TLDR: A new multi-robot system, inspired by swarm intelligence, uses a leader-follower approach to improve inpatient care. It integrates wearable sensors, AI-driven decision support (with ChatGPT), and RF communication for tasks like patient monitoring, medicine delivery, and fall detection in a simulated hospital environment. The system demonstrated high accuracy and reliability, offering a cost-effective solution for hospital automation and patient safety.
Hospitals today face significant challenges, including staff shortages, a growing number of patients, and the constant need for monitoring vital signs. These pressures can strain healthcare providers and affect the quality and speed of patient care. To address these issues, researchers have developed an AI-enhanced multi-robot system designed for inpatient care and diagnostic support.
This innovative system, detailed in the paper AI Enhanced Multi-Robotic Systems for inpatient care and Diagnostic support by Nakhul Kalaivanan, Girish Balasubramanian, and Senthil Muthukumaraswamy, is inspired by swarm intelligence principles. It uses a leader-follower configuration where different robots work together to perform various tasks in a simulated hospital environment. The system focuses on patient monitoring, medicine delivery, and emergency assistance.
How the System Works
The core of the system involves three main robotic units:
-
The Leader Robot: This robot acts as the central hub. It collects crucial physiological data from wearable health sensors, such as temperature, blood oxygen saturation (SpOâ‚‚), heart rate, and even detects falls. Equipped with a Raspberry Pi 5 and integrated with a ChatGPT-based AI model, it processes this data in real-time to make decisions and provide diagnostic support. It can display vital signs, communicate with patients via a chatbot on a 7-inch screen, and send automated email alerts to healthcare staff. The Leader Robot coordinates the other robots using RF-based communication.
-
The Corridor Robot (Assistant Robot): This unit is responsible for patrolling hospital corridors. It uses a HuskyLens AI camera for vision-related tasks like patient identification, corridor navigation, and fall detection. If it detects a fallen patient, it can recognize emergencies even if a patient isn’t wearing a sensor. The Corridor Robot also handles medicine delivery, picking up drugs and navigating to patient rooms. An LED tower light indicates its status (patrolling, delivering, or in emergency mode).
-
The Robotic Arm: This arm is designed for direct drug administration. It receives commands from the Leader Robot and can pick up and dispense medication according to a schedule set by doctors via the Blynk IoT platform. While initially considered for direct patient feeding, safety concerns led to its validation for precise and reliable medicine handling by transferring drugs to the Corridor Robot.
Key Technologies and Components
The system is built using cost-effective hardware components. Arduino Mega 2560 and Uno boards serve as the main controllers. NRF24L01 wireless modules enable reliable and fast RF communication between all robotic units and wearable sensors. Wearable devices include a pulse oximeter (MAX30102) for SpOâ‚‚ and heart rate, a temperature sensor, and a falling sensor. The HuskyLens AI camera provides advanced vision capabilities for object and face recognition, crucial for monitoring patient movement and detecting falls. DC motors (JGB37-520) with H-bridge drivers ensure smooth movement for the mobile robots, while a 5-way IR sensor array and PID control enable accurate line-following navigation.
AI-Driven Decision Support
A significant aspect of this system is its AI-enabled decision support. The Raspberry Pi 5, integrated with ChatGPT, analyzes patient vitals to provide early warnings of abnormal health conditions. This AI layer can offer medication recommendations (e.g., portable oxygen kit), environmental guidance (e.g., air out, move to cool area), and suggest action plans for the hospital (e.g., oxygen or antipyretic delivery). In simulations, the AI demonstrated high accuracy in providing the right treatment recommendations.
Testing and Performance
Due to ethical constraints, live patient trials were not conducted. Instead, validation was carried out through controlled self-testing with wearable sensors in a simulated hospital environment. The system achieved an overall sensor accuracy above 94%, a 92% task-level success rate (including medicine delivery, patient checks, and fall detection), and a 96% communication reliability rate. The Leader–Follower communication strategy using NRF24L01 achieved low latency, ensuring quick coordination. While fall detection using HuskyLens was 80% accurate for fallen patients, standing detection was lower, indicating an area for future improvement. Machine learning models like Random Forest and Decision Trees were used for offline analysis, achieving up to 100% accuracy for specific datasets.
Also Read:
- Agentic AI: How Independent Agents Learn to Coordinate in Multi-Agent Systems
- AI-Powered Robot Teams Master Exploration and Search in Unknown Environments
Advantages and Future Outlook
This multi-robot system offers a cost-effective and robust solution for hospital automation and patient safety, especially when compared to single-purpose, expensive commercial robots like TUG, Moxi, and Hospi. Its strengths lie in real-time vital sign integration, multi-tasking ability, low-cost design, and strong RF-based decentralized communication. Future work will focus on predictive path arbitration, expanding AI datasets, and conducting ethical hospital trials, with the ultimate goal of autonomous AI-based decision-making.
