A Study On Human Tracking Of Interactive Service Robots

The research on service robot aims to facilitate human's life and to service everyone. which is complex and dynamic. In order to apply the robot in human's living environmen-t,there are many task to be done, such as following a human or helping people to carry loads. Among them, human tracking in populated environments is one of the most active research areas of service robot. Because of its significant both in the theory and practice.This paper first introduces the research and development status of service robot tech-nology. Then studies on the problem of human tracking in an unknown environment. Two methods on human detecting are proposed based on laser and MESA camera. which are implemented into a multisensor SJPDAF framework. Finally, the algorithm and strategy are analyzed and summarized. Follow by some improvements for practical application.The main work of this paper includes:1.With laser and MESA camera, two human detecting algorithms are researched. Spe-cially, the curve-like detection algorithm is based on laser and a head-shoulder feature de-tection algorithm is proposed based on MESA camera.2.The improved SJPDAF algorithm is proposed, and a robust human following is achieved. The sensor detection models are added into the traditional SJPDAF to unify the measurement framework, realizing the multi-sensor fusion in human tracking. Besides, the efficiency of the multi-sensor SJPDAF is improved as well as the accuracy of tracking.3.An obstacle avoidance strategy which improves the edge detection algorithm is pro-posed. Using this strategy the robot can avoid obstacles on the ground in a narrow space.