A robust method to detect zero velocity for improved 3D personal navigation using inertial sensors.

Area of Science:

• Biomedical Engineering
• Robotics
• Sensor Fusion

Background:

• Accurate detection of stationary periods (zero velocity) in gait is crucial for pedestrian navigation.
• Traditional methods struggle with false detections, especially at higher speeds.
• Inertial sensor data combined with biomechanical knowledge offers potential for improved accuracy.

Purpose of the Study:

• To develop a robust zero velocity (ZV) detector algorithm for precise gait cycle analysis.
• To improve the accuracy of personal inertial navigation systems (PINS) by leveraging ZV detection.
• To enhance the reliability of gait phase detection using inertial sensor measurements and a Bayesian network.

Main Methods:

• A novel zero velocity (ZV) detector algorithm integrating gait cycle segmentation and a Bayesian network (BN) model.
• Utilizing measurements from inertial sensors and kinesiology knowledge for ZV period inference.
• Employing an Extended Kalman Filter (EKF) during detected ZV periods for error state estimation and position calibration.

Main Results:

• The proposed ZV detector significantly reduces false detections, with an 80% increase in removal rate compared to traditional methods at high walking speeds.
• Personal Inertial Navigation System (PINS) performance is enhanced when aided by the EKF-based ZV detection.
• Particular improvements in altitude estimation accuracy were observed with the proposed method.

Conclusions:

• The developed Bayesian network-based ZV detector offers a robust and accurate solution for identifying stationary phases in gait.
• The integration of this ZV detector with EKF significantly improves the performance of PINS, especially in altitude estimation.
• This approach provides a more reliable foundation for sensor-based human motion analysis and navigation.