Comparison between A* and RRT Algorithms for UAV Path Planning

Abstract

Unmanned Aerial Vehicles (UAVs) are being integrated into a wide range of indoor and outdoor applications. In this light, robust and efficient path planning is paramount. An extensive literature review showed that the A* and Rapidly{Exploring Random Tree (RRT) algorithms and their variants are the most promising path planning algorithms candidates for 3D UAV scenarios. These two algorithms are tested in different complexity 3D scenarios consisting of a box and a combination of vertical and horizontal plane obstacles with apertures. The path length and generation time are considered as the performance measures. The A* with a spectrum of resolutions, the standard RRT with different step{ size constraints, RRT without step size constraints and the Multiple RRT (MRRT) with various seeds are implemented and their performance measures compared. Results confirm that all algorithms are able to generate a path in all scenarios for all resolutions, step sizes and seeds considered, respectively. Overall A*'s path length is more optimal and generation time is shorter than RRT projecting A* as a better candidate for online 3D path planning of UAVs.