Research On Decentralized Event-triggered Interaction Control Method Of Reconfigurable Manipulator Considering Unknown Environmental Constraints

With the sustainable development of robot technology,various multi-functional robots have gradually become indispensable partners in human’s life and work.At present,the traditional manipulators can no longer satisfy the human requirements for intelligent manipulators as an increasing number of demands.Therefore,in order to complete multiple operational tasks under complex environments,the reconfigurable manipulators,with flexible configuration and excellent comprehensive performance,have attracted the extensive attention from researchers.The reconfigurable manipulators,composed of modules with standard interfaces,are able to change their configuration by increasing or decreasing the modules under the multitask objectives.The reconfigurable manipulators are widely applied to a variety of complicated space operation tasks owing to the remarkable mechanical structural characteristics with "modularization" and "reconfigurable",especially the interactive tasks in contact with the external environments.Currently,the reconfigurable manipulators with the structural advantages have been utilized in various fields that humans cannot directly participate,such as aerospace,earthquake rescue,deep-sea exploration,etc.Consequently,aiming at the reconfigurable manipulators with the environmental physical interaction,how to realize the comprehensive optimization between control accuracy and interference suppression is worth researching.Moreover,for the complicated contact tasks under extreme environments,it is a higher requirement that the system endurance is significantly improved through reasonable allocation of limited energy based on the stable system operation.In an effort to address the above practical problems,the decentralized event-triggered optimal interaction control method for the reconfigurable manipulator system considering the unknown environmental constraints is researched in this paper.Meanwhile,the specific research contents are divided into the following three parts:(1)Dynamic model establishment of reconfigurable manipulator system based on the joint torque feedback techniqueConsidering the mechanical characteristic of reconfigurable manipulators with variable configurations,it is necessary to establish the dynamic model for each joint module.Using the measurement information from joint torque sensors,the dynamic model of the reconfigurable manipulator subsystem considering the environmental collision and facing the environmental contact tasks are established,respectively.The model uncertainties are analyzed and processed,which lays the model foundation for the subsequent proposed decentralized control strategy facing the joint module space.(2)Event-triggered decentralized optimal tracking control method for reconfigurable manipulators facing collision in unknown environmentsConsidering the reconfigurable manipulator system facing the unknown spatial interference,an event-triggered decentralized optimal tracking control method under the radial basis function neural network(RBFNN)-based Romberg observer is proposed.Firstly,the RBFNN-based Romberg observer is designed with local state information to estimate the model uncertainties.Then,a performance index function including tracking errors,model approximation errors and collision interference in unknown environments is constructed to enhance the robustness of the manipulator system.Next,an event-triggered condition satisfying the Lyapunov stability theorem is devised.Simultaneously,the approximate event-triggered decentralized optimal tracking control strategy is obtained through establishing the critic neural network(NN)to solve the corresponding HamiltonJacobi-Berman(HJB)equation.Finally,the experimental verification of the designed control algorithm is carried out utilizing experimental platform of a 2 degree of freedom modular manipulator established in the laboratory.Simultaneously,the effectiveness of the proposed control method is further illustrated by the reasonable analysis of experimental results.(3)Decentralized interaction control approach under event-triggered mechanism for reconfigurable manipulators with multi-task constraintsFor the reconfigurable manipulator system facing multiple operational tasks,an eventtriggered decentralized optimal interaction control approach based on the nonlinear disturbance observer(NDOB)is presented.Firstly,the decentralized NDOB is designed utilizing the information that contains only the corresponding joint module to effectively map external spatial collision to each joint.Then,a performance index function containing interaction tracking accuracy information under the contact task constraints is constructed.Meanwhile,the HJB equation in the non-periodic sampling states with the critic NN is addressed online to obtain the decentralized interaction control strategy based on the eventtriggered mechanism.Next,considering the various interactive tasks from non-contact task to contact task,on the basis of the Lyapunov stability theorem,an event-triggered condition is designed to ensure stable system operation.Finally,two tracking tasks of manipulator,namely static interaction and dynamic interaction,are designed through the experimental platform of modular manipulator,and the feasibility and versatility of proposed approach are demonstrated through the experimental results.