Content uploaded by Jishiyu Ding
Author content
All content in this area was uploaded by Jishiyu Ding on Mar 05, 2018
Content may be subject to copyright.
Centralized Cooperative Intersection Control Under
Automated Vehicle Environment
Jishiyu DING1, Huile XU1,Jianming HU1and Yi ZHANG1,2,∗
Abstract— With the rapid development in vehicular commu-
nication technologies, cooperative driving of intelligent vehicles
can provide promising efficiency, safety and sustainability to the
intelligent transportation systems. In this paper, a centralized
cooperative intersection control (CCIC) approach is proposed
for the non-signalized intersections under automated vehicle
environment. The cooperative intersection control problem is
converted to a nonlinear constrained programming problem
considering vehicle delay, fuel consumption, emission and driver
comfort level. Furthermore, a simulation-based case study is
carried out on a four-legged, two-lane non-signalized intersec-
tion under different traffic volume scenarios to compare CCIC
with the actuated intersection control (AIC) system. The results
indicate that the CCIC approach shows significant potential
improvements on the traffic efficiency (i.e., nearly 14% of traffic
flow increase, nearly 90% of travelling time saving), emission
(nearly 60% of CO2reduction) and driver comfort level (nearly
2% of comfort level increase).
Index Terms— cooperative intersection control, automated
vehicle, collision avoidance, efficient and sustainable transporta-
tion network
I. INTRODUCTION
Efficiency, safety and sustainability have always been the
three main concerns in the transportation systems. With the
rapid increase of the number of the vehicles, traffic conges-
tion, traffic accidents and environmental pollution become
the critical problems for the government and urban traffic
planners. On one hand, Americans spent 6.9 billion hours of
extra travelling time and 3.1 billion gallons of extra gas due
to congestion in 2014–an increase of 23%–35% compared
with the previous decade [1]. On the other hand, there were
about 156 vehicular collisions kill and 10000 injure per day
on average due to the complicated traffic environment [2].
In addition, the amount of carbon dioxide (CO2) emission
produced increased 14% during the last decade [3]. Thus, it is
important and necessary for traffic planners to find innovative
solutions for those problems.
With the development of vehicle to vehicle (V2V) and
vehicle to infrastructure (V2I) communication technologies,
*Corresponding author. Tel.: +86 (010) 62796832
1Jishiyu DING, Huile XU and Jianming HU are with Department of
Automation, Tsinghua University, Beijing 100084, China; Tsinghua
National Laboratory for Information Science and Technology
{djsy15,hl-xu16}@mails.tsinghua.edu.cn, hujm@
mail.tsinghua.edu.cn
2Yi ZHANG is with Tsinghua - Berkeley Shenzhen Institute
(TBSI), ZhiYuan, Nanshan District, Shenzhen 518000, China,
zhyi@mail.tsinghua.edu.cn
connected vehicles and automated vehicles are able to pro-
vide a safe driving environment for the drivers by using in-
formation sharing and vehicle cooperation. Therefore, intel-
ligent transportation systems for cooperative vehicle control
have been the popular topics worldwide.
One of the most common scenarios in urban roads is
the intersection where one basic method towards collision
avoidance of vehicles and traffic efficiency is to install
traffic signal control systems. Although traffic signals are
not the optimal solutions, properly using them can improve
safety and efficiency in transportation systems. However,
because of the development of automated vehicles and V2X
communication technologies, many road intersections will
have no traffic lights in the future. This study focuses on
vehicle driving scenarios at such non-signalized intersec-
tions. In recent years, there are some studies focusing on
the cooperative driving or collision avoidance at the non-
signalized intersections. In 2006, Li et al. [4], [5] proposed
a concept named “safety driving patterns” to obtain the
allowable movement schedules of all encountered vehicles at
intersections. In 2012, Hafner et al. [6], [7] implemented an
active control Intersection Collision Avoidance (ICA) system
for the merging scenario of two vehicles. In 2012, Lee et al.
