Vision-based neural predictive tracking control for multi-manipulator systems with parametric uncertainty.

Area of Science:

• Robotics and Control Systems
• Artificial Intelligence

Background:

• Multi-manipulator systems face challenges in precise trajectory tracking due to parametric uncertainties.
• Existing control strategies often struggle to adapt to dynamic changes and external disturbances.

Purpose of the Study:

• To develop a robust two-layer control scheme for multi-manipulator trajectory tracking systems.
• To address parametric uncertainties and coordination problems in robotic manipulation.
• To enhance system stability and performance through advanced control techniques.

Main Methods:

• A two-layer control architecture combining image-based visual servoing and a recurrent neural network model predictive control (RNN-MPC).
• Development of a visual servoing error model using kinematic information.
• Design of an extended state observer (ESO) for disturbance estimation and compensation.
• Analysis of input-to-state practical stability under bounded disturbances.

Main Results:

• The proposed RNN-MPC approach generates reference velocities for the dynamic layer.
• The ESO effectively estimates and compensates for parametric uncertainties and disturbances.
• Simulations demonstrate the effectiveness and robustness of the proposed control algorithm.

Conclusions:

• The developed two-layer control scheme significantly improves trajectory tracking accuracy for multi-manipulator systems.
• The integration of RNN-MPC and ESO provides a robust solution for systems with parametric uncertainties.
• The approach offers enhanced stability and performance in complex robotic manipulation tasks.