Research On The Control Strategy For The Flexible-Joint Robot Arm

Construction and maintenance of on-orbit components is currently done primarily by extra-vehicular astronauts at great risk. Use of space robotics for these tasks provides the opportunity for both increased safety for the astronauts and major ground-crew cost savings. Owing to the characteristics such as large workspace and light weight, space robot will necessarily be quite flexible, which not only limits the dexterity and speed of the end-point, but also brings more complexity for the robot control. Due to the space robot's multilayer-task demands, precise position control and impedance control with much compliance are the promising control schemes. The dissertation is under supporting of the project"Study on the key technologies of satellite on-orbit self-servicing and its tele-operation", which aims to develop the research on how to perform the position control and the Cartesian impedance control and some associate issues for the robot with flexible joint.In this thesis the dynamical model of the flexible joint robot is built firstly which is based on the Lagrange energy method. The position control, Cartesian impedance control and some control issues with time-delay influence are investigated in detail based on the reduced model of the flexible joint robot.In view of the theoretical meaning, control methods for the flexible joint robot mainly consist of the cascaded system method, feedback linearization and singular perturbation technique, etc. The traditional singular perturbation approach can only be used in the robot system with weak joint flexibility. In order to solve this problem, a joint flexibility compensator is designed, which can increase the equivalent joint stiffness. So the limitation for the singular perturbation approach is removed and it can be extended to the robot system with normal joint flexibility. Besides these, an adaptive control law is designed for the slow subsystem, which can guarantee the trajectory error asymptotical tracking. The proposed control strategy can be applied to the engineering systems conveniently without any joint flexibility limitation and does not need the link acceleration or jerk information. The experimental results verify that whatever the PD, computed torque or adaptive control is used in the slow subsystem, the proposed singular perturbation strategy is more valid than the traditional one. Different from the traditional Cartesian impedance control schemes which are mostly based on the robot end's force/torque information, five Cartesian impedance control schemes including the force-based in Cartesian/joint space schemes, the position-based in Cartesian/joint space ones and the stiffness control scheme are considered, aiming at the robot with joint torque sensors in each joint. Among these, the force-based in joint space control scheme is analyzed in detail by using the passivity theory. In this approach an inner torque feedback loop is interpreted as a scaling of the motor inertia. Based on this physical interpretation of the joint torque feedback, the closed loop system can be regarded as a feedback interconnection of two passive subsystems, which guarantees the robustness against uncertainties in the model parameters. Besides these, aiming at the sensor signals in the practical system, the effect on the system's passivity owing to time delay is analyzed. Observer(Kalman filter or H∞filter) based control schemes are introduced in order to deal with the time delay influence caused by algorithms and low filter. But to other system, that is to say, an extensive application of the impedance control, such as the bilateral teleoperation system, the LMI based approach gives the problem a general solution, which assures the closed loop system's asymptotical stability and corresponding performance indexes during the derivation of the solution.Finally, corresponding experiments and applications are presented for evaluating the control method and theory. As expected from the analysis, the force-based in joint space control scheme turns out to be very robust against uncertainties in the model parameters and the observer based control scheme can overcome the time delay in the velocity computation, which enlarges the system's bandwidth and improves the control performance.