AR-enhanced near-miss incident alert system for onsite moving equipment
Journal
Safety Science
Journal Volume
191
Start Page
106912
ISSN
0925-7535
Date Issued
2025-11
Author(s)
Le, Thai-Hoa
Abstract
In recent years, significant studies have focused on enhancing the efficiency of construction safety monitoring by detecting and examining near-miss incidents to avoid potential accidents. Vision-based technologies have emerged because of their wide range of applications and sizeable onsite coverage with existing camera networks. However, challenges remain in raising alerts for near-miss events due to limitations in worker-centric observation, such as incomplete observation (e.g., blind spots, dynamic conditions) and insufficient information (e.g., overlooked signs, unrecognized information). Augmented Reality (AR) technology has shown the potential to mitigate these issues by providing visual indicator and instruction to improve onsite individuals’ awareness of approaching objects. This paper presents an AR-based alert system designed to detect and raise warnings for near-miss incidents caused by moving equipment in the workspace with camera networks. The proposed system is developed through a framework consisting of four modules: data registration, virtual scenario generation, location-based detection, and AR alert visualization. Through two experimental scenarios, the system demonstrated 91.3% accuracy in detection, achieving real-time monitoring with a 180-millisecond signal transmission. The results of nine qualitative interviews indicate that the AR application effectively visualized warnings, provided instructions to avoid near-miss incidents, and improved individual awareness in construction scenarios.
Subjects
Alert
AR
Construction safety
Near-miss
Warnings
Publisher
Elsevier BV
Type
journal article