The ability to hear auditory safety cues of mobile equipment while wearing hearing protection equipment (HPE) is critical to preventing injuries and deaths in construction. Existing collision hazard detection models using proximity technologies have limited applicability due to the need for an expensive and complex deployment of sensing devices on every piece of construction equipment. This study proposes a more affordable collision prevention technology that uses audio signals to detect the presence of mobile equipment. The study addresses the problem by improving the auditory situational awareness for construction workers exposed to loud noises with a novel sound detection model that uses artificial intelligence (AI) to detect the sound of collision hazards buried in a great deal of ambient noises. This study included three phases: (1) collecting audio data of construction equipment, (2) developing a novel audio-based machine learning model for automated detection of collision hazards, and (3) conducting field experiments to investigate the system’s efficiency and latency. The outcomes showed that the proposed model detects equipment correctly and can timely notify the workers of hazardous situations.

Safety-critical sounds at job sites play an essential role in construction safety, but hearing capability is often declined due to the use of hearing protection and the complicated nature of construction noise. Thus, preserving or augmenting the auditory situational awareness of construction workers has become a critical need. To enable further advances in this area, it is necessary to synthesize the state-of-the-art auditory signal processing techniques and their implications for auditory situational awareness (ASA) and to identify future research needs. This paper presents a critical review of recent publications on acoustic signal processing techniques and suggests research gaps that merit further research for fully embracing construction workers’ ASA of hazardous situations in construction. The results from the content analysis show that research on ASA in the context of construction safety is still in its early stage, with inadequate AI-based sound sensing methods available. Little research has been undertaken to augment individual construction workers in recognizing important signals that may be blocked or mixed with complex ambient noise. Further research on auditory situational awareness technology is needed to support detecting and separating important acoustic safety cues from complex ambient sounds. More work is also needed to incorporate context information into sound-based hazard detection and to investigate human factors affecting the collaboration between workers and AI assistants in sensing the safety cues of hazards.

Construction workplaces often face unforeseen collision hazards due to a decline in auditory situational awareness among on-foot workers, leading to severe injuries and fatalities. Previous studies that used auditory signals to prevent collision hazards focused on employing a classical beamforming approach to determine equipment sounds’ Direction of Arrival (DOA). No existing frameworks implement a neural network-based approach for both equipment sound classification and localization. This paper presents an innovative framework for sound classification and localization using multichannel sound datasets artificially synthesized in a virtual three-dimensional space. The simulation synthesized 10,000 multi-channel datasets using just fourteen single sound source audiotapes. This training includes a two-staged convolutional recurrent neural network (CRNN), where the first stage learns multi-label sound event classes followed by the second stage to estimate their DOA. The proposed framework achieves a low average DOA error of 30 degrees and a high F-score of 0.98, demonstrating accurate localization and classification of equipment near workers’ positions on the site.