Content uploaded by Sara El Hamdani
Author content
All content in this area was uploaded by Sara El Hamdani on Sep 14, 2020
Content may be subject to copyright.
Content uploaded by Sara El Hamdani
Author content
All content in this area was uploaded by Sara El Hamdani on Jun 06, 2020
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=gits20
Journal of Intelligent Transportation Systems
Technology, Planning, and Operations
ISSN: 1547-2450 (Print) 1547-2442 (Online) Journal homepage: https://www.tandfonline.com/loi/gits20
A protocol for pedestrian crossing and increased
vehicular flow in smart cities
Sara El Hamdani, Nabil Benamar & Mohmed Younis
To cite this article: Sara El Hamdani, Nabil Benamar & Mohmed Younis (2019): A protocol
for pedestrian crossing and increased vehicular flow in smart cities, Journal of Intelligent
Transportation Systems, DOI: 10.1080/15472450.2019.1683451
To link to this article: https://doi.org/10.1080/15472450.2019.1683451
Published online: 12 Nov 2019.
Submit your article to this journal
View related articles
View Crossmark data
A protocol for pedestrian crossing and increased vehicular flow
in smart cities
Sara El Hamdani
a
, Nabil Benamar
a
, and Mohmed Younis
b
a
Department of Mathematics and Computer Science, School of Technology, Moulay Ismail University of Meknes, Meknes, Morroco;
b
Department of Computer Science and Electrical Engineering, University of Maryland, Baltimore, MA, USA
ABSTRACT
With the technological drive for realizing smart cities, work on Autonomous Intersection
Management (AIM) protocols opts to replace the traditional traffic light system by using
cooperative Vehicle-to-Vehicle (V2V) communication in order to decrease road congestion
and increase vehicular throughput. However, these protocols simply ignore pedestrians as
road users and do not provision for safe pedestrian crossing. Basically, a road intersection
not only could be traffic bottleneck, but also nearly 23% of the total automotive related
fatalities and almost 1 million injury-causing crashes occur at or within intersections every
year. This article opts to fill such a technical gap. We present a novel system that prioritizes
pedestrians crossing and guarantees safety while preserving the efficiency of AIM based
approaches. Our system does not require additional infrastructure or pedestrian-carried devi-
ces, and works for both self-driving and human-derived vehicles. The simulation results
show that our proposed system for Autonomous Pedestrian Crossing (APC) protocol of non-
signalized intersections significantly decreases the vehicle’s delay and pedestrian walk dur-
ation compared to the conventional traffic light-based systems. The effectiveness of APC is
validated for traffic scenarios with (1) self-driving vehicles, and (2) mix of human-based and
self-driving vehicles.
ARTICLE HISTORY
Received 26 July 2018
Revised 18 October 2019
Accepted 18 October 2019
KEYWORDS
autonomous traffic
management; cooperative
driving; mixed traffic;
pedestrian crossing; V2V
communication
Introduction
Intelligent Transportation Systems (ITS) play a crucial
role in diminishing traffic jams, CO2 emissions, travel
delays and accident rates, while improving road safety,
traffic flow and passenger comfort. Vehicle-to-
Infrastructure (V2I) and Vehicle-to-Vehicle (V2V)
(Yang et al.) communications are considered to be the
foundations of current ITS. V2I communication ena-
bles wireless exchange of critical safety and traffic data
between vehicles and roadway infrastructure; mean-
while V2V communication utilizes short-range radio
links between vehicles to exchange information such
as motion speed and heading in order to avoid colli-
sions. Emerging techniques, such as Vehicle-to-
Pedestrian (V2P), Vehicle-to-Device (V2D) and
Vehicle-to-Grid (V2G) and Vehicle-to-Everything
(V2X), opt to grow the scope of interactions and sup-
port a broad range of vehicle types (Shladover, 2017).
The main objective of ITS is to mitigate congestion
and boost road safety. Road congestion is responsible
for significant losses of time and fuel, and thus has an
adverse effect on the economy. In 2014, road conges-
tion caused urban Americans to travel an extra 6.9
billion hours and purchase extra 3.1 billion gallons of
fuel for a congestion cost of $160 billion, which is
expected to grow to $192 billion in 2020 (Schrank
et al.). Furthermore, traffic congestion causes extra
emission of carbon dioxide, which is environmental
and health hazard. In addition, traffic congestion leads
to a high number of vehicle crashes threatening
human life and wellbeing. For example, in 2017 a total
of 1,793 people were killed and 24,831 were seriously
injured due to road accidents in the Great
Britain (Britain).
In urban settings, road intersections are considered
to be the traffic management bottleneck. Between
2011 and 2014, nearly 28% of fatal crashes in the US
occurred at intersections (Lombardi et al.). Hence,
research on intersection management in the realm of
self-driving vehicles has focused on possible elimin-
ation of traffic signals and instrumenting autonomy
among the vehicles in order to increase simultaneity
CONTACT Sara El Hamdani S.elhamdani@edu.umi.ac.ma Department of Mathematics and Computer Science, School of Technology, Moulay Ismail
University of Meknes, Meknes, Morroco.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gits.
ß2019 Taylor & Francis Group, LLC
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS
https://doi.org/10.1080/15472450.2019.1683451
of intersection crossing, which in turn will decrease
the travel delay (Bento et al., 2019). Such a non-sig-
nalized intersection concept proved its effectiveness in
increasing the intersection crossing rate for vehicles
(Namazi et al.). The majority of published work pur-
sued Vehicle to Infrastructure (V2I) and Vehicle-to-
Vehicle (V2V) communications to coordinate between
vehicles in order to avoid stopping at a non-signalized
intersection as much as possible.
However, the presence of traffic lights enables safe
crossing of pedestrians; thus supporting autonomous
vehicle passage is not sufficient justification of the
elimination of signals. In fact, the majority of new
autonomous intersection management protocols
assumed that only vehicles would use the road and
ignored the existence of other important road users,
namely, bicycles, motorcycles, and pedestrians
(Elhamdani & Benamar, 2017). Such an assumption is
not practical in the broad sense, especially in urban
setups. Therefore, despite the reduced travel time
advantage, a non-signalized intersection would not be
deemed as a suitable solution without supporting safe
pedestrian crossing. This is particularly important tak-
ing into account that pedestrians accounted for 26%
of Great Britain road fatalities according to reported
road casualties in GB in 2017 (Britain). Moreover,
according to NHTSA report, 76% of pedestrian fatal-
ities in 2016 occurred in urban areas (Brock et al.).
Previous work on pedestrian safety has focused on
improving the capabilities of self-driving vehicles for
detecting pedestrians (Franke et al.), and avoiding
them using Fuzzy Steering Controller (Fernandez
Llorca et al.). However, published solutions lack prac-
ticality since multiple pedestrians usually need to cross
the road simultaneously, and the vehicle cannot avoid
them one by one without stopping. Some techniques
rely on coordination between the vehicle and the ped-
estrian through V2P communication (Bagheri et al.).
Basically, instantaneous alarms about coming vehicles
are sent to the phones of pedestrians to alert them.
However, such solution strategy requires pedestrians
to carry smartphones, which may not always be pos-
sible. In fact, the pedestrian could be children, old
person, handicapped or simply has a smartphone with
dead battery.
Although V2V communication has been exploited
for autonomous traffic management, to the best of
our knowledge little attention has been paid to sup-
porting pedestrian crossing. Thus, we believe that this
article fills an important technical gap tackling the
shortcomings of published solutions and presenting a
complete coordination system for supporting
simultaneous passage of pedestrians and vehicles.
Fundamentally, we provision for safe pedestrian cross-
ing in autonomous traffic management system while
sustaining the efficiency of existing vehicular protocols
for the non-signalized intersections. We use V2V
communication for coordination between vehicles to
enhance the safety and to avoid unnecessary stopping.
Our proposed APC system defines a set of rules for
prioritizing pedestrian (especially vulnerable pedes-
trians) crossing of roads and guarantees safety without
the need for pedestrians to carry devices such as
smartphones. APC can be effectively applied for traffic
involving only self-driving vehicles and for cases
where there is a mix of human-driven and self-driven
vehicles. We have validated our system using the
SUMO simulator (Krajzewicz et al.). The simulation
results show that our APC system outperforms the
conventional traffic light model and significantly
decreases the vehicle’s time loss and pedestrian walk
duration due to pedestrian crossing. The effectiveness
of APC is also confirmed for cases when both self-
driving and human-driven vehicles are involved.
The remainder of this article is organized as fol-
lows. “Related work”section goes over related work.
