A 2D path-planning performance comparison of RRT and RRT* for unmanned ground vehicle

Abstract

In recent years, path-planning has gained significant attention as mobile robots are used in various applications. Several factors determine the optimal path for a mobile robot, including accuracy, length of path, execution time, and turns. Among all planners, sampling-based planners such as rapidly exploring random trees (RRT) and rapidly exploring random trees-star (RRT*) are extensively used for mobile robots. The aim of this paper is the review and performance of these planners in terms of step size, execution time, and path length. All planners are implemented on the Jackal robot in a static environment cluttered with obstacles. Performance comparisons have shown that the reduction of step size results in exploring a greater number of nodes in both algorithms, increasing the probability of each extension succeeding. However, this causes the tree to become denser in both algorithms due to the more explored nodes. The RRT planner requires less execution time when the step size and iteration count are equal to RRT* planners. Moreover, performance plots of both algorithms show that RRT* provides an optimal and smooth path than RRT.