Evolutionary Filter for Mobile Robot Global Localization

Abstract

Mobile robot global localization aims to determine the robot's pose in a known environment in absence of initial robot's pose information. This article presents an evolutive localization algorithm known as Evolutive Localization filter (ELF). Based on evolutionary computation concepts, the proposed algorithm search stochastically along the state space the best robot's pose estimate. The set of pose solutions (the population) represents the most likely areas according the perception and motion information received. The population evolves by using the observation and motion errors derived from the comparison between observed and predicted data obtained from the probabilistic perception and motion model. The resulting global localization module has been tested in a mobile robot equipped with a laser range finder. Experiments demonstrate the effectiveness, robustness and computational efficiency of the proposed approach.