Yutao Kang , Feng Liu , Weijiong Chen , Xin Li , Yajie Tao , Wei Huang
{"title":"Recognizing situation awareness of forklift operators based on eye-movement & EEG features","authors":"Yutao Kang , Feng Liu , Weijiong Chen , Xin Li , Yajie Tao , Wei Huang","doi":"10.1016/j.ergon.2024.103552","DOIUrl":null,"url":null,"abstract":"<div><p><span>Lack of situation awareness (SA) is a major source of human error in forklift operations. Effective assessment of SA levels is a critical link in improving the SA of forklift operators. Aiming at the intrusive, subjective and intermittent problems of current measurement methods, this paper proposed a SA recognition method based on eye movement and electroencephalography (EEG) features. A forklift operation experiment was designed in a real-life scenario, where eye movement and EEG data of forklift operators were collected, and the Situation Awareness Rating Technique (SART) method was used to calculate SA scores. Independent sample </span><em>t</em><span>-test and Mann-Whitney test were used to investigate the differences in eye movement and EEG indicators among participants with different SA levels. Finally, the classification models of K-Nearest neighbor (KNN), Random Forest (RF) and Support Vector Machine (SVM) were used to recognize the SA levels of forklift operators. The results indicated that the visiting time indicators, fixation time indicators and fixation count indicators in particular Areas of Interest (AOIs) are significantly different from the SA levels; The combined EEG indicators θ/β, (α+θ)/(α+β), θ/(α+β) in Frontal (F) lobe, Parietal (P) lobe and Central (C) lobe, and (α+θ)/β in P lobe and C lobe are significantly different from the SA levels; The average recognition accuracies of the models of KNN, RF and SVM are 90.61%, 94.18% and 91.15%, respectively, with the RF model having the highest recognition accuracy. The results confirmed that the method can be used to assess the SA of forklift operators in the real environment, which provides a new avenue for SA measurement.</span></p></div>","PeriodicalId":50317,"journal":{"name":"International Journal of Industrial Ergonomics","volume":null,"pages":null},"PeriodicalIF":2.5000,"publicationDate":"2024-01-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"International Journal of Industrial Ergonomics","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0169814124000088","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, INDUSTRIAL","Score":null,"Total":0}
引用次数: 0
Abstract
Lack of situation awareness (SA) is a major source of human error in forklift operations. Effective assessment of SA levels is a critical link in improving the SA of forklift operators. Aiming at the intrusive, subjective and intermittent problems of current measurement methods, this paper proposed a SA recognition method based on eye movement and electroencephalography (EEG) features. A forklift operation experiment was designed in a real-life scenario, where eye movement and EEG data of forklift operators were collected, and the Situation Awareness Rating Technique (SART) method was used to calculate SA scores. Independent sample t-test and Mann-Whitney test were used to investigate the differences in eye movement and EEG indicators among participants with different SA levels. Finally, the classification models of K-Nearest neighbor (KNN), Random Forest (RF) and Support Vector Machine (SVM) were used to recognize the SA levels of forklift operators. The results indicated that the visiting time indicators, fixation time indicators and fixation count indicators in particular Areas of Interest (AOIs) are significantly different from the SA levels; The combined EEG indicators θ/β, (α+θ)/(α+β), θ/(α+β) in Frontal (F) lobe, Parietal (P) lobe and Central (C) lobe, and (α+θ)/β in P lobe and C lobe are significantly different from the SA levels; The average recognition accuracies of the models of KNN, RF and SVM are 90.61%, 94.18% and 91.15%, respectively, with the RF model having the highest recognition accuracy. The results confirmed that the method can be used to assess the SA of forklift operators in the real environment, which provides a new avenue for SA measurement.
缺乏情况意识(SA)是叉车操作中人为失误的主要原因。有效评估态势感知水平是提高叉车操作员态势感知水平的关键环节。针对现有测量方法的侵入性、主观性和间歇性问题,本文提出了一种基于眼动和脑电图(EEG)特征的 SA 识别方法。在真实场景中设计了一个叉车操作实验,收集了叉车操作员的眼动和脑电数据,并采用情境意识评级技术(SART)方法计算 SA 分数。采用独立样本 t 检验和 Mann-Whitney 检验来研究不同 SA 水平的参与者在眼动和脑电指标上的差异。最后,使用 K-近邻(KNN)、随机森林(RF)和支持向量机(SVM)分类模型识别叉车操作员的 SA 水平。结果表明,特定兴趣区(AOIs)的访问时间指标、固定时间指标和固定次数指标与 SA 水平存在显著差异;额叶(F)、顶叶(P)和中央叶(C)的脑电综合指标θ/β、(α+θ)/(α+β)、θ/(α+β)以及P叶和C叶的(α+θ)/β均与SA水平有显著差异;KNN、RF和SVM模型的平均识别准确率分别为90.KNN、RF 和 SVM 模型的平均识别准确率分别为 90.61%、94.18% 和 91.15%,其中 RF 模型的识别准确率最高。结果证实,该方法可用于评估真实环境中叉车操作员的 SA,为 SA 测量提供了一条新途径。
期刊介绍:
The journal publishes original contributions that add to our understanding of the role of humans in today systems and the interactions thereof with various system components. The journal typically covers the following areas: industrial and occupational ergonomics, design of systems, tools and equipment, human performance measurement and modeling, human productivity, humans in technologically complex systems, and safety. The focus of the articles includes basic theoretical advances, applications, case studies, new methodologies and procedures; and empirical studies.