Robotic Nurse System

Abstract

This project presents the design and implementation of a Robotic Nurse System intended to assist with basic flu assessment and wound treatment procedures. The system integrates sensor-based hardware exe- cution, event-driven software coordination, and image-based wound classification within a unified, modular architecture. A Raspberry Pi is used for high-level decision-making, workflow control, and user interaction, while an Arduino Mega handles all low-level actuation and sensor verification to ensure deterministic and safe physical behavior. The system incorporates multiple sensing modalities, including temperature, pulse oximetry, proximity detection, and vision-based wound classification using a YOLO-based model trained externally on a GPU- enabled environment. All actions are triggered explicitly through user input or hardware-originated events, with no autonomous or background execution. Experimental results demonstrate that the system operates reliably within its design constraints and achieves the intended functional objectives using primarily hobby-grade components and a limited training dataset. While further refinement is required for medical deployment, the project validates the feasibility of an event-driven robotic assistance framework and provides a scalable foundation for future development.