Design of lateral and longitudinal tracking controller based on FA-DAPSO algorithm

Vehicle control is one of the most critical challenges in autonomous vehicles, critical in terms of vehicle safety, passenger comfort, transportation efficiency, and energy savings. Addressing the issues of low control accuracy and inefficient parameter tuning caused by difficulties in selecting the coefficient matrices Q and R in Linear Quadratic Regulator (LQR) and the three parameters of Proportional-Integral-Derivative (PID) controllers, this paper proposes a Dynamic Adaptive Particle Swarm Optimization (FA-DAPSO) algorithm based on swarm velocity and aggregation degree to solve for optimal parameters. The FA-DAPSO algorithm designs learning factors based on the average distance between particles, enhancing the search capability of the swarm. A dynamic inertia weight, referenced by the swarm velocity and particle aggregation degree, is devised to prevent the algorithm from falling into local optima. To validate this algorithm, a two-degree-of-freedom vehicle model is established, and a lateral LQR controller and a longitudinal PID controller are designed. The energy loss function and ITAE (Integral of Time-weighted Absolute Error) index are selected as cost functions to optimize the coefficients respectively. Compared with the best cost achieved by the traditional Particle Swarm Optimization (PSO) algorithm, the FA-DAPSO algorithm improves the performance by 11.2% for the LQR controller and 7.9% for the PID controller. Furthermore, the controllers solved by the FA-DAPSO algorithm can better track the preset trajectory. The simulation results indicate that the controllers optimized by the FA-DAPSO algorithm are capable of selecting distinct parameters in response to varying conditions, which in turn enhances tracking accuracy. Additionally, these controllers exhibit excellent stability and robustness.