Beyond Detection -- Orchestrating Human-Robot-Robot Assistance via an Internet of Robotic Things Paradigm

Hospital patient falls remain a critical and costly challenge worldwide. While conventional fall prevention systems typically rely on post-fall detection or reactive alerts, they also often suffer from high false positive rates and fail to address the underlying patient needs that lead to bed-exit attempts. This paper presents a novel system architecture that leverages the Internet of Robotic Things (IoRT) to orchestrate human-robot-robot interaction for proactive and personalized patient assistance. The system integrates a privacy-preserving thermal sensing model capable of real-time bed-exit prediction, with two coordinated robotic agents that respond dynamically based on predicted intent and patient input. This orchestrated response could not only reduce fall risk but also attend to the patient's underlying motivations for movement, such as thirst, discomfort, or the need for assistance, before a hazardous situation arises. Our contributions with this pilot study are three-fold: (1) a modular IoRT-based framework enabling distributed sensing, prediction, and multi-robot coordination; (2) a demonstration of low-resolution thermal sensing for accurate, privacy-preserving preemptive bed-exit detection; and (3) results from a user study and systematic error analysis that inform the design of situationally aware, multi-agent interactions in hospital settings. The findings highlight how interactive and connected robotic systems can move beyond passive monitoring to deliver timely, meaningful assistance, empowering safer, more responsive care environments.