Adaptation of Ultra Wide Band positioning system for Adaptive Monte Carlo Localization

Abstract

Self-localization problem of mobile robots is a crucial part of autonomous navigation, especially indoor environments, where the usage of GNSS technology is impossible. There is a lot of studies that focus on this issue, starting from those based on the Monte Carlo Localization (MCL) method, and ending on a variety of SLAM approaches. In majority of them robots explore their operation areas with their sensors and estimate their spatial pose by correlating measurements with known or estimated maps. The UWB (Ultra-Wide Band) technology allows for improvement of these methods since a local positioning system based on it becomes available in indoor environments. This paper proposes an innovative approach to Adaptive Monte Carlo Localization (AMCL) method, which adapts UWB transceivers as a local positioning system. The advantages of applying UWB in AMCL are proved by experimental tests with Automated Guided Vehicle (AGV) in indoor environment. The standard odometry source was substituted by the local positioning system based on UWB transceivers through a modification of original AGV's ROS (robot operating system) setup. Results present that the performance of UWB-AMCL is better than AMCL in most cases. Further advantage of the proposed approach is to enable the standard ROS navigation stack use in cases where no odometry information is available.