Task Acceptance in Hospital Logistics Robots Using Human-Robot Trust Cues: Agent Modeling

Abstract

The integration of autonomous logistics robots in hospital environments has the potential to significantly alleviate the physical and cognitive workloads of healthcare professionals. However, the successful deployment of these robotic systems is heavily dependent on the level of trust established between the human workers and the robots. This paper presents a comprehensive investigation into predicting task acceptance by leveraging human-robot trust cues through advanced agent modeling techniques. By analyzing implicit behavioral signals such as gaze duration, interpersonal distance, and interaction latency, this study constructs a computational agent model that continuously evaluates the psychological state of healthcare workers. The research was conducted in a highly realistic simulated hospital environment, involving nursing staff who interacted with a prototype logistics robot responsible for delivering medical supplies. The extracted behavioral features were integrated into a probabilistic framework to forecast whether a nurse would accept or reject a task delegated by or shared with the robot. The findings demonstrate a strong correlation between specific non-verbal trust cues and the likelihood of task acceptance, proving that agent-based modeling can serve as a highly accurate proxy for real-time trust assessment. The proposed system enables logistics robots to dynamically adapt their interaction strategies based on the predicted trust levels, thereby fostering a more harmonious and efficient collaborative healthcare ecosystem. The insights derived from this study provide a foundational framework for future designs of socially aware robotic systems in high-stress, safety-critical work environments.