The system model and approach overview are pro-
vided in the “System model and approach overview”
section. The APC protocol is described in detail in
both “Pedestrian collision detection”and “Pedestrian
crossing protocol”sections, covering pedestrian colli-
sion detection, and the pedestrian crossing protocol,
respectively. “APC in mixed traffic scenario”section
discusses the applicability of APC in mixed traffic
scenario. The safety and practicality of APC are dis-
cussed in “Protocol safety and practicality”section,
while “Performance evaluation”section reports the
performance validation results. Finally, “Conclusion
and future work”section concludes the article and
hints some directions for further research.
Related work
Vehicle collision avoidance at intersections
Contemporary Traffic Light Controller (TLC) based
intersection management mechanisms may not cope
with the high number of vehicles aiming to cross the
intersection. Thus, many researchers have focused on
improving the TLC operation by using sensors, as
well as V2V (Shladover, 2017) and V2I communica-
tion (Li & Shimamoto, 2011; Gradinescu, Gorgorin,
Diaconescu, Cristea, & Iftode, 2007; Murphy, Djahel,
Jabeur, Barrett, & Murphy, 2015). A TLC collects data
from vehicles and neighboring TLCs in order to
2 S. E. HAMDANI ET AL.
determine the optimal green light phase time and/or
to reroute vehicles to less congested roads based on
information gathered using Infrastructure to
Infrastructure (I2I) communication (Wang, Djahel, &
McManis, 2014). These advanced TLC systems have
been shown to outperform the traditional ones, yet
they require expensive infrastructure upgrade (Li &
Shimamoto, 2011; Murphy et al., 2015).
The abovementioned studies opt to sustain auton-
omy at the level of TLC operation by providing data
to enhance decision making. Some work eliminated
TLCs altogether. For example, the approaches of
(Dresner & Stone, 2008) and (Li, Chitturi, Yu, Bill, &
Noyce, 2015) apply a centralized strategy where a con-
troller manages the safe passage of vehicles in a non-
signalized intersection. In this case vehicles send
requests for crossing the intersection to a controller,
which grants permissions on a first-come-first-served
basis. However, the response time and scalability are
the major issues of a centralized controller and could
hinder the vehicle’s ability for reacting in real time.
On the other hand, the decentralized strategy
(Azimi, Bhatia, Rajkumar, & Mudalige, 2015) relies on
vehicles to manage the intersection autonomously
using V2V communication. The vehicles share infor-
mation, e.g., speed, position, and destination, with the
aim of detecting and avoiding collision by making
timely decisions for accelerating and decelerating.
Accordingly, autonomous (self-driving) vehicles can
cross the intersection without need for stopping,
which significantly reduces the congestion rate at the
intersection. Nevertheless, these intersection manage-
ment systems assume that autonomous vehicles are
the only road user and ignore the pedestrian’s need of
crossing the road.
Autonomous vehicle-pedestrian
collision avoidance
The incorporation of pedestrian detection technologies
is a fundamental requirement for autonomous vehicles
(Dollar, Wojek, Schiele, & Perona, 2012). An example
of these technologies is the monocular pedestrian
detection based on an intelligent real-time vision sys-
tem, e.g., zebra crossing detection (Brandstaetter,
Yannis, Evgenikos, & Papantoniou, 2011). However,
assuming that autonomous vehicles are able to detect
and stop for pedestrians would not address the con-
cern on vehicular flow and very much depends on
pedestrian behavior; this is particular a major concern
in crowded urban areas such as Manhattan in New
York City. To highlight how pedestrian behavior
could be influential, pedestrian crossing in the pres-
ence of driverless vehicle has been studies by
(Rodr
ıguez Palmeiro et al., 2018). They opt to capture
the pedestrian reaction during road crossing in the
presence of autonomous vehicles with dummies seated
in the driver seat. The study has found out that
pedestrians are stressed and do not attempt to cross
the road without establishing an eye-contact with
human driver.
The importance of pedestrian-driver gestural com-
munication was a motivation to involve interaction
between pedestrians and vehicles. To that end, the
approach of (Florentine, Ang, Pendleton, Andersen, &
Ang, 2016) consists of two methods of pedestrian
notification: turning on a LED strip when a close
obstacle is detected and broadcasting an audio
notification of vehicle intention.
The aforementioned techniques just focus on detec-
tion and stopping for pedestrians. An alternative
methodology is to schedule pedestrian crossing so that
the vehicular flow is minimally impacted. Earlier work
on pedestrian crossing scheduling relied on the TLC
of signalized intersections. For example, (Skikos,
Machia, & Christopoulos, 1993) aims at improving
the TLC so that it becomes able to emulate the deci-
sion process of an experienced crossing guard.
Nevertheless, TLC-based methodologies are deemed
inefficient for autonomous vehicles since it requires
signalized intersections and still cause frequent stop-
ping. Employing vehicle-centric pedestrian collision
avoidance schemes is deemed more appropriate for
supporting safe crossing while limiting the impact on
vehicular traffic.
The approach of (Franke et al., 1998) is an example
of vehicle-based solutions for supporting autonomous
pedestrian collision avoidance, where a self-driving
vehicle is assumed to be able to detect pedestrians on
its travel path. A fuzzy steering scheme is employed,
where the vehicle changes the lane and returns back
to the original lane after passing pedestrians.
Nonetheless, this approach considers a pedestrian as
an obstacle that should be avoided on the road and
not as a road user that has a right for safe crossing.
In other words, a pedestrian has no privilege on the
road. In addition, the approach could be chaotic since
changing a lane could limit a vehicle’s ability in
detecting a pedestrian in the new lane in a
timely manner.
Meanwhile, the approach of (Bagheri, Siekkinen, &
Nurminen, 2014) is based on cellular communication
between the vehicle and the pedestrian smartphones.
The mobile phones exchange awareness messages
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 3
periodically containing information such as speed and
position. Thus, a vehicle will receive an alert from
pedestrian smartphone. One of the concerns about
this approach is the impact on battery life and pedes-
trian position accuracy. (Bachmann, Morold, & David,
2017) opted to improve pedestrian position accuracy
and minimizing battery consumption using pedestrian
context information such as speed, location and direc-
tion. Nonetheless, this approach is not practical since
a pedestrian cannot be deprived from safe crossing a
road if not holding a smartphone. Unlike published
pedestrian collision avoidance schemes, our APC sys-
tem does not require any infrastructure or pedes-
trian devices.
System model and approach overview
Design goal
Our proposed APC system opts to manage safe pas-
sage of pedestrians on a road crossing while reducing
the vehicle’s travel delay and thus decreasing conges-
tion level mainly is urban highways. APC consists of
two modules, namely, pedestrian crossing detection
and vehicle motion coordination. The following sum-
marizes the features provided by our system:
APC does not require additional infrastructure for
the road.
APC supports autonomous intersection manage-
ment with no need of vehicle stopping
at junctions.
Contrarily to conventional AIM systems (Azimi
et al., 2015), APC considers a pedestrian as a road
user and provisions for safe pedestrian crossing.
APC privileges the pedestrian over other road
users and factors in variations in human’s
motion skills.
APC leverages on-vehicle sensors to detect pedes-
trians and avoid collision.
No devices, smartphone applications or V2P com-
munication are required for a pedestrian to cross
the road.
V2V communication is exploited to enable inter-
vehicle coordination and minimize vehicle stopping
at pedestrian crossing.
Proposed pedestrian crossing model
The conventional pedestrian crossing model relies on
the use of TLC, as illustrated in Figure 1(a).
According to this model, pedestrian crossing is
allowed at the end of each road segment. As
aforementioned, AIM systems outperform the TLC-
based management model because a vehicle avoids
stopping at the intersection, which wastes time not
only in waiting but also due to deceleration and accel-
eration. Nonetheless, current AIM systems support
non-signalized intersections without provision for
pedestrian crossing. To address this shortcoming, we
enable the pedestrian crossing in the middle of each
road, as illustrated in Figure 1(b). Hence, pedestrian’s
road crossing will be decoupled from intersections
and the efficiency of the AIM system’s handling of
vehicles at the intersection is sustained.
In addition, a refuge island will be added between
every pair of consecutive lanes. To accommodate the
addition of refuge islands, the shoulder of the road
will be narrower, or even eliminated. If necessary, a
lane may become narrower at the pedestrian crossing
subject to the minimum width standard (Dixon et al.,
2015; Karim, 2015).
Each pedestrian aiming to cross the road has to
walk on the pathway following these rules:
A pedestrian crosses the road on the pathway lane
by lane.
If any vehicle does not occupy the pathway of a
lane, the pedestrian can cross such a lane.
Otherwise, the pedestrian will wait safely at the ref-
uge island.
While in the middle of a crossing lane, a pedes-
trian has priority and walks freely without stopping
for coming vehicles.
