Enhancing Path Tracking for Autonomous Vehicles using Adaptive PID Control and Particle Swarm Optimisation

Abstract

A path tracking strategy that modifies the classical Proportional-Integral-Derivative (PID) controller with adaptive parameters optimised through a particle swarm optimisation algorithm and supplemented with a feed-forward controller is proposed for self-driving cars in this paper. Such a controller design aims to reduce errors in lateral displacement and heading while promoting smooth, optimal turns. As a part of this optimisation routine, parameters like a Proportionality Constant (Kp), Integration Constant (Ki), Derivative Constant (Kd) and heading error constant (Kh) and a look-ahead parameter would be taken into consideration. The mean absolute error of lateral error is minimised to 0.68993 m, with a root mean square error of 0.76227 m. Furthermore, the maximum lateral error is minimised to 2.0 m. The variance of the steering rate is minimised to 0.62233 (rad/s)2. It reveals that the steering response is pretty smooth. The mean steering angle is also 0.16967 rad. The presence of feedforward control terms leads to a remarkable decrease in oscillations in heading and lateral errors, with peak heading errors less than 5° and lateral errors tending towards zero. The results of the simulation show that the best-designed PID controller will enhance the accuracy of the path-tracking performance.