Gesture recognition based human–robot interactive control for robot soccer

This paper proposes an interaction method in human gestures for controlling robot play soccer. It aims to design a human–robot interactive control scheme let a humanoid robot to substitute who want but cannot such as disabled people without feet for playing soccer games. The scheme first adopts the Kinect based gesture recognition system to analyze human gestures, and then drives the humanoid robot to do respective motions for playing soccer or throwing ball according to what human gestures. The proposed gesture recognition scheme consists of two subsystems in which the Kinect-based 3D framework trace system provides the kernel recognition and the electromyography signal measurement assists the slight movement detection. It mainly makes all the motions of upper body classified as several types of gestures. Every different gesture is set in relative to a unique soccer motion of the humanoid robot. Therefore, a gesture is a control commend of a robotic motion. Based on such translation, a human can only use his/her hands and upper body to play soccer but in humanoid robot. From the experimental results, it demonstrates the satisfactory interactive human–robot soccer processes.