Robust Adaptive Learning Control of Space Robot for Target Capturing Using Neural Network

Area of Science:

• Robotics
• Control Systems Engineering
• Aerospace Engineering

Background:

• Space robot operations require sophisticated control for tasks like target capturing.
• Contact impact dynamics introduce significant nonlinearities and uncertainties.
• Adaptive and robust control strategies are crucial for handling unknown system dynamics.

Purpose of the Study:

• To develop a robust adaptive learning control strategy for space robots engaged in target capturing.
• To address the challenges posed by nonlinear dynamics during impact and unknown post-capture system behavior.
• To ensure fast convergence to desired states and stable system performance.

Main Methods:

• Utilizing momentum conservation theory to model impact dynamics.
• Designing robust control using nonsingular terminal sliding mode (NTSM) and fast NTSM.
• Implementing adaptive learning control with neural networks and disturbance observers for unknown dynamics.
• Employing a serial-parallel estimation model for adaptive law updates.

Main Results:

• The proposed control effectively handles nonlinear dynamics during contact impact.
• Adaptive learning control successfully manages unknown dynamics of the combined system post-capture.
• System signals remain bounded, and the sliding mode surface converges in finite time.
• Simulation studies validate the tracking and learning performance.

Conclusions:

• The robust adaptive learning control strategy is effective for space robot target capturing.
• The NTSM and adaptive learning components ensure fast and stable performance under uncertainties.
• This approach enhances the capability of space robots for complex manipulation tasks.