Endpoint Control of Manipulator Driven by Dual Neurodynamic Algorithm

In this paper, the time-varying neurodynamic algorithm is explained as a Recurrent Neural Network (RNN) with inputs. A collaborative computational method for neurodynamic algorithms is proposed via connecting neural networks by time-varying parameters as input of network and this method is used to compute a control strategy for the endpoint of the manipulator. The motion of the endpoint of the manipulator is planned through an optimization problem and solved using a neurodynamic algorithm. The state output of this neurodynamic algorithm is mapped to a distributed motor set at each joint through an optimal joint angular velocity programming problem. This distributed motor set assigns the required angular velocity of the joints to each motor through an allocation problem and the corresponding neurodynamic algorithm. The control of the entire manipulator is achieved through the cooperative use of two neural network. Benefiting from the time varying neurodynamic algorithm being a method with continuous dynamical behaviour and being able to respond to time varying inputs, the proposed multiple neurodynamic algorithms collaborative approach is a parallel computational method that can be implemented in hardware. Numerical simulation experiments were conducted to achieve endpoint control of a manipulator using dual neurodynamic algorithms.