Exploring eye-tracking searching strategies for construction hazard recognition in a laboratory scene

Abstract

Mental representations of hazard recognition are crucial to inspection strategies in construction safety management. However, few studies have investigated scan patterns as the means of mental representation of hazard recognition in the construction industry. This paper aims to explore visual searching strategies through eye-tracking scan patterns for construction safety inspection. An experimental study with 47 participants was designed to identify hazards in a pseudo construction site, and the scanning data were collected with a portable eye-tracking device. Artificial intelligence was then used to quantitatively define areas of interest based on the extracted fixation data of all participants. For two specific hazards, 23 and 29 scanpaths were generated and presented in the form of fixation sequences. Finally, the searching strategies of the participants who successfully recognized hazards were compared to the strategies of those who failed, and their scan patterns were summarized with eyePatterns®. The results show that the successful participants follow similar hazard searching patterns, concentrating on specific hazardous areas rather than unimportant distractors, following a logical and serial search pattern. They tend to observe one sub-area fully before a systematic shift to another one. This paper reports on a conceptual study of hazard searching patterns based on visual scanpaths, contributing to research of the searching strategies of mental representations for hazard recognition as well as providing useful practical implications for hazard inspection strategies in construction projects.