A multimodal person-following system for telepresence applications

Abstract

This paper presents the design and implementation of a multimodal person-following system for a mobile telepresence robot. A color histogram matching and position matching algorithm was developed for a person-recognition function using Kinect sensors. Robot motion was controlled by adjusting its velocity according to the humans position in relation to the robot. The robot was able to follow the targeted person in various person-following modes, such as the back-following mode, the side-by-side accompaniment mode as well as the front-guiding mode. An obstacle avoidance function was also implemented using the virtual potential field algorithm.