A LiDAR-Based Recommendation System to Improve Sitting Posture Respecting Privacy

Area of Science:

• Ergonomics
• Computer Science
• Public Health

Background:

• Sedentary work increases the prevalence of poor posture and associated long-term health issues.
• Current posture monitoring methods raise privacy and comfort concerns due to multiple sensors and cameras.
• Proper leg positioning is crucial for maintaining good overall posture.

Purpose of the Study:

• To develop a privacy-preserving system for monitoring leg positions using Light Detection and Ranging (LiDAR).
• To identify and alert workers about poor leg positions to improve posture and prevent health problems.
• To provide posture statistics for safety engineers to identify at-risk workers.

Main Methods:

• A LiDAR-based system captures horizontal leg outlines at knee height.
• A stacked autoencoder reproduces latent representations of leg outlines.
• One-class and multi-class support vector machines identify 15 distinct leg positions and reject unrecognized ones.

Main Results:

• The system achieved over 98% accuracy in identifying leg positions on new participants.
• The system effectively monitors leg positions while preserving user privacy and ergonomics.
• Alerts for poor leg positions and posture statistics dashboards were generated.

Conclusions:

• The developed LiDAR system offers a promising solution for improving desk workers' posture.
• This technology enhances workplace ergonomics and reduces long-term health risks associated with poor posture.
• The privacy-preserving nature of the system addresses limitations of current monitoring methods.