Autonomous vehicle path following system design using a combined control approach

Autonomous vehicles represent a central component of future automotive technologies. One of the critical challenges in enabling this technology is the development of path following control systems. This study presents a linear quadratic integral (LQI) optimal control approach improved by particle swarm optimization (PSO), reinforcement learning particle swarm optimization policy (PSO-P), and pure pursuit (PP) for the design of a control system for autonomous vehicle path following. A novel combined control system is proposed by integrating LQI, PSO-P, and PP methods. To further enhance system performance, a feed-forward steering control system is incorporated into all control strategies. The proposed and benchmark control systems are evaluated and compared using time-domain simulations in the MATLAB/Simulink and IPG CarMaker co-simulation environment. Statistical analyses were performed using performance metrics and data from two distinct path curvatures. The results demonstrated the efficacy of the method compared to alternative control strategies.