Research on the Application of Multi-Source Information Fusion in Multiple Gait Pattern Transition Recognition

Area of Science:

• Robotics and Human-Computer Interaction
• Signal Processing and Machine Learning

Background:

• Exoskeleton gait pattern recognition is crucial for human-robot interaction.
• Existing methods struggle with accuracy and robustness in complex environments.

Purpose of the Study:

• To improve the accuracy and robustness of exoskeleton gait pattern transition recognition.
• To explore a multi-source information fusion model for this purpose.

Main Methods:

• Proposed a hybrid fusion strategy combining feature-level and decision-level multi-source information fusion.
• Developed a multi-classifier fusion model using D-S evidence theory and three SVM classifiers (linear, RBF, polynomial).
• Selected optimal feature subsets via correlation feature extraction and selection algorithms.

Main Results:

• Achieved an average recognition accuracy of 99.70% for eight common gait pattern transitions.
• The fusion model demonstrated improved anti-interference and fault tolerance.
• Average recognition accuracy reached 97.47% even with missing feature data, indicating good robustness.

Conclusions:

• The developed multi-source information fusion model significantly enhances exoskeleton gait pattern transition recognition accuracy and robustness.
• The hybrid fusion strategy and optimized multi-classifier system provide superior performance compared to single classifiers.
• This approach offers a promising solution for reliable exoskeleton control in diverse environments.