“Excuse me! Please make way!” a nurse shouts urgently while pushing a patient on a gurney through the corridors of Mackay Memorial Hospital’s emergency department. Shortly afterward, an infant lying in an incubator is transferred to the pediatric ward. Nearby, patients seated in the waiting area watch quietly from their wheelchairs, accompanied by hospital volunteers or foreign caregivers. This is a daily scene in Mackay’s emergency department, and it vividly reflects the immense pressure faced by emergency medicine in Taiwan.

In this fast-paced, high-risk hospital environment, a robot shaped like a dog, with steady, confident steps, moves slowly through the crowd. It is a newly arrived patrol assistant—capable of detecting vital signs of a fallen person when no one else notices, and immediately notifying security.

This is not a scene from a technology exhibition. It is a reality that Mackay Memorial Hospital is about to realize. Cloud software company Ubitus is working with Mackay to experiment with and develop care-oriented robots that can truly integrate into medical environments.

The First Frontline for AI Deployment: The Emergency Department

This experiment has two main protagonists.

Mounted on the robot dog’s back are multiple sensors. Its primary mission is to patrol the emergency department, identify whether someone has collapsed and requires immediate security notification, and even detect vital signs such as heart rate. Because traffic flow and scenarios in the emergency department change rapidly, the Ubitus team built a 3D spatial model rather than relying on conventional 2D navigation systems. The robot dog can also detect water on the floor, avoid obstacles, reroute when necessary, and even wave, shake hands, or make heart gestures—combining practicality with approachability.

If a patient collapses inside the hospital, the robot dog can detect vital signs and immediately send alerts.

The other robot, a wheeled guidance robot named “Dr. Mackay,” features a clean, minimalist design and is equipped with a large touchscreen and voice recognition system. Visitors can use it to check symptoms, identify the appropriate department, or find destinations within the hospital. For example, if someone asks, “I have a headache—which department should I visit?” it can provide instant recommendations. If a user says, “I want to go to the pharmacy,” it can guide them directly, while also delivering health education information.

“The rhythm and culture of medical settings are very unique. Simply placing machines into hospitals is not enough—they must be deeply integrated into existing workflows to truly deliver value,” said Chang Wen-Han, Superintendent of Mackay Memorial Hospital, succinctly capturing the core challenge of introducing AI robots into healthcare.

As early as a decade ago, Mackay began laying the groundwork for digital healthcare, gradually establishing departments of applied medicine, data medicine, and imaging medicine. Through data integration and system-building, the hospital sought to optimize clinical judgment and medical decision-making. These years of accumulated foundations paved the way for the later adoption of AI technologies.

In 2024, Mackay took a further step by integrating NVIDIA’s AI platform, keeping sensitive patient data on-site to ensure information security. At the same time, the hospital opened up its most complex and fast-moving clinical environment—the emergency department—as a real-world training ground for AI models. Chang noted, “If robots can assist with patrols, transportation, and environmental checks, nurses can devote more time to direct patient care.”

Chang Wen-Han believes that for robots to work effectively in hospitals, they must be deeply aligned with hospital culture and workflows.

Before Robots Can Work, They Must Understand How Hospitals Operate

Mackay’s chosen partner, Ubitus, is one of the few Taiwanese startups to receive investment from NVIDIA. Leveraging its strengths in cloud streaming and GPU computing, the company has successfully entered the field of AI-driven smart healthcare.

“It’s not the machine that defines the task; it’s the task that drives the machine’s evolution,” explained Ubitus founder and CEO Ronnie Goh. Their most important work is not showcasing hardware performance, but clarifying what the medical frontline truly needs. Every function of the patrol robot dog and Dr. Mackay is distilled from real-world clinical scenarios, then repeatedly tested, refined, and trained by the technical team.

“A collapsed patient may be experiencing in-hospital cardiac arrest and requires immediate resuscitation. If the robot dog can detect this and alert staff in real time, it can greatly enhance patient safety,” said Dr. Huang Shu-Tien, senior attending physician in the Department of Emergency Medicine at Mackay Memorial Hospital.

Dr. Huang emphasized that different medical settings have different scenarios and needs, requiring continuous adjustment in robot development.

