ABB Robotics and NVIDIA white paper defines transformative impact of physical AI on manufacturing

For decades, industrial robots have been built to do exactly what they are programmed to do. But as factories become more complex and production environments continue to change, fixed automation is facing a new challenge: adaptability. ABB Robotics and NVIDIA are exploring a different future with a new joint white paper on Physical AI, outlining how intelligent robots could understand, adapt and learn in real time. Published in July 2026, the paper focuses on bringing AI into high-precision manufacturing through a combination of robotic vision, digital twins, synthetic data and real-world feedback.

At the heart of the approach is the “sim-to-real” challenge making sure what a robot learns in a virtual environment translates accurately to the physical world. ABB and NVIDIA propose a continuous Real2Sim2Real process, where robots can be trained and tested virtually before deployment, while real-world operational data feeds back into the digital model for ongoing improvement.

The shift could fundamentally change how manufacturing automation is designed. Instead of repeatedly programming robots for predefined tasks, manufacturers could move towards Autonomous and Versatile Robotics (AVR) capable of responding to changing conditions.

With physical AI moving closer to the factory floor, the future of manufacturing may belong not simply to robots that can work, but to robots that can learn how to work better.

Adapted from the original press release published on ABB Robotics News and Media: ABB Robotics and NVIDIA white paper defines transformative impact of physical AI on manufacturing.