Right now, AI simulation is changing the world of industrial manufacturing. Robots used to struggle to move from virtual training to real factory environments without breaking down but now, Nvidia and ABB have finally come up with a lasting solution through a physically accurate simulation.
Together, they’re building an accurate simulation so the virtually trained robots can perform perfectly in a physical environment. The rest of the industry is observing closely.
Why AI Simulation for Industrial Robotics Was Broken
For decades, robotics engineers encountered the same problem. They trained robots in virtual environments and those robots failed the moment they were exposed to physical environments.
For example, changes like lighting differences, unpredictable surfaces, and material variations caused robots to misread their environment and malfunction. The industry came to call this the sim-to-real gap.
As a result, manufacturers lost weeks to physical recalibration and spent fortunes on prototypes that should never have existed. The sim-to-real gap cost the industry dearly and for a long time nobody closed it.
How ABB and Nvidia Fixed It
To conquer this problem, ABB and Nvidia decided to join forces.
Firstly, ABB embedded Nvidia Omniverse libraries directly into RobotStudio, its flagship simulation platform. This gave engineers a physically accurate virtual factory that replicated real-world conditions with enough precision to replace physical prototyping entirely.
Furthermore, ABB’s virtual controller runs the exact same firmware as the physical robot, enabling simulation results that correlate with real-world behavior at up to 99% accuracy.
Beyond that, synthetic images generated in the simulation feed directly into machine vision models so engineers no longer need massive real-world datasets to teach robots how to see.
Additionally, ABB’s Absolute Accuracy calibration technology also reduces robot positioning errors from around 15mm down to just 0.5mm.
This means manufacturers can now cut deployment costs by up to 40%, reach production up to 50% faster, and slash commissioning times by up to 80%.
The Engineers Say This Changes Everything
Notably, the people building this technology are direct about what it means.
Marc Segura, President of ABB Robotics, puts it plainly. “We are offering a platform where you can close the sim-to-real gap at industrial grade. Now with Nvidia, we’re enhancing that and expanding that beyond the robot and the environment.”
Deepu Talla, VP of Robotics and Edge AI at Nvidia, reinforces that point. “The industrial sector needs physically accurate simulation to bridge the gap between virtual training and the real-world deployment of AI-driven robotics at scale.”
Together, both leaders describe a technology that completely eliminates the old workflow.
AI Simulation for Industrial Robotics Meets the Assembly Line
To be fair, Foxconn put this to the test first. Its assembly robots trained entirely inside the simulation and handled multiple device variants with delicate metal components, delivering 99% real-world accuracy before touching a single product on the line.
Dr. Zhe Shi, Chief Digital Officer of Foxconn, captures the significance clearly. “Precision is everything in consumer electronics manufacturing and until now, this level of accuracy and fidelity just wasn’t possible in simulation and digital twins.”
Meanwhile, Workr took the same technology to smaller manufacturers, building a platform that lets businesses onboard new parts in minutes without any programming expertise.
Therefore, advanced robotics no longer belongs solely to companies with huge budgets.
AI Simulation for Industrial Robotics Has Already Won
Ultimately, ABB and Nvidia moved first but the rest of the industry is already responding.
At GTC 2026, Nvidia confirmed it is working with Fanuc and Yaskawa alongside ABB to deploy next-generation intelligent robotics across manufacturing. Also, Samsung targets fully AI-driven factories by 2030 and is investing in digital twin simulation to get there.
Moreover, a recent Deloitte survey found that 58% of companies already use physical AI and that figure climbs to 80% within two years.
Jensen Huang, CEO of Nvidia, said it best at GTC 2026. “Every industrial company will become a robotics company.” The factory of the future runs on simulation first and ABB and Nvidia have already proven it works.
