Noise-resistant Gradient Neural Network for Kinematic Control of Redundant Manipulators

Redundant manipulators have been used in the industrial field due to its high maneuverability. This paper constructs a noise-resistant gradient neural network (GNN) model tailored for redundant manipulators. The model reformulates the tracking problem of the manipulator into a generalized nonlinear equation, addressing the challenges posed by periodic noise and superimposed noise interference. Based on a compensatory approach, the model integrates dynamic terms to offset these noise effects, thereby enhancing the precision and stability of motion control. Numerical simulations using the PUMA 560 manipulator in MATLAB validate the model’s effectiveness. Results demonstrate that the GNN model successfully maintains trajectory accuracy under varying noise conditions, underscoring its potential for reliable operation in complex, noise-affected environments.