Untitled

Abstract

Detecting and tracking a particular person are considered the main tasks of a mobile robot. In this paper, we propose a real-time mobile robot system using 3D Kinect sensor for automatically detecting, tracking, and following humans. This method is based on depth information, skeleton, and color of humans from 3D camera. Firstly, the depth image is taken from 3D Kinect to segment the individual region. After that, we calculate the body length, shoulder length, and arm length in combination with the color of target's clothes to gather as the material for the tracking task. Finally, the mobile robot which is controlled by voice command can recognize and follow the single target person. The effectiveness and robustness of the proposed method are evaluated in comparison with the method based on single skeleton or color of objective. Moreover, our proposed method can identify the target again when it disappears and appears again in the frame. All experiments are implemented with the support of model designed to integrate 3D Kinect camera on a wheeled mobile robot. The velocity and direction of the wheeled mobile robot are controlled by a proportional-integral-derivative controller to keep a constant velocity all the time. The experiment result is shown that the proposed system has worked effectively, stably, and flexibly and the success rate is more than 90%.