Performance comparison of RRT-series algorithms for mobile robot path planning

Abstract

This paper presents a comparative performance evaluation of three sampling-based path planning algorithms—RRT, RRT-Connect, and RRT*—for mobile robot navigation in complex environments. Experiments are conducted on three obstacle-avoidance maps with two distinct path-length scenarios, using path length, planning time, number of nodes, and tracking error as key metrics. The results show that RRT-Connect achieves the highest computational efficiency with significantly reduced planning time, while RRT* consistently produces the shortest, smoothest paths with the lowest tracking error at the cost of higher computational overhead. Further analysis on different step sizes reveals that larger step sizes improve planning speed but may reduce path optimality or even lead to planning failures in complex mazes. This study highlights the inherent trade-offs between optimality, efficiency, and robustness of RRT-series algorithms, providing practical guidance for selecting appropriate path planning methods and tuning parameters for different robotic navigation tasks.