Kinect-based Human Body Tracking System Control of Medical Care Service Robot

In order to realize the human body following of medical care service robot, Kinect v2 is used to detect the target person in real time, and the improved Gaussian filter algorithm is used to denoise the depth image. In this paper, Kalman filtering algorithm is used to predict human movement information, thus ensuring safer and more reliable real-time tracking. When the target character is blocked, the robot can be stopped safely and effectively. This system uses the gesture of the target person to start or stop following, and it can realize good human-computer interaction. Finally, a series of human body following test experiments are designed to verify the effectiveness of the proposed method and the feasibility of the robot system to achieve human body following in hospital environment.