Development of a practical sensor fusion module for environment modeling

Abstract

The goal of this study is to develop a sensor module that models environment for mobile robots from ultrasonic and infrared sensor data in an unknown circumstance. We can obtain distance information from sensors to obstacles using this module that discriminate various obstacles on a segment that is composed of one ultrasonic sensor and two infrared sensors. Infrared sensors are used to reduce defects of ultrasonic sensors such as poor directionality and frequent misreading. With this module, we propose algorithms, the modified HPF sensor fusion algorithm and the geometric sensor fusion algorithm, to represent instant environment from the fused data on obstacles. For this study, we have made a semicircular sensor module that has an array of 6 ultrasonic and 7 infrared sensors. Finally we evaluate the usefulness of the sensor module with the proposed sensor fusion algorithms through several experiments in indoor environment.