Posture-related data collection methods for construction workers: A review

Yantao Yu, Waleed Umer, Xincong Yang, Maxwell Fordjour Antwi-Afari

Research output: Journal article publicationReview articleAcademic researchpeer-review

48 Citations (Scopus)

Abstract

Construction workers' posture-related data is closely connected with their safety, health, and productivity performance. The importance of posture-related data has drawn the attention of researchers in construction management and other fields. Accordingly, many data collection methods have been developed and applied to collect posture-related data. Despite the importance of workers' posture-related data, there lacks a review of previous data collection methods in the construction industry. This paper fills the research gap by reviewing previous methods to collect posture-related data for construction workers via 1) summarizing working principles and applications of posture-related data collection in construction management, which demonstrates the extensive use of motion sensors and Red-Green-Blue (RGB) cameras in posture-related data collection, 2) comparing the above methods based on data quality and feasibility on construction sites, which reveals the reason why motion sensors and RGB cameras have been prevalent in previous studies, 3) revealing research gaps of posture-related data collection tools and applications, and providing possible future research directions.

Original languageEnglish
Article number103538
JournalAutomation in Construction
Volume124
DOIs
Publication statusPublished - Apr 2021

Keywords

  • Behavior-based safety (BBS)
  • Computer vision
  • Construction worker
  • Deep learning
  • Motion sensor
  • Occupational safety and health (OSH)
  • Pose estimation

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Civil and Structural Engineering
  • Building and Construction

Fingerprint

Dive into the research topics of 'Posture-related data collection methods for construction workers: A review'. Together they form a unique fingerprint.

Cite this