A Real-time Moving Target Following Mobile Robot System with Depth Camera

Abstract

An intelligent system that could obtain a certain moving target human and could follow the target in real-time by a mobile robot could be widely applied in manufacturing as assistant or service industry to provide a human robot interaction experience. In this work, an improved efficient human following algorithm for a mobile robotic system is proposed. The system consists by a mobile robot and a depth camera which overcomes target scale variation, missing and occlusion. An improved three-dimensional tracking method is presented with the depth camera. Besides, the depth camera also provides information to avoid obstacles in robot's path. The omnidirectional mobile robot in the system provides a flexible and fast motion responding solution. In the experiments, the proposed system is tested in different scenes and the results show the reliability of the 3D tracking system in terms of accuracy and robustness to the environment.