State Awareness and Collision Risk Assessment Algorithm for Tower Crane Based on Bidirectional Inverse Perspective Mapping and Skeleton Key PointsSource: Journal of Construction Engineering and Management:;2025:;Volume ( 151 ):;issue: 002::page 04024205-1DOI: 10.1061/JCEMD4.COENG-15723Publisher: American Society of Civil Engineers
Abstract: To address the current issues of frequent tower crane accidents, imperfect supervision systems and the low adaptability of existing algorithms, a novel collision risk warning model for tower cranes is proposed. The model comprises a state-awareness module and a collision risk assessment module. The skeleton key points of the tower crane are proposed and constructed in the state-awareness module. On this basis, the You Only Look Once (YOLO) v8 algorithm detects the tower crane and its skeleton key points. The ByteTrack algorithm is used to track skeleton key points in real time. In the collision risk assessment module, the speed analysis method for the skeleton key points of the tower crane and the minimum safety distance assessment method between different tower cranes are proposed and compiled. Finally, the effectiveness and robustness of this collision risk warning model are verified by taking three practical projects as examples. The research results demonstrate that the precision of the box, the precision of key points, the mAP@0.5 of the box, and the mAP@0.5 of key points are 96.18%, 96.10%, 92.85%, and 92.53%, respectively. This paper’s method implements an intelligent assessment of tower crane collision risk. The model exhibits accurate visualization information in practical engineering applications and has broad application prospects.
|
Show full item record
contributor author | Fan Zhang | |
contributor author | Hongbo Liu | |
contributor author | Longxuan Wang | |
contributor author | Zhihua Chen | |
contributor author | Qian Zhang | |
contributor author | Liulu Guo | |
date accessioned | 2025-04-20T10:05:06Z | |
date available | 2025-04-20T10:05:06Z | |
date copyright | 12/11/2024 12:00:00 AM | |
date issued | 2025 | |
identifier other | JCEMD4.COENG-15723.pdf | |
identifier uri | http://yetl.yabesh.ir/yetl1/handle/yetl/4303954 | |
description abstract | To address the current issues of frequent tower crane accidents, imperfect supervision systems and the low adaptability of existing algorithms, a novel collision risk warning model for tower cranes is proposed. The model comprises a state-awareness module and a collision risk assessment module. The skeleton key points of the tower crane are proposed and constructed in the state-awareness module. On this basis, the You Only Look Once (YOLO) v8 algorithm detects the tower crane and its skeleton key points. The ByteTrack algorithm is used to track skeleton key points in real time. In the collision risk assessment module, the speed analysis method for the skeleton key points of the tower crane and the minimum safety distance assessment method between different tower cranes are proposed and compiled. Finally, the effectiveness and robustness of this collision risk warning model are verified by taking three practical projects as examples. The research results demonstrate that the precision of the box, the precision of key points, the mAP@0.5 of the box, and the mAP@0.5 of key points are 96.18%, 96.10%, 92.85%, and 92.53%, respectively. This paper’s method implements an intelligent assessment of tower crane collision risk. The model exhibits accurate visualization information in practical engineering applications and has broad application prospects. | |
publisher | American Society of Civil Engineers | |
title | State Awareness and Collision Risk Assessment Algorithm for Tower Crane Based on Bidirectional Inverse Perspective Mapping and Skeleton Key Points | |
type | Journal Article | |
journal volume | 151 | |
journal issue | 2 | |
journal title | Journal of Construction Engineering and Management | |
identifier doi | 10.1061/JCEMD4.COENG-15723 | |
journal fristpage | 04024205-1 | |
journal lastpage | 04024205-16 | |
page | 16 | |
tree | Journal of Construction Engineering and Management:;2025:;Volume ( 151 ):;issue: 002 | |
contenttype | Fulltext |