Zero-Effort Camera-Assisted Calibration Techniques for Wearable Motion Sensors.

Area of Science:

• Human-computer interaction
• Wearable computing
• Biomedical engineering

Background:

• Activity recognition with wearable motion sensors is crucial for healthcare and wellness monitoring.
• Existing algorithms require known sensor orientation, necessitating manual calibration upon displacement.
• Manual calibration is inconvenient, requiring users to perform specific movements or input sensor placement data.

Purpose of the Study:

• To propose a camera-assisted calibration method for wearable motion sensors.
• To eliminate the need for manual user effort during sensor recalibration.
• To enable seamless recalibration during everyday activities.

Main Methods:

• Utilized a camera (Kinect) to assist in sensor calibration.
• Developed an approach for automatic detection of sensor location and orientation.
• Calibration occurs passively as the user performs arbitrary activities in front of the camera.

Main Results:

• Demonstrated the effectiveness of the camera-assisted calibration approach.
• Showcased that recalibration can be achieved without user-initiated actions.
• Validated the approach through experimental results.

Conclusions:

• Camera-assisted calibration offers a user-friendly solution for wearable motion sensor recalibration.
• This method simplifies the deployment and maintenance of activity recognition systems.
• The proposed approach enhances the practicality of wearable sensors in pervasive health monitoring.