Acceleration Level Control of Redundant Manipulators with Physical Constraints Compliance and Disturbance Rejection under Complex Environment

Jinglun Liang,Guoliang Ye,Yisheng Rong,Xiaoxiao Li,Zhenzhen He,Jianwen Guo,Ning Wang

doi:10.1155/2020/8844209

Abstract

Investigation of joint torque constraint compliance is of significance for robot manipulators especially working in complex environments. A lot of which is attributed to that, on the one hand, it is beneficial to the improvement of both safety and reliability of the mission execution. On the other hand, the energy consumption required by the robot to complete the desired mission can be reduced. Most existing schemes do not take the joint torque limit and other inherent physical structure limits in a manipulator into account at the same time. In addition, many unavoidable uncertainties such as the external environmental disturbance and/or electromagnetism interferences in the circuit system may influence the accuracy and effectiveness of the task execution for a robot. In this study, we cast light on the acceleration level control of redundant robot manipulators considering both four physical constraint limits and interference rejection. A robust unified quadratic-programming-based hybrid control scheme is proposed, where the joint torque constraints are converted as two inequality constraints based on the robots’ dynamics equation. A recurrent-neural-network-based controller is designed for solving the control variable. Numerical experiments performing in PUMA 560 manipulator and planer manipulator illustrate that a rational torque distribution is obtained among the joints and the considered physical structural vectors are all restricted to the respective constraint range. In addition, even disturbed by the noise, the manipulator still successfully tracks the desired trajectory under the proposed control scheme.

Highlights

With the gradually mature robotic technology, the robot is being applied to all kinds of complicated or dangerous tasks such as deep-sea exploring, search, and rescue tasks in quake-hit areas [1]
A manipulator is considered to be redundant if its degrees of freedom (DOFs) is more than the minimal ones required by the robot to complete the desired end-effector task [4,5,6,7,8]
E literature [19] investigated the inverse kinematics problem of redundant manipulators subject to torque limit, where minimum torque infinity norm (MIN) was chosen as the objective function that was to be minimized, the primary task was described as an equality constraint. erefore, a time-varying QP formulation was obtained, which was solved by the recurrent neural network (RNN) or called Lagrange neural network

Summary

Introduction

With the gradually mature robotic technology, the robot is being applied to all kinds of complicated or dangerous tasks such as deep-sea exploring, search, and rescue tasks in quake-hit areas [1]. E literature [19] investigated the inverse kinematics problem of redundant manipulators subject to torque limit, where minimum torque infinity norm (MIN) was chosen as the objective function that was to be minimized, the primary task was described as an equality constraint. Among the above-mentioned control scheme, they did not take physical limits including joint torque, joint angle, velocity, and acceleration into account at the same time. This paper investigates the inverse kinematics control problem of redundant manipulator considering both the interference rejection and the above-mentioned four physical constraints compliance and proposes a robust unified QP-based hybrid optimization scheme. (1) Based on RNN, this paper investigates the acceleration level inverse kinematics control problem of redundant manipulators with physical constraints compliance and disturbance rejection.

Methods

QP Problem Formulation

RNN Solver

Numerical Experiments

Conclusion

Conflicts of Interest