The Design and Implementation of Natural Human-Robot Interaction System Based on Kinect Sensor

In this paper, researchers propose a method to build a natural human-robot interaction system based on Kinect sensor. Taking the advantage of skeleton tracking technology, researchers can easily get the depth data from Kinect sensor and capture body movements, which greatly simplify the recognition algorithms. After establishing the 3D coordinates of human joints, the rotation angles of human waist, shoulder, elbow and wrist can be calculated according to the spatial geometry, and these joints correspond with the joints of the robot arm one to one. Further, researchers need to extract the palm characteristic information to recognize hand gesture based on Hu Moment. Finally, the results of identification will be converted to commands and sent to controller via Bluetooth to manipulate the robot arm. Experiments show that by this method, researchers can effectively control the rotation of the 5-DOF robot arm to grab object. As the relationship between people and robots becomes more and more closely, this kind of human-robot interaction interface will play an important role in the future development of robot.