Longitudinal Control of Autonomous Electric Vehicles via Genetic Algorithm tuned PID and Lyapunov-based Adaptive Control: A Robustness Evaluation Under Dynamic Uncertainties

Abstract

The proposed work consists in a significant contribution to the field of motion planning and control of autonomous electric vehicles. Sinusoidal and S-curve speed profiles are developed and discussed to enable smooth transitions while minimizing jerks that could affect vehicle stability and passenger comfort. These assessments build a foundation for longitudinal control. A conventional PID controller is first implemented to achieve basic speed tracking performance. While it performs adequately, it demonstrates notable limitations. To enhance this latter, the PID gains are optimized using a metaheuristic optimization technique based on Genetic Algorithms (GA). This approach automatically tunes the controller parameters, resulting in improved dynamic response and reduced tracking errors. However, due to their fixed-parameter structure, both classical and optimized PID controllers encounter difficulties in adapting to system variations. To overcome these challenges, Lyapunov-based Model Reference Adaptive Control (MRAC) mechanism is cautiously designed to dynamically adjust the system parameters, compensating for the uncertainties, disturbances, and non-linearities under which real vehicles inherently operate under. Simulation results are finally conducted to validate the proposed control system.