施楊洋 楊家富 布升強 朱林峰

摘? ?要：針對RRT算法隨機性大、收斂速度慢和偏差性的問題，采用雙向隨機樹和多棵局部隨機樹的探索與合并。增加引力分量，使雙向隨機樹朝著各自目標(biāo)方向生長，減少了算法的隨機性?；谡系K物周圍均勻生成若干根節(jié)點，對根節(jié)點增加斥力分量，生成多棵局部隨機樹?？焖賹ふ铱赏ㄐ械穆窂剑瑴p少擴展過程中對障礙物的檢測時間，加快算法的收斂速度，改善了算法的偏差性。用MATLAB進(jìn)行虛擬仿真，驗證了該算法的正確性。

關(guān)鍵詞：智能車輛;快速搜索隨機樹;路徑規(guī)劃;障礙物斥力函數(shù)

中圖分類號：TP242? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 文獻(xiàn)標(biāo)識碼：A

Improved Intelligent Vehicle Pathing Planning Algorithm Based on RRT

SHI Yang-yang，YANG Jia-fu？覮，BU Sheng-qiang，ZHU Lin-feng

（College of Mechanical and Electronic Engineering，Nanjing Forestry University，Nanjing，Jiangsu 210037，China）

Abstract：Aimed at the problems of large randomness，slow convergence rate and deviation of RRT algorithm，the exploration and merging of bidirectional random tree and multiple local random tree are proposed，which increases the gravitational component and makes bidirectional random tree grow in the direction of the respective target，reducing the randomness. Several root nodes are evenly generated around obstacles，and the repulsion component is added to the root node to generate multiple local random trees. Through this method，the accessible path can be quickly searched，the detection time of obstacles in the expansion process can be reduced，the convergence speed of the algorithm can be accelerated，and the deviation of the algorithm can be improved. The improved algorithm is simulated by MATLAB software，which verifies its correctness.

Key words：intelligent vehicle;RRT（Rapidly-Exploring Random Tree）;path planning;obstacle repulsion function