Candidate: Mahmoud S Faraj
Title: Cooperative Autonomous Vehicle Speed Optimization Near Signalized Intersections
Date: November 30, 2018
Time: 9:30 AM
Place: EIT 3142
Supervisor(s): Gaudet, Vincent - Fidan, Baris (Mechanical & Mechatronics)
Road congestion in urban environments, especially near signalized intersections, has been a major cause of significant fuel and time waste. Various solutions have been proposed to solve the problem of increasing idling times and number of stops of vehicles at signalized intersections, ranging from infrastructure-based techniques, such as dynamic traffic light control systems, to vehicle-based techniques that rely on optimal speed computation. However, all of the vehicle-based solutions introduced to solve the problem have approached the problem from a single vehicle point of view. Speed optimization for vehicles approaching a traffic light is an individual decision-making process governed by the actions/decisions of the other vehicles sharing the same traffic light. Since the optimization of other vehicles' speed decisions is not taken into consideration, vehicles selfishly compete over the available green light; as a result, some of them experience unnecessary delay which may lead to increasing congestion. In addition, the integration of dynamic traffic light control system with vehicle speed optimization such that coordination and cooperation between the traffic light and vehicles themselves has not yet been addressed.
As a step toward technological solutions to popularize the use of autonomous vehicles, this thesis introduces a game-theoretic-based cooperative speed optimization framework to minimize the idling times and number of stops of vehicles at signalized intersections. This framework consists of three modules to cover issues of autonomous vehicle individual speed optimization, information acquisition and conflict recognition, and cooperative speed decision making. It relies on a linear programming optimization technique and game theory to allow autonomous vehicles heading toward a traffic light cooperate and agree on certain speed actions such that the average idling times and number of stops are minimized. In addition, the concept of bargaining in game theory is introduced to allow autonomous vehicles trade their right of passing the traffic light with less or without any stops. Furthermore, a dynamic traffic light control system is introduced to allow the cooperative autonomous vehicles cooperate and coordinate with the traffic light to further minimize their idling times and number of stops. Simulation has been conducted in MATLAB to test and validate the proposed framework under various traffic conditions and results are reported showing significant reductions of average idling times and number of stops for vehicles using the proposed framework as compared to a non-cooperative speed optimization algorithm. Moreover, a platoon-based autonomous vehicle speed optimization scheme is posed to minimize the average idling times and number of stops for autonomous vehicle connected in platoons. This platoon-based scheme consists of a linear programming optimization technique and intelligent vehicle decision-making algorithm to allow vehicles connected in a platoon and approaching a signalized intersection decide in a decentralized manner whether it is efficient to be part of the platoon or not. Simulation has been conducted in MATLAB to investigate the performance of this platoon-based scheme under various traffic conditions and results are reported, showing that vehicles using the proposed scheme achieve lower average values of idling times and number of stops as compared to two other platoon scenarios.
200 University Avenue West
Waterloo, ON N2L 3G1