[8] proposed an algorithm to manipulate individual vehicles’
maneuvers by eliminating the potential overlaps of vehicular
trajectories coming from all conflicting approaches at the
intersection. However, the overlaps of vehicular trajectories
do not mean real collision occurrence, so this method needs
further improvement. Zohdy et al. [9] used “The Intersection
Cooperative Adaptive Cruise Control System Concept” to
minimize the total delay of the intersection. Campos et
al. [10] proposed a velocity-based negotiation approach for
intersection crossing. These studies mainly considered the
efficiency and the safety at the intersection and ignored the
sustainability (i.e., environmental impacts) and the comfort
level for the drivers.
In this paper, a centralized cooperative intersection control
(CCIC) approach is proposed for the non-signalized intersec-
tions under automated vehicle environment. The cooperative
intersection control problem is converted to a nonlinear con-
strained programming problem. Compared with the recent
studies, sustainability measure and comfort level for drivers
are taken into account to cooperate vehicles at the non-
signalized intersections.
The remaining of the paper is organized as follows.
Section 2 describes the methodology including optimization
model and collision avoidance model. A simulation-based
case study and corresponding results are presented in Section
3. Finally, concluding remarks are presented in Section 4.
II. METHODOLOGY
This paper focuses on a four-legged, two-lane approach
non-signalized intersection where the links are labeled from
1 to 8 in Fig.1. The objective of cooperative intersection
control is to guide vehicles passing the intersection more
efficiently, sustainably, safely and comfortably by using
V2X communication technology. Thus, quantitative indexes
of efficiency, sustainability, safety and comfort level are
discussed and the cooperative intersection control problem is
converted to a multi-objective optimization problem. In this
section, the optimization model and the collision avoidance
model are presented.
1
2
34
5
6
78
lane1
lane2
Fig. 1. Diagram of a four-legged two-lane intersection
A. Optimization Model
The objective, decision variables and constraints of the
optimization model are discussed in detail.
1) Objective: The objective of centralized cooperative
intersection control is to minimize the intersection delay, fuel
consumption and emission as well as the discomfort level of
drivers.
The intersection delay describes the differences between
actual passing time and optimal passing time (passing with
a maximum speed) through the intersection which can be
formulated as Eq.(1)-Eq.(3).
di(t) = ATi(t)−OTi(t)(1)
Li(t) = vi(t)×ATi(t) + 1
2×ai(t)×AT 2
i(t)(2)
OTi(t) = Li(t)
vmax
(3)
where di(t)is the intersection delay of vehicle iat time t,
ATi(t)the actual passing time of vehicle iat time t,OTi(t)
the optimal passing time of vehicle iat time tand Li(t)is
the remaining distance to passing the intersection of vehicle
iat time t.vi(t)and ai(t)denote the speed and acceleration
of vehicle iat time trespectively.
Thus, the total intersection delay is the summation of the
delay of each vehicle described as Eq.(4).
D(t) =
N
X
i=1
di(t)(4)
where D(t)represents the total intersection delay at time t
and Nis the total number of vehicles.
Some fuel consumption and emission models such as
Virginia Tech comprehensive power-based model [11] are
so complicated due to considering vehicle characteristics,
road conditions and so on. Since vehicles in our experiments
are of the same type and run on the same road condition,
the only differences are their speed and acceleration. Thus,
VT-Micro model (Rakha et al., [12]) is incorporated for
measuring the fuel consumption and emission during the
cooperative driving in this paper. In addition, the choose
of model has little effect on the experiments results in our
experiments. The VT-Micro Model estimates emission and
fuel consumption using instantaneous vehicular speeds and
accelerations formulated as Eq.(5).
ln ei(t) =
3
X
k=0
3
X
j=0
Le
k,j ×vk
i(t)×aj
i(t)(5)
E(t) =
N
X
i=1
ei(t)(6)
where ei(t)is the CO, CO2, NOx, HC and fuel at time t,
Le
k,j the model coefficients for ei(t)and E(t)denotes the
total vehicle fuel consumption and emission at time t.
