Detecting and measuring construction workers' vigilance through hybrid kinematic-EEG signals

Safety management in construction is crucial to the success of a project. A large amount of accidents is the results of construction workers unsafe behaviors, which associated with inappropriate risk detection and perception. Recently, researchers proposed to implement electroencephalograph (EEG) to measure construction workers' perceived risks based on their vigilance status. However, the EEG signals are often contaminated by the artifacts that caused by muscle movements and traditional stationary measurement metrics are not suitable for wearable implementation in the construction industry. To fill in this research gap, this study proposed a new hybrid kinematic-EEG data type and adopted wavelet packet decomposition to compute the vigilance measurement indices with redefined the EEG sub-bands. A validation experiment was conducted to examine thirty candidate vigilance indicators and two mature measurement metrics were compared to select the most proper and consistent indicators. The experiment results suggested three indices with highest correlation coefficients can be applied in vigilance detection. These quantitative vigilance indicators can provide a new perspective to understand the construction workers' risk perception process and improve the safety management on construction sites.