Abstract: Aimed at the problem that it is difficult for unmanned ground vehicles to track complex road conditions precisely at high speed in off-road environment, a feedforward compensation controller with self-learning parameters is designed, which can form a feedforward-feedback controller framework with the traditional model predictive controller. In the framework, the learning coefficients of the feedforward controller can be updated online according to the tracking errors of the real-time states. The controller framework can effectively consider the effects of nonlinear dynamics which cannot be accurately modeled during the high-speed movement of the vehicle, as well as the constantly changing curvature and road conditions, etc. The proposed control method can quickly reduce tracking errors while ensuring stability. The actual vehicle experiments of high-speed path tracking along S-shape and right-angle shape paths are carried out in the off-road scenes, and the results show that compared with the traditional model predictive controller, the designed controller has smaller tracking errors and yaw, and the tracking accuracy and vehicle stability have been greatly improved.

Key words: unmanned ground vehicles, path tracking, parameters self-learning, model predictive control, off-road environment

摘要： 针对无人车在越野环境下难以高速、高精度地跟踪复杂路况的问题，设计了一种参数自学习的前馈补偿控制器，与模型预测控制方法构成前馈-反馈的控制结构。在该控制结构中，前馈控制根据实时状态的跟踪误差在线更新学习系数，有效考虑车辆高速运动过程中无法精确建模的非线性动力学特性以及复杂路况不断变化的曲率和路面条件等的影响，在保证稳定性的同时快速减小跟踪误差。在越野场景进行了高速的S型与直角弯路径跟踪实车实验来验证参数自学习控制器的有效性，结果表明，所设计的参数自学习控制器相比传统的模型预测控制器跟踪误差和横摆都较小，在跟踪精度和车辆稳定性上都有较大改善。

关键词: 无人车, 路径跟踪, 参数自学习, 模型预测控制, 越野环境