人机共驾型车道保持辅助控制策略

Lane Keeping Assistant Control Strategy for Human-Machine Co-driving

车道保持辅助系统对于汽车横向安全具有重要意义,然而在实际使用中往往存在人机冲突等问题,为此,文章提出了人机共驾型车道保持辅助控制策略。基于实车平台搭建了驾驶员驾驶数据采集系统并对驾驶数据进行实时采集;根据学习向量量化神经网络理论,利用实车驾驶数据训练并生成驾驶员驾驶意图辨识模型;设计人机共驾型车道保持辅助控制系统构架,基于偏离预警模型确定辅助系统的介入条件,基于单点预瞄最优曲率驾驶员模型输出能够使车辆快速回正的最优转向盘转角,设计模糊控制系统实现对共驾系数的实时计算;搭建驾驶员在环平台并进行测试验证。结果表明,所提出的车道保持辅助控制策略能够在对驾驶员驾驶意图进行准确辨识的基础上,根据车辆偏离状态,提供合理的辅助的机器转角以实现人机共驾型车道保持辅助控制。

Lane keeping assistant system is of great significance to vehicle lateral safety.However,there are many problems such as human-machine conflict in actual use.This paper proposes a lane keeping assistant control strategy under human-machine cooperation.Based on the theory of learning vector quantization neural network,the driver's driving intention identification model is constructed.The lane keeping assistant control system framework of human-machine cooperation is designed,determining the intervention conditions of the assistant system based on the deviation early warning model,using optimal curvature driver model of single point preview to calculate the optimal steering wheel angle,designing the fuzzy control system to realize the real-time calculation of co-driving coefficient.The test is carried out by the driver in the loop platform.The results show that the lane keeping assistant control strategy proposed in this paper could accurately identify the driving intention and provide a reasonable assisted machine corner according to the vehicle's deviation status to realize the lane under human-machine cooperative driving.