Local Path planning for an Autonomous Vehicle in a Non-Deterministic Environment

Abstract

One of the biggest problems that face the production of an autonomous vehicle is finding a clear path where the vehicle steers away from static and moving obstacles to prevent any possible accidents, and yet continues moving toward the required destination to fulfill its goal of reaching a desired location. This thesis focuses on the problem of finding a feasible trajectory and a collision-free path for an autonomous vehicle with the presence of many moving obstacles in the environment. We approach this issue by using an analytical method where we consider the kinematic model of the car and the surrounding environment to develop a feasible trajectory with the corresponding steering control and the collision avoidance conditions. The collision-free path generated can be updated in real time once the environment is changed and detected by the car. Matlab simulation is used to provide the results for the analytical method used where obstacles were sometimes fed into the simulation and sometimes randomly generated. Also, an intersection crossing scenarios were simulated to show the ability of the analytical method to be used in a more constrained environment.