Digital Twin System with IoT for Healthcare AMRs

What it does

The system creates virtual twins of healthcare robots, helping hospitals and nursing homes automate tasks like medication delivery and patient monitoring, reducing workload and easing staff shortages.

Your inspiration

Seeing hospitals and nursing homes struggle with staff shortages and overwhelmed caregivers inspired us to find a practical way to ease their burden. We noticed healthcare robots were not flexible enough to fully support daily operations. Our idea came from seeing digital twins transform manufacturing by improving efficiency and responsiveness. We realised this technology could similarly help healthcare, giving staff more time to focus on patients, ensuring better care, and ultimately improving lives.

How it works

Our design works by creating a virtual replica, or "digital twin", of a physical healthcare robot, using real-time data collected by sensors on the actual robot. A LIDAR sensor localises the robot and detects obstacles, helping the robot navigate safely through hospitals or nursing homes. If the robot faces an unexpected obstacle, the digital twin quickly calculates a new route. Information from the robot, such as its location and battery status, is synced to its digital twin. The system also uses AI facial recognition to correctly identify patients, ensuring medications reach the right people. Together, this approach keeps robots efficient, reduces the need for manual oversight, and if needed, provides a clearly conceptualised overview of the scene. It also improves safety, allowing caregivers to spend more time directly supporting patients.

Design process

1. Identified hospital and nursing home challenges. 2. Sketched initial healthcare robot concept. 3. Built 3D design model in Fusion360. 4. Converted model for use in NVIDIA Isaac Sim. 5. Developed physical robot prototype with IoT sensors. 6. Added AI facial recognition for accurate patient identification. 7. Enhanced obstacle avoidance and battery monitoring through tests. 8. Refined the system based on simulation feedback to achieve current robust solution.

How it is different

Our design uniquely integrates real-time digital twin technology directly with physical healthcare robots, unlike typical solutions that rely heavily on static, pre-deployment simulations or manual oversight. Traditional healthcare robots usually perform fixed tasks like medication delivery with limited adaptability, often failing to address sudden changes effectively. In contrast, our robots instantly respond to dynamic conditions — like unexpected obstacles or changing patient needs — through continuous synchronisation with their digital twins. Additionally, while most existing systems depend on proprietary software, our use of open-source platforms (ROS2, Isaac Sim) significantly lowers development costs and enhances flexibility. Moreover, our AI-based patient identification system, which combines facial recognition and real-time reaction analysis, is more comprehensive compared to conventional barcode or wristband identification methods.

Future plans

Our next steps involve refining the robot’s AI capabilities, especially enhancing facial recognition accuracy beyond 90% to ensure safer medication delivery. We plan to conduct extensive real-world testing in hospitals and nursing homes to gather user feedback and improve system usability. Additionally, we aim to expand our digital twin platform to manage entire robot fleets seamlessly, supporting larger healthcare institutions. Ultimately, we envision integrating more advanced predictive analytics and remote operation capabilities, bringing us closer to fully autonomous healthcare robots capable of independently managing complex care tasks.

Awards

ViTrox Tech 4 Good Challenge 2025 finalists: CAT 2 Healthcare - Team HealthTech