ESA GNC Conference Papers Repository

Envisat is a former Earth remote sensing satellite being now space debris. Due to its high mass and collision probability with other satellites in sun-synchronous orbits it is on top of the list of space debris objects which need to be removed urgently in order to avoid the Kessler syndrome. The e.Deorbit mission is devoted to remove Envisat safely from its orbit. The chaser approaches the target satellite on a safe trajectory, synchronises its motion with the target motion and grasps Envisat at its launch adapter ring. After stabilisation and establishment of a rigid connection between the two satellites, the compound is de-orbited into the South Pacific Ocean. The major challenges in the close range are the motion synchronisation between chaser and target and the coupled control during capture employing the robotic arm. This paper is devoted to the coupled control phase, during which the chaser performs station keeping at the so-called capture point which is a point relative to the target in the target body frame. Due to Envisat's tumbling motion the chaser has to follow a trajectory which is determined by the angular rate and by the moments of inertia of the target. The chaser has basically to compensate the centrifugal forces along this trajectory. Furthermore it has to compensate the forces and torques from the robot arm acting at the arm base. The robot arm has to position the end-effector at the launch adapter ring while compensating the station keeping errors of the chaser platform. The end-effector trajectory has to respect several constraints: The launch adapter ring must remain in view of the end-effector stereo camera throughout the complete manoeuver, which implies an inequality constraint on the orientation of the robot end-effector. The robot joints shall not exceed position and velocity limits. This also guarantees that robot singularities are avoided. The end-effector velocity shall not exceed image processing requirements. Collision between the robot and the target and with itself must be avoided. Furthermore the contact forces between end-effector and target shall be limited in order to avoid significant transfer of energy and impulse to the target as well as bouncing off after first contact. Therefore the positioning of the end-effector is performed in impedance control mode. The coupled controller consists of two separate control elements, one being the Chaser GNC and the other being the robot controller. The first operates in relative coordinates with respect to the target, the second operates in robot operational space coordinates defining the pose of the robot end-effector with respect to the grasping point at the launch adapter ring and also directly controls the robot joint configuration, to account for the robot kinematic redundancy stemming from the 7 joints for a 6 DoF grasping task. The two controllers exchange information like estimated relative pose and commanded forces and torques which is then used in a feedforward term. The sampling frequencies and the controller bandwidths of the two controllers as used in this study differ by more than two orders of magnitude, allowing independent control design of the two. The overall performance of the coupled control in terms of station keeping performance for the chaser and positioning performance of the end-effector is demonstrated in extensive Monte Carlo simulations.