摘要
为满足无人车对轨迹的安全可执行要求,设计了一种高实时连续规划算法.三次B样条的轨迹生成,实现结果的连续平滑;结合执行器约束,限制轨迹的曲率,实现轨迹可执行性;考虑安全距离约束,将轨迹优化转化为最短控制线段长度和夹角的求解,实现算法加速.千次循环的平均耗时低于0.040 s,并且峰值不超过0.100 s,表明算法具有高实时性;实车峰值综合误差小于0.62 m,并且平均综合误差小于0.25 m,表明算法能满足避障安全的需求.
In order to meet the safe and executable requirements of unmanned vehicles,a high real-time continuous planning algorithm is designed.Cubic B-spline curve generates continuous and smooth trajectory.Combining with the actuator constraints,the curvature of trajectory is limited to achieve the track enforceability.Considering the constraint of safe distance,the optimization of trajectories is transformed to solve the shortest control line length and included angle,which accelerates the solution of the algorithm.The average time consumption of thousands of cycles is less than 0.040 s and the peak is no more than 0.100 s,which means that the algorithm has high real-time performance.The peak comprehensive error of real vehicles is less than 0.62 m and the average comprehensive error is less than 0.25 m,which means that the algorithm can meet the requirements of obstacle avoidance safety.
作者
曾德全
余卓平
张培志
陈慧勇
ZENG Dequan;YU Zhuoping;ZHANG Peizhi;CHEN Huiyong(School of Automotive Studies,Tongji University,Shanghai 201804,China;Clean Energy Automotive Engineering Center,Tongji University,Shanghai 201804,China;Zhengzhou Yutong Bus Co.,Ltd.,Zhengzhou 450016,China)
出处
《同济大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2019年第S01期159-163,共5页
Journal of Tongji University:Natural Science
基金
国家重点研发计划(2018YFB0105103)
科技部国家重点研发专项(17092430047)
上海市工业强基专项(GYQJ-2017-4-08).
关键词
无人车
避障
轨迹规划
三次B样条曲线
unmanned vehicle
obstacle avoidance
trajectory planning
cubic B-spline curve
作者简介
第一作者:曾德全(1988-),男,博士生,主要研究方向为无人车决策规划。E-mail:zdq1610849@126.com
