Simulation of decentralized algorithm for collision avoidance between autonomous vehicles at road intersection
Date of Issue2018-09-24
School of Electrical and Electronic Engineering
This dissertation involves implementation of collision avoidance technique for autonomous vehicles at road intersections using V2X (Vehicle to everything) technology. In the modern traffic scenario, road intersections where multiple roads meet, are most prone to accidents. Special care needs to be taken to prevent accidents at these locations. One way of doing that is use of traffic signaling which involves different states of traffic light to control the vehicle flow at intersection points. However, the use of traffic signals at busy intersections leads to increase in fuel consumption as well as waiting time. V2X technology can be used to replace the traffic signals resulting in less fuel consumption and better traffic flow. Recent research in collision avoidance using V2X technology has resulted in use of either centralized or decentralized algorithm to prevent collision at intersection. In centralized approach, a road side unit (RSU) can be used at the intersection to schedule the vehicles approaching intersection and adjust the vehicle speed accordingly. In decentralized approach, the on-board unit (OBU) is placed in each vehicle and the vehicles communicate their information among each other to prevent collision at intersection. The vehicles adjust their speed accordingly to avoid collision at intersection by minimizing the control input in the form of acceleration or brake and by minimizing the difference between the velocity of the vehicle at different time instants and the desired velocity. In this dissertation project, a decentralized collision avoidance strategy at intersection has been implemented. An intersection model is created using VISSIM. Vehicle parameters are passed from VISSIM to MATLAB using the COM interface. MATLAB executes the collision avoidance algorithm and computes the required control inputs for the vehicles. These inputs are then used by VISSIM to visualize the performance of the algorithm.
DRNTU::Engineering::Electrical and electronic engineering