Motion Planning Of Space Robots For Target Capture

Along with the developments of aerospace science and technology, On-Orbit Servicing operated with assist of space robots is a hotspot of researches in the field of aerospace, and the target capture attracts much attention of researchers as the key techniques of the OOS. The motion planning of free-floating space robots for target capture is studied in this dissertation. Based on the analysis of kinematics and dynamics, four problems are discussed. The singularity-consistent trajectory tracking are discussed firstly, and then the coordinated planning for the manipulator and the base orientation is discussed. Thirdly, the zero-disturbance planning during the target capture is discussed. At last, the adaptive planning with dynamic parameter uncertainties is discussed during the post-impact phase.There are many differences between the motion characteristics of the space robot and that of the fixed-base robot. The position equation and velocity equation are presented at first. And the generalized Jacobian is derived with the momentum conservation equation. Then the dynamics of free-floating space robot is established. At last, space robot systems studied in the dissertation are presented and some characteristics of the system such as equivalences, singularities, and the nonholonomy are discussed, which lay the groundwork for the researches in the following chapters.Singularities are the inherent properties of the manipulator, and there are no exceptions for the space robot. With the traditional planning method, the trajectory of the end-effector might be departed from the expected trajectory when the desired path passes through singularities. In this dissertation, based on the singularity-consistent null space approach, conditions for prisely tracking passing through singularities are analyzed firstly, and then trajectory generation is transformed to a nonlinear control problem. The frame for joint velocity planning is constructed as"Null-space Approach + Nonlinear Control + Command Filter". The control law and the command filter are constructed respectively. Simulation results show that the path following by the end-effector is realized precisely under the limitation of joint velocities whenever the desired path is in or out of the singular region.In the process of point-to-point planning, not only the configuration of the manipulator but also attitude disturbances of the base need to be considered. In this dissertation, the coordinated planning for the configuration of the manipulator and the base orientation is studied firstly. An invariant manifold with a linear state feedback is constructed, and its physical meaning is explained. A piecewise controller is designed, stabilizing the system states to the invariant manifold and then to the desired states along the manifold. And then combined with the Virtual Manipulator approach, the planning method is extended to the coordinated planning of the configuration of the end-effector and the base orientation. Simulation results show that the configuration of the manipulator and the base orientation can reach their desired values simultaneously with the designed controller.During the capture of a dynamic target, the momentums of the target are transferred to the robot system. The change of angular momentum might bring the change of angular velocity, and the base orientation might be disturbed. In this dissertation, the disturbance of the angular velocity of the base is discussed. The contact dynamics is established firstly. Then the relationship between the relative velocity of the contact points and the variation of the base angular velocity is analyzed. And then the zero-disturbance planning method is designed. At last, the momentum transformation during the capturing phase is discussed, and the full bias momentum approach is applied to optimizing the zero-disturbance planning method. Simulation results show that angular velocity variation of the base is avoided, and the angular momentum transformation is quickened. This facilitates the motion control of the manipulator during the post-impact phase.Different from the fixed-base manipulator, the kinematics of the space manipulator should take dynamic parameters into consideration. The dynamic parameters are changed after the capture operation. This brings difficulties to the motion planning of the space robot. In this dissertation, the linearization for the kinematics and the momentum conservation law with respect to dynamic parameters is discussed firstly. Then the output state is constructed as the combination of the end-effector position and the base orientation. The adaptive planning algorithm is designed without the acceleration measurement. In order to improve the convergence, with an additional predicting module, a composite adaptive planning method is presented and the stability and convergence are analyzed. Simulation results show that the tracking error in task space is asymptotically stabilized to zero with the designed planning method.Finally, main contents of this dissertation are summarized and key points for future researches are discussed.