Indoor mobile robot localization using probabilistic multi-sensor fusion

This paper presents a probability fusion methodology with a decision rule for a feature extraction of indoor environment. Two range sensors with complementary property are equipped on a mobile service robot. One is a servo sonar ring composed by Polaroid 6500 ultrasonic rangers and the other is a Hokuyo infrared range-finder. For a real indoor environment is usually composed of different objects with variant material characteristics such as sound absorbed or light refraction. Theses may cause the sensor measurement failure to imperil the localization estimation procedure of an indoor service robot. Thus, multi-sensor fusion with a decision methodology is necessary for feasibility and reliability while service mobile robot is working in a compound indoor environment.