Remote Interactive Control System Design And Implementation Based On NAO Robot

In real life,the coverage of wireless networks has increased.And the share of service robots in the robot market has been increasing year by year.Remote interaction of intelligent robots has become a hot topic in the field of robotics research.In domestic services,medical information,entertainment,industrial production,the robot has been used widely in the field.The human-computer interaction of home service robots is a hot spot of research in recent years.Many universities and robot research institutions are throwing themselves into it.The monitoring of special groups in the family has always been a blind spot in real life.Active emergency response to unexpected situations in the family is an urgent problem to be solved.The system applies a new Web server framework,front-end development technology,Ajax data exchange technology and HTTP application layer network protocol to realize a remote interaction system of NAO robot vision under the B/S network model.The system consists of three parts: application,server browser client,and robot.The server side uses non-blocking asynchronous communication Tornado framework to solve high concurrent access and asynchronous data communication.Binding interface label ID and controlling command string in server program,parsing control instructions,it is a necessary condition that realize remote control of robot under wide area network environment.The face are targeted and tracked by loading cascading classifier,Handler video images.The browser client uses the CSS,jQuery,and Bootstrap integrated JavaScript plug-in tools to write a humanized operation interface.It monitors the user interface event request and build a two-dimensional pixel coordinates in interface video window.The robot's head is controlled by calculating the difference between the centers coordinates of the face rectangle and the center of the video window.Face tracking is realized based on face detection.In the framework of the robot NAOqi,the agent is created to edit the action of the robot.The internal video module is rewritten and called to return the video.In addition,by adding weight thresholds and improving weight updating rules,we can achieve the purpose of optimizing training classifier's AdaBoost algorithm,and train a cascade classifier with excellent performance under specific environment.The system test can realize the remote control of NAO robot.All the performance indexes meet the requirements of real-time system.