ABB and Nvidia collaborate to enhance factory robot training
ABB has partnered with Nvidia to improve how industrial robots are trained by making virtual simulations closer to real factory conditions, the companies said on Monday.
Under the partnership, ABB’s robotics division will use Nvidia’s Nvidia Omniverse simulation platform to create more realistic training environments for robots. The technology adds detailed elements such as lighting, shadows and textures, allowing robots to better understand and react to real-world conditions before they are deployed in factories.
Marc Segura said robots often operate with limited awareness of their surroundings, which can reduce their accuracy, speed and reliability. He explained that environmental factors, such as vibrations from nearby machinery, can affect robot performance.
Segura gave the example of a factory robot working near a stamping machine that produces strong vibrations. Normally, the robot would need time to learn how to handle these conditions. With the new simulation technology, the robot can be trained virtually to deal with such issues from the start.
He said this approach could save companies significant time and money by reducing the need for trial-and-error adjustments on the factory floor.
The partnership reflects a broader trend in manufacturing where companies increasingly use digital simulations to plan production and configure robots before installing them in real factories. ABB said the new system, delivered through its robot control software, could lower costs and speed up product development by reducing the need for physical prototypes of products and assembly lines.
The technology is expected to launch in the second half of 2026 and will mainly target industries such as automotive and consumer electronics.
Electronics manufacturer Foxconn is already testing the system to help robots install side buttons in consumer devices. ABB said this task was previously difficult because shadows interfered with robot vision systems.
Deepu Talla said the industrial sector needs highly accurate physical simulations to close the gap between virtual training and real-world deployment of AI-powered robots at scale.
