Visual tracking of a moving person for a home robot

Abstract

This paper presents a visual tracking system for a home robot to pursue a person. The system works by detecting a human face and tracking a person via controlling a two-degree-of-freedom robot head and the robot body. An image processing system has been developed to extract facial features using a complementary metal-oxide semiconductor (CMOS) web camera. An algorithm is proposed to recognize a human face by using skin colour and elliptical edge information of a human face. A digital signal processing (DSP)-based motor control card is designed and implemented for robot motion control. The visual tracking control system has been integrated on a self-constructed prototype home robot. Experimental results show that the robot tracks a person in real-time.