System model and assumptions
Our proposed system makes the following assump-
tions about the road configuration, pedestrians, and
vehicle’s capabilities:
Road: We assume bidirectional road segments.
Traffic in each direction can be over one or mul-
tiple lanes, as illustrated in Figure 2. A vehicle con-
siders some safety distance away from the crossing
to stop at, if a pedestrian is walking through its
lane; hence, we assume that a line is drawn on the
road to mark such a safety distance. Lanes are sep-
arated at the crossing with refuge islands where
pedestrian can stand on before crossing next lane.
Pedestrian: The typical walking speed for a healthy
adult is about 1.3 m/s (Mohler, Thompson, Creem-
Regehr, Pick, & Warren, 2007). However, a pedes-
trian can also be an elder, child, or handicapped,
holding a pet, or pushing a stroller. Therefore, in
4 S. E. HAMDANI ET AL.
our APC system we consider that pedestrian cross-
ing speed is variable. We do not force the pedes-
trian to accelerate while crossing the road.
Moreover, we do not require a pedestrian to carry
a smartphone or any other device.
Vehicles: We assume that vehicles are equipped
with front cameras and other sensors to detect
pedestrians, recognize pedestrian crossing, and
safety line marks before arriving at them. Likewise,
we assume that a vehicle has access to a digital
map database and is equipped with a Global
Positioning System (GPS) receiver. Moreover, a
vehicle is to have sufficient computation resources
to process the collected data and make decision
on whether to stop or move in real time.
Furthermore, the vehicle can interact with other
vehicles using V2V communication and can share/
receive information, i.e., position, speed, decision
(cross, accelerate, or decelerate) and actual status
(stopping, or moving). In APC, a vehicle takes into
account only the part of the crossing that belongs
to the lane on which it is traveling.
Communication: We assume that vehicles can com-
municate and broadcast Cooperative Awareness
Messages (CAMs) (ETSI, 2011) to vehicles in their
vicinity. To do so, vehicles are to be equipped with
wireless technologies, e.g., dedicated short-range
communication (DSRC) or Wireless Access in
Vehicular Environments (WAVE) (SAE, 2009).
Pedestrian collision detection
In our APC system, vehicles use the front camera to
detect pedestrians at the designated crossing area
Figure 1. Illustration of pedestrian crossing models: (a) Four-way signalized intersection managed by a TLC. Traditional pedestrian
crossings are surrounding the intersection; (b) The crossing model of the proposed APC system. The intersection is non-signalized,
and pedestrian crossings are located in the middle of each road segment and their number are halved to decrease vehicles flow
on the road and mainly at the intersections areas.
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 5
before arriving at it. A vehicle needs to only detect
pedestrians in the crossing area of the lane where it is
currently traveling. On the other hand, we assume
that a pedestrian detects an approaching vehicle in the
lane before stepping inside such a lane; this is more
or less an intuitive precaution that is contemporarily
followed. Contrariwise, when in the middle of the
lane, the pedestrian does not take into account any
approaching vehicles.
The crossing is a critical area where the safety of
pedestrians could be risked. Since a vehicular hit of a
pedestrian could be severe and may cause injuries or
even fatalities, we add a safety zone before the pedes-
trian crossing area. The width of such safety zone
(SZ) is based on the distance a vehicle will travel
after activating the brake until it completely stops.
To calculate SZ, first we consider the kinetic
energy E:
Figure 2. Illustrations of APC road model for a sample configuration with two directions: (a) One lane per direction. (b) the case
of three lanes per direction. The crossing is provisioned in the middle of the segment. A Safety Line (SL) is drawn in each road dir-
ection to mark the Safety Zone (SZ) for coming vehicles. The crossings are provided with Refuge Islands (RI) to allow pedestrian to
stand on them safely.
6 S. E. HAMDANI ET AL.
E¼1=2mv2(1)
where mis the mass of the vehicle (kg), and vis the
speed at the start of braking (m/s). On the other
hand, the work Wdone by braking is determined by:
W¼lmgd (2)
where lis the coefficient of friction (unit less), gis
acceleration due to gravity (m/s
2
), and dis the trav-
eled distance (m). The braking distance given in an
initial driving speed vis found by putting W¼E,
from which the width of the safety zone is derived as
follows:
lmgd ¼1=2mv2(3)
SZ ¼d¼v2
2lg(4)
Since the vehicle’s speed could significantly vary in
practice, we consider the speed limit to calculate the
road safety zone. The coefficient of friction l, also
changes according to the road condition, i.e., icy, wet
or normal. Therefore, this coefficient should be calcu-
lated based on the typical conditions of the concerned
road. The value of acceleration due to gravity is fixed
and equal to 9.80 (m/s
2
).
In order to avoid collision, the SZ is delineated on
the road by the Safety Lane (Figure 2) as a landmark
for both vehicles and pedestrians. If a coming vehicle
V
i
decide to stop it will stop at the safety line and not
at the crossing. Therefore, V
i
has enough space and
time to stop before the crossing in critical circumstan-
ces where a stopping decision is taken late or a cross-
ing decision is taken simultaneously from the vehicle
and pedestrian. For instance, Figure 3 illustrates the
case when both a pedestrian and a vehicle detect sim-
ultaneously that the collision region of the crossing is
free and then decide to cross. Therefore, the vehicle
cannot stop at the safety line because it detects the
pedestrian on the crossing lane after arriving to the
line. In such situation, the vehicle applies an emer-
gency brake and stops in the area between the cross-
ing and the safety line and guarantees a safe crossing
for the pedestrian.
APC requires a pedestrian to cross the roads in
strides, lane by lane. As outlined in pedestrian cross-
ing process diagram in Figure 4, a pedestrian waits at
the refuge island between two consecutive lanes and
proceeds if it is safe. Such safety assessment is made
visually, e.g., seeing no approaching vehicle close to
the safety Line (SL).
For vehicles, we define the collision region of a
lane Li as the part of the pedestrian Crossing D
k
–CL
i
in Li combined with the safety zone D
k
–SZ
i
:(D
k
–SZ
i
þD
k
–CL
i
) which is delineated by SL. In the example
in Figure 5, the collision region for L
2
of direction D
1
is shaded (D
1
–SZ
2
þD
1
–CL
2
). Thus, a vehicle travel-
ing on D
1
–L
2
will stop only if a pedestrian exists in
the collision region of the current lane. When yielding
to a pedestrian is deemed necessary, a vehicle slows
down to stop and broadcasts a CAM to alert the
vehicles that follow on the same lane. The CAM con-
tains the vehicle’s ID, direction, the lane number,
vehicle-speed, vehicle GPS coordinates, and the
planned action. When the front vehicle announces its
stop, all CAM recipients check their position rela-
tively, e.g., GPS coordinates, the lane number and
travel direction, and decide on their response
accordingly.
It is worth noting the effect that a pedestrian cross-
ing may have on vehicular collision. A recent study
(Arvin, Kamrani, & Khattak, 2019) has pointed out
that the probability of rear-end crashes could be
impacted by the intersection geometry and traffic
density at the intersection. Specifically, lateral acceler-
ation volatility may contribute to increased rear-end
collision at intersections. However, the pedestrian
crossing proposed in this article is placed in the mid-
dle of a road segment and involve refuge islands;
therefore, lateral acceleration volatility is not an issue
and only longitudinal control of the vehicle is
required. Moreover, advanced systems based on
Variable Speed Limit (VSL) control (Wu, Abdel-Aty,
Wang, & Rahman, 2019) could be further employed
to mitigate rear-end collisions.
Pedestrian crossing protocol
Detailed APC protocol
APC opts to improve the vehicular throughput and
travel time while supporting safe pedestrian crossing.
The main idea is to determine when a vehicle should
Figure 3. An illustration of emergency braking at the safety
zone due to late stopping decision. The vehicle decides to
cross after arriving to the safety line and stop before the cross-
ing safety.
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 7
stop, slow down, or continue traveling when
approaching a pedestrian crossing. APC is lane-based
and relies on the vehicle’s on-board cameras (and/or
other sensing device on board a vehicle) to detect
existing pedestrians in its current lane. Figure 6 shows
a flow diagram summary of the steps. First, the
approaching vehicle detects the status of the con-
cerned Crossing Lane (D
k
–SZ
i
þD
k
–CL
i
) using the
front camera. If there is no pedestrian, the vehicle
proceeds and crosses the pathway. Otherwise, i.e., the
lane-crossing is occupied by pedestrians, the vehicle
tries to avoid stopping by decelerating while
approaching the Safety Line. If the crossing-lane is
still busy, the vehicle brakes at the Safety Line and
stays waiting as long as there are pedestrians in this
crossing-lane.
