An Incremental Broad-Learning-System-Based Approach for Tremor Attenuation for Robot Tele-Operation.

Area of Science:

• Robotics
• Control Systems
• Machine Learning

Background:

• Physiological hand tremor hinders high-precision tele-operation tasks such as tele-surgery.
• Existing methods like deep learning face slow training due to hyperparameters, while support-vector-machine methods may sacrifice accuracy.

Purpose of the Study:

• To propose and evaluate a novel tremor filter for tele-operation systems that overcomes limitations of existing methods.
• To achieve fast tremor elimination with high accuracy in high-precision robotic applications.

Main Methods:

• A broad-learning-system-based tremor filter integrating incremental learning algorithms was developed.
• A sliding mode control approach was employed to enhance the closed-loop system's control performance.
• The system utilizes incremental learning for continuous parameter updates and stops when error converges.

Main Results:

• Simulation experiments demonstrated the feasibility and effectiveness of the proposed broad-learning-system-based tremor filter.
• The filter exhibits a fast learning rate and maintains accuracy due to its unique network structure.
• The integrated sliding mode control improved the overall performance of the slave robot system.

Conclusions:

• The proposed broad-learning-system-based tremor filter is a viable solution for mitigating physiological hand tremors in tele-operation.
• This approach offers a significant advancement in achieving precise and stable robotic operations, particularly in sensitive fields like tele-surgery.