Research On Human-robot Interaction Based On Somatosensory Sensor

With the German "Industrial 4.0",Japanese "intelligent manufacturing system" and American "Industrial Internet" proposed,China also put forward the " manufacturing in China 2025" program,the purpose is intelligent manufacturing that an intelligent network is so composed of sensors,intelligent control system,communication facilities etc that form interconnection between man and machine,so as to realize a high degree of horizontal integration.In this paper,somatosensory sensor as the research object,a Human-Robot interaction system based on somatosensory sensor is designed.The system can directly control the robot by gesture,and accurately identify the identity of the designated person and carry on the position tracking.The advantage of the somatosensory sensor is that the operator can directly realize the interactive between man and machine only through the body without wearing any auxiliary equipment.Firstly,the development trend of robot and technology of the Human-Robot interaction are briefly introduced.The working principle and data structure of Kinect and Leap Motion are studied in detail.The application of somatosensory sensor is introduced based on the correlation of basic theory.Secondly,for the target recognition tracking,the depth image and color image are collected by using Kinect,the designated person is identified according to the color histogram matching result,and the human skeleton is extracted according to the depth image and the designated person is tracked.For the gesture recognition,the joint angle of the finger is obtained by constructing the finger kinematic model,and then the joint angle,as feature vector of gesture recognition,is sent to the robot controller to control the robot to execute the corresponding action.Finally,the experimental platform of Human-Robot interaction system based on somatosensory sensor is built,and the communication between PC and robot is established by C/S structure.The control system software is developed using the C++ in the Visual Studio 2013.During the experiment,the target person identification,the position tracking are accomplished by the somatosensory sensor and control the robot working by the gesture.