车道保持系统的车辆状态预测模型

摘要:

针对目前车道保持系统中车速变化较大时，被控车辆偏移量较大的问题，通过状态预测方法，设计了用于单目视觉车道保持系统的状态预测模型。单目摄像头识别预瞄点处车-路偏差，考虑车辆经过预瞄时间后的状态变化量，设计基于车辆状态的预瞄点处车-路偏差模型。通过仿真与实车试验验证，说明提出的车辆预测模型能够准确预测出预瞄点处的车-路偏差，减小预测模型误差，提高车道保持系统控制精度和准确性，一般工况下，能够使车-路偏移量控制在10 cm以下。与目前采用的简化车路偏差模型相比较，能够提高被控车辆对于车道中心线的跟随性能。

Abstract:

To deal with the large offsets when the vehicle speed largely changes in lane keeping system, a vehicle lateral control model was designed based on states preview. Using the car-road deviation detected by monocular camera and the state variation after preview time, the car-road deviation model has been design at preview point. Simulation experiments and the real vehicle tests showed that by using the proposed model, lane keeping system could predict the preview of car-road deviation accurately, decrease preview model error. In general conditions, it could enable vehicles road offset control to maintain control within 10 cm. The algorithm could improve precision and accuracy of lane keeping control system, and improve the performance of follow the road centerline with the current model.