Combined Control of the Servicing Spacecraft and Robotic Manipulator with Model Predictive Control Architecture

Kurzfassung

The growing satellite congestion in the Earth orbits increases the risk of Kessler Syndrome, that could potentially hinder humanity’s activities in space. One of the ways to tackle the problem is Active Debris Removal (ADR) and On­Orbit Servicing (OOS) missions with the capture phase of another spacecraft performed using a satellite equipped with robotic manipulator. Robotics solutions are good candidates for the application in on­orbit servicing and active debris removal missions. Due to the heritage in previous missions in which the robotic arms were used mainly operated by astronauts or in a semi­autonomous mode, and also given the possible technology transfer from terrestrial robotics autonomous systems, the space robotics are foreseen to have high potential in space for applications in close proximity operations between two spacecraft. The GNC system for such a mission is one of the main challenges due to the strict safety requirements and precision, moreover the complexity increases with any uncertainty of the target spacecraft state. Agility and precision in coupled position and attitude control are critical to ensure an operation free of collisions between the servicer spacecraft and the target. The challenge lies mainly in the close rendezvous, reach and capture phase, as all the mission phases prior to close rendezvous phase function are similar to single satellite mission without robotic subsystem. This research project aims to propose a control system solution of a robotic spacecraft specifically for the reach phase of the OOS space mission, when the close rendezvous phase is successfully finished and the spacecraft starts the reach maneuver towards the target. Majority of proposed solutions for control system concerns freefloating mode in which the controller of the s/c base is turned off and only the manipulator’s controller is active. This project aimed to investigate the combined controller approach for OOS mission reach phase in which both the base and robotic manipulator are actively controlled by a single control system that coordinates all sensor data and the actuation. The project boundaries were set such that the initial condition is assumed that the chaser spacecraft is few meters away from the target and the initial relative position and velocity with respect to the target is kept constant, the end of the reach phase is when the end­effector reaches the grasping point. In order to focus uniquely on control system design certain simplifying assumptions were taken with respect to the guidance and navigation capabilities. Moreover, the capture maneuver itself is out of scope of this work, for this reason contact dynamics were not included in the model. The multi­body dynamics were defined with the SpaceDyn toolbox in Matlab and the advanced control strategy was chosen, Model Predictive Control (MPC). The objective of the project was to investigate the application of MPC for the design of the combined controller. For this purpose firstly a variety of MPC architectures was studied in order to preliminary choose the best candidate for the robotic s/c system. The final decision was also supported by a trade­off study. The advantage of MPC is the optimization­based strategy, a straightforward definition of the constraints and the dynamics prediction within the future horizon. A nonlinear model predictive control (NMPC) based on successive linearization approach was developed, the linear dynamics prediction model is obtained in every simulation step by linearizing the nonlinear dynamics around the current operating point. The optimization problem was constructed as quadratic problem with the primary goal of end­effector position reference tracking and other secondary goals and it was modelled with Yalmip toolbox interface. Finally, the NMPC controller and the plant model were put together in a feedback loop and tested for different scenario cases. The final results show that the reach maneuver can be successfully accomplished as long as the weights tuning procedure is performed carefully. The final position error of end­effector is very small and remains in the acceptable performance limits. The constraints imposed by the definition of the MPC problem are well respected, and the optimization problem is converging in every iteration. The main improvement point of the designed control system is that it requires to be re­tuned for changing initial conditions, the s/c base position and attitude and manipulator configuration. Further extensions of the project would be very interesting, including complete integration with a proper guidance and navigation capabilities, adding low­level control level and extending the dynamics model such that it accounts for the contact dynamics and capture maneuver. All in all, the project results are a good starting point for the future development of the combined controlled strategies for OOS missions. This study was performed in collaboration with the German Aerospace Center DLR within RICADOS (Rendevous, Inspection, Capture and Detumbling by Orbital Servicing) project.