GNC architecture solutions for robust operations of a free-floating space manipulator via image based visual servoing

On-orbit servicing often requires the use of robotic arms, and a key asset in this kind of operations is autonomy. In this framework, the use of optical devices is a solution, already analyzed in many researches both for autonomous rendezvous and docking and for the evaluation of the control of the manipulator. In the present paper, simulations for assessing the controller performance are realized in a high-fidelity purposely developed software architecture, in which not only the selected 6 DOF space manipulator is modeled, but also a virtual camera, acquiring in the loop images of the target CAD model imported, is included in the GNC loop. This approach allows to emphasis several problems that would not emerge in simulations with images characterized by easily-identifiable, purposely-created markers. At the scope, a specific GNC architecture is developed, based on finite-state machine logic. According to this approach, two different Image Based Visual Servoing strategies are alternatively performed, commanding only linear or angular velocity of the camera, switching between the two control techniques when the “stack” or “divergence” condition is triggered. In this way a stable and robust accomplishment of the tasks is achieved for many configurations and for different target models.