Mobile Robot Global Localization Using Particle Swarm Optimization with a 2D Range Scan

Abstract

This paper presents a novel approach based on the particle swarm optimization (PSO) for globally localizing a mobile robot with a single laser scan, under the assumption that the initial pose of the robot is unknown. The environment map is first converted with a signed fitness function that encodes the distance to the nearest obstacle from a given location. Using the end-point model of a laser beam, captured sensor data are associated with the world model without data association or feature extraction. The PSO is then performed to explore the pose space to search for the correct robot pose iteratively, in which the potential solutions are optimized by scan matching technique to get more accurate pose estimation. The proposed approach performs better than the popular particle filter based approach with regard to convergence speed, estimation precision and computational cost. Experiment results based on public domain dataset demonstrate the effectiveness of proposed algorithm.