基于权值惩罚法自适应人机协同避撞策略

针对协同避撞过程中人机权值难以适应不同驾驶员与工况的问题,提出了一种自适应人机协同避撞策略。基于单点预瞄驾驶员模型,根据驾驶员决策意愿搭建了表征驾驶风格的驾驶员模型。基于距离策略,提出避撞模式切换策略,设计模型预测控制器跟踪规划路径。提出驾驶员权值惩罚法,首先根据驾驶员参数与道路条件对驾驶员权值进行先验条件的惩罚,其次引入碰撞、侧滑和越出车道三种事故风险度,设计模糊控制器,依据事故风险度对避撞过程中的驾驶员权值进行实时惩罚。仿真结果表明所提出的人机协同避撞策略可实时调整人机权值,显著降低事故风险,从而在协调人机有效避撞同时保证行驶安全性与稳定性。

An Adaptive Human-machine Cooperative Collision Avoidance Strategy Based on Weight Penalty Method

In order to solve the problem that the man-machine weight was difficult to adapt to different drivers and working conditions in collaborative collision avoidance, an adaptive human-machine collaborative collision avoidance strategy was proposed. Based on a single-point preview theory, the driver model of the driving style was built according to the willingness to the driver"s decision. Based on the distance strategy, the collision avoidance mode switching strategy was proposed, and the model prediction controller was designed to track the planned path. The driver weight penalty method is proposed. Firstly, the driver"s weight was penalized according to the driver"s parameters and road conditions. Secondly, three accident risk degrees of collision, sideslipping and out of the lane were introduced, and a fuzzy controller was designed. The simulation results show that the proposed human-machine cooperative collision avoidance strategy can adjust the weight in real time and significantly reduce the accident risk, so as to effectively coordinate human-machine collision avoidance and ensure the safety and stability of driving.