Accordingly, the approaching vehicle broadcasts a
CAM for every decision it takes; either it is stopping
or crossing. Generally, a vehicle will need to send and
receive two types of CAM:
Passing-Message: this is for a vehicle V
x
to inform
its neighbors that it is entering the collision region
in its lane without slowing down.
Stopping-Message: this CAM is to announce that
vehicle V
x
will stop at the safety line.
When the next coming vehicle V
y
in the same lane
receives the CAM of V
x
,V
y
reacts based on the pro-
cedure in Figure 7. Accordingly, V
y
reads CAM and
checks first if V
x
is traveling in the same Direction
and Lane (if it is not V
x
’s CAMs do not concern V
y
).
Then, V
y
checks if it is very close to V
x
, which means
that the distance between the two vehicles is equal to
the Safety Distance (SD). SD is the minimum gap
required between two consecutive vehicles and it is
calculated based on the same formula described in the
previous section. Thus, if V
x
and V
y
are not close
enough; V
y
ignores V
x
’s actions; hence, it decelerates
and execute the first algorithm. In the other case, V
y
takes the decision based on V
x
’s action and does not
Figure 4. Flow diagram of pedestrian crossing procedure. A pedestrian crosses the road in steps according to the number of lanes.
A stop is made at the Refuge Island (RI) between every pair of lanes to check whether the next Crossing Lane (CL) and Safety
Zone (SZ) area is free before crossing the Lane (L).
Figure 5. Illustration of pedestrian-vehicle collision region on each lane. The first direction of the road (D
1
) is composed of three
lines (L
1
,L
2
and L
3
). The collision region in each lane includes the Crossing-Lane (CL) plus the Safety Zone (SZ); for instance,
(D
1
–SZ
1
þD
1
–CL
1
) represents the collision region of the first line of this direction. The collision regions for the other lanes are
marked in the same way.
8 S. E. HAMDANI ET AL.
Figure 6. Flow diagram of the APC protocol execution at a vehicle Vi while approach a pedestrian crossing. The approaching
vehicle detects the crossing state using Front Camera (FC) and check the collision region (Safety Zone þCrossing Lane), takes
accordingly a decision of crossing or decelerate in the aim of avoiding stopping, and informs the following vehicles by broadcast-
ing a CAM message.
Figure 7. Flow diagram of next coming vehicle V
iþ1
process. The next coming vehicle takes its decision (crossing, decelerating or
stopping) based on the approaching vehicle CAM and based on the distance between the two vehicles.
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 9
need to check the collision region using its own cam-
era. Accordingly, V
y
crosses after V
x
and stops if it is
stopping that it can decrease the delay of passing the
crossing. V
y
; in its turn, broadcasts a CAM periodic-
ally for each action it takes to aware other com-
ing vehicles.
Figure 8 illustrates APC operation for all possible
scenarios. To simplify the figure reading, we assume
that all pedestrians are crossing in one direction
(from D
1
to D
2
). In the first lane, L
1
, of direction D
1
,
vehicle V
1
is still far from the safety line. When
detecting the pedestrian P
1
,V
1
decelerates while
approaching to avoid stopping and broadcasts a CAM
to announce such deceleration. In lane L
2
of the same
direction, V
3
has to stop and stays still until pedes-
trian P
2
crosses the crossing pathway of the lane. V
3
broadcasts a “Stopping Message”; hence, V
2
will stop
as well since the two vehicles are close to each other.
On the other hand, V
4
will continue and pass through
the pedestrian crossing since lane L
3
is free of
pedestrians.
Likewise, in direction D
2
, vehicle V
5
crosses and
does not wait for pedestrian P
3
because P
3
is stand-
ing on the refuge island RI
4
.V
5
broadcasts a crossing
message; yet V
7
, which follows V
5
on the same lane,
will not take any action because it is relatively far. P
3
will eventually cross before V
7
arrives at the crossing.
Meanwhile, vehicle V
6
is decelerating to avoid stop-
ping while pedestrian P
4
is crossing. On the other
hand, P
5
will cross immediately the last lane and no
action is needed because there are no com-
ing vehicles.
Pedestrian freedom from blocking situation
Pedestrians could wait for long time at a refuge island
if the next lane is experiencing high vehicular traffic.
According to APC, each vehicle takes the action of
the one ahead if they are close to each other.
Therefore, if we assume that a lane is congested or
slow for any reason, i.e., due to turn left down the
road, vehicles will follow each other, and the pedes-
trians will be queuing as shown in Figure 9 and even-
tually blocking the other lanes one after the other
when they cannot find a room to stand on the Refuge
Island. Thus, the crossing-pathway will be in a block-
ing situation.
According to the manual on Uniform Traffic
Control Devices (UTCD) of U.S. Federal Highway
Administration, the width of a crossing shall be not
less than 3 m (The Federal Highway Administration
(FHWA), 2010). On the other hand, the average
of men’s shoulder width is about 0.465 m ’0.5 m
(Pamela Buxton, 2015). Thus, about six pedestrians
(3 m 0.5) can stand. If we consider the presence of a
gap between each two pedestrians, five could be the
maximum number of pedestrians that could wait at a
refuge island. Accordingly, if the number of waiting
pedestrians exceeds five, they will be queuing in the
middle of a lane. Since pedestrians queuing should
not block the pathway and consequently the lane,
APC imposes the following rule (1):
Rule (1): If an AV
i
detects more than five waiting
pedestrians (wP) in its lane, it cannot cross the
Figure 8. An illustration of different scenarios of pedestrians
on the crossing, along with various positions of vehicles that
are willing to pass the crossing.
Figure 9. An illustration of pedestrians causing a blocking situ-
ation due to a congested Lane. Pedestrians are standing on
Refuge Island (RI) and waiting to cross the congested Lane (L).
Pedestrians that are willing to cross (D1–CL3) and cannot find
a room to stand on Refuge Island will stand on the collision
region (D1–CL2) and thus cause a blocking situation for
vehicles traveling on D1–L2.
10 S. E. HAMDANI ET AL.
pathway and has to stop and wait at the SL until the
pedestrians cross.
Pdetected wP >5)Action ¼“Stop”
Correspondingly, we add rule (2) to the pedes-
trian process:
Rule (2): If the waiting pedestrians are more than
five, they can cross the lane without waiting.
This rule ensures pedestrians privilege in crossing
the road while enabling vehicular flow. Pedestrians
will not pileup as the rule controls the flow pedes-
trians and prevents them from blocking any lane
while trying to cross the next. In addition, pedestrians
will not wait indefinitely if the road is congested. It is
important to note that an autonomous vehicle is typ-
ically capable of monocular vision, is thus able to
count pedestrians at both side of its lane
(Mitzel, 2013).
APC protocol takes on consideration vulnerable
pedestrians that are not capable to stand on RI for a
long while such as pedestrian with a baby, pedestrian
carrying a pet or pedestrian using a wheelchair.
Accordingly, APC protocol ensures that such vulner-
able pedestrians cross immediately without waiting on
refuge islands. Furthermore, the protocol guarantees
that strollers or wheelchairs would not congest refuge
islands. Thus, APC imposes rule (3) for AVs:
Rule (3): If a coming AV
i
detects any object on the
two neighboring RIs that is distinctive from pedestrian
shape, it has to stop at SL until the RI become free.
Respectively, we add rule (4) to the pedes-
trian process:
Rule (4): If the pedestrian is using a wheelchair,
pushing a stroller or carrying a pet, they can cross the
lane without waiting.
APC in mixed traffic scenario
Autonomous vehicles are seen as the future of ITS,
and is envision to solve many pressing issues related
to road safety and traffic congestion. APC is designed
mainly for a Connected and Autonomous Vehicles
(CAV) system where each vehicle is equipped with
computational resources and can cooperate based on
V2V communication. However, transitioning to full
CAV system is expected to be gradual and AVs will
have to share the road with human-driven vehicles
(Olia, Razavi, Abdulhai, & Abdelgawad, 2018).
Therefore, cooperative traffic management systems,
such as APC, should be applicable to traffic scenario
of mixed autonomous and human-driven vehicles. In
this section, we discuss the possibility to apply APC
protocol in the special scenario where the system is
not 100% CAV and we discuss the limitations of
human-driver based vehicles in this context.