The comfort level of drivers is closely related to the
speed, acceleration and headway to the preceding vehicle
[13]. Traditional methods of measuring drivers comfort level
are proportional to the headway and affected by the current
speed. According to the study of Vos et al. [14], the regres-
sion comfort model can be formulated as Eq.(7).
ci(t) = b0+b1hi(t) + b2h2
i(t) + b3h3
i(t)(7)
where ci(t)is the comfort level of vehicle iat time t,hi(t)
the headway of vehicle iat time t,b0,b1,b2and b3are the
regression coefficients.
The shortcoming of the above model is leaving out of
consideration with the frequency of acceleration. Since too
often accelerations or decelerations make the drivers un-
comfortable, it is reasonable to consider the historical value
of acceleration to describe the comfort level which can be
described as Eq.(8).
ci(t)=(b0+b1hi(t)+b2h2
i(t)+b3h3
i(t))×(1−1
KX
k
|ahik|
amax
)
(8)
where ahik(t)is the kth historical value of acceleration and
amax is the maximum acceleration.
Therefore, the discomfort level can be formulated as the
summation of the reciprocal of the comfort level.
UC(t) =
N
X
i=1
uci(t) =
N
X
i=1
1
ci(t)(9)
where UC(t)denotes the total discomfort level and uci(t)
is the discomfort level of vehicle iat time t.
Considering intersection delay, fuel consumption, emis-
sion and comfort level, the comprehensive objective can be
formulated as Eq.(10).
Z(t) = α
N
X
i=1
di(t) + β
N
X
i=1
ei(t) + γ
N
X
i=1
uci(t)(10)
where α,βand γare the weighting coefficients.
2) Decision variables: Since the vehicles are all automat-
ed, the lateral control can be completed by the vehicle itself
(along the guide lines) and the longitudinal control can be
adapted to perform cooperatively. Namely, the acceleration
(or the desired speed) can be controlled by the control
center at the intersection. Thus, the decision variable is the
acceleration (or the desired speed) of each vehicle.
3) Constraints: The cooperative intersection control need
to meet some constraints. First of all, all of the vehicles
need to cooperate to avoid collisions with each other and the
collision avoidance model is presented in the next subsection.
Second, the speeds and accelerations of vehicles should be
within a reasonable range described as Eq.(11) and Eq.(12).
vmin ≤vi≤vmax (11)
amin ≤ai≤amax (12)
where vmin,vmax are the minimum and maximum speed,
amin,amax are the minimum and maximum acceleration.
B. Collision Avoidance Model
The collision avoidance model is discussed in detail in this
part. As shown in Fig.2, there are many potential conflict
points which can be divided into four conflict patterns:
through-through (i.e., 1-1 to 5-1 vs. 4-2 to 8-2), right-through
(i.e., 1-1 to 5-1 vs. 4-1to 5-1), left-through (i.e., 1-1 to 5-1
vs. 4-2 to 2-2) and left-left (i.e., 1-2 to 8-2 vs. 4-2 to 2-2).
(Note: 1-1 to 5-1 denotes that a vehicle is driving from lane1
of link1 to lane1 of link5.)
1
2
34
5
6
78
lane1
lane2
Fig. 2. Illustration of potential conflict points
1
4
5
lane1
lane2
A
C
B
Intersection Area
Fig. 3. Example of right-through conflict pattern
We take the right-through conflict pattern as an example
to illustrate the collision avoidance model. The scenario of
right-through conflict pattern (i.e., 1-1 to 5-1 vs. 4-1 to 5-1)
is shown in Fig.3. The vehicle trajectories are plotted in dash
to show the potential collision.
The curves in Fig.4 indicate the predictive trajectories
of individual vehicles maintaining their current accelera-
tion/deceleration at t= 0. As noted, the x-axis denotes the
time, and the y-axis stands for the remaining distance from a
vehicle to the potential collision point. The individual vehicle
predictive trajectory is formulated as Eq.(13).
d=d0−(vt +1
2at2)(13)
where dis the predicted remaining distance to the potential
collision point, d0the current remaining distance to the
potential collision point, vand aare the current speed and
acceleration/deceleration of the vehicle.
