e-bikes Conflicts between an e-bike and a car

Conflict levels
Threshold Range Threshold Range
Serious conflicts 0.66 (0.66, 1] 0.36 (0.36, 1]
Less serious conflicts 0.18 (0.18, 0.66] 0.06 (0.06, 0.36]
Slight conflicts / [0, 0.18] / [0, 0.06]

Besides, if considering the threshold determi­ The method of determining ICI thresholds for
nation process conversely, the proportions of three three conflict levels (serious, less serious and slight),
conflict levels can be derived from ICI data with and two conflict types (conflicts between two e-bikes,
the unified thresholds, and they are the same as the and conflicts between an e-bike and a car) is put for­
proportions derived from the questionnaire data. ward based on the questionnaire data about safety
This indicates that the measure of ICI data for con­ perceptions of e-bike riders, which is regarded as the
flict levels is in accordance with that of question­ criterion of e-bike safety conditions at intersections.
Field surveys at four-arm signalized intersections
naire data (which is regarded as the criterion). So
about conflicts involving e-bikes were conducted to
it can be concluded that ICI and its thresholds for
determine ICI thresholds and to verify them. ICI was
determining conflict levels are valid and practical. converted into ICI-index ranging in [0, 1] to make
its meaning more intuitive. The thresholds at three
6. CONCLUSION intersections were calculated, and they are close to
Although there are numerous conflict measures each other, which indicates that the ICI-index thresh­
or indicators, they are not specially designed for olds are independent of intersections. It is verified
that the measure of ICI data for conflict levels is in
e-bikes, thus needing improvement. The analysis of
accordance with that of the questionnaire data, so ICI
driving characteristics of e-bikes reveals that they
and its thresholds meet the criterion of e-bike safety
differ from cars greatly. E-bikes have high flexibility
conditions at intersections.
to change speed and trajectories, leading to a change­
This work can be used in the selection of intersec­
able conflict point, more uncertainty, and more time tions to be improved for e-bikes. A larger proportion
for conflict resolution. Thus, this study focuses on a of serious conflicts suggests that there is greater need
new measure for traffic conflicts involving e-bikes at to take improvement measures at this intersection.
intersections. The improvement can be specific to the conflict type.
The traffic conflict measure for e-bikes is estab­ Future work can aim at increasing the amount of
lished based on the considerations of both conflict intersections to further verify the thresholds proposed
risk and conflict severity. CPR is proposed to mea­ by this study. Moreover, it can be tested whether this
sure the possibility of a conflict converting into a measure can be used for different geometries of inter­
crash with TTC, and CPH is proposed to measure sections in different countries.
the severity degree of a possible crash with relative
kinetic energy. The combination of CPR and CPH is ACKNOWLEDGEMENT
CPRH, which measures the conflict at a moment. The The research was supported by the National Nat­
maximum CPRH during a complete conflict process ural Science Foundation of China (No. 61773288)
is defined as ICI, which is finally used to measure and the National Key Research and Development
conflicts involving e-bikes. Program of China (No. 2018YFB1600805).

318 Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

^XM,W±1 Compared to Classic Bicycles. Accident Analysis & Pre­

4>?W^: wuzhizhou@tongji.edu.cn vention. 2014;73: 174-180.
eax^xma1 [6] Gorenflo C, Rios I, Golab L, Keshav S. Usage Patterns
4>?W^: zengxin13579@qq.com of Electric Bicycles: An Analysis of the WeBike Project.
x^,«±1 (»WftO Journal of Advanced Transportation. 2017; Article ID
^TW^: wang_ling@tongji.edu.cn 3739505.
1 ^»^misosom [7] Hertach P, Uhr A, Niemann S, Cavegn M. Characteristics
+ ffl±Wm>£EW££< 4800^ of Single-vehicle Crashes with E-bikes in Switzerland.
Accident Analysis & Prevention. 2018;117: 232-238.

ffififcwwxn^mw [8] Bai L, Liu P, Wang W, Chen Y. Characteristics of Traffic

Conflicts between Electric Bicycles and Motor Vehicles
at Unsignalized Intersections. The 12th COTA Interna­
imTOWH»mraH£ tional Conference of Transportation Professionals, 3-6
^SiJ^ttoXW^K^MxmX^^fflW August 2012, Beijing, China; 2012.

