Neural network based diagonal decoupling control of powered wheelchair systems

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence in Engineering

Background:

• Powered wheelchair systems often exhibit complex multivariable dynamics with significant coupling effects.
• Existing control methods may struggle with independent control design and Jacobian calculation.
• Parameter uncertainties can challenge the robustness and performance of control systems.

Purpose of the Study:

• To propose an advanced diagonal decoupling control method for powered wheelchair systems.
• To simplify the independent control design procedures for multivariable systems.
• To eliminate the need for plant Jacobian calculation in neural network control design.

Main Methods:

• The proposed method combines systematic diagonalization techniques with neural network control design.
• A dynamic model of the powered wheelchair system is obtained for control design.
• The control strategy is implemented and verified in a real-time setting.

Main Results:

• The diagonal decoupling control effectively reduces coupling effects in the powered wheelchair system.
• The method eliminates the requirement for Jacobian calculation within the neural network control design.
• Real-time implementation demonstrated guaranteed robustness and desired system performance.

Conclusions:

• The advanced diagonal decoupling control method offers a robust and effective solution for powered wheelchair systems.
• This approach simplifies control design and enhances performance under parameter uncertainty.
• The findings validate the practical applicability and benefits of the proposed technique.