ND voxel localization using large-scale 3D environmental map and RGB-D camera

We propose an efficient 3D global localization and tracking technique for a mobile robot in a large-scale environment using 3D geometrical map and a RGB-D camera. With the rapid development of high-resolution 3D range sensors, highspeed processing of a large amount of 3D data is becoming an urgent challenge in robotic applications such as localization. To tackle this problem, the proposed technique utilizes a ND (Normal Distributions) voxel representation. Firstly, a 3D geometrical map represented by point-clouds is converted to a number of ND voxels, and local features are extracted and stored as an environmental map. In addition, range data captured by a RGB-D camera is also converted to the ND voxels, and local features are calculated. For global localization and tracking, the similarity of ND voxels between the environmental map and the sensory data is examined according to the local features or Kullback-Leibler divergence, and optimum positions are determined in a framework of a particle filter. Experimental results show that the proposed technique is robust for the similarity in a 3D environmental map and converges more stable than a standard voxel-based scan matching technique.