| The aim of study is to develop the mobile robot handling system using natural gesture. In this system, five gestures are used for handling: one of gesture is to stop robot, and others are to move it. The gestures we use this system are some continuous motions of right arm. The sequences of 3D position data of right arm motion are acquired by Kinect sensor system, and the data of right arm motions are Fourier-transformed. DC component and difference of phases estimated from the Fourier transform are features for the gesture recognition. The input gesture is determined by the similarity of features considering the learning data. In the experiments, gesture recognition has time delay, but all five gestures to handle the mobile robot can be recognized correctly. |