A Safe Topological Waypoints Searching-Based Conservative Adaptive Motion Planner in Unknown Cluttered Environment

Abstract

Autonomous navigation of unmanned aerial vehicles (UAVs) in unknown and complex environments is still a challenge. Because the environment is partially observable to the drone, it is hard to consider trajectory safety and exploration efficiency simultaneously in autonomous navigation. In this paper, we present a motion planning method composed of a geometrically topological waypoints searching method and an adaptive trajectory replanning framework, which improves trajectory safety without sacrificing navigation efficiency. Our waypoint searching approach considers the safety distance and reduces pathfinding’s search space by extracting some feasible path points on both sides of the obstacle. And this is based on the ESDF gradient and geometry information of a given obstacle. Besides, the found waypoints keep a safe distance from the obstacles, making the method work well in a scene that contains large obstacles. Based on the waypoint searching method, we proposed an adaptive trajectory replanning framework to improve trajectory safety and navigation efficiency further. The replanning procedure is event-triggered. When the planned trajectory is too close to an obstacle according to our safe condition, the trajectory will be re-planned. The proposed method is tested extensively in various simulation environments. Results show that the trajectory safety of our method is improved by 27.8%, and the computing time for replanning is reduced by 90.8% compared to the state-of-the-art method.