ROBOTIC AUTONOMY IN SPACE: VISION-BASED CONTROL OF ROBOTIC CAPTURE OPERATION

On-orbit servicing (OOS) consisting of assembly, repair, and maintenance tasks of spacecraft using satellite robot manipulators is an emerging technology and promises to be a key element in the future of space exploration as missions are becoming more complex and expensive. The OOS usually employs a robotic manipulator mounted onto a satellite to capture, service, and refuel other orbiting satellites. One of the critical phases of OOS is the capture of the object satellite, where the chase satellite's robotic manipulator approaches the free-floating object satellite to capture it. If not properly controlled, the object satellite and the robot may be pushed away from each other or the end-effector/object may be damaged by the contact force. The key examination point of this paper is the vision based guidance and control of the robotic manipulator. The system employs pure photogrammetry as guidance and performs the capture of a non-cooperative target. In order to improve, both in terms of time and strains, the capture operation, several strategies are examined and implemented including, dynamic digital damping, visual compensation, and weighted capture criteria. Kalman filter is used in the vision system in order to capture a target in motion. The first step is the approach which aligns the gripper in the most likely orientation to result in a successful capture. The second is the capture which involves bringing the gripper within range and determining whether or not the target is within its grasp. The final stage is bringing the target to a relative halt with minimal force and disturbances to the target. Once captured, a hybrid speed-force controller is developed to limit the amount of force applied to the grasping bar of the target while propagating the speed and direction of the target in order to smoothly bring the target to a halt. Testing results have demonstrate the proposed control system is effective and robustness.