BeSafe B2.0 Smart Multisensory Platform for Safety in Workplaces

Area of Science:

• Engineering
• Computer Science
• Occupational Health

Background:

• Wearable technologies, including smartwatches, helmets, and belts, are increasingly vital for personal health monitoring and workplace safety.
• These devices offer real-time data on vital signs and environmental factors, functioning as advanced Personal Protective Equipment (PPE).
• The integration of Artificial Intelligence and the Internet of Things (AIoT) presents opportunities to significantly improve safety protocols in hazardous work environments.

Purpose of the Study:

• To conduct a comprehensive review of electronic systems for human activity behavior monitoring.
• To develop and present a smart multisensory bracelet and control platform designed to enhance operator security in industrial settings.
• To leverage AIoT for real-time health and safety monitoring, specifically targeting high-risk environments like construction sites and power stations.

Main Methods:

• A review of existing literature on electronics for human activity behavior.
• Development of a smart multisensory bracelet integrated with other devices into a control platform.
• Implementation of a hybrid machine learning system combining Gaussian Mixture Model (GMM) for unsupervised learning and Long Short-Term Memory (LSTM) for supervised human activity classification.
• Real-time data transmission to a central server for processing and alarm generation.

Main Results:

• The Gaussian Mixture Model achieved performance rates of 80%, 85%, 70%, and 80% for four distinct real-time classified activities.
• The Long Short-Term Memory model demonstrated high accuracy in activity recognition, with specific results of 0.769 for carrying-displacing, 0.892 for falls, and 0.921 for walking-standing activities.
• The developed platform successfully processed data in real-time, enabling an effective alarm system for detected anomalies.

Conclusions:

• The proposed AIoT-based system, utilizing a hybrid machine learning approach, significantly enhances operator safety through real-time health and activity monitoring.
• The smart multisensory bracelet and integrated platform provide a robust solution for early anomaly detection, contributing to accident prevention and improved occupational health.
• This technology holds substantial potential for application in various high-risk industries, offering a proactive approach to worker safety and well-being.