摘要
针对自动泊车算法对自车初始位姿要求较高的问题,设计了一套自动泊车决策规划算法,对自车初始位姿无要求。设计了库位跟踪算法,在泊车过程中不断更新库位坐标,并以此在关键点进行后续规划,保证最终泊车姿态理想;针对平行库位长度不满足一次规划入库最短需求及垂直泊车中停车通道宽度不足的情况分别提出了多次规划算法和动态调整算法。通过Simulink和Pre Scan联合仿真及实车试验进行了验证,证明了上述算法的有效性。
An autonomous parking system was designed that had no requirement on vehicle's initial position.Parking spot tracking algorithm was designed,which could continuously update coordinate of parking spot,and make subsequent planning at critical points,to ensure an ideal parking position.In case the actual length of parallel spot does not satisfy the minimum required length of planning path,and parking channel has insufficient width in vertical parking,the multiplanning algorithm and dynamic adjustment algorithm were proposed.These above algorithms were validated in Simulink and PreScan simulations as well as real vehicle tests.
作者
熊璐
严森炜
余卓平
张培志
Xiong Lu;Yan Senwei;Yu Zhuoping;Zhang Peizhi(Tongji University,Shanghai 201804)
出处
《汽车技术》
CSCD
北大核心
2018年第8期18-26,共9页
Automobile Technology
基金
国家自然科学基金项目(U1564207)
上海市科委项目(17DZ1100202
16DZ1100700)
16DZ1100700
关键词
自动泊车
库位跟踪
多次规划
动态调整
卡尔曼滤波
Autonomous parking
Closed-loop update of parking spot
Multi-planning,Dynamic adjustment
Kalman filter
