Revealing Hidden Safety Hazards Using Workers' Collective Bodily and Behavioral Response Patterns

目前安全管理中的危险识别工作主要受到人类识别危险的能力和/或其对已知危险的现有知识的限制。因此，许多危险是无法识别的，造成了无法管理的风险。为了增强危险识别能力，本研究侧重于理解和开发人类在与物理环境系统相互作用时的身体和行为反应。众所周知，系统内的潜在危险可能会导致行动的不稳定，并最终导致事故。然而，人类行为的不可预测性在利用对这种行为的分析来识别不稳定的系统条件方面构成了一个关键的挑战。这项研究中开发的方法将能够评估与人类反应相关的集体模式，以估计整个建筑工地发生危险位置的可能性。因此，从这项研究中获得的知识将提高我们利用响应信息预防事故的能力，从而减少与建筑相关的事故造成的伤亡。研究成果将被整合到工程课程开发、本科研究活动、行业研讨会和针对K-12学生和代表性不足的学生群体，特别是女性和少数族裔的外展活动中。本研究的目标是检查工人的集体身体和行为反应模式是否、如何以及在多大程度上识别已识别/未识别的危险，以提高建筑环境中的安全性能。这项研究的重点是使用从可穿戴式惯性测量传感器捕获的工人运动学传感数据，检测导致跌倒事故的危险，这是建筑业中最危险的伤害事件。这项研究假设，多名工人在一个位置的平衡和步态明显的集体异常与该位置存在已识别/未识别的跌倒危险的可能性(和/或风险)相关。为了验证这一假设，该项目将：1)确定适当的指标，以表征已识别和未识别的危险对工人平衡和步态造成的扰动；2)建立评估工人步态和平衡异常的险情指数(NMI)；3)调查集体NMI模式与每个位置危险的存在和风险之间的关系；4)识别和检查适当的传感器网络平台，以实现该方法的可扩展实施；以及5)通过在建筑工地中的应用，验证所开发方法的有效性和有用性。

Current hazard-identification efforts in safety management are mostly limited by humans' abilities to recognize hazards and/or by their existing knowledge of known hazards. Consequently, numerous hazards go unidentified, creating unmanageable risks. To enhance hazard recognition capabilities, this research focuses on understanding and exploiting humans' bodily and behavioral responses in their interaction with the physical environmental system. It is well recognized that a potential hazard within the system may cause instability in actions, and ultimately accidents. However, the unpredictable nature of human behavior poses a critical challenge in utilizing the analysis of such actions for the identification of unstable system conditions. Methodologies developed in this research will enable the evaluation of collective patterns associated with human responses to estimate the likelihood of hazard locations across a construction site. Thus, knowledge gained from this research will advance our ability to utilize response information for accident prevention, leading to reduced injuries and fatalities from construction-related accidents. Research outcomes will be integrated into engineering curriculum development, undergraduate research activities, industry workshops, and outreach activities for K-12 students and underrepresented student groups, especially women and minorities.The objective of this research is to examine whether, how, and to what extent workers' collective bodily and behavioral response patterns identify recognized/unrecognized hazards for the purpose of enhancing safety performance in construction environments. This research focuses on detecting hazards that causes fall accidents, a single most dangerous injury event within the construction industry, using workers - kinematic sensing data captured from wearable inertial measurement sensors. This research hypothesizes that the collective abnormalities apparent in multiple workers' balance and gait in one location is correlated with the likelihood of the presence (and/or the risk) of a recognized/unrecognized fall hazard in that location. To test this hypothesis, this project will: 1) identify appropriate metrics that characterize the perturbation to workers' balance and gait caused by recognized and unrecognized hazards; 2) model a near-miss index (NMI) that evaluates the abnormalities of workers' gait and balance; 3) investigate the relationship between the collective NMI patterns and the presence and risk of a hazard in each location; 4) identify and examine appropriate sensor network platforms for the scalable implementation of the approach; and 5) validate the efficacy and usefulness of the developed approach through its application within construction sites.