Study On Configuration Optimization And Active Disturbance Rejection Control Forreconfigurable Modular Robots

With the continuous development and maturity of robot technology, the applicationsof modern high-performance robot are expanding. The aspects of traditional industrialproduction, manufacturing, deep space exploration, deep-sea exploration, dangerous andadverse conditions of operations and so on have large numbers of applications withremarkable results. At the same time, in a certain environment, fixed configuration of therobots need to change the corresponding parameters for the specific tasks to the overalldesign in order to perform different tasks. However, the external environment is oftenunknown and varied in the actual task such as nuclear power stations, lunar space station,deep sea polar research station and so on. And the design of a fixed configuration robotsystem which is used to accomplish many tasks is very difficult or even impossible.Therefore, we urgently need a kind of robot which can adapt to different tasks bychanging the parameters to be reconstructed. Reconfigurable manipulator is a kind ofmanipulator which can change its configuration according to the variety of the task orenvironment. Through combine the modules, reconfigurable manipulator can beassembled into a suitable geometric configuration simply and quickly that adapt todifferent tasks. This kind of combination is not only includes a simple mechanicalreconstruction, but also the recombination of control system. After reconstruct, themanipulator is competent for new working environments and tasks, and also possesses agood flexibility. Although the study of reconfigurable modular robot has beenconsiderable part of the outcome, there are still many challenging issues to be furtherin-depth study. Such as the configuration automatically generation, optimization andevaluation methods of reconfigurable modular robot; solving method of inversekinematics; processing of model of the unknown dynamics of subsystem andinterconnection term; fault identification and fault-tolerant control method, and so on. So,there are far-reaching theoretical and practical significance of more in depth study ofreconfigurable modular robot.In this paper, the problems of reconfigurable modular robots have been researched.Through establishing the dynamic model, automatic generation of reconfigurable modularrobot configuration, optimization and evaluation methods have been discussed. And forthe robot dynamics trajectory tracking control problem, a backstepping decentralizedcontrol method based on ESO and sliding mode decentralized ADRC control methodbased on VGSTA-ESO have been proposed. The full text of the work and research are asfollows:Firstly, a new multi-objective-based configuration auto-generation, optimization and evaluation method was proposed for reconfigurable modular robot. The modules of robotwere assigned and designed, and the way of module link was defined. Genetic-simulatedannealing algorithm was used for the automatic generation of configuration andmulti-objective optimization. The resulting configuration is described by configurationcoupling matrix. Then base on rigid body Newton-Euler iterative algorithm, a iterativeNewton-Euler function of the system is built by iterative of generalized velocitygeneralized acceleration and inverted iterative of generalized force. The dynamicsfunction of reconfigurable and modular robot is set up.Secondly, Based on Extended State Observer (ESO) and Lyapunov stability theory, abackstepping sliding mode decentralized control algorithm for reconfigurablemanipulators is proposed. For the nature of repetitive motion and nonlinear strongcoupling of reconfigurable manipulator, the backstepping technology is used to design thesliding mode decentralized controller. Because of the configuration of reconfigurablemanipulator has been changed frequently so that it is very difficult to solve the exactsubsystem dynamics model. ESO is used to model the unknown dynamics of subsystemand the interconnection term can be constructed using the state estimations. The stabilityof system and trajectory tracking performance have been proved by Lyapunov stabilitytheory.Thirdly, a new ESO based decentralized ADRC control scheme for reconfigurablemanipulators is proposed. This method no longer depends on nonlinear function withfixed form. It combines VGSTA with ESO to estimate the overall uncertainty, and theestimation error can converge to zero in finite time. In view of the nature of repetitivemotion and strong coupling of nonlinear of reconfigurable manipulators, we design adecentralized sliding mode ADRC controller by terminal sliding mode and backsteppingscheme. Because the configuration of reconfigurable manipulators is variable, thus it isdifficult to solve the exact dynamic model of the subsystem. To solve this problem, weuse VGSTA-ESO to estimate the uncertainty items of subsystems and interconnectionterms. By Lyapunov stability theory, we prove that the system of VGSTA-ESO has thestability and tracking performance. This approach will not only effevtively improve thesystem response speed, but also track the desired trajectories better, and enhance thestability of the system. Then, simulation examples are presented to illustrate theeffectiveness of the proposed decentralized ADRC control scheme.Finally, the results obtained and the lessons learned are summarized, and future workis discussed.