Supporting traditional vehicle in APC
APC relies on the on board sensors and autonomy of
AVs. Nonetheless, human driver is naturally able to
sense the surrounding environment, to detect possible
obstacles and to have quick reaction. Furthermore, the
APC algorithms are designed based on what naturally
a driver would do to avoid stopping in non-signalized
roads without hurting a pedestrian. Thus, a human-
driven vehicle is able to apply the majority of the
instructions of APC protocol except sending and
receiving messages, which make our solution more
flexible to cover mixed traffic situations. We note that
in the context of our solution CAMs are mainly
exchanged to reduce traffic delay and minimize stop-
ping rather than exchanging safety and security warn-
ings. The following discusses the two distinct
scenarios related to how a Human-driven Vehicle
(HV
i
) affects the operation of APC.
Approaching a pedestrian crossing
In the case where APC is applied in mixed traffic
scenario, the human driver of an approaching would
visually sense the collision region related to the cur-
rent lane (SZ
i
þCL
i
). Upon detecting pedestrians inside
(SZ
i
þCL
i
), the driver will decelerate and prepare to
stop at the safety line. Otherwise, the driver continues
on and avoids stopping. The driver of HV
i
applies
rule (2) when seeing a neighboring refuge island fully
occupied with pedestrians, and rule (3) in the pres-
ence of strollers or wheelchairs. In these two cases,
the driver stops immediately despite of the fact that
no pedestrian is present inside (SZ
i
þCL
i
).
Nonetheless, HV
i
may not broadcast CAMs informing
about the taken decision unless such a feature is sup-
ported. We note that a driver is expected to abide by
APC rules as part of the traffic regulation, where vio-
lation will be subject to penalties.
Effect on successors on a lane
When HV
i
follows any vehicle, whether autonomous
or human-driven, no APC rule would specifically
apply since until it is the turn for HV
i
to pass the
pedestrian crossing. The immediate successor of HV
i
on the lane may be either a human-driven vehicle
HV
iþ1
or an autonomous one HV
iþ1
. Regardless the
type, such a successor will not receive CAM broad-
casted by HV
i
, unless HV
i
is equipped with such tech-
nology. In case of HV
iþ1
, its driver will see HV
i
and
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 11
will react according what HVi does, e.g., deaccelerate,
stop, etc. analyze action taken by it. Human driver
is able to know clearly if the front vehicle is stopping
or moving. Accordingly, when HV
iþ1
and HVi are
very close to each other (d(HV
iþ1
,HV
i
)¼SD), HV
iþ1
keeps going or stops depending on HV
i
visually recog-
nized action.
If HV
iþ1
is not close enough to HV
i
(d(HV
iþ1
,
HV
i
)>SD), HV
iþ1
slows down and approaches to
(SZ
i
þCL
i
) and applies Algorithm (1) as an approach-
ing normal vehicle HV
i
. As for HV
i
, driver of HV
iþ1
is not required to apply instructions related to
receiving and broadcasting CAMs.
Safety distance for mixed traffic
Human eye especially at night can detects obstacles
only at a distance of 75 m compared to AV sensors
that can detects up to 250 m afar using long-range
radar (Yan, 2016) which gives less time to human
driver to react. Moreover, based on the observation of
human driver behavior, 85% of drivers need up to
2.5 s as perception-reaction time (Layton & Dixon,
2012). Thus, safety distance should be increased
in a mixed traffic to ensure safety of passengers in
normal vehicles.
Furthermore, Safety Sight Distance (SSD) is the
minimum distance available on a highway at any spot
having sufficient length to enable the human driver
to detect an obstacle, react, e.g., stop, safely without
collision (Layton & Dixon, 2012). In essence, SSD is
the sum of the Lag Distance (LD) and the Braking
Distance (BD) as determined by:
SSD ¼LD þBD (5)
where LD is the distance the vehicle travels during
the reaction time t and is given by (6), where vis the
velocity in m=s
2
.
LD ¼vt (6)
BD can be determined using (4), similar to
the safety distance discussed in Section 4. Thus, SSD
is determined by (7), where lis coefficient of
friction (unite less), and g is acceleration due to
gravity (m/s
2
):
SSD ¼vtþv2
2lg(7)
Since SSD is bigger than SD (SSD >SD þvt) and
in order to ensure safety, we increase the safety
distance in mixed traffic for both AVs and human
driver-based vehicles. Thus, the Safety Line (SL) is
drawn based on SSD where both kinds of vehicles are
supposed to stop.
Positives and limitations
Under APC, a HV can navigate safely in a mixed traf-
fic with AVs and pass the pedestrian crossing without
colliding with pedestrians, since a human driver is
able to detect vehicles and pedestrians visually, analyze
situations, and react accordingly. Furthermore, a
human driver can recognize full refuge island, strol-
lers, wheelchairs and other obstacles better than
AVs systems.
Accordingly, normal vehicle is able to simplify the
operation of the APC protocol. Moreover, like AVs, a
human-based vehicle is not obliged to stop at the
crossing as long as its lane is free, even if there are
pedestrians walking in other parts of the crossing.
However, an HV does not apply instructions
related to sending receiving CAMs. Consequently, it is
not able to communicate with other vehicles and can-
not detect the state of the collision zone when it is
not in the front. Such a limitation could slightly
extend the crossing time as a human driver would be
more cautious.
Protocol safety and practicality
In this section, we analyze how APC will sustain its
design goal under non-ideal scenarios, namely, when
there is a failure in the vehicle and when pedestrians
do not comply with the traffic rules mandated by
APC. The scenarios could be experienced in practice.
Pedestrian safety
Safety is the biggest issue related to ITS and
autonomous vehicles, since an error in the system
could affect human wellbeing or even life (ElHamdani
& Benamar, 2018). The presentation so far assumes
that autonomous vehicles work perfectly and
pedestrians follow the crossing process. However, we
explain in the following how APC remains safe even
in the absence of these assumptions:
Communication Fault: APC relies on V2V commu-
nication in order to provide vehicles with the
necessary information to be able to adjust speed
and to avoid stopping. Actually, V2V is fundamen-
tally required to realize cooperative driving among
autonomous vehicles and not just for applying
APC. Although communication failure could affect
12 S. E. HAMDANI ET AL.
the efficiency of the cooperative driving, it does
not have any impact on pedestrian safety. It is
worth noting that a self-driving vehicle is equipped
with sensors, e.g., Lidar, to detect possible
collisions; therefore, communication failure can be
mitigated safely by vehicles.
Faulty Camera: APC relies on the vehicle’s ability
to detect pedestrians on their way and determine
their proximity. The front camera on a self-driving
vehicle is typically able to detect obstacles up to 80
m ahead, which allows the vehicle to calculate
the deceleration rate to avoid hitting a pedestrian.
A vehicle is often equipped with sensors to com-
plement and mitigate camera failure. For example,
a vehicle will be able to detect pedestrians by
Thermal Infrared sensor and Lidar (Ger
onimo,
L
opez, Sappa, & Graf, 2010). Thus, pedestrian
safety will not be affected.
Standing in the middle of road: In APC, pedestrians
cross the road lane by lane; thus a pedestrian
would need sometimes to stand within the road.
APC advocates the addition of refuge islands,
which is popular in residential areas and around
schools. They are meant not only to provide a safe
waiting spot for pedestrians but also motivate
vehicles to slow down. The pictures in Figure 10
show an example of a refuge island on an
existing road in Ellicott City, Maryland USA.
Installing islands on roads has been demonstrated
to decrease the pedestrian crashes and casualties
rate by 57 to 82% (U.S. Department of
Transportation, n.d.).
Pedestrian Misbehavior: One of the possible safety
issues is pedestrian’s compliance with the traffic
regulations. This issue is of concern even when
traffic lights are present. Thus, good pedestrian
Figure 10. Pictures (a) and (b) of real examples of Refuge Islands “B”installed on Pedestrian Crossing “A”. Pictures are for
a crossing in a road in Ellicott City, Maryland, USA.
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 13
behavior is not guaranteed in our system. We
argue though that we are not causing increased
risk of casualties. Fundamentally, our proposed
pedestrian crossing process matches the natural
reflex of humans. A recent study of pedestrian
behavior while crossing the street (Rasouli,
Kotseruba, & Tsotsos, 2018) has shown that pedes-
trians trend to move their head (checking the
road) even in the presence of traffic lights.
Furthermore, pedestrians are more cautious in
busy roads and in wider streets. This study has
also concluded that pedestrians anticipate the
vehicle action based on its speed and on the traffic
condition. Moreover, the insertion of refuge islands
will lower the risk since a pedestrian will have a
safe waiting area after a short walk across a lane.
