Fixed-time neural network control of a robotic manipulator with input deadzone

Area of Science:

• Robotics
• Control Theory
• Artificial Intelligence

Background:

• Robotic systems often face uncertainties, actuator saturation, and time-delayed constraints.
• Achieving precise and rapid control in such complex systems remains a significant challenge.
• Existing control methods may struggle with the combined effects of these issues.

Purpose of the Study:

• To develop a novel fixed-time control strategy for uncertain robotic systems.
• To address actuator saturation and time-varying constraints within the control framework.
• To ensure finite-time convergence of tracking errors for improved robotic performance.

Main Methods:

• A fixed-time control framework is employed for guaranteed convergence time.
• Model-based and neural network-based control approaches are utilized.
• Neural networks are used for uncertainty handling and adaptive compensation of input deadzones.
• A novel stabilizing function combined with an error shifting function is introduced.

Main Results:

• The proposed control method ensures that all tracking errors converge to compact sets near zero in fixed-time.
• System stability and boundedness of all error signals are rigorously proven using Lyapunov stability theory.
• The effectiveness of the fixed-time control algorithm is validated through simulations and experiments.

Conclusions:

• The developed fixed-time control method effectively manages uncertainties and constraints in robotic systems.
• The approach offers a robust solution for achieving fast and stable robotic control.
• The findings are experimentally verified on multi-joint robot manipulators, demonstrating practical applicability.