Research On Trajectory Planning And Intelligent Control Of Uncertain Space Robot

Nowadays, space robots have been applied in satellite maintenance and space station construction successfully. Consequentially, space robots will soon take the place of astronauts to fulfill the extra-vehicular activities, which can greatly reduce the labor intensity and risks of the astronauts. In this dissertation, the dynamic characters of space robot, point to point trajectory planning and continuous trajectory planning are discussed. In addition, the control methods of trajectory tracking in both joint space and task space are introduced in detail.First, the fundamental theories and control methods of space robot, as well as some influential engineering systems are surveyed. The kinematic and dynamic models for attitude controlled free-flying space robot and free-floating space robot are deduced respectively.Trajectory planning of space robot is foundation to complete the tasks, point to point trajectory planning and continuous trajectory planning are studied in this dissertation. The approach for the point to point trajectory planning only uses the direct kinematic equations, so it is not affected by the dynamic singularity. The joint trajectories are parameterized by sine functions. Then, the system of equations about the parameters is established by integrating the differential kinematics equations. Lastly, the parameters are solved by the BFS rank 2 algorithm. The method settles the shortcomings of similar approaches, i.e., it satisfies the limits on angle, angular velocity and angular acceleration of the manipulator, inverse of the updating matrix and the skill for choosing initial parameters is unneeded. The continuous trajectory planning of free-floating space robot is very important theoretical and practical to complete tasks. The problem is transformed into a solution of coupled differential equations based on the velocity kinematic equations and conservation angular momentum equations, predictor-corrector method is proposed for the end-effector tracking, the accuracy is improved from the first-order of step count to the sixth-order.While working in space, the total mess of the spacecraft decreases with gas jet thruster control. The change of the manipulator's load will also affect the dynamic parameters of the system. Thus, the uncertainties of the space robots are analyzed quantificationally. These uncertainties cause many dynamical control methods unstable, such as coordinated control and computational torque control. To solve the problem of trajectory tracking in joint space for space robot with uncertainties, neural network adaptive controllers are proposed. The first method is compensating scheme based on neural networks. By using a RBF neural network to learn the upper bound of the uncertainty and disturbance, the proposed method can compensate influence of uncertainty and guarantee the tracking error converges to zero in finite time. It solves the unknown upper bound problem of neural network. The second method utilizes RBF neural networks to approximate the nonlinear function and uncertainty of the system, the knowledge of system parameters and uncertainty is unneeded, that can effectively escape from complicated dynamic computation.The attitude of the free-floating space robot is not controlled while the manipulator is working, which can save limited fuel. However, the coupling of the manipulator and base attitude makes it difficult to position the end-effector in task space. Furthermore, the generalized Jacobian matrix of free-floating space robot is too complex to compute on-orbit. In this dissertation, a neural network adaptive controller for trajectory tracking in task space is given,that using generalized Jacobian matrix substituted by simple Jacobian matrix to reduce the complexity of the arithmetic, and the knowledge of system parameters and uncertainty is unneeded. Further, an indirect fuzzy adaptive controller based on backstepping is presented; the output of controller is adjusted according to fuzzy rules and experts'knowledge. Both the two controllers can ensure the stable trajectory tracking in task space with uncontrolled base.