Performance analysis of mobile robot self-localization based on different configurations of RFID system

Abstract

This paper analyzes the MCL (Monte Carlo Localization) performances of an omnidirectional vehicle based on an RFID system with multiple readers at the bottom of a vehicle and the tags on the floor with different configurations of RFID system. This paper also proposes a method of reinitializing of the particles in the MCL and a likelihood function as the measurement model that are specialized to the RFID based self-localization. The simulations demonstrate that how accurately the self-localization works according to the changes of the RFID system's configuration and how the definition of the likelihood function affects the accuracy. The results show that the self-localization based on MCL works accurately enough for realistic situation even using less RFID readers on the vehicle and lower density IC tag textiles on the floor.