摘要
针对无人驾驶汽车,提出了一种基于改进快速搜索随机树(RRT)的轨迹规划算法。首先,基于车辆转向约束,在传统RRT算法的基础上,限制采样可行区域,基于目标偏向策略,以一定的概率将目标点作为随机点进行随机树扩张;其次,对生成路径采用三次B样条曲线平滑处理,降低轨迹曲线的曲率及曲率变化率;最后,基于Matlab/CarSim联合仿真平台构建直线路况与直角路况,并使用提出的算法在不同路况下进行轨迹规划的仿真与分析。结果表明:改进RRT算法下规划轨迹的曲率均值、极大值与方差小,相比于无约束控制,15°转角约束下在直线路况中分别降低了22. 9%,51. 7%与43. 3%,在直角路况中分别降低了29. 5%,59. 3%与40. 3%;同时,改进RRT算法的平均采样点数及平均树节点数少、规划时间短,相比于无转角约束,15°转角约束下在直线路况中分别降低了62. 9%,38. 2%与78. 5%,在直角路况中分别降低了34. 0%,41. 9%与27. 4%,路径规划质量和速度显著提高。
A trajectory planning algorithm based on the improved RRT algorithm was proposed for autonomous vehicle.Firstly,with the vehicle steering kinematics constraints and the traditional RRT algorithm,the feasible region was restricted.With the target bias strategy,the target point was regarded as random point with a given probability to expand the random tree.Secondly,cubic B-spline curve smoothing was applied to the path to reduce the curvature and curvature change rate of the trajectory curve.Finally,based on the Matlab/CarSim co-simulation platform,the straight road conditions and the right angle road condition were constructed,and the proposed algorithm was used to simulate and analyze the trajectory planning under different road conditions.The results showed that the mean,maximum and variance of the planned trajectory curvature of the improved RRT algorithm were smaller.Compared with unconstrained algorithm,the three values under 15 steering angle constraint on straight road were reduced by 22.9%,51.7%and 43.3%respectively,and on right angle road were reduced by 29.5%,59.3%and 40.3%respectively.At the same time,with the improved RRT algorithm,the average number of sampling points and the average number of tree nodes of were fewer,and the planning time was shorter.Compared with unconstrained algorithm,this three values under the 15°steering angle constraint on straight road condition were reduced by 62.9%,38.2%and 78.5%respectively,and on right angle road were reduced by 34.0%,41.9%and 27.4%respectively.The quality and the speed of path planning were improved significantly.
作者
贺伊琳
高奇
赵丹
刘伟
HE Yilin;GAO Qi;ZHAO Dan;LIU Wei(School of Automobile,Chang′an University,Xi′an 710064,China)
出处
《西北大学学报（自然科学版）》
CAS
CSCD
北大核心
2018年第5期651-658,共8页
Journal of Northwest University（Natural Science Edition）
基金
国家自然科学青年基金资助项目(51507013)
中国博士后基金资助项目(2017M613034)
陕西省博士后基金资助项目(2017BSHEDZZ36)
关键词
无人驾驶汽车
转向约束
改进RRT算法
目标偏向
三次B样条曲线
autonomous vehicle
steering constraint conditions
improved RRT algorithm
target bias
cubic B-spline curve
