ORB-SLAM based humanoid robot location and navigation system

Abstract

Abstract. Aiming at the indoor location and navigation problem of humanoid biped robot with complex motion structure, a humanoid biped robot localization and navigation system based on ORB-SLAM is designed. Firstly, the working principle of ORB-SLAM is analyzed, and it is improved to realize the function of missing map reading and generating dense point cloud map. Secondly, the dense point cloud map is converted to octomap, and then the conversion of 3D map to 2D map is completed. The SBPL planning library is improved to carry out the path planning of the robot, and the path planning based on the boundary exploration is realized. Finally, the experimental verification is carried out on the biped robot to verify the effectiveness of the location and navigation system design in the indoor environment.