AI-Powered Cobot Applications Courtesy of Teradyne Robotics-Nvidia Collaboration

Teradyne Robotics, parent company of Universal Robots (UR), and its partners have unveiled artificial-intelligence (AI)-driven robotics applications at Nvidia GTC 2025, now underway in San Jose, CA. This represents the first public demonstration of the AI Accelerator, resulting from a collaboration between UR and Nvidia, in commercially viable applications.

“Physical AI equips robots with the capacity to perceive and respond to the real world providing the versatility and problem-solving capabilities that are often required by complex use cases that have been out of scope until now,” says James Davidson, chief AI officer for Teradyne Robotics. “Instead of merely executing pre-programmed instructions, robots empowered by AI gain the ability to learn, adapt and make informed decisions grounded in their sensory input. The AI Accelerator helps our cobots better understand their surroundings, plan optimal paths, and execute tasks safely and efficiently in previously unmanageable spaces.”

A toolkit that enables development of AI-powered applications by bringing AI acceleration to UR's PolyScopeX software platform, the AI Accelerator is powered by Nvidia Isaac accelerated libraries and AI models.

• UR-partner applications employing the AI Accelerator on display at Nvidia GTC:
  Dynamic metrology from 3D Infotech: A UR3e cobot scans workpieces, compares them with CAD models, and highlights dimensional inaccuracies by projecting them onto the workpiece surface. The AI Accelerator enhances the perception capabilities of the cobot, allowing it to more accurately locate the workpiece for inspection.
• GenAI-driven programming from T-Robotics (pictured): A UR5e cobot performs CNC-machine tending. With T-Robotics ActGPT, users can describe an application setup in natural language and have it translated into a robot program with the help of the AI Accelerator.
• Reinforcement-learning assembly from Aica: A UR5e cobot executes single-arm gear assembly using reinforcement learning. The cobot locates a part using AI Accelerator-based perception and then employs a reinforcement-learning skill to complete a contact-rich assembly process.
• Bimanual assembly from Acumino: The AI Accelerator facilitates a bimanual UR5e cobot’s ability to learn complex manipulation tasks from human demonstrations. The system completes an electrical-cable-handling task as attendees collaborate with the cobot to solve the task.
• Workpiece detection for streamlined picking from Groundlight: The AI Accelerator trains a model to detect a workpiece and generate a robot program for picking, then verifies that the robot workspace is ready, thus ensuring a robust and simple-to-deploy solution.

While primarily focused on UR cobots, the underlying AI Accelerator technology also benefit Mobile Industrial Robots (MiR), another Teradyne company. For example, the MiR1200 Pallet Jack uses 3D vision for pallet detection, allowing the pallet jack to identify, pick up and deliver pallets with precision in complex environments.

See also: Universal Robots, Mobile Industrial Robots

Technologies: Pressroom Automation

Subscribe to the Newsletter

Start receiving newsletters.