Human tracking and following using sensor fusion approach for mobile assistive companion robot

Abstract

The ability to track and follow target person in intelligent service mobile robot is indispensable. A robust method for tracking and following a target person with a small size mobile robot by integrating single vision sensor and laser range finder is proposed. Instead of stereo-vision, we acquire the distance between mobile robot and target person by single camera. The laser range finder and vision sensor have their respective drawbacks. To compensate the drawbacks of each sensor we present the complementary data fusion approach - covariance intersection, it will complement the uncertainty of each sensor measure and enhance the reliability of human's position information. The virtual spring model is the control rule of mobile robot that can smoothly tracking target person. Experimental results validate the robust performance of the method.