Practicality
As discussed, APC relies on the cooperation of
autonomous vehicles and require limited additional
infrastructure support. This subsection argues the
practicality of APC’s assumptions:
Road Architecture: Our proposed pedestrian cross-
ing model can be applied to both new and existing
roads. For existing roads, small changes are
required by redrawing crossing lines in the middle
of the road and adding refuge islands between the
lanes. It is worth mentioning that our model ena-
bles the designation of multiple crossing areas and
not necessarily in the middle of road segments;
this particularly useful for large blocks and will be
advantageous for pedestrians who otherwise have
to walk all the way to the intersection. It is import-
ant to note that adopting autonomous vehicles will
eliminate traffic lights (Badger, 2015); the changes
required by APC can be viewed as part of the
adjustment in the road infrastructure to support
such vehicular autonomy.
Road Width: Shifting lanes at pedestrian crossing is
a very popular safety mechanism. The rationale is
to force drivers to slow down. Figure 10 shows a
picture of one of the crossing in the US, where the
shoulder becomes narrower to make room for a
refuge island in the middle of the road. Although
the boundaries of a lane shift at the crossing, the
lane does not necessarily become narrower and
continues to be consistent with the standard width.
For self-driving vehicles, we do not envision even
any effect of lane shifting on the vehicle speed
given the autonomous control of the vehicle
motion. In other words, there is no reason for a
self-driving vehicle to slow down if no pedestrians
are present at the crossing. Our simulation results
confirm that APC does not diminish the through-
put when pedestrian traffic is very low, e.g., during
late night hours.
Pedestrians Convenience: As pointed out earlier,
the pedestrian crossing in APC matches natural
human behavior (Rasouli et al., 2018); a pedestrian
will cross as an opportunity arises and is not
obliged to wait for a sequence of traffic light
phases in the conventional TLC system. The lane-
by-lane progress will facilitate crossing wide and
busy roads. Moreover, according to the US Federal
Highway Administration (U.S. Departement of
Transportation, n.d.), refuge islands permit pedes-
trians to be concerned with only one direction of
traffic at a time and reduce exposure time on the
vehicle travel path. Furthermore, APC system pri-
oritizes vulnerable pedestrians with a stroller a pet
or a wheelchair and allows them to cross the road
immediately.
Performance evaluation
We have validated the performance of APC through
simulation. This section discussed the simulation
setup, performance metrics and the obtained results.
Simulation tools
We have implemented APC using the open source
traffic simulator SUMO (Simulation of Urban
Mobility). SUMO is one the most widely used stand-
ard simulation platforms that are used for purely
microscopic modeling whereby each vehicle is mod-
eled explicitly and moves individually through the
network. We built our road network within XML files
and we implemented APC within Python scripts using
Traffic Control Interface (TraCI) API. To provide
access to SUMO (acting as a server), TraCI uses a
TCP based client/server architecture.
As aforementioned, APC protocol is mainly dedi-
cated for urban highways with low pedestrian density.
Thus, we consider a simulation area composed of one
road of two directions. In addition, we deactivated
randomization for vehicles speed since in this context
we rely on speed as a factor of congestion. Thus, we
have set the maximum allowed speed for vehicles on
the road to 45 MPH, i.e., 20.11 m/s, which is consist-
ent with contemporary urban environments, so the
vehicle travel with this speed unless it need to slow
14 S. E. HAMDANI ET AL.
down for a specific reason. Likewise, we define pedes-
trian walking speed as 1.39 m/s (Mohler et al., 2007).
Based on the maximum allowed speed (45 MPH)
and the coefficient of friction of 0.8 on a dry asphalt,
the safety distance (SD) is set to 20 m. For the third
set of experiment we set a safety distance equal to
Stopping Sight Distance (SSD) of 70 m. SSD is the
sum of Braking Distance which is 20 m, and Lag
Distance 50 m which is calculated as follows in (8):
SSD ¼BD þLD;
BD ¼SDfSD 20 m for v ¼80 Km=hg;
LD ¼vtfv¼80 Km=h;t is perception time 2:5sg;
SSD ¼20 þ80 2:5¼70 m
(8)
The vehicles front camera is 360and assumed to be
able to detect pedestrians up to 80 m (Information, 2013).
In the experiments, we vary traffic flow and cross-
ing rates of pedestrian to assess the efficiency of our
protocol in term of reducing waiting time of vehicles.
The pedestrian crossing rate reflects the average num-
ber of pedestrians who walk through the crossing
pathway in both directions per time unit. Table 1
summarizes the simulation parameters. The following
metrics are used to evaluate vehicles’experience:
Vehicle Travel Duration: The time that a vehicle
takes to travel the considered road from start to
end. To qualify the travel duration, we compare it
with the smallest travel time t
Tmin
, which reflects
the case when the vehicle travels at the maximum
allowed speed on an empty road with no other
vehicles or pedestrian crossing.
VehicleTimeLost: Time wasted due to stopping or
driving with a lower speed than the maximum
allowed. It equals the Vehicle Travel Duration –t
min
.
Vehicle Waiting Time: This is the time wasted by a
vehicle due to stopping only.
To evaluate pedestrians’experience, we used the
following metrics:
Walk Duration: The time that a pedestrian takes to
walk the considered path from start to end. To
qualify the walk duration, it is compared with the
smallest walk time t
Wmin
, which reflects the case
when the pedestrian walks at the maximum speed
(value set in simulation parameters) on an empty
path with no other vehicles or pedestrian traveling.
Walk Time Loss: Time wasted due to stopping or
walking with a lower speed than the value set in
simulation parameters. It equals the Pedestrian
Walk Duration –t
Wmin
.
Simulation results
We conducted two sets of experiments. In the
first, we fixed the average pedestrian crossing rate
at 100 p/h and varied the traffic volume. We changed
the pedestrian crossing rate in the second set of
experiments while fixing vehicles traffic volume at 600
v/h (per direction). We compare the performance of
APC to the conventional TLC approach. We consider
two configurations for the TLC: (i) long green dur-
ation (traditional Traffic light model of 30 s) which is
known to be more efficient during heavy traffic, and
(ii) short green duration (traditional Traffic light
model of 10 seconds) which suits low traffic volume,
i.e., more privilege for pedestrians. All results reflect
the average over 20 simulation runs. The error bars
for 95% confidence intervals are plotted along with
the results.
Exp. Set 1: Performance under increased
traffic volume:
In this set of experiments, we fixed the pedestrian
density at 480 p/h and increased traffic flow rate from
100 to 1200 v/h (per direction). Figure 11(a) shows
that our protocol outperforms traffic light models in
term of vehicle travel duration. Moreover, the effi-
ciency of APC is almost stable and does not decrease
with traffic flow volume. Meanwhile the performance
of the conventional traffic signal model degrades with
increased traffic.
Figure 11(b) shows the average lost time due to
waiting for pedestrians and in essence explains the
rise of the duration in traffic light models and the
steady performance of APC. Time lost in traffic light
models is quite high even in a low traffic and it wor-
sens with the increase in traffic volume. TLC with
long green duration performs better in high traffic
due to the time “wasted”at every signal switching;
Meanwhile TLC with short green duration performs
well in low traffic because each road segment is served
Table 1. Simulation parameters.
Parameter Value
Simulator SUMO 0.32.0
Algorithms Code language Python 3.6.3
Simulation time 3600 s
Number of road directions 2
Number of lanes in each road direction 3
Vehicle’s Speed 45 MPH ’20.11 m/s
Pedestrian’s Speed 1.39 m/s
Road length 200 m
Pedestrian flow (1st set of experiments) 480 p/h
Traffic volume (2nd set of experiments) 600 v/h (per direction)
Safety Distance 20 m
Stopping Sight Distance (SDD) 70 m
Camera detection distance <80 m
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 15
more frequently. However, APC outperforms both
traffic light models whether the vehicular traffic is
high or low traffic; Such distinct performance is due
to the negligible waiting time spent when APC is
applied, as shown in Figure 11(c). Consequently, the
time lost in case of APC in Figure 11(b). is domin-
antly due to deceleration and not to stopping.
Figure 12 evaluates pedestrian’s experience under
increased vehicular traffic flow. Figure 12(a) shows
that pedestrian walk duration at the crossing under
APC protocol increases with the traffic flow contrarily
to TLC models. However, the walk duration is very
high under Traffic Light 10 s and higher under Traffic
Light 30 s even when vehicles flow is very low (100 v/
h). Figure 12(b) shows that the walk time loss under
APC protocol is negligible especially in low density
(less than 5 s) and very low in high traffic volume
ranging from 4.62 s at 600 v/h (Per Direction) to
11.76 s at 1200 v/h. Compared to APC, walk time loss
under TLC models is very huge starting from 11.19 s
under Traffic Light 10 s and from 16.03s under
Traffic Light 30 s in very low traffic.
