Localization for mobile robots with a multi-ToF camera system

Abstract

The localization of mobile robots is the core function for the various solutions utilizing autonomous mobile robots. By comparison with 3D LiDAR which is expensive and bulky, the time-of-flight (ToF) camera is inexpensive and can be miniaturized. So it can greatly contribute to the diffusion of autonomous mobile robots. However, due to the limitations, such as mixed pixel effect and multipath interference, the measurement of ToF camera is noisy and inaccurate. Therefore, in order to obtain high-quality pose estimation with the low-cost ToF camera, a preprocessing process and localization algorithm considering the characteristics of the ToF camera are essential. We experimented with the possibility of localization using a multi-ToF camera system. In consideration of the FoV of the ToF camera, the scan data was acquired using a mobile robot with a multi-ToF camera system. The scan data was analyzed by comparing it with the high-resolution map acquired with the high-precision 3D LiDAR, and a preprocessing process for localization was implemented, taking account of the analyzed characteristics. By building the localization algorithm utilizing the iterative closest point (ICP) algorithm, the possibility of localization using the multi-ToF camera system was demonstrated.