Artificial Intelligence Platform for Monitoring and Risk Prevention in Confined Spaces

Abstract Confined spaces represent challenging work environments due to the combination of significant risks, such as toxic atmospheres, inadequate ventilation, and restricted access. These factors make occupational safety a priority in such environments, especially in high-risk sectors like the oil and gas industry. Despite regulatory standards and traditional safety practices, the limitations of conventional technologies hinder continuous monitoring and rapid emergency response. In this context, the present study introduces the development, integration, and validation of a technological platform for monitoring confined spaces, combining wearable devices, environmental sensors, and cameras into a centralized and scalable solution. The primary objective was to enhance occupational safety and operational efficiency in high-risk environments, such as those found in the oil and gas industry. The platform was structured into three main layers: data collection, local processing, and management with visualization, enabling real-time monitoring, anomaly detection, and automatic alert generation. Tests conducted in laboratories and refineries demonstrated the solution's effectiveness in real-world conditions, with results highlighting the accuracy of data collection, the system's robustness in challenging scenarios, and the usability of its graphical interface. Continuous improvements, such as optimizations in communication protocols and integration of PTZ cameras, ensured greater reliability and flexibility of the solution. The use of active learning proved essential for refining detection models, increasing the precision in identifying risks such as changes in workers' vital signs and critical atmospheric conditions. The results underscore the importance of emerging technologies in confined space safety management, overcoming the limitations of traditional approaches and meeting regulatory requirements, such as NR-33 and ISO 45001. This study contributes to the technical literature by exploring the integration of IoT devices and artificial intelligence into a unified system, promoting proactive risk management and reducing operational costs. It is concluded that the developed platform offers a robust and efficient solution for high-risk industrial operations, with potential applications in various sectors.