Lin, J; Xing, R and Cai, Y (2026) Advancing safer construction hris: A multisourced approach for diagnosing unsafe behavior tendencies. Journal of Construction Engineering and Management, 152(2): 04025247, ISSN 0733-9364
Abstract
Human-robot interactions (HRIs) are expected to become a key component of future construction scenarios. However, the unsafe behaviors of construction workers, one of the leading causes of accidents and injuries, must be addressed in HRIs. Accurately diagnosing tendencies toward unsafe behavior is crucial for preventing such behavior in HRIs. This study focuses on diagnosing the unsafe behavior tendencies of construction workers in HRIs based on multisourced data of the cognitive processes and inherent factors influencing unsafe behavior. The key tendencies identified in this study include hyposensitivity tendency, safety complacency tendency, egotistical self-efficacy tendency, potentially profitable tendency, and risk avoidance deficiency tendency. A questionnaire and virtual reality system for HRIs were developed to collect data related to the identified influencing factors and cognitive processes. The multisourced data of 62 participants was first analyzed by the decision tree algorithm to evaluate feature importance and then modeled by five machine learning (ML) algorithms to diagnose unsafe behavioral tendencies, including logistic regression, k-nearest neighbor, support vector machine (SVM), and random forest (RF). The accuracy, precision, F1, confusion matrix, sensitivity, and specificity of the 25 ML models for five tendencies classified by the five ML algorithms were analyzed and discussed. The accuracy of the 25 ML models varies from 0.462 (SVM) to 0.846 (RF) with the precision between 0.467 (SVM) and 0.846 (RF). The results indicate that the model was generally effective in identifying true positives while keeping false positives under control. Based on the analysis of SHapley Additive exPlanations, risk perception in cognitive processes was identified as a crucial feature in diagnosing unsafe behavior tendencies. The findings emphasize the importance of multisource data for accurate diagnostics to enhance safety management in construction HRIs.
| Item Type: | Article |
|---|---|
| Uncontrolled Keywords: | behavioral tendency; cognitive mechanisms; construction workers; diagnosis; machine learning; safety management; unsafe behavior; virtual reality |
| Index terms: | machine learning, safety management, questionnaire, risk perception, accuracy, forest, unsafe behaviour, injury, interaction, self-efficacy, construction worker, virtual reality, decision tree, influencing factor, logistic regression |
| Subjects: | human factors and perception, practitioner, environmental science, financial risk, environmental hazards, behavioral psychology, data collection methods, risk assessment, virtual reality, artificial intelligence, professional development, statistical analysis, occupational health and safety management, health conditions and diseases, decision analysis |
| Topics: | Health and Safety, Risk Management, Sustainability, Cost Management, Information Management, Research Practice, Roles and Professions, Digital Applications |
| Descriptive scope: | 3 PCA |
N.B. Descriptive scope is a count of how many of the five facets of empirical research are indicated by the words used in title, abstract and keywords. It is not intended as a judgement on the research; merely a count of the kind of word we would expect to indicate Phenomenon, Concepts, Theoretical framing, Empirical techniques, Analytical techniques. If all five are present, then a code of “5 PCTEA” will indicate this. If you feel the coding for this record is questionable, we welcome discussion around the terms we matched or the way we categorized them. The facet you would expect may not be coded, or a facet may be coded inappropriately. This can also bear on a larger question, of which facets should be treated as defining in construction management research. Please get in touch, and we will look at it. More details here