To address stability challenges in intelligent vehicle lateral control, this study proposes a feedforward predictive LQR controller optimized via hybrid Particle Swarm Optimization and Genetic Algorithm. First, an LQR controller with feedforward compensation was designed to eliminate steady-state errors. Second, to resolve LQR parameter tuning difficulties, a PSO-GA-based intelligent optimization framework was used to optimize the LQR weighting matrices, establishing an optimized LQR controller. Third, considering the limited predictive capability of conventional LQR controllers for future states, a fuzzy predictive timing regulation module was integrated with the optimized LQR controller, yielding an optimized fuzzy-LQR controller to enhance dynamic response velocity and stability. Finally, co-simulation experiments were conducted using CarSim and Matlab/Simulink platforms under three operational velocities: 10 m/s, 20 m/s, and 30 m/s. The results demonstrate that the optimized fuzzy-LQR controller achieves superior path-tracking performance across all tested velocities. At 10 m/s, both optimized controllers achieved maximum lateral errors of 0.05 m, representing a 50% reduction compared with the 0.10 m error of the traditional LQR controller. At 20 m/s, the optimized fuzzy-LQR controller restricted maximum error below 0.08 m, outperforming the conventional and optimized LQR controllers, exhibiting 0.12 m fluctuations by 33.3%. Under 30 m/s high-speed conditions, the optimized fuzzy-LQR maintained errors at 1.07 m, measuring 8.5% lower than the optimized LQR controller value of 1.17 m and 13.7% lower than the traditional LQR controller value of 1.24 m. Experimental evidence confirms that while both PSO-GA optimized controllers deliver competent low-speed performance, the fuzzy predictive enhanced controller achieves unequivocal superiority in tracking precision, system stability, and ride comfort during medium to high-speed operations.
Wang et al. (Thu,) studied this question.