V1 V2
d
t
O(B)
C
A
tt
1
2
Fig. 4. Illustration of collision avoidance
The collisions occur when vehicles arrive at the potential
collision point at the same time. Considering the vehicle
length and the safe headway, the collision avoidance con-
dition is formulated as Eq.(14)-Eq.(16). Eq.(16) denotes that
the difference between the arriving time to the collision point
of two vehicles need to be greater than the safe headway.
d01 −(v1t1+1
2a1t2
1)=0 (14)
d02 −(v2t2+1
2a2t2
2)=0 (15)
|t1−t2| ≥ Hsafe (16)
where t1,t2are the arriving time to the potential collision
point and Hsafe denotes the safety headway.
Thus, the multi-objective optimization model can be sum-
marized as Eq.(17)-Eq.(20).
min
A
A
Aα
N
X
i=1
di(t) + β
N
X
i=1
ei(t) + γ
N
X
i=1
uci(t)(17)
s.t. |ti−tj| ≥ Hsafe ∀(i, j)∈CP S (18)
vmin ≤vi≤vmax ∀i∈[1, N ](19)
amin ≤ai≤amax ∀i∈[1, N ](20)
where A
A
Ais the control policy, (i, j)denotes the conflict
pattern pair and CP S denotes the conflict pattern set.
Finally, the whole control policy is a series of accelerations
for each vehicle arriving at the intersection region described
as Eq.(21) and Eq.(22).
A
A
A={a1, a2, . . . , ai, . . . , aN
a1, a2, . . . , ai, . . . , aN
a1, a2, . . . , ai, . . . , aN}i= 1,2, . . . , N (21)
a
a
ai={ai1, ai2, . . . , aiti, . . . , aiTi}ti= 1,2, . . . , Ti(22)
where A
A
Ais the acceleration command matrix of the control
center, a
a
aiis the acceleration command vector of vehicle i,
aitiis the acceleration command to vehicle igiven by the
intersection control center at time tiand Tiis the total time
of vehicle iin the intersection region.
C. Solving Algorithm
The cooperative intersection control problem mentioned
above is considered as a nonlinear constrained programming
problem. To solve this problem, this paper employed interior
point method [15]. This algorithm–also referred to as a bar-
rier algorithm–attempts to solve the problem as a sequence
of approximate minimization problems where the bounds
(constraints) are satisfied at all iterations. Consequently, the
algorithm can solve nonlinear constrained problems efficient-
ly and accurately.
III. CAS E STU DY
In this section, a simulation-based case study and corre-
sponding results are presented in different traffic volume sce-
narios under an integrated simulation test bed incorporating
VISSIM and C++.
A. Assumptions
•All vehicles are automated vehicles which means it can
follow the instructions absolutely. In addition, the type
of vehicles are all private cars.
•All vehicles are assumed to be equipped with V2X
communication device and controlled by the control
center at the intersection. The communication condition
is assumed to be perfect.
•In the scenarios of turning left and right, the vehicles
are assumed to follow the guide lines strictly.
B. Simulation test bed
This paper developed an integrated simulation test bed
incorporating VISSIM 6.0 for microscopic-level vehicular
simulation and C++ for the implementation of cooperative
control algorithm and optimization through the VISSIM’s
COM interface. The hypothesis intersection for the exper-
iment is shown in Fig.5 and the experiments parameters
are summarized in Table I. The weighting coefficients of
objective function can be set for different considerations. For
example, traffic planners want to achieve high efficiency with
a larger αwhile the government wants to reduce the emission
and consumption with a larger β.
Fig. 5. Hypothesis intersection for simulation
TABLE I
SIMULATION PARAMETER SETTING
Parameter CCIC setting
Maximum Speed 45km/h
Maximum Acceleration 3m/s2
Minimum Acceleration -3m/s2
Safety Headway 2s
Simulation Time 1h
α, β, γ 0.4,0.4,0.2
To measure the performance of CCIC approach, some
quantitative indexes are summarized in Table II where three
types of measures are selected to measure the efficiency,
sustainability and comfortability of the CCIC approach.
