The landscape of industrial robotics is undergoing a significant shift, driven by advancements in artificial intelligence. This week at GTC 2026, Universal Robots unveiled its new UR AI Trainer, a system designed to streamline the process of developing and deploying AI-powered robotic solutions. Developed in collaboration with Scale AI, the UR AI Trainer aims to bridge the gap between laboratory research and real-world factory applications, offering a more efficient pathway for robots to learn and adapt to complex tasks. This development signals a move away from pre-programmed automation towards robots capable of continuous learning and improvement through AI.
For years, training robots has been a complex and often fragmented process. Traditionally, data collection occurred using research robots that weren’t necessarily suited for the demands of a production environment. Many systems relied heavily on visual feedback, limiting their ability to handle tasks requiring delicate manipulation or tactile interaction. The UR AI Trainer directly addresses these challenges by providing a platform for capturing high-fidelity, synchronized robot and vision data directly on the robots intended for deployment. This approach, according to Universal Robots, represents the industry’s first “direct lab-to-factory” solution for AI model training.
Accelerating AI Model Training with the UR AI Trainer
The core innovation of the UR AI Trainer lies in its ability to facilitate AI-ready data capture. The system allows human operators to guide Universal Robots’ collaborative robots – or “cobots” – through tasks in a leader-follower setup. This process captures high-quality, synchronized multimodal data during real-time demonstrations, creating the structured datasets necessary for training Vision-Language-Action (VLA) models. These models are crucial for enabling robots to understand and respond to complex instructions, and environments. The UR AI Trainer operates on the company’s AI Accelerator platform, integrating Universal Robots’ hardware with Scale AI’s software to enable scalable data capture within production settings, ultimately supporting the continuous optimization of physical AI systems.
Anders Beck, VP of AI Robotics Products at Universal Robots, emphasized the changing needs of their customer base. “Our customers, ranging from large enterprises to AI research labs, are no longer just asking for AI features,” Beck stated. “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.” This highlights a growing demand for practical, deployable AI solutions within the industrial sector.
Leveraging Direct Torque Control and Scale AI’s Expertise
A key component of the UR AI Trainer’s functionality is Universal Robots’ Direct Torque Control and force feedback features. These capabilities provide developers with direct influence over how the robot physically interacts with its environment, allowing for training on the same robust hardware used in over 100,000 industrial deployments. Direct Torque Control enables precise control of the robot’s movements, crucial for tasks requiring delicate handling or precise force application. What we have is particularly important for applications in industries like electronics assembly, food processing, and pharmaceuticals.
The partnership with Scale AI is central to the UR AI Trainer’s capabilities. Scale AI provides the software infrastructure necessary for managing and processing the large datasets generated during the training process. Ben Levin, General Manager, Physical AI at Scale AI, explained the synergy between the two companies. “Universal Robots is a leader in industrial robotics, and its global footprint offers the ideal foundation for data capture and AI deployment,” Levin said. “Together, we’ve created an integrated robotics data flywheel, allowing customers to train, deploy, and improve their AI models faster than ever before.” This “data flywheel” refers to the continuous cycle of data collection, model training, deployment, and refinement, leading to increasingly sophisticated and capable robotic systems.
A Large-Scale Dataset on the Horizon
Universal Robots and Scale AI are planning to release a large-scale industrial dataset collected on UR robots later this year. This dataset will be a valuable resource for researchers and developers working on AI-powered robotics applications, providing a standardized benchmark for evaluating and comparing different algorithms and approaches. The availability of such a dataset is expected to accelerate innovation in the field and foster the development of more robust and reliable robotic systems. The dataset will likely include data from a variety of industrial tasks and environments, offering a comprehensive view of real-world robotic applications.
Demonstrations at GTC 2026
Visitors to Universal Robots’ booth at GTC 2026 were able to experience the UR AI Trainer firsthand. Demonstrations featured two UR3e “leader” robots providing haptic input to control two UR7e “follower” robots. This setup allowed attendees to perform advanced smartphone packaging with haptic feedback, showcasing the potential of imitation learning and VLA training. The demonstration data was recorded in real-time on Scale’s stack and was immediately replayable on the AI Trainer, providing a tangible example of the system’s capabilities. The employ of haptic feedback – the ability to transmit tactile sensations – is particularly significant, as it allows robots to learn from human demonstrations even in tasks requiring fine motor skills and precise force control.
The demonstrations at GTC also included an embodied foundation model demo with Generalist AI and a haptics-based training demo with Haply Robotics, further illustrating the versatility and potential of the UR AI Trainer platform. These collaborations highlight the growing ecosystem of companies working to advance the field of AI-powered robotics.
Key Takeaways
- Accelerated AI Training: The UR AI Trainer significantly reduces the time and complexity associated with training robots for real-world applications.
- Direct Lab-to-Factory Transfer: The system enables training on the same hardware used in production environments, ensuring seamless deployment.
- Enhanced Data Capture: High-fidelity, synchronized data capture, including force feedback, provides richer training data for AI models.
- Strategic Partnership: The collaboration between Universal Robots and Scale AI combines robotics expertise with AI data infrastructure.
The development of the UR AI Trainer represents a significant step forward in the evolution of industrial robotics. By simplifying the process of AI model training and enabling robots to learn from real-world data, Universal Robots and Scale AI are paving the way for a new generation of intelligent, adaptable, and efficient robotic systems. The release of the large-scale industrial dataset later this year is expected to further accelerate innovation in the field, fostering the development of even more sophisticated and capable robotic solutions.
Universal Robots and Scale AI have not yet announced a specific timeline for wider availability of the UR AI Trainer beyond the GTC 2026 demonstrations. Further updates on product release and pricing are expected in the coming months. Readers interested in learning more about the UR AI Trainer and its capabilities are encouraged to visit the Universal Robots website and follow the company’s announcements for future updates.
What are your thoughts on the future of AI-powered robotics? Share your comments below and let us grasp how you spot these technologies impacting your industry.