Research On Optimal Force/position Control For Reconfigurable Manipulator Systems Under Environmental Constraints

The reconfigurable manipulator can reconfigure and combine its configuration according to different task requirements,showing that traditional manipulator do not possess many advantages."Reconfigurable" and "modular" are two basic requirements for the design of reconfigurable manipulators.The main idea is to decompose a complex manipulator system into multiple subsystems with high portability and maintainability,therefore effectively shortening the design and manufacturing cycle of the manipulator system.The reconfigurable manipulator system inevitably completes its task under environmental constraints,and comprehensively considers and ensures the stability,robustness,accuracy,energy saving and other indicators of the system.Therefore,the appropriate control strategy is very necessary under uncertain environment.Optimal control is an important part of modern control theory.The core issue of its research is to select a suitable control strategy for a given controlled system to optimize certain performance index.For a reconfigurable manipulator system,to obtain its optimal control strategy is needed to solve the HamiltonJacobi-Bellman equation,which is a type of nonlinear partial differential equation,and it is difficult to obtain the optimal solution by analytical methods.The adaptive dynamic programming method is a powerful tool to solve the optimal control problem of the nonlinear system.In the adaptive dynamic programming algorithm,the neural network is designed to approximate the performance index function and estimate the solution of Hamilton-JacobiBellman equation.In order to improve the control performance of the reconfigurable manipulator system and reduce the energy consumption cost required by the controller,this paper combines the adaptive dynamic programming method with the optimal control theory to study the optimal force/position control method of the reconfigurable manipulator system.The specific content includes:(1)Dynamics modeling of reconfigurable manipulator systemsEstablished a Newton-Euler iterative manipulator dynamic model and extended into the subsystem reconfigurable manipulator.Considering the force/torque acting on each joint and the coupling force/torque between the joints,the dynamic model of reconfigurable manipulator is established.On this basis,based on the local joint dynamics information,the dynamic model of the manipulator system is decomposed into several dynamic subsystems,the dynamic characteristics of the subsystems are deeply analyzed,and the dynamic model of each module is established.(2)Optimal control of sliding mode force/position of reconfigurable manipulator system based on adaptive dynamic programming under uncertain disturbanceBased on dynamic model of the reconfigurable manipulator has been established in this paper,a sliding mode-based optimal force/position control algorithm is studied,which solves fast and stable trajectory tracking as well as the constraint force tracking problem of the manipulator.In this paper,the adaptive dynamic programming method is extended to the sliding mode force/position algorithm,and combined with the online policy iteration.The Hamilton-Jacobi-Bellman equation is approximated by critic only neural network and the optimal admissible control strategy is obtained.(3)Critic-observer structure-based reconfigurable manipulator decentralized optimal force/position control under unknown environmental constraintsBy adopting the dynamics of the reconfigurable manipulator based on the NewtonEuler iterative subsystem form,an optimal force/position control algorithm for the reconfigurable manipulator based on critic-observer is studied,which solves the tracking problem of position,velocity and contact force under unknown environmental constraints.In this paper,the critic-observer structure is integrated into the adaptive dynamic programming algorithm,and the optimal control law is obtained through the designed fusion error function as well as the policy iteration algorithm,and the closed-loop manipulator system is proved to be asymptotically stable through the Lyapunov theorem.The simulation experiment further proves the effectiveness of the algorithm.