An Audio-Based Emotion Monitoring System for Enhancing Construction Worker Safety

The high suicide rate among construction workers in the U.S. is a concerning issue. For every 100,000 workers in this field, 53.3 are reported to commit suicide, compared to the national average of 12.93. This statistic highlights the importance of addressing the mental well-being of construction workers, which directly impacts their performance and safety on the job. In this research, an audio-based system is proposed to monitor the emotions of construction workers in real-time, aiming to enhance their safety and address potential concerns proactively. The system involves a novel stimulus mechanism that encourages workers to interact with an AI, which records their voices. An emotion classification model analyzes these recordings to assess the workers' emotional states. Considering the noisy environment of construction sites, the model is trained on mixed human voices with construction noises, ensuring its effectiveness in real conditions. This proposed system allows for continuous emotional monitoring without the need for intrusive equipment like EEGs. Although the system is currently focused on detecting emotional states, a framework is proposed for a proactive policy that could recommend preventive actions based on the workers' emotional histories. By implementing advanced technologies, this research aims to improve the emotional well-being of construction workers, creating a safer and more productive work environment, which could lead to better project outcomes and reduced downtime.