ABSTRACT
KF-based sensor fusion is applied to obtain improved performance in this chapter, and a Kinect sensor is utilized to capture the motion of the operator's arm with a vector approach. By selecting five out of seven joints on each arm, the vector approach can precisely calculate the angular data of human arm joints. The continuous-time KF method outputs the designed data with less error, after that, the data will be applied to the joints of a simulated robot for teleoperation. A teleoperation system is developed based on visual interaction, and we put forward a robot teaching method based on ELM. Kinect is used to control the robot in V-REP by human body motion in the teleoperation system. The robot can reproduce the trajectory, which is provided by the RBF network through learning and training. The experimental results show that robot can learn the motion from human demonstration in a natural manner and does not require analytical modeling of a robot.
