Mobile Robot Localization Using an Evolutionary Particle Filter

Abstract

This paper presents an evolutionary particle filter method for mobile robot localization. Particle filter suffers from the impoverishment problem and needs a large number of particles to represent the system posterior probability density function (PDF). In order to improve the performance of PF, the Selection, Crossover and Mutation operations in evolutionary computation are introduced into PF to optimize the samples and make them move towards regions with large value of PDF, as well as improve the diversity of samples and reduce sample impoverishment phenomenon. Simulation results show that, when compared with PF, the evolutionary particle filter algorithm needs fewer particles and is more precise and robust for mobile robot localization.