Design of Kinect-based human robot interaction systems for a RoboCup middle size league soccer robot

In order to launch a soccer game between human and robots, one of the crucial problems for robots is to effectively identify human and their actions. In this paper, based on human skeleton tracking information from Kinect, GR (Gesture Recognition), FVGR (Front View Gait Recognition) and NNVGR (Non-Normative View Gait Recognition) are carried out to develop the HRI (Human Robot Interaction) systems for a middle size league soccer robot. This method is used to identify the human body language and obtain information of body movements, so that the HRI system can generate different command to control the movements of the middle size league soccer robot. The performances of the proposed HRI systems are verified by different experiments. The experimental results show that this method can quickly and safely identify human and their movements. This research can provide a practical solution for Mid-Sized League competition between human and robots in the future.