[Human action and road condition recognition based on the inertial information]

Area of Science:

• Biomedical Engineering
• Robotics
• Signal Processing

Context:

• Intelligent prostheses require accurate human motion and road condition recognition for self-control.
• Inertial sensors in artificial limbs provide crucial data for motion analysis.

Purpose:

• To develop and validate a robust system for recognizing human motion modes and road conditions using inertial sensor data.
• To enhance the control capabilities of intelligent prostheses through accurate environmental and user state perception.

Summary:

• Inertial sensor data from lower limbs is processed using wavelet packet transform, fast Fourier transform, and principal component analysis (PCA).
• Gaussian mixture models (GMM) and hidden Markov models (HMM) are employed to classify human motion and road conditions.
• The proposed method achieves high recognition rates for various activities, including walking (96.25%) and running (92.5%).

Impact:

• The system demonstrates superior performance over Support Vector Machine (SVM) methods, offering higher recognition accuracy.
• This research provides a novel approach for improved monitoring and control of intelligent prosthetic limbs.
• Enhanced prosthetic functionality can lead to greater user independence and mobility.