Abstract: The lane centering control (LCC), one of the advanced assisted driving system (ADAS) feature, enables the vehicle to drive in the center of the lane by requesting torque or angle to the electric power steering (EPS) system. LCC has strict requirements of the EPS response, but the steering system is complex and has many influencing factors. These parameters change often. Research has found that when the control parameters of LCC are determined, changes of the response parameters of EPS will affect the performance of LCC. To reduce the impact of EPS parameter changes on LCC performance, a self-learning algorithm for EPS response deviation is proposed, and proportion, integral and differential (PID) compensation is used on the selflearning to output requests of LCC. After CarSim/Simulink simulation and actual vehicle testing, it is found that the impact of EPS response deviation on LCC performance is significantly improved, which is comparable to the acceptance results at the completion stage of development, and state the strategy is efficient.

Key words: lane centering control; response deviation; self-learning; compensation algorithms

摘要： 高级辅助驾驶系统（ADAS）中的车道居中辅助（LCC）系统，通过 LCC 控制器向电 动助力转向（EPS）系统请求扭矩或转角，再由 EPS 系统响应 LCC 的请求，实现车辆行驶于 车道中心区域。LCC 对 EPS 响应有严格要求，但转向系统复杂、影响因素众多，EPS 响应能 力常有变化。研究发现，当 LCC 的控制参数确定后，如果 EPS 的响应性能变化，会不同程度 地影响 LCC 的表现。为减小车辆的 EPS 响应参数变化对 LCC 的性能影响，提出了对 EPS 响 应偏差的自学习算法，并通过比例-积分-微分（PID）补偿至 LCC 的请求输出，经 CarSim/ Simulink 模拟及实车测试后，发现 EPS 的响应偏差对 LCC 的影响表现，有明显的改善，与开 发完成阶段验收结果相当，证明算法有效。

关键词: 车道居中控制；响应偏差；自学习；补偿算法