The effects of automation and transparency on human psychophysiological states and perceived system performance in construction safety automation: an electroencephalography experiment

Abstract

<p>Automation technology has experienced explosive growth in recent decades. However, the construction industry is still among the least digitized industries globally, let alone the field of construction safety. Although many types of automation have proven effectiveness in enhancing productivity and accuracy, adoption and acceptance by construction professionals are still limited. Human factors are essential for the success of automation adoption. Therefore, we investigated the effects of different levels of system automation and transparency on human states (trust, mental workload, situational awareness) and perceived performance (explainability, satisfaction, usability). An experiment was conducted using a scaffolding design safety assessment tool with varying automation and transparency levels. A between-group design was adopted in which participants were assigned into four groups (a 2 × 2 matrix of automation and transparency). In a multi-methods approach, human states were measured through questionnaires and electroencephalography, and system performance was measured through a questionnaire. The results indicate that i) automation level does not have significant impact on human states or perceived system performance, ii) a highly transparent automation system is associated with significantly higher trust and better perceived system performance, and iii) the positive impacts of transparency tend to be more obvious in low-automation systems. Due to perceived complexity, one might think that transparency would be emphasized more in highly automated systems. However, the results of this study shed light on the importance of providing a highly transparent interface for explaining system logic in all automated design safety assessment systems for construction, particularly those with low automation levels.<br></p>