A Comparison of Motion Planning Methods for Autonomous Ground Vehicle Exploration and Search
Apoorva Khairnar, Shathushan Sivashangaran, Azim Eskandarian · 2023
Abstract Navigation of an Autonomous Ground Vehicle (AGV) is a major challenge for complex applications such as exploration of uncharted terrain, search and rescue (SAR) missions, and military assignments. An AGV is subjected to numerous uncertainties such as unexpected obstacles, uneven terrain, and hazardous regions. Motion planning algorithms are responsible for the safe, quick, and energy-efficient autonomous navigation of an AGV in the surrounding environment. In this paper, we compare the performances of conventional motion planning algorithms that include A*, RRT, RRT*, pRRT, and APF for exploration and search tasks. The comparison is carried out in a realistic environment simulated in AutoVRL, a high-fidelity simulator built using open-source tools. This study assumes the presence of a high-level planner that provides the algorithms with environmental information, and application-specific goal positions. The algorithms are compared on the basis of computation time, and efficiency of the generated trajectories, to navigate to goal positions determined by the high-level planner. The results indicate that A* and pRRT perform significantly better than the other tested algorithms, yielding the most efficient trajectories with the least computation time.