IMU Sensor-Based Worker Behavior Recognition and Construction of a Cyber-Physical System Environment

Area of Science:

• Occupational Safety and Health
• Wearable Sensor Technology
• Cyber-Physical Systems

Background:

• Construction industry faces high rates of worker injuries and fatalities annually.
• Existing AI safety systems have limitations due to CCTV blind spots.
• Current wearable sensors often lack comprehensive multi-point monitoring capabilities.

Purpose of the Study:

• To develop a novel miniaturized inertial measurement unit (IMU) sensor for multi-point attachment on construction workers.
• To integrate physiological and environmental sensors for comprehensive worker monitoring.
• To establish a real-time cyber-physical system (CPS) for immediate safety alerts.

Main Methods:

• Developed a miniaturized IMU sensor integrating accelerometers, gyroscopes, and barometric pressure sensors.
• Incorporated PPG, body temperature, and acoustic sensors for multi-modal data acquisition.
• Utilized Kalman and extended Kalman filters for data preprocessing and developed a safety evaluation algorithm.
• Established a cyber-physical system (CPS) with internal filtering, communication, and server connectivity.

Main Results:

• The IMU sensor achieved over 90% accuracy in classifying work activities at a low sampling rate (15 Hz).
• Demonstrated improved miniaturization, measurement accuracy, and server integration compared to commercial sensors.
• Successfully established a CPS enabling real-time monitoring and alert transmission to safety managers.

Conclusions:

• The novel multi-point wearable IMU sensor significantly enhances construction worker safety monitoring.
• The integrated CPS provides real-time health status evaluation and immediate accident alerts.
• This technology offers a promising solution for reducing injuries and fatalities in the construction sector.