How can robots safely patrol emergency departments, identify fallen patients, and send instant alerts? How should they respond to patients’ spoken questions, guiding them to appropriate departments and navigating hospital routes? And even more, during moments of extreme anxiety for patients and families, could robots offer multilingual prayers to help bring emotional comfort? Whenever workflows change or needs evolve, both teams reconvene to reassess and redesign the robots’ task logic.

Dr. Huang further explained, “Different campuses and different spaces—lobbies, wards, operating rooms—each have unique application scenarios. That’s why we must continuously test and adjust.”

While other companies ask, “What can this machine do?” Ubitus works with hospitals to define, “What should this machine do, how should it do it, and who needs it?” Through this process, Ubitus has gradually accumulated modular, replicable know-how—this is its true core product.

Not Selling Hardware, but Continuous Optimization Through Subscription

Because Ubitus does not sell equipment outright, its operating model differs fundamentally from traditional robot manufacturers. Instead, it offers subscription-based services, providing continuously evolving care robots optimized according to real-world needs.

Goh explained that healthcare institutions pay a one-time project development fee to design task logic and functional modules based on their needs. Instead of purchasing expensive hardware, they lease the robots and subscribe to ongoing maintenance services, receiving continuous model upgrades and technical support. Every improvement is driven by data and experience gathered from actual use.

Goh believes that only by clearly understanding real clinical needs can truly effective robots be developed.

For healthcare institutions, this subscription model significantly lowers adoption barriers and makes risk management easier. For Ubitus, each deployment becomes a co-creation opportunity with medical professionals—accumulating data and refining models.

The Key to Going Global: Field-Based Co-Creation

Today, this system is preparing to expand beyond Taiwan. Ubitus is in discussions with healthcare institutions in Japan and Southeast Asia, aiming to export this “field-based co-creation” framework to regions with different languages, cultures, and healthcare systems. There, AI will no longer be a cold technological showcase, but a care revolution that starts from real needs and provides ongoing companionship.

The Ubitus team continues testing and refining the robot dog to further enhance its capabilities.

“The functions of care robots should adapt to their environments—outpatient clinics, diagnostics, inpatient care—all require different workflows,” Chang emphasized. By connecting Taiwan’s medical expertise, national health insurance data, technology industry, and AI computing power, the Taiwan Model emerges—not theoretical, but operational in real clinical settings.

The collaboration between Mackay and Ubitus not only highlights Taiwan’s leading position in smart healthcare, but also offers a valuable reference for healthcare institutions worldwide, demonstrating how advanced technology can be seamlessly integrated with real-world needs to deliver practical, actionable innovation in medical care.

(Producer / Wu Yun-Ru; Special Correspondent / Chu Chih-Yi; Editor-in-Chief / Wang Li-Hua)

Profile: Ubitus

Founded: 2012
Founder & CEO: Ronnie Goh
Headquarters: Tokyo, Japan
Core Businesses: Cloud game streaming, AI solutions (image generation, multilingual applications, etc.)
Technical Strengths: Proprietary GPU virtualization and cloud streaming technology (GameCloud®), enabling rapid cloud deployment of AI applications with low latency and high-quality visuals across devices

Further Analysis: What Technologies Has Ubitus Deployed at Mackay?

B200 GPU Platform: NVIDIA’s latest-generation high-performance GPU, designed for large-scale AI model training and cloud computing. It delivers stable performance under massive computational demands and is a core piece of hardware in the global AI and cloud race.
Meta Llama 4: Meta’s next-generation large language model, offering faster inference, multilingual generation, and stronger contextual understanding. Ubitus builds cross-language, cross-environment smart healthcare solutions based on Llama 4.

Trend Series: Care Robots Go Live—Who Pays, and How Do They Make Money?

Hospitals Build Their Own Nursing Robots! China Medical University Leads × Changlian R&D, Realizing “Healthcare Sovereignty AI”
A New Assistant for ER Heroes: Robot Dogs—Mackay Teams Up with Ubitus to Use AI in the Most Complex Medical Settings
NT$30 Billion in Computing Power for a Single Nursing Robot! Foxconn’s 3+3+3 Strategy, Starting from Nursing Stations
Hardware Drivers: From Tax Systems to Smart Hospitals—Bojun’s 100% Taiwan-Made Robots Go Global
Interview with Taiwan’s “Father of Robotics,” Lo Ren-Chuan: For Robots to Leave the Lab, Commercialization Must Successfully “Take the Baton”