An automated UAV-based vision system for detection of construction hazardous areas -

Abstract

Construction is one of the most dangerous industries whereby many hazardous tasks and conditions occur. As such, construction safety management processes are needed and primarily entail implementing health and safety regulations to maintain a safe environment. Yet a major component of these processes comprises identifying potentially hazardous areas, and consequently taking remedial actions. If left undetected, such areas would be a primary risky factor to create injuries and cause fatalities. Having said that, traditional practices have been re-lying on either the workers capabilities in identifying hazardous zones, or safety officers circulating and recognizing on-site hazards and notifying workers of un-safe conditions. However, the former practice is considered to be incompetent and not very reliable, while the latter practice of relaying information is considered manual, tedious, and time-consuming. Therefore the purpose behind this study is two-fold: (1) assessing the safety awareness of engineers, foremen and workers on different construction sites in Lebanon, by conducting a computer-based interactive survey, and (2) highlighting the potential of using camera-equipped Un-manned Aerial Vehicles coupled with computer vision techniques to conveniently identify and visualize on-site hazardous areas in indoor construction environments. Survey results revealed deficiencies in risk perception among construction personnel, and experimental results contribute in filling a gap towards detecting indoor hazards in a rapid, automated and infrastructureless way