Modeling Trust in human–Robot Collaborative Construction: An Improved Cloud Bayesian Network

Abstract

Trust developed by workers towards robotic systems is critical to the successful implementation of human-robot collaboration (HRC) in construction, directly influencing operational efficiency and safety outcomes. To accurately evaluate trust risks within HRC scenarios, this study proposes an integrated method combining an improved Cloud Model (CM) with Bayesian Networks (BNs) for dynamic trust risk analysis. Initially, key factors influencing trust risks in HRC were identified through literature review and expert elicitation. The improved CM was then employed to capture inherent uncertainties and fuzziness in trust state definitions, facilitating the discretization of continuous expert evaluations into appropriate risk states. Subsequently, the BN was developed to perform forward reasoning, sensitivity analysis, and backward diagnosis, enabling proactive trust risk prediction, critical factor identification, and targeted interventions. The primary contributions of this research include: (a) identifying 11 trust factors from human, organizational, and robotic perspectives, offering a comprehensive basis for analyzing HRC trust risk in construction; (b) employing an optimized cloud entropy approach to accurately capture fuzziness and randomness in expert evaluations, thereby producing robust prior probabilities; and (c) developing a hybrid CBN framework to assess HRC trust risk in construction, demonstrating superior performance in risk perception, analysis, and control. Overall, this study provides valuable insights into safer and more effective HRC through dynamic evaluation of trust risk.