A Non-Contact Privacy Protection Bed Angle Estimation Method Based on LiDAR

TL;DR

This paper introduces a privacy-friendly LiDAR-based system for accurately measuring bed angles in healthcare settings without using cameras.

Contribution

The novel contribution is a non-contact LiDAR system that uses advanced algorithms to estimate bed angles while preserving patient privacy.

Findings

The system achieves an average angle detection error of less than 3 degrees in ICU environments.

The method uses a combination of coordinate transformation, plane fitting, and a deep learning framework for accurate estimation.

Abstract

Accurate bed angle monitoring is crucial in healthcare settings, particularly in Intensive Care Units (ICUs), where improper bed positioning can lead to severe complications such as ventilator-associated pneumonia. Traditional camera-based solutions, while effective, often raise significant privacy concerns. This study proposes a non-intrusive bed angle detection system based on LiDAR technology, utilizing the Intel RealSense L515 sensor. By leveraging time-of-flight principles, the system enables real-time, privacy-preserving monitoring of head-of-bed elevation angles without direct visual surveillance. Our methodology integrates advanced techniques, including coordinate system transformation, plane fitting, and a deep learning framework combining YOLO-X with an enhanced A2J algorithm. Customized loss functions further improve angle estimation accuracy. Experimental results in ICU environments demonstrate the system’s effectiveness, with an average angle detection error of less than 3 degrees.

Figures