Closing the Sim-to-Real Gap
ABB Robotics announced a major integration of NVIDIA's Omniverse platform into its RobotStudio software, aiming to dramatically reduce the gap between simulated and real-world robot performance. The new capability, branded RobotStudio HyperReality, will launch as a subscription service in the second half of 2026.
The integration allows robotics engineers to export complete robot cell configurations — including robots, sensors, parts, and lighting — into Omniverse environments for physics-based simulation and photorealistic rendering. ABB says its Absolute Accuracy calibration technology combined with Omniverse simulation can reduce robot positioning errors to approximately 0.5 millimeters in calibrated systems.
"RobotStudio HyperReality makes industrial-grade physical AI ready for real-world deployment at scale," said Marc Segura, president of ABB Robotics. The company claims the platform can cut setup times by up to 80 percent, reduce costs by 40 percent, and speed time-to-market by around 50 percent for complex automation products.
Why This Matters for Manufacturing
RobotStudio is already one of the most widely used robot programming environments in industry, with more than 60,000 robotics engineers using it to design robot cells, program robots offline, and simulate production processes. The addition of Omniverse's physics engine and rendering capabilities addresses one of the persistent challenges in industrial robotics: the gap between how a robot behaves in simulation and how it performs in the real world.
This sim-to-real gap has historically been a major bottleneck in robot deployment. Engineers often spend weeks or months fine-tuning robot programs after installation because simulated behavior does not perfectly match physical reality. Factors like gravity, friction, sensor noise, and mechanical tolerances all introduce discrepancies that can be difficult to predict.
By incorporating NVIDIA's physics simulation engine, ABB is betting that higher-fidelity simulation will shrink these discrepancies enough to allow engineers to deploy robots with minimal on-site adjustment. The approach aligns with a broader industry trend toward digital twins — virtual replicas of physical systems that can be tested and optimized before any hardware is installed.
