APG-RRT: Sampling-Based Path Planning Method for Small Autonomous Vehicle in Closed Scenarios

Abstract

To address the shortcomings of the classical RRT (Rapidly exploring Range Tree) path planning algorithm, such as long planning time and path curvature in some narrow and complex environments, an improved APG-RRT (Adaptive Path Guide RRT) algorithm is proposed. First, a guiding path is introduced in the sampling stage, and a node on the preset guiding path is selected first to expand the node tree so as to guide the algorithm to plan the exploration process. Secondly, the selection weight of the loading guidance path is dynamically adjusted according to the probability of collision between obstacles during the exploration process, and a safe and feasible path point is generated in the path expansion stage by combining the expansion information of the obstacles. Finally, in the path post-processing stage, combined with the vehicle kinematic constraints, the triangular inequality method is used to remove redundant path points, making the path more smooth, so as to fulfill the specific operational needs of the vehicle. The results of the simulation experiment demonstrate that the suggested method exhibits superior planning efficiency compared to the previous algorithm, resulting in a higher quality final path. At the same time, the verification of the algorithm's feasibility and effectiveness is conducted through real car testing.