Abstract
This paper describes about a control scheme that enables industrial robots to grasp moving objects. In the developed control system, the target position is measured at a constant frequency by processing images acquired from two cameras. To consider the delay of image acquisition and robot motion, the move destination of the target object is predicted by using the state of movement estimated based on the Extended Kalman Filter (EKF). Results of the experiment that estimates the state of movement of a pendulum target are provided in order to validate the developed system to predict the move destination, and to detect the sudden change of target motion that affects the final accuracy of grasping.
