Research On Repetitive Motion Planning Of Redundant Robot Manipulators Based On Neurodynamics

For redundant robot manipulators,repetitive motion planning is a significant index reflecting its performance.In industrial engineering practice,such as assembly,welding,painting and some special situations where robots are used,it is required that manipulators itself have extra degrees of freedom to perform corresponding motion planning,specific path obstacle avoidance and singular configuration avoidance tasks in the workspace.However,in practical redundant robot manipulators control,joint angle drift often occurs(When redundant manipulator performs closed path planning,the end-effector closes the path in the workspace,but the trajectory of each joint angle is not closed in the joint space.This phenomenon is called Joint angle drift).For Joint angle drift,redundant manipulator may have unknown joint configuration when it is working in industrial production.It causes serious damage to the motor and joint links,which is not conducive to the operation control and popularization of redundant robot manipulators.Based on the joint angle drift phenomenon,that the redundant manipulator can complete the repetitive motion is researched.The main research works of this paper are as follow:Firstly,the differences between conventional manipulators and redundant robot manipulators in repetitive motion planning are analyzed.Taking the joint angle drift of redundant robot manipulators as the cost function,considering the motion planning,joint angle constraint and joint velocity constraint of redundant robot manipulators,the quadratic programming mathematical model of redundant robot manipulators repetitive motion planning is established.Secondly,in order to solve the quadratic programming problem of repetitive motion planning model,considering the previous dual problems and convex optimization mathematical model,the dual neural network and linear variational inequality based on primal-dual neural network are established,and the solver of the neural network is deduced according to the dynamical differential equation.In order to verify that the neural network solver can solve the quadratic programming problem and the established neural network solver has high accuracy,operational performance is shown by the Matlab experiment data.Finally,at the end of the paper,the mathematical motion models based on the planar 3-Dof redundant robot manipulator and the existing industrial robots PUMA560 and SCR5 are established.The motion information of closed path is planned,which is used to investigate the application of the neural dynamics network solver in repetitive motion planning of redundant robot manipulators.The feasibility and effectiveness of the scheme are verified by experimental results.In this paper,the neurodynamic model,dual neural network and linear variational inequality based on primal-dual neural network,is established on the basis of Hopfield neural network model.Therefore,in the subsequent hardware implementation,according to different motion control information,different neural network controllers can be established to facilitate practical engineering application.