改进RRT算法在复杂环境下智能车路径规划中的应用

采用快速搜索随机树(RRT)算法进行路径规划时,在存在大量随机障碍物的复杂环境下,规划出的路径曲折且算法无法快速收敛,不能满足智能车路径规划的要求.为了实现智能车路径规划,提出一种基于RRT的运动规划算法——同心圆RRT算法.该算法在RRT算法的基础上结合智能车行驶时自身运动学约束,引入同心圆采样策略和邻近点选择方法....

Application of Improved RRT Algorithm in Intelligent Vehicle Path Planning Under Complicated Environment

When a rapidly-exploring random tree(RRT)algorithm is used for path planning in a complicated environment with many random barriers,the convergence is slow and the obtained path is usually twisted.To meet the requirements of path planning of an intelligent vehicle in a complicated environment,a motion planning algorithm,named homocentric circles RRT algorithm,based on a fast searching random tree is proposed.Based on basic RRT and combined with the kinematic constraints of the intelligent vehicle,the homocentric circles sampling strategy and adjacent point selection method were introduced in the proposed algorithm.The homocentric circles sampling considers the target point as the center;the homocentric circles coefficient m was used to adjust the density of the homocentric circles to generate random points to determine the next path point.Considering the vehicle kinematic constraints and target distance factor,the adjacent point selection method was adopted to calculate the proximity coefficient,and the random tree node corresponding to the minimum proximity coefficient was taken as the adjacent point.For the planned path,a path processing method based on vehicle kinematic constraints was used to simplify the obtained path,and the cubic B-spline curve was employed to optimize the path to generate a smooth and executable path.The results show that the algorithm takes the least time to find the path when the coefficient of homocentric circles is in the range of 0.5-1.5.A larger constraint value for the angle of vehicle attitude and next path point implies that less time is used to find the path and it tends to be stable when the angle is 35°.Under the same environment,the quality of the planned path obtained using the proposed RRT improves considerably compared with the basic RRT,target bias RRT,and updated RRT.Compared with the RRT,target bias RRT,and updated RRT algorithms,the required time and length of the planned path of the proposed RRT algorithm are lower by 43.1%and 18.7%,7.3%and 15.5%,and 29.6%and 7%respectively.Finally,the effectiveness and practicability of the algorithm were verified through the intelligent vehicle experiment.