Collective sensing of workers’ gait patterns to identify fall hazards in construction

Kanghyeok Yang, Changbum R. Ahn, Mehmet C. Vuran, Hyunsoo Kim

Research output: Contribution to journalArticlepeer-review

53 Scopus citations


Current hazard-identification efforts in construction mostly rely on human judgment, a reality that leaves a significant number of hazards unidentified or not well-assessed. This situation highlights a need for enhancing hazard-identification capabilities in dynamic and unpredictable construction environments. Given the fact that hazards cause disruptions in workers’ behaviors and responses, capturing such disruptions offers opportunities for identifying hazards. This study proposes a collective sensing approach that senses and assesses workers’ gait abnormalities in order to identify physical fall hazards in a construction jobsite. Laboratory experiments simulating an ironworkers’ working environment were designed and conducted to examine the feasibility of the proposed approach. A wearable inertial measurement unit (WIMU) attached to a subject's ankle collected kinematic gait data. The results indicated that the aggregated gait abnormality score from multiple subjects have a strong correlation with the existence of installed fall hazards such as obstacles and slippery surfaces. This outcome highlights the opportunity for future devices to use workers’ abnormal gait responses to reveal safety hazards in construction environments.

Original languageEnglish (US)
Pages (from-to)166-178
Number of pages13
JournalAutomation in Construction
StatePublished - Oct 2017


  • Construction safety
  • Gait analysis
  • Hazard identification
  • Slips, trips, and falls
  • Wearable inertial measurement units

ASJC Scopus subject areas

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

Fingerprint Dive into the research topics of 'Collective sensing of workers’ gait patterns to identify fall hazards in construction'. Together they form a unique fingerprint.

Cite this