A UAV Global Planner to Improve Path Planning in Unstructured Environments

Abstract

A good performance for path planning is essential to carry out real-world missions. In this paper, the path planning consists of a global planner, that finds the optimal path, and in a local planner, recalculates the path to avoid obstacles. The main focus is to improve the performance of local planning techniques decreasing the complexity. There are two main ways to do it: bidirectional algorithms, improving time, and global planners, improving time and completeness. Thus, we propose a global planner algorithm that generates auxiliary nodes, backtracking by the goal node. We perform a comparison among A*, Bi A*, Artificial Potential Field (APF), Bi APF, Rapid Exploring Random Tree (RRT), and Bi RRT with and without the global planner through statistical metrics of time, path length, CPU, and memory. The results show the advantages of using bidirectional algorithms and the proposed global planner. The bidirectional algorithms decrease the time to return to the trajectory and sometimes assist in the algorithm's completeness. The proposed global planner reduced the planning time by 91.6% and improved the completeness of all algorithms in an unstructured indoor environment.