Autonomous Visual Servo Robotic Capture of Non-cooperative Target

This doctoral research develops and validates experimentally a vision-based control scheme for the autonomous capture of a non-cooperative target by robotic manipulators for active space debris removal and on-orbit servicing. It is focused on the final capture stage by robotic manipulators after the orbital rendezvous and proximity maneuver being completed. Two challenges have been identified and investigated in this stage: the dynamic estimation of the non-cooperative target and the autonomous visual servo robotic control. First, an integrated algorithm of photogrammetry and extended Kalman filter is proposed for the dynamic estimation of the non-cooperative target because it is unknown in advance. To improve the stability and precision of the algorithm, the extended Kalman filter is enhanced by dynamically correcting the distribution of the process noise of the filter. Second, the concept of incremental kinematic control is proposed to avoid the multiple solutions in solving the inverse kinematics of robotic manipulators. The proposed target motion estimation and visual servo control algorithms are validated experimentally by a custom built visual servo manipulator-target system. Electronic hardware for the robotic manipulator and computer software for the visual servo are custom designed and developed. The experimental results demonstrate the effectiveness and advantages of the proposed vision-based robotic control for the autonomous capture of a non-cooperative target. Furthermore, a preliminary study is conducted for future extension of the robotic control with consideration of flexible joints.