Exp. Set 2: Performance under growing pedestrian
density:
In this set of experiments, we fixed traffic flow rate at
600 v/h (per direction) and we increased pedestrian
density from 60 p/h to 2400 p/h to evaluate the effect-
iveness of APC compared to Traffic Light 10 s and
Traffic Light 30 s.
Figure 13(a) compares the efficiency of APC to that
with the two baselines in term of vehicle travel dur-
ation. The results in the figure show that the trip time
slightly increases with the density of pedestrians.
Nevertheless, APC outperforms the traffic light mod-
els in all pedestrian density volumes.
As shown in Figure 13(b), APC does not waste the
vehicles time under low pedestrian density and scales
nicely with the rise in pedestrian density. Meanwhile
the traffic light model extends the travel time for
vehicles significantly even with little pedestrian cross-
ing; as expected the travel time grows with the
extended green time for pedestrians.
Figure 13(c) shows that the majority of the time
lost under the TLC models is for waiting due to
unnecessary stopping during red light phases.
Contrarily, the figure confirms that the waiting time
in APC does not grow with increased pedestrian
crossing, where the average waiting time for APC
varies between 0.63 s in low pedestrian density, and
6.03 s in high density. Such waiting time almost equals
to the stopping duration for one or few pedestrians to
cross one lane in a low traffic volume and a low dens-
ity volume.
APC involves vehicle stopping at the crossing and
increases the travel duration compare to some AIM
Figure 11. Comparison of vehicles delays under increased traf-
fic volume based on: (a) Average Travel duration, (b) Average
Vehicle Time Lost and (c) Average Waiting Time.
16 S. E. HAMDANI ET AL.
systems. However, those systems are based on consid-
ering autonomous vehicles as the only road user.
Thus, they could not be relevant for the real word’s
roads in urban areas. Accordingly, we assume that
vehicle stopping on the road should be minimized but
cannot be avoidable. Our approach solves many prob-
lems related to pedestrian management on the road.
On the one hand, it prioritizes the pedestrian and
guarantees safe passage. On the other hand, the sys-
tem decreases vehicle-stopping time at the crossing.
According to the simulation, the TLC model forces
a waiting vehicle to wait 30 s or 10s even in the
absence of pedestrians in the crossing, which causes
vehicles to pile up. On the other hand, our APC
protocol decreases the waiting time to one lane pedes-
trian crossing time. Thus, the vehicle does not stop
unless the pedestrian is crossing its lane. Moreover,
Figure 12. Comparison of pedestrian delays under increased
traffic volume based on: (a) Average Walk Duration and (b)
Average Vehicle Time Loss.
Figure 13. Comparison of vehicles delays under increased traf-
fic volume based on: (a) Average Travel duration, (b) Average
Vehicle Time Lost and (c) Average Waiting Time.
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 17
the cooperation between the vehicles allows them to
avoid stopping by decelerating.
Exp. Set 3: Performance in mixed traffic scenario:
In this set of experiments, we have run simulation
for three scenarios of mixed traffic. The first,
namely “APC_25%HV”, has 25% of HV vehicles and
75% are AVs. Such a ratio is reversed in third
scenario “APC_75%HV”, where 75% of the vehicles
are human-driven. In the second scenario
“APC_50%HV”, the number of HV and AV vehicles
are equal. Meanwhile, we have fixed the pedestrian
density at 480 p/h and increased traffic flow rate from
100 to 1200 v/h (per direction).
Figure 14(a) compares the travel duration for the
three mixed traffic scenarios to that of a full CAV
configuration and of traffic light 10 s and 30 s. The
results show that performance in mixed traffic is
slightly lower (travel duration is longer) compared to
the full CAV scenario. The travel duration also grows
with the increased HV population. For instance,
the average travel duration under heavy traffic of 1000
v/h (per direction) increases from 16,65 s in “APC”,to
19,9 s in “APC_25%HV”, to 22,5 s in “APC_50%HV”
and to 24,6 s in “APC_75%HV”. The results are very
much expected and is attributed to the decreased level
of coordination as the vehicles approach pedestrian
crossing when more HVs are involved. Nonetheless,
APC still significantly outperforms the traffic signal
models even with 75% of the vehicles are HV.
Figure 14(b) shows the average lost time; the
results are consistent with Figure 14(a) and in fact
explains the difference in travel duration among the
full CAV APC, APC in mixed traffic, and traffic light
models. Basically, the time lost in traffic light
increases significantly, especially with traffic light 10 s,
as the traffic gets heavier. Meanwhile, the performance
of APC is very much independent of vehicle density
for both full CAV and mixed traffic scenarios.
Figure 14(c) shows that the average waiting time
under APC is very small and is independent of the
type of vehicles on the road. Meanwhile traffic lights
models are experiencing a very high waiting time
which is increasing with traffic. Figure 14(c) explains
that the difference in time loss between different the
full CAV and mixed vehicle types scenarios is due to
traveling with lower speed (decelerating) and not to
stopping time. On one hand, the safety distance has
increased from 20 m in case of CAV to 70m (SSD) in
mixed traffic which obliges vehicles in mixed traffic to
decelerate in for longer time. On the other hand, the
lack of communication among human-driven vehicles
force them to decrease speed when approaching the
pedestrian crossing, especially when there is another
vehicle in front and they are not able to know
its decision.
Figure 14. Performance of different mixed scenarios traffic
under increased traffic volume compared to full CAV APC and
traffic light models based on: (a) Average Travel duration, (b)
Average Vehicle Time Lost and (c) Average Waiting Time.
18 S. E. HAMDANI ET AL.
In summary, the results for mixed traffic scenarios
confirm the APC is still as effective and that the pres-
ence of human-based vehicles does not impact the
coordination at the pedestrian intersection.
Conclusion and future work
AIM systems manage the traffic at the intersection
efficiently and are expected to replace the conven-
tional TLC model. Although V2V communication has
been exploited for autonomous traffic management,
little attention has been paid to supporting pedestrian
crossing. Thus, this article fills an important technical
gap in AIM. The goal of this article is to include ped-
estrian crossing in autonomous traffic management.
We have developed APC, a cooperative protocol for
pedestrian crossing management. APC pursues a new
pedestrian avoidance scheme using V2V communica-
tion. We have also studied and solved the pedestrian
blocking situation on the pedestrian crossing.
Compared to TLC models, our protocol diminishes
the vehicle stopping time due to pedestrian crossing
and thus increases the road throughput and decreases
congestion rate in both full CAV system and in mixed
traffic scenarios.
The next research steps will be related to classifying
pedestrians crossing in clusters of different sizes and
estimate how different scenarios will influence delays
for both pedestrians and vehicles. Future work
includes considering more vulnerable road users as
bicycles (ElHamdani & Benamar, 2019) and motor-
cycles in autonomous traffic management. Autonomous
vehicles should undertake the existence of vulnerable
users and prioritize their safety while increasing traffic
flow as possible as it is.
Acknowledgments
This work was supported by the Grant Project ITIC-
TRANSPORT Moulay Ismail University of Meknes.
ORCID
Sara El Hamdani http://orcid.org/0000-0002-5358-835X
References
Arvin, R., Kamrani, M., & Khattak, A. J. (2019). How
instantaneous driving behavior contributes to crashes at
intersections: extracting useful information from con-
nected vehicle message data. Accident Analysis &
Prevention,127, 118–133.
Azimi, R., Bhatia, G., Rajkumar, R., & Mudalige, P. (2015).
Ballroom intersection protocol: Synchronous autonomous
driving at intersections. Proceedings—IEEE 21st
International Conference on Embedded and Real-Time
Computing Systems and Applications, RTCSA 2015
(pp. 167–175). doi:https://doi.org/10.1109/RTCSA.2015.20
Bachmann, M., Morold, M., & David, K. (2017). Improving
smartphone based collision avoidance by using pedestrian
context information. 2017 IEEE International Conference
on Pervasive Computing and Communications
Workshops, PerCom Workshops 2017 (pp. 2–5). doi:
10.1109/PERCOMW.2017.7917507
Badger, E. (2015). 5 confounding questions that hold the
key to the future of driverless cars. Retrieved from
https://www.washingtonpost.com/news/wonk/wp/2015/01/
15/5-confounding-questions-that-hold-the-key-to-the-future-
of-driverless-cars/?noredirect=on&utm_term=.33b1906e93d9
Bagheri, M., Siekkinen, M., & Nurminen, J. K. (2014).
Cellular-based vehicle to pedestrian (V2P) adaptive com-
munication for collision avoidance. 2014 International
Conference on Connected Vehicles and Expo, ICCVE
2014 - Proceedings, November (pp. 450–456). doi:10.