Traffic flow, average delay, queue length and average speed
are selected to measure the performance of traffic efficiency.
TABLE II
PERFORMANCE MEASURE INDEX
Parameter Performance Index Unit
Efficiency Measure
Traffic Flow Vehicles/h
Average Delay s
Average Queue length m
Average Speed km/h
Sustainability Measure Fuel Consumption kg
Vehicle Emission kg
Comfortability Measure Comfort Level /
To examine the efficiency, sustainability and comfortabil-
ity under different traffic volume conditions, four different
scenarios are presented and simulated. Table III shows the
specific details about each scenario. Ten repetitions of each
TABLE IV
COMPARISON OF CCIC AND AIC UNDER DIFFERENT SCENARIOS
Scenario Measure AIC CCIC Gain p-value
1
Traffic flow (vehicle/h) 2581 2998 16.16% /
Average Delay (s) 25.3 1.17 95.38% 0
Average Queue length (m) 27.06m1.46m94.60% 0
Average speed (km/h) 28.87 32.34 12.02% 0.021
Fuel Consumption (kg) 13.25 4.293 67.60% 0.001
Emission (kg) 926.15 300.09 67.60% /
Comfort Level 0.8514 0.8133 -4.47% 0.214
2
Traffic flow (vehicle/h) 2183 2412 10.49% /
Average Delay (s) 22.11 2.53 88.56% 0
Average Queue length (m) 19.21 0.86 95.52% 0
Average speed (km/h) 30.14 34.14 13.27% 0.04
Fuel Consumption (kg) 9.8295 3.396 65.45% 0.002
Emission (kg) 687.08 237.46 65.44% /
Comfort Level 0.8425 0.8913 5.79% 0.052
3
Traffic flow (vehicle/h) 1805 2022 12.02% /
Average Delay (s) 19.755 1.54 92.20% 0
Average Queue length (m) 13.07 0.71 94.57% 0
Average speed (km/h) 32.69 36.48 11.59% 0.033
Fuel Consumption (kg) 8.328 2.847 65.81% 0
Emission (kg) 582.13 199.04 65.81% /
Comfort Level 0.8971 0.9145 1.93% 0.076
4
Traffic flow (vehicle/h) 1204 1416 17.61% /
Average Delay (s) 19.32 1.18 93.89% 0
Average Queue length (m) 8.495 0 100.00% 0
Average speed (km/h) 33.44 36.54 9.27% 0.048
Fuel Consumption (kg) 5.3745 2.178 59.48% 0
Emission (kg) 375.68 146.04 61.13% /
Comfort Level 0.8843 0.9130 3.25% 0.068
TABLE III
VOLUME SETTINGS OF DIFFERENT SCENARIOS
Scenario E-W Volume N-S Volume Overall ratio
1 900 600 1.01
2 800 400 0.90
3 600 400 0.78
4 400 300 0.61
scenario are simulated on the test bed. The actuated intersec-
tion control (AIC) is also implemented to make comparison
with the CCIC approach and the corresponding optimal
signal timing plans are developed by Husch et al. [16].
C. Results
Table IV summarizes the efficiency, sustainability and
comfortability benefits of the centralized cooperative inter-
section control (CCIC) approach applied to the hypothesis
intersection under different scenarios. The histograms of gain
(traffic flow, average delay, queue length, speed, comfort lev-
el and consumption) under different scenarios are presented
in Fig.6.
For the traffic efficiency measure, CCIC approach shows
significant advantages over the AIC system. As shown in
Fig.6(A), the total traffic flow shows an improvement of
16.16%, 10.49%, 12.02% and 17.61% respectively for each
scenario. As clearly shown in Fig.6(B) and Fig.6(C), the
intersection delay and average queue length show a signifi-
cant improvement compared with AIC. Since CCIC guides
the vehicles passing the non-signalized intersection with
no collision, the improvement in efficiency is predictable.
