Machine Learning Methodology in a System Applying the Adaptive Strategy for Teaching Human Motions

Area of Science:

• Biomechanics
• Rehabilitation Engineering
• Machine Learning

Background:

• Automatic teaching of motion activities is crucial in rehabilitation, sports, and professional settings.
• Fast motion teaching requires adaptive systems that respond to dynamic learning conditions.
• Existing systems may lack the adaptability needed for complex motion training.

Purpose of the Study:

• To present a prototype automatic teaching system employing online motion signal classification.
• To evaluate the system's effectiveness compared to non-classification-based approaches.
• To propose a standardized structure for adaptive teaching systems.

Main Methods:

• Utilizing multidimensional motion signals captured by Micro-Electro-Mechanical Systems (MEMS) sensors.
• Employing machine learning to acquire expert knowledge for signal classification.
• Implementing an adaptive system with online classification to select teaching algorithms.
• Integrating vibrotactile actuators for learner feedback.

Main Results:

• Statistical tests confirmed the superior effectiveness of the proposed adaptive teaching system.
• The system demonstrated enhanced learning process control through online signal classification.
• The study validated the hypothesis that classification-based teaching is more effective.

Conclusions:

• The developed adaptive system significantly improves the effectiveness of motion activity teaching.
• Standardizing system architecture facilitates easier configuration and implementation of specialized algorithms.
• This approach offers a promising direction for intelligent motion training systems.