A Hybrid Learning and Optimization-Based Path Tracking Control Strategy for Intelligent Electric Vehicles

This paper proposes a hierarchical control framework designed to enhance the path tracking accuracy of intelligent electric vehicles under diverse operating conditions. For lateral control, an improved model predictive control strategy is developed, utilizing a fuzzy inference system for parameter initialization and a Deep Deterministic Policy Gradient algorithm for online adaptive tuning. For longitudinal control, a proportional–integral–derivative controller is optimized via a hybrid genetic algorithm–particle swarm optimization method. Co-simulations conducted in CarSim/Simulink under straight-line, double-lane-change, and double-sine-wave maneuvers demonstrate that the proposed framework significantly reduces lateral deviation and heading error while ensuring smoother actuator response. Compared to conventional MPC and PID controllers, the proposed method reduces maximum lateral error by over 50% and settling time by 60%, confirming its effectiveness and robustness in complex tracking scenarios.