Decentralized Guaranteed Cost Control For Reconfigurable Manipulators With Environmental Constraints Based On Adaptive Dynamic Programming

The reconfigurable manipulator is a modular robot constructed from standardized components such as joint modules,link modules,communication units and control units,with standard mechanical and electrical interfaces that can reconstruct multiple configurations.With the increasing demand of robot industry in China,many kinds of robots,such as reconfigurable manipulator,industrial robot,guide robot,rehabilitation robot and so on,have been developed unprecedentedly,but in the face of the changeable task object and complex working environment,appropriate control algorithms are needed to meet the robot's demand for stability and accuracy.This paper takes the reconfigurable manipulator as the research object,considering the inevitable contact or collision with the external environment when the reconfigurable manipulator operates,and introduces adaptive dynamic programming(ADP)algorithm to design the decentralized controller to reduce the design cost,shorten the design cycle and reduce the energy consumption.This paper takes ADP algorithm as the theoretical basis,and combines the method of guaranteed cost control(GCC)to study the decentralized guaranteed cost control with environmental constraints.The main contents of this paper include:1.Introduce the topic and research significance of the paper,the research status of the reconfigurable manipulator,adaptive dynamic programming and guaranteed cost control method,and gives the chapter arrangement of this paper.2.Based on the Newton-Euler iterative algorithm,the reconfigurable manipulator with free state space and environmental constraints are modeled dynamically.Based on this,the dynamic model of the reconfigurable manipulator is decomposed into a separate subsystem with interconnection dynamic coupling(IDC),which lays a theoretical foundation for the research work of chapter 3 and chapter 4.3.A decentralized guaranteed cost control method based on ADP algorithm is studied for reconfigurable manipulators under environmental constraints.The dynamic subsystem of the reconfigurable manipulators is represented as a state space.By constructing a new guaranteed cost control method,a policy iterative algorithm is introduced,a cost function approximately modified by critic neural network(critic neural network)is used,and a compensation controller is designed for joint friction items,then,the designed decentralized guaranteed cost control strategy can be derived.The stability of the reconfigurable manipulator system under the designed decentralized guaranteed cost controller is proved by Lyapunov theory,and the effectiveness of the designed decentralized guaranteed cost control method is simulated by Matlab software.4.Based on the third chapter,a decentralized neural optimal guaranteed cost control method is designed based on ADP algorithm for reconfigurable manipulators with uncertain environments.First,for the uncertain terms in the subsystem model,including the friction term and the IDC term,the compensation term based on robust control and the identifierbased learning control are designed respectively.Further,the decentralized neural optimal guaranteed cost control problem of reconfigurable manipulator can be transformed into an optimal control problem for nonlinear systems.Then,the derivation of the optimal control strategy can be realized by policy iterative algorithm and critic neural network.The stability proof of the robot closed-loop system is completed by using Lyapunov theory,and the effectiveness and superiority of the controller designed in this chapter are verified by comparing with the existing decentralized guaranteed cost control strategy under the Matlab simulation platform.Finally,the research contents of this paper are summarized,and the future research contents of the reconfigurable manipulator are prospected in combination with the problems encountered in the research process and the current challenges of the reconfigurable manipulator.