t&iSlRm^TW»^.«( [9] Hayward JC. Near-Miss Determination through Use of

a Scale of Danger. Highway Research Board. 1972;384:
24-34.
tt,$ww(m,ww^g [10] Hyden C, Linderholm L. The Swedish Traffic-Conflicts
ffw%#wm.*tmww Technique. In: Asmussen E. (ed.) International Calibra­
^ft.^T-mnnfcW^W tion Study of Traffic Conflict Techniques. Berlin, Heidel­
^ttm.efflfW£ berg: Springer; 1984. p. 133-139.
asiWM.iswmimiiT^ [11] Hirst S, Graham R. The Format and Presentation of Col­
riSS.4iWrafflf«l»^RS;ffl^^t lision Warnings. In: Ergonomics and safety of intelligent
fflfwt»^^igfi.^^»^®«^yaaa driver interfaces. Mahwah, USA: Lawrence Erlbaum As­
«^^^^».*^^<^Sff^Mff<^^®S sociates; 1997. p. 203-219.
m^mMmwmo® [12] Allen BL, Shin BT, Cooper P. Analysis of Traffic Con­
m&^fW.^Tffii^ftw flicts and Collisions. Transportation Research Record:
Journal of the Transportation Research Board. 1978;667:
Y(rs^>i&^>ew^)^w^^ 67-74.
^(tamwm^'&am [13] Hyden C. A Traffic-Conflicts Technique for Determining
^n^^)i^^<^<«^&.« Risk. Bulletin 15B. Lund, Sweden: Department of Traffic
^Wnom^WOfflTfl Planning and Engineering, Lund Institute of Technology;
W^lWim,I^TafflXX^xXnK^ 1977.
i^w^wsm^m^ [14] Minderhoud M, Bovy PH. Extended Time-to-collision
w^xa^w^.aim^fflx«>w^ Measures for Road Traffic Safety Assessment. Accident
ffcW$$mmo. Analysis & Prevention. 2001;33: 89-97.
[15] Chen P, Zeng W, Yu G, Wang Y. Surrogate Safety Analy­

^» sis of Pedestrian-Vehicle Conflict at Intersections Using

Unmanned Aerial Vehicle Videos. Journal of Advanced
^mt^w^w Transportation. 2017; Article ID 5202150.
^am» io [16] Kloeden C, Ponte G, Mclean J. Travelling Speed and the
Risk of Crash Involvement: The South Australian Expe­
rience. National Speed and Road Safety Conference, 23­
REFERENCES 24 August 2001, Adelaide, Australia; 2001.
[1] Bai L, Liu P, Chen Y, Zhang X, Wang W. Comparative [17] Shelby SG. Delta-V as a Measure of Traffic Conflict Se­
Analysis of the Safety Effects of Electric Bikes at Sig­ verity. The 3rd International Conference on Road Safety
nalized Intersections. Transportation Research Part D: and Simulation, 14-16 September 2011, Indianapolis,
Transport and Environment. 2013;20: 48-54. USA; 2011.
[2] Perkins SR, Harris JL. Traffic Conflict Characteristics: [18] Chen Y, Liu P, Wang W. Using Kinetic Energy to Eval­
Accident Potential at Intersections. Highway Research uate the Severity of Different Types of Traffic Conflict
Record. 1968;225: 35-43. at Signalized Intersections. The 11th International Con­
[3] Baker WT. An Evaluation of the Traffic Conflicts Tech­ ference of Chinese Transportation Professionals, 14-17
nique. The 51st Annual Meeting of the Highway Research August 2011, Nanjing, China; 2011.
Board, 17-21 January 1972, Washington, D.C.; 1972. [19] Sobhani A, Young W, Logan D, Bahrololoom S. A Kinet­
[4] Lin S, He M, Tan Y, He M. Comparison Study on Operat­ ic Energy Model of Two-Vehicle Crash Injury Severity.
ing Speeds of Electric Bicycles and Bicycles: Experience Accident Analysis & Prevention. 2011;43(3): 741-754.
from Field Investigation in Kunming, China. Transpor­ [20] Fazekas A, Hennecke F, Kallo E, Oeser M. A Novel Sur­
tation Research Record: Journal of the Transportation rogate Safety Indicator Based on Constant Initial Accel­
Research Board. 2008;2048: 52-59. eration and Reaction Time Assumption. Journal of Ad­
[5] Schepers JP, Fishman E, Hertog PD, Wolt KK, Schwab vanced Transportation. 2017; Article ID 8376572.
AL. The Safety of Electrically Assisted Bicycles [21] Alhajyaseen WKM. The Development of Conflict Index

Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320 319
 Wu Z, Zeng X, Wang L. A New Traffic Conflict Measure for Electric Bicycles at Intersections

for the Safety Assessment of Intersections Considering and Engineering, Lund Institute of Technology; 1987.
Crash Probability and Severity. Procedia Computer [25] Svensson A. A Methodfor Analysing the Traffic Process
Science. 2014;32: 364-371. in a Safety Perspective. Bulletin 166. Lund, Sweden:
[22] Chen J, Xie Z, Qian C. Traffic Conflict Models on Department of Traffic Planning and Engineering, Lund
Shared-Use Paths Used by Pedestrians, Cyclists, and Institute of Technology; 1998.
Electric Bicycle Riders. The 10th International Confer­ [26] Lowry M, McGrath R, Scruggs P, Paul D. Practitioner
ence of Chinese Transportation Professionals, 4-8 Au­ Survey and Measurement Error in Manual Bicycle and
gust 2010, Beijing, China; 2010. Pedestrian Count Programs. International Journal of
[23] Dong B-J. The Study of Characteristics of Electric Bicy­ Sustainable Transportation. 2016;10(8): 720-729.
cle. Master’s thesis. Tongji University, Shanghai; 2008. [27] Wang Y, Xing F, Zhang L. Research on the Vehicle-Bi­
[24] Hyden C. The Development ofa Methodfor Traffic Safety cycle Conflict Model at Signalized Intersection. Interna­
Evaluation: the Swedish Traffic Conflicts Technique. Bul­ tional Conference on Green Intelligent Transportation
letin 70. Lund, Sweden: Department of Traffic Planning System and Safety, 1-4 July 2016, Nanjing, China; 2016.

320 Promet - Traffic & Transportation, Vol. 32, 2020, No. 3, 309-320