Although the improvement in average speed is not obvious
in Fig.6(D), it is necessary to note that the maximum speed
limited passing the intersection is set as 45 km/h which
means that average speed of CCIC almost reaches the speed
limit.
(A). Traffic flow gain under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
0
10%
20%
Gain
(B). Delay gain under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
0
50%
100%
Gain
(C). Queue length gain under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
0
50%
100%
Gain
(D). Speed gain under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
0
5%
10%
15%
Gain
(E). Consumption gain under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
0
50%
100%
Gain
(F). Comfort level gain under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
-5%
0%
5%
10%
Gain
Fig. 6. Measure index gain under different scenarios
For the sustainability measure, CCIC approach is fuel e-
conomy and environmentally friendly. As shown in Fig.6(E),
the fuel consumption reduces 67.6%, 65.45%, 65.81% and
59.48% respectively for different scenarios. Moreover, the
CCIC reduces the vehicle emission between 61% and 67%
depending on the different volume scenarios. Thus, consid-
ering the fuel consumption and emission in the objective of
optimization model has a big advantage in sustainability.
As shown in Fig.6(F), the CCIC approach does not show
significant improvements over the AIC system for the com-
fort level measure. For the first scenario, the comfort level of
CCIC even reduces 4.47% compared with AIC. The probable
reason for that is the short headway for the heavy traffic
scenario may lead to discomfort for drivers. Moreover, the
weighting parameter γfor comfort level is smaller than the
others and with the increase of γthe comfort level will be
enhanced.
Moreover, t-test (α= 0.05) is carried out there to
investigate the difference of means between CCIC and AIC
system. The histograms of p-value (intersection delay, speed,
comfort level and consumption) under different scenarios
are presented in Fig.7. As clearly shown in Fig.7(A) and
Fig.7(D), the intersection delay, speed and fuel consumption
benefits in CCIC are statistically significant under different
scenarios. Especially, CCIC approach shows larger impacts
on average speed under heavy traffic condition compared
with light traffic condition. Since the traffic flow volume
is relatively small under light traffic conditions, thus the
impacts on average speed do not show such an improvement.
However, the comfort level does not achieve statistically
significants under level α= 0.05. The probable reason for
that is discussed above. The p-value of comfort level under
heavy traffic condition is relatively high, since large traffic
volume resulting in traffic congestion which makes drivers
feel uncomfortable.
(A). Delay p-value under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
0
0.0005
0.0010
0.0015
p-value
(B). Speed p-value under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
0
0.01
0.02
0.03
0.04
0.05
p-value
(C). Comfort level p-value under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
0
0.05
0.1
0.15
0.2
0.25
p-value
(D). Consumption p-value under different scenarios
0.61 0.78 0.90 1.01
Overall Ratio
0
0.0005
0.0010
0.0015
0.0020
0.0025
0.0030
p-value
Fig. 7. Measure index P-value under different scenarios
IV. CONCLUSIONS
In this paper, a centralized cooperative intersection con-
trol (CCIC) approach is proposed under automated vehicle
environment. CCIC approach can provide promising traffic
efficiency, safety, sustainability and comfortability by co-
operating vehicles passing the non-signalized intersection
through V2X communication. Based on a simulation-based
case study at a four-legged two-lane non-signalized intersec-
tion, the potential improvements of the CCIC approach over
the traditional actuated intersection control (AIC) system are
evaluated. Furthermore, different traffic volume scenarios are
investigated on the test bed. Compared with the AIC system,
the proposed CCIC approach shows significant potential
improvements on efficiency (i.e., 10.49%-17.61% of traffic
flow increase, 88.56%-95.38% of travelling time saving) and
sustainability (i.e., 17.78%-37.81% of gas saving, 61.13%-
67.60% of CO2reduction) to the transportation system. With
such promising benefits in traffic efficiency and sustainabil-
ity, CCIC also provides a strict safe environment when the
vehicles are cooperated to pass the intersection.
