UAV- Navigation Using Map Slicing and Safe Path Computation

Abstract

This paper is concerned with the safe path planning of a drone while exploring an unknown space. The drone is localized by fusing measurements from sensors including an IMU, RGB-D sensor, and an optical flow system, while executing an RTAB-Map SLAM algorithm. The 3D-occupancy Octomap is generated online and a slicing algorithm is employed to compute 2D-maps. The maps' traversible coordinates are identified and used as potential points for the drone intermediate navigation to the destination. The final segment corresponds to a shortest Chebyshev-length path between all frontier pixels and the endpoint over the unexplored map region. The drone's path is computed using a skeletal path between the identified map boundaries so that the drone moves from its current location through the free map coordinates to the destination point. Simulation studies using within the interior of an apartment indicate the efficiency and effectiveness of the proposed method.