Autonomous and Intervention-Based Trust Calibration in Construction Robots: Impact on Unsafe Behavior

As the construction industry increasingly adopts robotic systems, ensuring effective and safe human–robot interactions (HRIs) becomes critical. Trust in robots plays a critical role in shaping the safety and performance of workers. However, little is known about how trust evolves in dynamic construction environments or its impact on unsafe behavior in HRIs. This study examines the calibration of trust in robots within construction HRIs and its impact on unsafe behavior, focusing on the effects of both autonomous and intervention-based trust calibration methods. A mixed experimental design was conducted with 50 participants, incorporating two types of trust tasks (trust repair and dampening), two phases of autonomous calibration, and three intervention strategies (virtual reality–based training, textbook learning, and incentive and punishment). Multisource data, including questionnaires, electrodermal activity, and human–robot collision data in virtual reality, were analyzed using regression models. The findings demonstrated that both autonomous calibration and intervention-based methods significantly influence workers’ trust in robots. Virtual reality–based training can either repair or dampen trust in robots, depending on the training setup. Overtrust was found to lead to unsafe behavior, whereas autonomous calibration helped mitigate collisions, confirming the critical role of calibrated trust in enhancing safety. These results highlight the importance of trust calibration in promoting safer and more efficient HRIs in construction.