1109/ICCVE.2014.7297588
Bento, L. C., Parafita, R., Rakha, H. A., Nunes, U. J., Conde,
L., Parafita, R., …Urbano, J. (2019). A study of the
environmental impacts of intelligent automated vehicle
control at intersections via V2V and V2I communica-
tions control at intersections via V2V and V2I communi-
cations. Journal of Intelligent Transportation Systems,
0(0), 1–19. doi:10.1080/15472450.2018.1501272
Brandstaetter, C., Yannis, G., Evgenikos, P., & Papantoniou,
P. (2011). Traffic safety basic facts 2010: Single vehicle
accidents.
Buxton, P. (2015). Metric handbook: Planning and design
data (5th ed.). Buxton: Routledge.
Dixon, K., Fitzpatrick, K., Avelar, R., Perez, M., Ranft, S.,
Stevens, R., …Voigt, T. (2015). Reducing lane and
shoulder width to permit an additional lane on a freeway
(Technical Report). 7(2), p. 104. doi:10.3141/2588-10
Dollar, P., Wojek, C., Schiele, B., & Perona, P. (2012).
Pedestrian detection: An evaluation of the state of the
art. IEEE Transactions on Pattern Analysis and Machine
Intelligence,34(4), 743–761. doi:10.1109/TPAMI.2011.155
Dresner, K., & Stone, P. (2008). A multiagent approach to
autonomous intersection management. Journal of Artificial
Intelligence Research,31, 591–656. doi:10.1613/jair.2502
Elhamdani, S., & Benamar, N. (2017). A comprehensive
study of intelligent transportation system architectures for
road congestion avoidance. In International Symposium
on Ubiquitous Networking (Vol. 10542, pp. 95–106). doi:
10.1007/978-3-319-68179-5
ElHamdani, S., & Benamar, N. (2018). Autonomous traffic
management: Open issues and new directions. 2018
International Conference on Selected Topics in Mobile
and Wireless Networking (MoWNeT) (pp. 1–5). doi:10.
1109/MoWNet.2018.8428937
ElHamdani, S., & Benamar, N. (2019). DBDA: Distant
bicycle detection and avoidance protocol based on V2V
communication for autonomous vehicle-bicycle road share.
2019 International Conference on Wireless Technologies,
Embedded and Intelligent Systems, WITS 2019 (pp. 1–6).
doi:10.1109/WITS.2019.8723866
ETSI. (2011). ETSI TS 102 637-2 vehicular communications;
Basic set of applications; Part 2: Specification of cooperative
JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 19
awareness basic service. History,1,1–18. doi:10.1145/
2656877.2656887
Florentine, E., Ang, M. A., Pendleton, S. D., Andersen, H.,
& Ang, M. H. (2016). Pedestrian notification methods in
autonomous vehicles for multi-class mobility-on-demand
service. Proceedings of the Fourth International
Conference on Human Agent Interaction –HAI ’16
(pp. 387–392). doi:10.1145/2974804.2974833
Franke, U., Gavrila, D., Gorzig, S., Lindner, F., Puetzold, F.,
& Wohler, C. (1998). Autonomous driving goes
downtown. IEEE Intelligent Systems,13(6), 40–48. doi:
10.1109/5254.736001
Ger
onimo, D., L
opez, A. M., Sappa, A. D., & Graf, T.
(2010). Survey of pedestrian detection for advanced
driver assistance systems. IEEE Transactions on Pattern
Analysis and Machine Intelligence,32(7), 1239–1258. doi:
10.1109/TPAMI.2009.122
Gradinescu, V., Gorgorin, C., Diaconescu, R., Cristea, V., &
Iftode, L. (2007). Adaptive traffic lights using car-to-car
communication. IEEE Vehicular Technology Conference
(1, pp. 21–25). doi:10.1109/VETECS.2007.17
Information, P. (2013). Mercedes-Benz intelligent drive: The
intelligent car. Retrieved from Daimler Communications,
70546 Stuttgart/Germany - Mercedes-Benz - A Daimler
Brand website: http://media.daimler.com/marsMediaSite/
en/instance/ko/Mercedes-Benz-Intelligent-Drive-The-intelligent-
car.xhtml?oid=9904196
Karim, D. M. (2015). Narrower lanes, safer streets. CITE
Conference, June (pp. 1–21). Retrieved from https://www.
academia.edu/12488747/Narrower_Lanes_Safer_Streets_
Accepted_Paper_for_CITE_Conference_Regina_2015_
Layton, R., & Dixon, K. (2012). Stopping sight distance -
Discussion Paper #1. (April). Retrieved from http://cce.
oregonstate.edu/sites/cce.oregonstate.edu/files/12-2-stopping-
sight-distance.pdf
Li, C., & Shimamoto, S. (2011). A real time traffic light con-
trol scheme for reducing vehicles CO
2
emissions. The 8th
Annual IEEE Consumer Communications and Networking
Conference –Emerging and Innovative Consumer
Technologies and Applications, June (2010, pp. 855–859).
doi:10.1109/CCNC.2011.5766627
Li, Z. (Richard), Chitturi, M. V., Yu, L., Bill, A. R., &
Noyce, D. A. (2015). Sustainability effects of next-
generation intersection control for autonomous vehicles.
Transport,30(3), 342–352. doi:10.3846/16484142.2015.
1080760
Mitzel, D. (2013). Taking mobile multi-object tracking
to the next level (PhD Thesis), 566–579.
Mohler, B. J., Thompson, W. B., Creem-Regehr, S. H., Pick,
H. L., & Warren, W. H. (2007). Visual flow influences
gait transition speed and preferred walking speed.
Experimental Brain Research,181(2), 221–228. doi:
10.1007/s00221-007-0917-0
Murphy, J., Djahel, S., Jabeur, N., Barrett, R., & Murphy, J.
(2015). Toward V2I communication technology-based
solution for reducing road traffic congestion in smart
cities toward V2I communication technology-based
solution for reducing road traffic congestion in smart
cities. International Symposium on Networks, Computers
and Communications (ISNCC), September, 1–6. doi:10.
1109/ISNCC.2015.7238584
Olia, A., Razavi, S., Abdulhai, B., & Abdelgawad, H. (2018).
Traffic capacity implications of automated vehicles mixed
with regular vehicles. Journal of Intelligent Transportation
Systems,22(3), 244–262. doi:10.1080/15472450.2017.
1404680
Rasouli, A., Kotseruba, I., & Tsotsos, J. K. (2018).
Understanding pedestrian behavior in complex traffic
scenes. IEEE Transactions on Intelligent Vehicles,8858(c),
1–1. doi:10.1109/TIV.2017.2788193
Rodr
ıguez Palmeiro, A., van der Kint, S., Vissers, L., Farah,
H., de Winter, J. C. F., & Hagenzieker, M. (2018).
Interaction between pedestrians and automated vehicles:
A wizard of Oz experiment. Transportation Research Part
F: Traffic Psychology and Behaviour,58, 1005–1020. doi:
10.1016/j.trf.2018.07.020
SAE. (2009). Dedicated short range communications
(DSRC) message set dictionary. Surface Vehicle Standard,
2735. doi:10.4271/J2735_201603
Shladover, S. E. (2017). Connected and automated
vehicle systems: Introduction and overview. Journal
of Intelligent Transportation Systems,2450,1–11. doi:
10.1080/15472450.2017.1336053
Skikos, G. D., Machia, A. V., & Christopoulos, S. A. (1993).
Application of fuzzy logic to the control of wind energy.
Proceedings. Joint International Power Conference Athens
Power Tech,2(98), 30–38. doi:10.1109/APT.1993.673945
The Federal Highway Administration (FHWA). (2010).
Mini-roundabouts. Retrieved from https://safety.fhwa.dot.
gov/intersection/innovative/roundabouts/fhwasa10007/
fhwasa10007.pdf
U.S. Department of Transportation. (n.d.). Traffic Calming:
25. Crossing Islands. Retrieved from https://safety.fhwa.
dot.gov/saferjourney1/Library/countermeasures/25.htm
Wang, S., Djahel, S., & McManis, J. (2014). A multi-agent
based vehicles re-routing system for unexpected traffic
congestion avoidance. 2014 17th IEEE International
Conference on Intelligent Transportation Systems, ITSC
2014, 2541–2548. doi:10.1109/ITSC.2014.6958097
Wu, Y., Abdel-Aty, M., Wang, L., & Rahman, M. S. (2019).
Combined connected vehicles and variable speed limit
strategies to reduce rear-end crash risk under fog condi-
tions. Journal of Intelligent Transportation Systems,1–20.
Yan, C. (2016). Can you trust autonomous vehicles:
Contactless attacks against sensors of self-driving vehicle.
doi:10.1145/1235
20 S. E. HAMDANI ET AL.
