AbstractβThis paper proposes a Vehicle-to-Vehicle (V2V)
communication-based forward collision avoidance algorithm by
alarming the driver for the normal driver mode and controlling
the driving wheel for the self-governed (autonomous) driving
mode. The proposed algorithm benefits from the information
exchange between the host vehicle and the leading vehicle to
calculate the safe distance between host vehicle and leading vehicle
to guarantee the avoidance of the collision. The proposed system
gives advisory and imminent warnings according to the predicted
accident levels, using a three different levels of collision avoidance
for the driver mode. Also, in the autonomous driving mode, it
follows an alternative optimal path to avoid the collision. The
simulation results are implemented using Prescan and MATLAB.
The simulation results show that the proposed collision avoidance
system makes a composite analysis of the collision risk and
provides an accurate real-time warning and an alternative path
for dual driving modes.
KeywordsβCollision Avoidance, Vehicle-to-Vehicle (V2V),
Prescan, Autonomous.
I. INTRODUCTION
Automobile industry occupies a pivotal position in
technological research around the world, Billions of dollars are
spent yearly on the field of research and development of cars
capacities either in terms of luxury or safety. This opens the
way for many of global corporations to invest Billions of dollars
every year into research for the huge financial revenue and
technological gains in favor of the corporations and economies
around the world.
Safety element in cars is a basic pillar in the reduction of
accidents around the world. According to the World Health
Organization (WHO) 2016 statistics, 1.25 Million deaths
worldwide were caused by car accidents [1], meaning that car
accidents are the ninth leading cause of death around the globe,
and the number is expected to rise to 1.9 Million by 2030,
meaning that car accidents will be the fifth leading cause of
death around the globe by 2030 [2]. This paper main objective
is to reduce accidents resulting from driver error or road
problems by avoiding the forward collision.
This work develops a forward collision avoidance algorithm
that controls the vehicle to change or keep the presented lane to
avoid a forward collision that may be caused by another vehicle.
The algorithm is based on the optimization problem provided
by the same authors in [3,4] in a symbolic numerical and well-
defined mathematical model. Some parameters are treated as a
given input from an integrated Vehicle-to-Vehicle (V2V) or
Vehicle-to- Any device (V2X) systems, such as roadβs obstacle
dimensions and information on roadβs velocity [4]. This work
provides practical simulations for the mathematical model in
[3,4] in collision avoidance by using Prescan tool and
MATLAB.
The paper is organized as follows: Section II gives descriptions
of the proposed forward collision algorithm. Simulation results
are given in Section III. Finally, the whole work is concluded
in Section IV.
II. PROPOSED FORWARD COLLISION ALGORITHM
In this paper, the system model is based on three scenarios
including the duality between the autonomous mode system and
the warning system for the driver mode. The basic scenario
contains two vehicles which are Leading Vehicle (LV) and
Following Vehicle (FV). Every vehicle is provided by Global-
Positioning-System (GPS) device and Radio Frequency (RF)
[5]. Antenna System is used to transfer positioning data
between LV and FV and is used in couple of autonomous and
warning system that uses Basic Safety Message (BSM).
According to the Society of Automotive Engineers (SEA
J2735) protocol [6], the BSM is the primary message set that
contains the basic information to secure the driver life and to
send data among vehicles.
The Following vehicle updates the safe braking distances as a
function of the speed, latitude, and longitude (position values).
Using this information combined with the data received from
the BSM sent by Leading Vehicle, the host vehicle calculates
the Distance-To-Collision based on the acceleration (DTCa)
between the two vehicles.
DTC is frequently used in literature as a descriptor of how
urgent a situation has become, as well as potentially how a
driver perceives stimuli during an event [7]. DTC is determined
using various measures and theories. In an event that includes
an FV and LV and when the LV is moving at a constant rate
(zero acceleration), the DTC is computed based on the velocity
only [8] as follows:
π·ππΆ = βπ
πππππ£ (1)
Dual-Mode Forward Collision Avoidance Algorithm
Based on Vehicle-to-Vehicle (V2V) Communication
Mohamed Yousef1, Ahmed Hosny1, Wessam Gamil1, Mohamed Adel1,2, Hazem M. Fahmy3, M. Saeed Darweesh1,4,
Hassan Mostafa2,4
1Institute of Aviation Engineering and Technology, Giza, Egypt 2Faculty of Engineering, Cairo University, Giza, Egypt
3Germany University in Cairo, Cairo, Egypt 4Nanotechnology Department, Zewail City of Science and Technology, Egypt
978-1-5386-7392-8/18/$31.00 Β©2018 IEEE 739
Where r is the range between the vehicles based on the
Haversine formula [7], and ππ is the relative velocity, which is
defined as:
ππ = πππ£ β πππ£ (2)
Where πππ£ and πππ£ are the speeds of the LV and the FV,
respectively.
