Theses and Dissertations

ORCID

https://orcid.org/0000-0001-8500-8631

Advisor

Wang, Jun

Committee Member

Ma, Junfeng

Committee Member

Gutbuz, Ali Cafer

Date of Degree

8-13-2024

Original embargo terms

Visible MSU Only 2 Years

Document Type

Graduate Thesis - Campus Access Only

Major

Civil Engineering

Degree Name

Master of Science (M.S.)

College

James Worth Bagley College of Engineering

Department

Department of Civil and Environmental Engineering

Abstract

Identifying construction hazards is a safety priority. Eye-tracking features have been used in prevalent research to assess hazard identification skills. However, confusion exists over their effectiveness due to certain features being more common. To address this gap, a machine learning-based approach is suggested to identify the most effective eye-tracking features for assessing hazard identification skills, thereby improving workplace safety. Four hazard types and eleven eye-tracking features were identified; accordingly, eighteen strategies were developed using these features. Support vector machines (SVM) and artificial neural networks (ANN) models were developed to assess these strategies. Virtual reality was used for the data collection. The findings showed that these models effectively analyzed eye-tracking features for hazard identification, suggesting the most impactful features for each hazard type and overall. This approach offers construction professionals a more precise assessment of hazard identification skills, enabling targeted improvements and reducing accident risks on construction sites.

Download

Share

COinS