基于有限状态机的代客泊车决策规划系统研究

Research on the Decision and Planning System of Automated Valet Parking Based on Finite State Machine

针对当前代客泊车研究中对泊车规划起始点的位姿要求较高的问题,提出一种基于分层有限状态机的代客泊车系统决策方法,设计车辆上层功能状态机与下层行为状态机,根据车辆所处环境与定义规则,建立车辆各行为状态之间的逻辑切换关系,完成车辆代客泊车与定点召回功能。为确保车辆在进入泊车状态前处于不确定状态与位姿的情况下,针对垂直车位完成车辆泊入路径的规划,采用多段路径规划,同时对行车与泊车切换区域进行计算,保证车辆在不同位姿情况下,均可以规划出泊车路径。最终通过实车试验,验证了该代客泊车系统方案的可行性与可靠性。

For the problem of high requirements of the posture of the starting point of parking planning in the current automated valet parking research,a decision-making method of valet parking system based on the hierarchical finite state machine is proposed.To realize the functions of valet parking and fixed-point recall of the vehicle,the upper-level functional state machine and the lower-level behavior state machine are designed,and the logical switching relationship between the behavior states of the vehicle is established according to the environment and the defined rules of the vehicle.In order to ensure completion of vehicle parking path planning for the vertical parking space under the condition of entering the parking state in an uncertain state or position and orientation,multi-stage path planning is used,and the switching area between driving and parking is calculated to make sure that the parking path can be planned for vehicles in different positions.The feasibility and reliability of the valet parking system are verified through real-vehicle tests finally.