Natural Tele-manipulation for Robot Movement based on Motion Pattern of Arm Posture

摘要

There have been tremendous efforts on robot telemanipulation by the human arm through various sensors. Some approaches use mechanical or electronics sensors attached to the human arm. Other uses non-touch sensor such as camera, Kinect, or Motion Leap. The extracted features of the human arm are often mapped to corresponding joints of the manipulated robots. This condition leads humans to stick to stiff movements. In contrast, there is a need for an actor doing telemanipulation on stage with natural movement. The more natural the movement during telemanipulation, the more attractive to the audience. In which, the audience feels as if the actor communicates with the robot. And the robot understands what being instructed by the actor. Indeed, this paper considers the natural movements of the actor by examining the motion pattern of the human arm posture as the manipulation reference. A Myo Armband sensor is used in this paper to extract the arm posture. At the same time, the movement is observed to obtained the motion pattern. Later both information is then used for manipulating a robot. In order to validate the proposed method, a serial manipulator robot is utilized. Some experimental results show that the arm posture and the pattern of changing from one posture to another can be extracted and used for manipulating robot movement.