Inertial Sensor-Based Step Length Estimation Model by Means of Principal Component Analysis.

Area of Science:

• Biomechanics and Human Motion Analysis
• Sensor Technology and Data Fusion
• Indoor Positioning Systems

Background:

• Inertial sensor-based step length estimation is crucial for pedestrian dead reckoning (PDR) indoor positioning.
• Existing models often lack accuracy and do not fully utilize human walking kinematics or actual step lengths.
• There is a need for robust and orientation-independent step length estimation models.

Purpose of the Study:

• To introduce a novel step length estimation model utilizing acceleration magnitude.
• To leverage Principal Component Analysis (PCA) for deriving the model from experimental data.
• To evaluate the model's performance across various smartphone positions and walking conditions.

Main Methods:

• Collected walking data from anatomical landmarks using an optical measurement system.
• Developed a new step length estimation model based on acceleration magnitude.
• Applied Principal Component Analysis (PCA) to characterize experimental data for model derivation.
• Evaluated the model using long-term walking data from four typical smartphone positions.

Main Results:

• The proposed model achieved a mean absolute stride length estimation error of 6.44 cm.
• Outperformed all compared acceleration-based step length estimation models.
• Demonstrated minimal sensitivity to walking speed and smartphone position.
• Showed no sensitivity to smartphone orientation.

Conclusions:

• The novel model offers accurate and reliable step length estimation for PDR-based indoor positioning.
• Its independence from smartphone orientation simplifies practical application.
• The model's robustness to variations in walking speed and position enhances its utility.
• Publicly available dataset facilitates further research and validation.