A Multi-Resolution Frontier-Based Planner for Autonomous 3D Exploration

In this paper we propose a planner for 3D exploration that is suitable applications using state-of-the-art sensors such as lidars, which produce large point clouds with each scan. The based on the detection of frontier - boundary between explored and unknown part environment consists algorithm detecting points, followed by clustering points selecting best to be explored. Compared existing frontier-based approaches, more scalable, i.e. it requires less time same data set size while ensuring similar time. Performance achieved not relying obtained directly from sensor, but mapping algorithm. order cluster use properties Octree representation, allows easy analysis different resolutions. tested in simulation an outdoor test area UAV equipped lidar sensor. results show advantages approach.