Sensor Fusion for Vision and Sonar Based People Tracking on a Mobile Service Robot

Abstract

Service robots intended to interact with people must be able to localize and continuously track their users. A method is described which integrates information from visual and sonar based tracking path- ways while updating hypotheses about the position of the robot's human user. Each tracking method uses information from the other to generate a more robust measure of the user's position, and thus a more robust behavior generation is achieved.