无人驾驶车辆路径跟踪控制预瞄距离自适应优化

基于道路信息，使用驾驶员预瞄模型产生执行器输入是无人驾驶车辆在路径跟踪中使用的主要方法之一，但对于车速较高与转弯半径小等工况，模型误差会导致较差的驾驶舒适性，车辆甚至失去稳定性。为提高无人驾驶车辆路径的跟踪精度，同时兼顾转向频度和车辆稳定性，提出基于粒子群多目标优化（Particle swarm optimization，PSO）算法的预瞄距离自适应驾驶员模型，并将之应用于路径跟踪控制。首先，基于单点预瞄偏差模型，采用滑模变结构设计转向控制器；其次，以路径跟踪精度、转向频度和车辆稳定性为综合性能指标，设计了PSO优化算法，实现了驾驶员模型预瞄距离的自适应寻优。最后，在搭建的CarSim-Simulink联合仿真平台与台架试验上，对所提出的预瞄距离自适应驾驶员预瞄模型进行了仿真和硬件在环试验验证。结果表明，经优化后的预瞄距离能够适应不同车速和道路曲率，驾驶员预瞄模型能兼顾路径跟踪精度、转向频度和车辆稳定性等需求。预瞄距离自适应驾驶员模型结合道路与车速信息，增大对路况与车况适应性，为无人驾驶车辆路径跟踪控制提供可靠的输入。

Preview Distance Adaptive Optimization for the Path Tracking Control of Unmanned Vehicle

Based on the road information, using the driver preview model to generate the actuator input is one of the main method in unmanned vehicle path tracking, but for higher speed and small turn radius, the error of the driver preview model may induce less driving comfort and the vehicle even lose stability. In order to improve the accuracy of path tracking and ensure its steering frequency and stability, an adaptive optimization control strategy for preview distance is proposed, which is based on particle swarm optimization(PSO) algorithm. Firstly, based on the tracking error, designed a driver preview model by using slide mode control. Secondly, an adaptive algorithm based on PSO is proposed, which takes into account of tracking accuracy, steering frequency and stability. Finally, the proposed algorithm is tested on co-simulation platform of CarSim and Simulink software and driving simulator test bench. The results show that the algorithm is effective and it can ensure path tracking accuracy, steering controllability and stability. The optimized preview distance driver model combined the information of road and speed, and provides reliable input for unmanned vehicle path tracking control.