Collision Hazard Prevention and Notification for Construction Worker Safety Using Audio Surveillance
Date of Award
12-2024
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Civil Engineering
Committee Chair/Advisor
Dr. Tuyen (Robert) Le
Committee Member
Dr. Kalyan R. Piratla
Committee Member
Dr. Kapil Chalil Madathil
Committee Member
Dr. Chao Fan
Abstract
The construction industry faces significant safety challenges, with collision hazards ranking as the second highest cause of annual fatalities and injuries in the United States, as reported by the Occupational Safety and Health Administration (OSHA). Current collision detection methods predominantly rely on proximity technologies, which necessitate costly and complex installations on each construction equipment piece. Furthermore, auditory situational awareness declines among workers due to hearing loss and intricate construction noises, which further heightens collision risks. This research introduces an innovative, low-cost, audio-based collision prevention technology aimed at enhancing auditory situational awareness for construction workers exposed to high noise levels. The primary objective is to develop and evaluate a sound-based framework for detecting, localizing and notifying workers of heavy mobile construction equipment, thereby preventing collision hazards. This study addresses the limitations of existing collision prevention systems by proposing a cost-effective technology that leverages auditory signals, eliminating the need for extensive sensor installations and improving effectiveness in low-visibility conditions. The research aims to advance the understanding of auditory situational awareness among construction workers and explore opportunities for integrating AI-based and non-AI-based sound sensing methods. The proposed framework is comprised of four main model algorithms: (1) development of a collision hazard detection system using audio surveillance to differentiate between "abnormal" and "normal" equipment sounds in noisy environments. (2) designing a wearable device for collision hazard identification and localization, utilizing a multi-channel audio signal. (3) simulating multi-channel equipment audio scenario dataset for training a convolutional recurrent neural network (CRNN) to classify equipment types and estimate their spatial information. (4) study investigates how the integration of environmental sounds and warning audio cues through a smart headset system affects construction workers’ auditory situational awareness (ASA) and their ability to detect and respond to potential collision hazards in noise-intensive environments
Recommended Citation
Elelu, Kehinde, "Collision Hazard Prevention and Notification for Construction Worker Safety Using Audio Surveillance" (2024). All Dissertations. 3784.
https://open.clemson.edu/all_dissertations/3784
Included in
Civil Engineering Commons, Computational Engineering Commons, Construction Engineering and Management Commons
