Automated Rendezvous & Docking (AR&D) without Impact Using a Reliable 3D Vision System Farhad Aghili Space Technologies, Canadian Space Agency, Saint-Hubert, Quebec J3Y 8Y9, Canada There are several space missions that will require capabilities for Automated Rendezvous & Docking (AR&D) without human intervention. Advances in sensor technology and on-orbit flight control are critical to the success of such missions. This paper describes AR&D of a free-floating object with uncertain dynamics by integrating a vision system, a dynamic estimator, and an optimal controller. A Kalman filter (KF) estimator and an Iterative Closest Point (ICP) algorithm in a closed-loop configuration are used for estimating not only the states of the target object but also its dynamics parameters. Subsequently, the estimated states and parameters are used by an optimal controller to derive a chaser spacecraft to intercept the target object with zero relative velocity at a rendezvous point. This arrangement allows smooth and impactless docking to a free-floating object with uncertain dynamics. Furthermore, continuous pose estimation for the purpose of the feedback is possible even with temporally failure of the vision sensor as a consequence of the prediction estimate capability of the dynamic estimator. Experiment results obtained from robotic docking to a tumbling satellite using a laser camera system are appended. I. Introduction Automated rendezvous and docking (or capture) of spacecraft is critical to the success of many space missions such as in-orbit construction and assembly, refueling of satellites, repairing or rescuing failed satellites, active removal of defunct satellites or space debris, autonomous re-supply and crew exchange of space stations, and sample return in planetary exploration. AR&D involves two or more spacecraft or space objects. The docking spacecraft usually has its own active control system, which uses information received from proximity sensors, e.g., vision, laser range finder or radar, to dock to another spacecraft or a free-floating space object, see Fig. 1. Capturing a tumbling free-floating object is one of the most challenging task in servicing of failed satellites or removal of space debris. The servicing satellite is usually equipped with a manipulator to be guided by a vision system so as to capture a grasping point along the tumbling satellite. This problem is closely related to the application of automated docking and rendezvous of spacecraft. A smooth docking or capture can be accomplished only if the two spacecraft arrive at a rendezvous point with the same velocity. Otherwise, there will be impact at the instant of capture or docking that can cause mechanical damage to the spacecraft. In order to plan adequate trajectories for a smooth docking, the motion of the target should be predictable. When the dynamics of the target spacecraft is not known beforehand, not only its states but also its inertia parameters must be accurately estimated in real time, so that the docking spacecraft can reliably predict the pose (position and orientation) of the target in near future. The other challenge is how to guide the docking spacecraft (or a space manipulator) optimally to rendezvous and capture the non-cooperative target satellite. 1, 2 There are different vision systems capable of estimating the pose of moving objects. Among them, an active vision system such as the Laser Camera System (LCS) of Neptec 3 is preferable because of its robustness in face of the harsh lighting conditions of space. 3 As successfully verified during the STS-105 space mission, the 3-D imaging technology used in the LCS can indeed operate in space environment. The use of laser range data has also been proposed for the motion estimation of free-floating space objects. 1, 2, 4, 5 Taking advantage of the simple dynamics of a free-floating object, which is not acted upon by any external force or moment, observers predicting the motion of a target satellite are proposed in. 6 There has been extensive research conducted in the area of autonomous satellite capture. 1, 7–9 In all these works, however, the dynamics of the 1 of 11 American Institute of Aeronautics and Astronautics AIAA Guidance, Navigation, and Control Conference 2 - 5 August 2010, Toronto, Ontario Canada AIAA 2010-7602 Copyright © 2010 by Canadian Government. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.