Semi-supervised near-miss fall detection for ironworkers with a wearable inertial measurement unit

Kanghyeok Yang, Changbum R. Ahn, Mehmet C. Vuran, Sepideh S. Aria

Research output: Contribution to journalArticle

51 Scopus citations

Abstract

Accidental falls (slips, trips, and falls from height) are the leading cause of occupational death and injury in construction. As a proactive accident prevention measure, near miss can provide valuable data about the causes of accidents, but collecting near-miss information is challenging because current data collection systems can largely be affected by retrospective and qualitative decisions of individual workers. In this context, this study aims to develop a method that can automatically detect and document near-miss falls based upon a worker's kinematic data captured from wearable inertial measurement units (WIMUs). A semi-supervised learning algorithm (i.e., one-class support vector machine) was implemented for detecting the near-miss falls in this study. Two experiments were conducted for collecting the near-miss falls of ironworkers, and these data were used to test developed near-miss fall detection approach. This WIMU-based approach will help identify ironworker near-miss falls without disrupting jobsite work and can help prevent fall accidents.

Original languageEnglish (US)
Pages (from-to)194-202
Number of pages9
JournalAutomation in Construction
Volume68
DOIs
StatePublished - Aug 1 2016

    Fingerprint

Keywords

  • Anomaly detection
  • Fall accident
  • Ironworker
  • Machine learning
  • Near-miss fall

ASJC Scopus subject areas

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

Cite this