Adaptive Reinforcement Learning-Enhanced Motion/Force Control Strategy for Multirobot Systems

Phuong Nam Dao,Duy Khanh Do,Dinh Khue Nguyen,Hussein Abulkasim

doi:10.1155/2021/5560277

Phuong Nam Dao, Duy Khanh Do + Show 2 more

Open Access

https://doi.org/10.1155/2021/5560277

Copy DOI

Abstract

This paper presents an adaptive reinforcement learning- (ARL-) based motion/force tracking control scheme consisting of the optimal motion dynamic control law and force control scheme for multimanipulator systems. Specifically, a new additional term and appropriate state vector are employed in designing the ARL technique for time-varying dynamical systems with online actor/critic algorithm to be established by minimizing the squared Bellman error. Additionally, the force control law is designed after obtaining the computation of constraint force coefficient by the Moore–Penrose pseudo-inverse matrix. The tracking effectiveness of the ARL-based optimal control is verified in the closed-loop system by theoretical analysis. Finally, simulation studies are conducted on a system of three manipulators to validate the physical realization of the proposed optimal tracking control design.

Highlights

In recent years, the tracking control problem for robotic systems has been remarkable, considered from academia and industrial automation [1, 2]
In the literature of cooperating mobile manipulators (CMMs)’ control objectives, they can be classified into two categories, i.e., multiple mobile robot manipulators in cooperation carrying a common object with unknown parameters and disturbances [7, 10,11,12,13,14,15,16,17,18]; one of them tightly holds the object by the end effector, and the remaining mobile manipulators’ end effector follows a trajectory on the surface of the object [19, 20]
We investigate the adaptive reinforcement learning- (ARL-)based motion/force tracking control problem for a multimanipulator system in the presence of disturbance. e control goal is to obtain the unification of the optimality principle and tracking problem

Summary

Introduction

The tracking control problem for robotic systems has been remarkable, considered from academia and industrial automation [1, 2]. For designing the ARL scheme in linear dynamical systems [31] and in nonlinear systems [32, 40], the methods are realized to find the optimal control input being Kronecker product and approximating neural networks (NNs), respectively This technique is extended for several situations, such as goal representation heuristic dynamic programming (GRHDP) with the multivariable tracking scheme [35] and uncertain discrete-time systems by using NNs [33]. (1) First, it is obviously different from [26, 27, 34, 41, 42] studying optimal control for time-invariant systems; we propose ARL-based optimal control design in the situation of the time-varying nonlinear dynamical system under the influence of the time-varying desired trajectory by utilizing the online actor/critic technique for the motion dynamic model of multimanipulator systems.

Preliminaries and Problem Statement

D12 E2 λ12 q4

Simulation Results