Universal Robots introduced the UR AI Trainer at GTC 2026 in Silicon Valley, marking the first integrated lab-to-factory AI model training system. The platform adopts a leader-follower configuration, recording synchronized motion, force feedback and vision data on production-grade UR robots to accelerate high-fidelity dataset creation for Vision-Language-Action models.

In collaboration with Scale AI, the UR AI Trainer combines robots with Scale’s software to establish an integrated robotics data flywheel. Human operators guide a leader robot through tasks while a follower replicates movements, automatically capturing multimodal datasets. A large-scale industrial dataset collected on UR robots will be released later this year to support ongoing model optimization.

At its GTC booth, Universal Robots showcased two UR3e leader robots with haptic input controlling UR7e followers in a complex smartphone packaging demonstration. The setup highlighted real-time data capture via NVIDIA’s Isaac Sim and Omniverse for virtual trials, illustrating seamless transition from simulated environments to physical deployments using advanced foundation models.

By leveraging direct torque control and force feedback, the AI Trainer addresses limitations of research-only robots, enabling delicate, contact-rich tasks on over 100,000 installed cobots. This shift from pre-programmed automation to AI-driven operations positions Universal Robots to expand its collaborative robot offerings across manufacturing sectors at scale.