NVIDIA GTC 2026: Universal Robots, Scale AI launch imitation learning system to accelerate AI model training

Universal Robots (UR), in collaboration with Scale AI, unveiled the UR AI Trainer at NVIDIA GTC 2026. 

The companies said that the AI Trainer represents a tectonic shift as robots move from pre-programmed applications to fully AI-driven tasks. UR and Scale AI said that these systems are powered by robust data generated in AI training cells where robots imitate humans.

Alongside the new AI Trainer at GTC 2026, Universal Robots is also showcasing what it calls a state-of-the-art robotic foundation model from Generalist AI, a UR preferred model partner.

The company said that by utilizing this model, two UR robots will complete a complex smartphone packaging task, which was previously impossible without recent advances in the field of physical AI. 

AI-ready data capture with force feedback, direct torque control  

“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.”

Universal Robots said that AI robotics training is often hindered by fragmented hardware and low-fidelity data capture. UR added that because many of today’s training data is collected on research robots not suited for production environments, many systems rely only on visual feedback, which makes delicate or contact-rich tasks difficult. 

“The AI Trainer directly addresses these barriers,” Beck said. “By utilizing our unique Direct Torque Control and force feedback features, we give developers direct influence over how the robot physically interacts with the world, training on the same robust hardware used in over 100,000 industrial deployments.”

Scale AI partnership enables a flywheel of integrated robotics data  

The AI Trainer allows human operators to guide UR robots through tasks in a leader-follower setup while automatically capturing high-quality multimodal data for robotics AI development. UR said that operators physically guide a “leader” robot through a task while a synchronized “follower” robot mirrors the motion in real time. During each demonstration, the system records synchronized motion, force and visual data, producing the structured datasets required to train Vision-Language-Action (VLA).

Deploying on UR’s AI Accelerator platform, the UR AI Trainer combines UR robots with Scale AI software to enable data capture on UR robots in production and at scale creating continuous feedback that drives ongoing optimization of physical AI systems.

"Universal Robots is a leader in industrial robotics, and its global footprint offers the ideal foundation for data capture and AI deployment,” said Ben Levin, general manager, physical AI at Scale AI. “Together, we’ve created an integrated robotics data flywheel, allowing customers to train, deploy, and improve their AI models faster than ever before.” 

As part of this collaboration, UR and Scale AI said that they will release a large-scale industrial dataset collected on UR robots later this year.

GTC encounters with AI trainer

UR said that GTC attendees can experience the system first-hand at the company’s booth as they guide two UR3e ‘leader’ robots providing haptic input to control two UR7e ‘follower’ robots.

Universal Robots added that the process of capturing robot training data for AI models is further showcased through a demo that illustrates the same smartphone packaging task - just trained virtually.

Built in NVIDIA Omniverse and utilizing, the simulated setup allows attendees to control a virtual bi-manual UR3e system with real-time haptic feedback using two Haply Inverse3 devices as ‘leaders,’ providing a physics-accurate simulation.

"The shift toward Physical AI requires a fundamental move from rigid, pre-programmed automation to generalist robots that can perceive, reason, and learn through human-like interaction," said Amit Goel, head of robotics and edge AI ecosystem at NVIDIA. "By leveraging the NVIDIA Isaac simulation frameworks, Universal Robots is building a scalable engine for high-fidelity data capture and generation, providing the essential infrastructure to train the next generation of autonomous systems at scale."

Generalist AI demonstrates real-world robotic foundation model performance 

Complementing the two data-capture demonstrations, UR said that Generalist’s showcase highlights how advances in data collection and AI models translate into real-world robotic performance.

In the first public demonstration of Generalist’s embodied foundation models, UR said that two UR7e robots autonomously executed a complex smartphone packaging task, demonstrating dexterity, coordination and contact-rich manipulation in a real-world environment. The demonstration showed how scaled, high-quality training data combined with frontier model architectures can enable robust physical AI systems beyond the lab.

“Generalist is building embodied foundation models that deliver industry-leading dexterity and reliability,” said Pete Florence, co-founder and CEO of Generalist AI. “This demonstration on Universal Robots’ trusted industrial platform shows how physical commonsense can be translated into real-world capability, paving the way for deployment across industries at scale.”