Hierarchical Path Planning Method for UAVs Based on Multi-Submap Exploration Strategy and Octree-Topological Hybrid Map

To address the issues of invalid re-entry and significant decline in later exploration efficiency in traditional autonomous exploration algorithms, this paper focuses on rotor UAVs equipped with small depth cameras. When facing an unknown scene without prior information, the UAV relies solely on its own sensors for environmental exploration. Based on a multi-map collaborative exploration strategy, we propose a corresponding UAV hierarchical path planning algorithm. A local-global path planning method using a local grid-topology hybrid map is employed to collaboratively plan the path to the target point. By constructing a wave array diagram coupled with the wavefront detection algorithm, the backtracking node is selected to solve the local path. Real-time map information is utilized to extract topological nodes in real time, construct the topological map, generate the topological numerical model, and plan the global path. We set up an autonomous exploration system for drones in both simulation and real-world scenarios, analyzing the exploration process based on a multi-map collaborative exploration algorithm. The trajectory optimization method that converts points into lines effectively reduced the number of turns and the length of the trajectory, indirectly accelerating the drone's exploration speed and improving flight safety.