Bilateral Elimination Rule-Based Finite Class Bayesian Inference System for Circular and Linear Walking Prediction.

Area of Science:

• Robotics and Human-Machine Interaction
• Biomechanics and Assistive Technology
• Machine Learning for Predictive Systems

Background:

• Predicting walking transitions is crucial for lower limb assistive device safety and usability.
• Current systems may struggle with complex, non-linear walking patterns.
• Accurate prediction enhances the seamless interaction between assistive devices and users.

Purpose of the Study:

• To develop a novel system for predicting transitions between linear and circular walking.
• To utilize inertial measurement units for enhanced walking activity recognition.
• To improve the safety and functionality of lower limb assistive devices.

Main Methods:

• Implementation of a bilateral elimination rule-based finite class Bayesian inference system (BER-FC-BesIS).
• Leveraging bilateral human body motion data for improved prediction accuracy.
• Utilizing inertial measurement units for real-time motion capture.

Main Results:

• BER-FC-BesIS achieved a prediction accuracy of 93.98%.
• Mean predicted times for upcoming walking activities were 119.32 ± 9.71 ms (left) and 113.75 ± 11.83 ms (right).
• Low mean time differences between predicted and actual times (14.22 ± 3.74 ms left, 13.59 ± 4.92 ms right).

Conclusions:

• BER-FC-BesIS accurately predicts upcoming steady walking activities, including linear and circular patterns.
• The system offers innovative capabilities for enhancing lower limb assistive devices.
• This research aids in improving non-linear walking activity prediction for daily living applications.