RGB-D Camera Based Walking Pattern Recognition by Support Vector Machines for a Smart Rollator.

Area of Science:

• Robotics
• Biomechanical Engineering
• Computer Vision

Background:

• Smart rollators require advanced gait analysis for user assistance.
• Accurate walking pattern detection is crucial for personalized mobility support.
• Existing methods often lack precision in complex gait analysis.

Purpose of the Study:

• To develop and validate a robust walking pattern detection method for smart rollators.
• To extract key gait features from lower extremity movement data.
• To improve the accuracy of gait analysis in real-world assistive device applications.

Main Methods:

• Utilizing depth data from an RGB-D camera to detect lower extremities.
• Segmenting 3D point data to create a skeletal representation of the gait.
• Employing K-means clustering and Markov chains for gait feature analysis.
• Training Support Vector Machines (SVMs) for multi-class walking pattern classification.

Main Results:

• The proposed method successfully extracts gait features from skeletal data.
• Clustering identified 6 key gait features, enabling gait sequence modeling.
• The Markov chain model effectively represents walking patterns via stationary distribution.
• Experimental results show superior performance compared to seven existing methods.

Conclusions:

• The developed method provides an effective approach for smart rollator-based walking pattern detection.
• The integration of computer vision, machine learning, and biomechanics offers significant advantages.
• This technology has the potential to enhance the functionality and user-centricity of assistive devices.