Adaptive control for shape memory alloy actuated systems with applications to human-robot interaction.

Area of Science:

• Robotics and Control Systems
• Materials Science
• Biomedical Engineering

Background:

• Shape memory alloy (SMA) actuators offer unique advantages for robotics but face challenges in precise control for human-robot interaction.
• Existing control methods often rely on complex mechanism models, limiting adaptability and robustness.

Purpose of the Study:

• To develop a novel adaptive control strategy for SMA actuators using a gray-box model.
• To enhance the precision, robustness, and applicability of SMA actuators in human-robot interaction, particularly for hand rehabilitation robots.

Main Methods:

• A gray-box model was employed, measuring only input current and output displacement.
• An adaptive algorithm based on the multi-innovation concept with a dead-zone weighted factor was developed.
• A PI controller with gains determined by pole assignment was utilized for the control law.

Main Results:

• The adaptive PI control strategy demonstrated broad applicability and robustness, especially under load variations.
• A hand rehabilitation robot integrated with an SMA actuator showed improved position tracking accuracy and system stability.
• The system successfully accommodated various hand rehabilitation movements and gestures.

Conclusions:

• The proposed adaptive controller balances computational complexity and control accuracy, proving practical and reliable.
• The controller shows significant potential for human-machine interaction in neural rehabilitation applications.
• This approach enhances SMA actuator performance for sophisticated robotic tasks.