Path planning for mobile robots in dense spaces using rapidly-exploring random trees (RRT) algorithm

Abstract

Highly dense spaces such as narrow pathways are a great challenge for path planning. The goal of this study is to develop feasible paths in highly constrained spaces for a mobile robot. In this study, the rapidly-exploring random trees (RRT) algorithm was used to generate paths. The RRT was subjected to two workspaces of the same dimensions but with different obstacle size. By varying the obstacle size, the free regions available will adjust depending on the obstacle.  The step size of the RRT algorithm was varied incrementally from 20, 40, 60, 80 and 100. For each step size variation, the path length and path coordinates were retrieved. Small step size always produced successful search results. Null searches were encountered in the workspace with large obstacles, most especially at high step size. The step size 100 has the highest number of null searches with 21 out of 40 trials. Smaller mean free path length and standard deviation were obtained in the environment with smaller obstacles. Practical applications of the study are in closed-loop path planning in a workspace wherein robot-robot and robot-obstacle interactions are the dominant features.