An Improved Particle Filter for Service Robot Self-Localization

Abstract

Mobile robot localization is a problem of determining a robot's pose in an environment, which is also one of the most basic problems in mobile robot applications. Recently, introduction of particle filters becomes the most popular approach in mobile robot localization and has been applied with great success to a variety of state estimation problems. In this paper, an particle filter is applied in the authors' service robot position tracking and global localization. Moreover, the posterior distribution of a robot pose in global localization is usually a multi-model due to the symmetry of the environment and ambiguous detected features. Considering these characteristics, we propose an improved cluster particle filter to increase the global localization robustness and accuracy. On-line experiments based detailed analysis of coordinate errors and algorithm efficiency are given. On-line experimental results also show the efficiency and robustness of the approach in the authors' service robot ApriAlpha™ Platform.