System design of a 9-DOF robot capable of fast and flexible rally task

Abstract

Dynamic interactions between a robot and environments have been an active field for years. A table tennis robot can serve as a comprehensive platform for these researches. Investigating previous work, neither articulated robot nor Cartesian one can solely provide satisfying solutions on this topic. This paper proposes a 9-DOF table tennis robot consisting of a flexible arm, a fast-moving body and highspeed cameras. The high-speed vision system identifies and predicts the target trajectory. The 7-DOF articulated arm can accomplish complicated and flexible operations while the 2-DOF fast-moving body in Cartesian space helps the arm to form an enough workspace. To perform rally task, a distributed and hierarchical control system based on RT-Linux which can be divided into ball trajectory identification, ball trajectory prediction, paddle trajectory planning and a real-time distributed motion controller for joints control is constructed. Experiments demonstrate the validity of the system design.