This paper proposes a dual mode forward collision avoidance
system (driver mode and autonomous mode) based on the
acceleration to be more practical and realistic model by
calculating the DTCa when the FV acceleration is assumed to
be zero and when the LV is deaccelerating. The LV acceleration
is included in the equation of DTCa as follows:
π·ππΆπ = βππββππ
2+2πππ£π
πππ£ πππ£ (3)
Where πππ£ and βππββππ
2+2πππ£π
πππ£ are the acceleration of the LV,
and the time-to-collision, respectively.
In this paper there are three levels of thresholds defined for the
warning distances considering the speed and acceleration of
both vehicles (LV and FV) and the distance between them to
calculate DTCa from (3). Warnings are triggered when the
distance between the two vehicles are less than these defined
thresholds.
The First threshold warning distance π·π€1 is calculated from (4):
π·π€1 = 0.5( πππ£
2
πππ£β
πππ£2
πππ£) + π‘ππππ£ + π·π (4)
Where πππ£ and πππ£ are the accelerations of the LV and the FV,
respectively. The additional term π·π is the distance between the
host and the leading vehicle when they stop, it is used to inform
the driver that there is a vehicle on the same lane.
The second threshold warning distance π·π€2 calculated from (5)
and it warns the driver to take a decision to change the lane
(autonomous) or to decelerate the vehicle (driver mode).
π·π€2 = πππ£
2
19.6(πππ£
π+π+πΊ)
+ π‘ππππ£ + π·π (5)
Equation (5) is commonly used in road design for establishing
the minimum stopping sight distance required on a given road
according to the American Association of State Highway and
Transportation Officials (AASHTO) which gives the formula
for calculating the stopping distance [9]. If the actual vehicle
spacing drops below this threshold, then the distance-to-
collision is less than the total distance travelled by the vehicle
during the delay time (td) then the third threshold warning
distance will be activated.
According to the second equation of motion [10], this work
proposed forward collision avoidance system by proposing the
equation of the third threshold warning range in an LV situation
after adding the acceleration as:
π·π€3 = πππ£π‘π β 0.5πππ£π‘π2 + π·π (6)
The third threshold warning range warns the driver urgently to
take a decision to avoid the collision immediately. In case of
no-response from the driver, the autonomous mode makes the
car brake or change the lane if there is no coming car in the
opposite lane.
At every incoming BSM, the algorithm checks the difference
between the DTCa and the warning distances (π·π€1, π·π€2, and
π·π€3) then calculates the time interval to reach the warning
distance considering the current speed. If the obtained time to
warning (π‘π€) is smaller than the GPS update period (π‘πΊππ), a
collision warning is triggered after π‘π€ seconds.
Algorithm 1 shows the proposed forward collision algorithm.
Algorithm 1: Proposed Forward Collision Algorithm
Input: GPS update period(π‘πΊππ), Distance to collision
regrading acceleration (π·ππΆπ), Safe and
automatic braking distances
(π·π€1, π·π€2, π·π€3) and Host vehicle current speed
(πππ£).
Output: Time to warning (π‘π€1, π‘π€2, ππ π‘π€3) begin
βπ·1 β DTC β π·π€1
βπ·2 β DTC β π·π€2
βπ·3 β DTC β π·π€3
π‘π€1 β βπ·1 Vππ£β
π‘π€2 β βπ·2 Vππ£β
π‘π€3 β βπ·3 Vππ£β
if π‘π€3 β€ π‘πΊππ then
return (π‘π€3)
else if π‘π€2 β€ π‘πΊππ then
return (π‘π€2)
else if π‘π€1 β€ π‘πΊππ then
return (π‘π€1)
end
end
The proposed optimization and control algorithm is shown in
Fig. 1 where π‘π π‘πππ‘ and π‘πππ are the limits of the whole
simulation interval, π‘π and π‘π are the limits of each optimization
iteration based on the discrete sampling. π·πππ‘ππππ€ and
π΅ππ‘ππππ€ are the new optimized steering angle and braking ratio
for the current iteration that corresponds to a minimum πΆ which
is the cost function. π·ππ£ denotes the number of divisions that
the whole interval should be sampled to. π ππππ denotes the ideal
line that the optimization should refer to when calculating the
minimum cost function. ππππ₯ and ππ€π denotes the maximum
allowed velocity and constraints violation penalty weights
respectively [4]. Full details of the optimization problem is
available in [3,4] by the same authors.
740
Fig. 1: Optimization algorithm for collecting the data of each iteration and
interpolate it into a vector that can be visualized after the process is done
III. SIMULATION RESULTS
In this work, the Prescan simulation tool connected to
MATLAB to verify the performance of the proposed forward
collision avoidance algorithm. The traffic scenario is
implemented in the Prescan as shown in Fig. 2.
Fig. 2: Traffic Scenario in Prescan
The performance metric in this work is the success of
generating a warning message for the driver then enter the
suitable mode (driver mode or autonomous mode) to decelerate
or brake or change the lane before the collision occurs.
