Tracking and Catching of an In-Flight Ring using a High-Speed Vision System and a Robot Arm

Robot-catching of in-flight objects is a challenging task, requiring a high-frequency sequence of pose estimation, trajectory prediction, catching point determination, and motion planning. Considering the small working space and visual occlusion in the natural environment, we investigate robot-catching with a short-distance and partially observable trajectory in this paper. We introduce a marker-based high-speed visual tracking method to collect sufficient data from the limited trajectory. Besides, we design a new catching point selection strategy to achieve a timely and stable response of the robot arm. Based on the proposed method and the dynamics of a thrown ring, we get a success rate of 90% in experiments to catch the in-flight rings using a collaborative robot arm.