Mobile robot global localization using particle filters

Abstract

Mobile robot global localization is the problem of determining a robotpsilas pose in an environment by using sensor data, when the initial position is unknown. Particle filter based Probabilistic algorithm called Monte Carlo localization is the current popular approach to solve the robot localization problem. In this paper we introduce the multi-sensor based Monte Carlo Localization (MCL) method which represents a robotpsilas belief by a set of weighted samples and use the laser range finder (LRF) sensor to measurement update. We also proposed likelihood based particle filter to solve the kidnapped problem. The experiment results illustrate the efficiency and robustness of particle filter approach for our mobile robot.