The proposed forward collision avoidance algorithm is tested in
different scenarios (driver and autonomous) as follows:
A. Driver Mode Scenario
β’ The LV deaccelerate, the driver of the FV is informed by the
three levels of warnings before it takes the action of braking
to stop before the collision occurs as shown in Figs. 3 and 4.
In case of no-response from the driver, the autonomous mode
makes the car brake or change the lane if there is no coming
car in the opposite lane.
Fig. 3: The first threshold warning
Fig. 4: The second threshold warning
β’ When the LV stops, the FV brakes before the collision occurs
as shown in Fig. 5.
Fig. 5: The third threshold warning
B. Autonomous Mode Scenario
β’ The LV is moving slowly, while the FV is moving in higher
speed and accordingly, the FV should pass the LV by
changing the current lane then back to the original lane. In
doing this lane change action, the algorithm ensures that there
is no coming vehicle in the other lane to avoid the collision
as shown in Figs. 6 and 7.
Fig. 6: The Following Vehicle acts to skip the Leading Vehicle
741
Fig. 7: The Following Vehicle after skipping the Leading Vehicle
β’ The same previous scenario is repeated in case there is
another vehicle coming in the other lane. Thus, the FV should
deaccelerate and never change the lane to avoid the collision
with the coming vehicle in the other lane as shown in Figs. 8
and 9.
Fig. 8: The autonomous mode with coming vehicle in another lane
Fig. 9: The Following Vehicle decelerate and keep its lane
β’ The same previous scenario and still there is another vehicle
coming in the other lane, but in this scenario the FV wants to
skip the LV. Therefore, the FV should wait the other lane
vehicle to pass first and then take the action of skipping the
LV by changing the lane as shown in Figs 10, 11 and 12.
Fig. 10: The autonomous mode with coming vehicle in another lane
Fig. 11: The Following Vehicle waits the other lane vehicle to pass then changes the lane
Fig. 12: The Following Vehicle after skipping the Leading Vehicle
IV. CONCLUSION
In this work, a proposed forward collision avoidance algorithm
based on the vehicle acceleration which is used for warning the
vehicle driver by three warning levels. Also, support
autonomous mode for lane change and skip the Leading Vehicle
to avoid the collision in all scenarios. The proposed algorithm
has been tested for different scenarios to ensure the ability to
avoid the collision in all these scenarios using the Prescan
simulation environment and MATLAB.
ACKNOWLEDGMENT
This work was performed on a project at Institute of Aviation
Engineering and Technology (IAET) funded by the Academy
of Scientific Research and Technology (ASRT). This research
was partially funded by ONE Lab at Cairo University, Zewail
City of Science and Technology, and KAUST.
REFERENCES
[1] βRoad traffic deaths,β in World Health Organization, [online] Available:
http://www.who.int/gho/road_safety/mortality/en/
[2] Christopher J.L. Murray and Alan D. Lopez, eds., βThe Global Burden of Disease: A Comprehensive Assessment of Mortality and Disability from
Diseases, Injuries, and Risk Factors in 1990 and Projected in 2020,β Harvard
School of Public Health, Boston, 1996.
[3] H. Fahmy, G. Baumann, M. AbdelGhany, βVehicle Risk Assessment and
Control for Lane-Keeping and Collision Avoidance at Low-Speed and High-Speed Scenarios,β IEEE Transactions on Vehicular Technology, 2018.
[4] H. Fahmy, G. Baumann, M. AbdelGhany, and H. Mostafa, βV2V-Based
Vehicle Risk Assessment and Control for Lane-Keeping and Collision
Avoidance,β IEEE International Conference on Microelectronics (ICM 2017), Beirut, Lebanon, pp. 61-65, 2017.
[5] Mohamed Saeed, Hanan Kamal, Mona El-Ghoneimy, βNovel Type-2 Fuzzy Logic Technique for Handover Problems in a Heterogeneous Network,β
Engineering Optimization Journal, 2017, DOI:
10.1080/0305215X.2017.1402012.
[6] SAE Std. J2735 SAE Int., DSRC Committee, βDedicated Short Range Communications (DSRC) Message Set Dictionary,β Nov. 2009.
[7] F. Ivis, βCalculating geographic distance: Concepts and method,β 2006. [Online]. Available: http://www.lexjansen.com/nesug/nesug06/dm/.
[8] Knipling, R., Mironer, M., Hendricks, D., Tijerina, L., Everson, J., Allen,
J., & Wilson, C., βAssessment of IVHS countermeasures for collision
avoidance: Rear-end crashes,β DOTHS-807-995. Washington, DC: National Highway Traffic Safety Administration, 1993.
[9] American Association of State Highway and Transportation Officials. βA
Policy on Geometric Design of Highways and Streets,β 2001.
[10] R.G. Lerner, G.L. Trigg, Encyclopedia of Physics (2nd Edition), VHC
Publishers, ISBN (Verlagsgesellschaft) 3-527-26954-1 (VHC Inc.) 0-89573-
752-3, 1991.
742