ACKNOWLEDGMENT
This work is supported by National Natural Science Foun-
dation of China under Grant No. 61673233, Beijing Munic-
ipal Science and Technology Program(D15110900280000),
National Key R&D Program in China (2016YFB0100906)
and National Science and Technology Support plan Project
under Grant No. 2014BAG03B01.
REFERENCES
[1] D. Schrank, B. Eisele, T. Lomax, and J. Bak, “2015 urban mobility
scorecard,” 2015.
[2] N. H. T. S. Administration et al., “Traffic safety facts, 2012 data:
Pedestrians,” Annals of Emergency Medicine, vol. 65, no. 4, p. 452,
2015.
[3] A. EPA, “Inventory of us greenhouse gas emissions and sinks: 1990-
2009,” EPA 430-R-11-005, Tech. Rep., 2011.
[4] L. Li, D. Wen, and D. Yao, “A survey of traffic control with vehicular
communications,” IEEE Transactions on Intelligent Transportation
Systems, vol. 15, no. 1, pp. 425–432, 2014.
[5] L. Li and F.-Y. Wang, “Cooperative driving at blind crossings using
intervehicle communication,” IEEE Transactions on Vehicular tech-
nology, vol. 55, no. 6, pp. 1712–1724, 2006.
[6] M. R. Hafner, D. Cunningham, L. Caminiti, and D. Del Vecchio,
“Cooperative collision avoidance at intersections: Algorithms and ex-
periments,” IEEE Transactions on Intelligent Transportation Systems,
vol. 14, no. 3, pp. 1162–1175, 2013.
[7] M. Hafner, D. Cunningham, L. Caminiti, and D. Del Vecchio, “Au-
tomated vehicle-to-vehicle collision avoidance at intersections,” in
Proceedings of World Congress on Intelligent Transport Systems, 2011.
[8] J. Lee and B. Park, “Development and evaluation of a cooperative
vehicle intersection control algorithm under the connected vehicles
environment,” IEEE Transactions on Intelligent Transportation Sys-
tems, vol. 13, no. 1, pp. 81–90, 2012.
[9] I. H. Zohdy and H. Rakha, “Game theory algorithm for intersection-
based cooperative adaptive cruise control (cacc) systems,” in 2012 15th
International IEEE Conference on Intelligent Transportation Systems.
IEEE, 2012, pp. 1097–1102.
[10] G. R. de Campos, P. Falcone, and J. Sj¨
oberg, “Autonomous coop-
erative driving: a velocity-based negotiation approach for intersection
crossing,” in 16th International IEEE Conference on Intelligent Trans-
portation Systems (ITSC 2013). IEEE, 2013, pp. 1456–1461.
[11] H. A. Rakha, K. Ahn, K. Moran, B. Saerens, and E. Van den Bulck,
“Virginia tech comprehensive power-based fuel consumption model:
model development and testing,” Transportation Research Part D:
Transport and Environment, vol. 16, no. 7, pp. 492–503, 2011.
[12] K. Ahn, H. Rakha, A. Trani, and M. Van Aerde, “Estimating vehicle
fuel consumption and emissions based on instantaneous speed and
acceleration levels,” Journal of transportation engineering, vol. 128,
no. 2, pp. 182–190, 2002.
[13] M. Taieb-Maimon and D. Shinar, “Minimum and comfortable driving
headways: Reality versus perception,” Human Factors: The Journal
of the Human Factors and Ergonomics Society, vol. 43, no. 1, pp.
159–172, 2001.
[14] A. De Vos, J. Theeuwes, W. Hoekstra, and M. Co¨
emet, “Behavioral
aspects of automatic vehicle guidance: Relationship between headway
and driver comfort,” Transportation Research Record: Journal of the
Transportation Research Board, no. 1573, pp. 17–22, 1997.
[15] S. Mehrotra, “On the implementation of a primal-dual interior point
method,” Journal on optimization, vol. 2, no. 4, pp. 575–601, 1992.
[16] D. Husch and J. Albeck, “Trafficware synchro 6 user guide,” Traf-
ficWare, Albany, California, vol. 11, 2004.
