By Matthew Loxton
Universal Robots’ (UR new UR AI Trainer could become a key enabler for Scotland’s fast‑growing robotics and AI ecosystem. Speaking to Silicon Scotland, Mark Gray, UK Sales Manager for Universal Robots, said the company sees its AI‑focused offering as a ready‑made base for Scottish R&D teams.
The system, launched at NVIDIA’s GTC 2026 in Silicon Valley and developed with Scale AI, is designed to bridge the “lab‑to‑factory” gap by letting robots learn directly from human demonstrations on the same hardware used in production. “Our customers, ranging from large enterprises to AI research labs, are no longer just asking for AI features,” said Anders Beck, VP of AI Robotics Products at Universal Robots. “They need a way to collect high-fidelity, synchronized robot and vision data to train AI models on the same robots they intend to deploy. Our AI Trainer is the industry’s first direct lab-to-factory solution for AI model training.”
The UR AI Trainer, developed with Scale AI and unveiled at NVIDIA’s GTC 2026, is designed to let robots learn directly from human demonstrations and then deploy those skills on the factory floor. Instead of relying on rigid pre‑programming, the system collects rich data as people guide the robots through tasks, and uses that data to train advanced AI models.
At its core, the Trainer uses a “leader–follower” approach: an operator moves one robot through a job while another robot copies the motion in real time. During this process, the system records movement, force and visual signals in a tightly synchronised way, creating high‑quality datasets for training Vision‑Language‑Action models and other Physical AI techniques. Deployed on UR’s AI Accelerator platform and integrated with Scale AI’s software stack, the setup is intended to support a continuous loop of data collection, model training and redeployment on the same industrial cobots.
For Scotland, with strengths in sectors like energy, advanced manufacturing and offshore engineering, this kind of platform offers a way to test and industrialise new AI‑driven automation more quickly. Research centres such as the National Robotarium in Edinburgh and innovation hubs in Aberdeen and Glasgow already work with collaborative robots; a system like the UR AI Trainer could give those organisations a more direct route from lab experiments to real production use.
Mark Gray on Scottish research and tech hubs
Speaking to Silicon Scotland, Mark Gray, UK Sales Manager for Universal Robots, said the company sees its AI‑focused offering as a ready‑made base for Scottish R&D teams:
“At UR we see our AI Accelerator package as a platform that valuable research can be carried out on by the tech hubs, without the need for external and costly orchestration speeding up the solutions required for industry.”
By supplying a standardised cobot and data platform, UR aims to remove some of the engineering overhead that typically slows collaboration between academia, innovation centres and industry. Instead of building bespoke control and data pipelines around one‑off research robots, Scottish teams can experiment on the same type of hardware that’s already deployed in thousands of factories worldwide, making it easier to scale promising prototypes into production.
Physical AI for Scottish industry use cases
Asked about how Scottish companies might apply imitation learning and Physical AI, Gray highlighted inspection and assembly as early opportunities, particularly where vision is crucial:
“Its possible to use our systems as a base for Physical AI to enable tasks that would normally include significant development time for applications such as inspection and assembly by connecting to vision systems that can give real world orientation and perception to our cobots. Advanced manufacturing sectors such as medical device manufacturing or electronics assembly can benefit greatly from AI based robotic solutions.”
That vision‑driven approach is directly in line with what UR and its partners demonstrated at GTC, where cobots executed a highly dexterous smartphone packaging task based on data‑driven models rather than traditional point‑to‑point programming. For Scotland, similar setups could be adapted to automate visual inspection of offshore components, repeatable assembly of electronics, or tightly controlled medical device production, while still allowing human operators to teach and refine tasks on the fly.
Changing roles on the Scottish factory floor
Gray also addressed how more adaptive robots are likely to change work on the shop floor rather than simply remove jobs. He framed the shift as moving people away from undesirable tasks and into roles where human judgement and problem‑solving matter more:
“As manufacturing starts to embrace AI and robotics the workforce will see benefits in changing the nature of their roles by letting robots take over the 3D tasks (Dirty, Dangerous & Dull) and allow employees to add value with their human skill.”
For Scotland’s manufacturing strategy, which balances productivity gains with a focus on high‑quality employment and skills, that message is likely to resonate. If platforms like the UR AI Trainer are adopted by Scottish firms and research centres, they could support a transition where cobots handle repetitive, hazardous work, while Scottish engineers, operators and technicians focus on configuring, improving and supervising increasingly capable Physical AI systems.
