Identification of Safety Hazards Using Wearable EEG

Identifying hazards on construction worksites is a critical component in safety management. Traditional practice relies on the experience and knowledge of safety managers to identify jobsite hazards, which is subjective and leaves many hazards unidentified. A promising alternative is to use wearable biosensors, which can detect physiological responses such as electroencephalogram (EEG), heart rate (HR), and electrodermal activity (EDA) to identify hazards in real time. Since physiological signals from human body show abnormal patterns when a person encounters/perceives a hazard, understanding such patterns may lead to improved detection of jobsite hazards. In this context, this paper examines whether and how the analysis of workers’ emotional responses allows the identification of a safety hazard. The proposed method involves three steps: collecting and preprocessing EEG raw data, measuring workers’ emotional states using a bipolar dimensional emotion model, and establishing the relationship between the workers’ emotional states (e.g., fear, nervous, and displeasure) and the existence of safety hazards. Laboratory experiments are performed to simulate hazards on the jobsite. The results demonstrate the feasibility of utilizing workers’ collective EEG data to identify safety hazards in construction environments, particularly in the context of valence levels.