ABB Robotics is integrating simulation libraries from NVIDIA’s Omniverse platform into its RobotStudio software suite in an effort to support the deployment of physical artificial intelligence in industrial robotics applications. The integration combines ABB Robotics’ programming, design and simulation environment with NVIDIA’s physically based simulation technologies and accelerated computing infrastructure. The collaboration is intended to address the longstanding gap between virtual simulation and real-world robotic deployment, often referred to as the “sim-to-real” gap.
The combined system allows developers to create digital twins of production environments and generate synthetic data to train artificial intelligence models that control industrial robots. By enabling robots to be trained in simulated environments before being deployed on factory floors, manufacturers can test and optimize workflows virtually before implementation. According to ABB Robotics, the technology can achieve simulation accuracy of up to 99 percent compared with real-world performance.
Marc Segura, president of ABB Robotics, said the collaboration aims to enable industrial adoption of AI-driven robotics by improving the reliability of simulations used in development and testing. “The last barriers to making industrial and physical AI a reality at a global scale have been removed by closing the sim-to-real gap,” Segura said.
RobotStudio HyperReality
The integration will be delivered through a new capability called RobotStudio HyperReality, which ABB Robotics plans to make available to RobotStudio users in the second half of 2026. The system combines ABB’s virtual controller, which runs the same firmware used in its physical robots, with NVIDIA’s simulation tools. These models are designed to continuously refine themselves through feedback from real-world operational data, improving accuracy over time.
Deepu Talla, vice president of robotics and edge AI at NVIDIA, said physically accurate simulation is necessary to support large-scale deployment of AI-driven robotics systems in industrial settings. “Integrating NVIDIA Omniverse libraries into RobotStudio brings advanced simulation and accelerated computing to ABB Robotics’ virtual controller technology,” Talla said.
ABB Robotics said the technology is intended to allow manufacturers to design, test and optimize production lines in virtual environments before physical deployment. By simulating real-world lighting conditions, materials and environments, developers can validate production processes and robotic workflows prior to installation, reducing the need for physical prototypes and shortening commissioning timelines.
The collaboration also includes exploration of integrating NVIDIA’s Jetson edge computing platform into ABB’s OmniCore robot controller to support real-time artificial intelligence processing at the edge. ABB Robotics has previously incorporated NVIDIA Jetson modules into its autonomous mobile robots using visual simultaneous localization and mapping (VSLAM) technology.
Early Industry Use Cases
Several companies are testing the simulation capabilities ahead of broader availability. Foxconn is conducting a pilot project focused on consumer electronics assembly, where robots must perform precise pick-and-place operations on small components across multiple device variants. In the pilot, Foxconn is using simulated training environments and synthetic data to train robots for these tasks before deployment on production lines.
Dr. Zhe Shi, chief digital officer at Foxconn, said improvements in simulation accuracy could enable faster engineering cycles and production ramp-ups in electronics manufacturing. “Precision is everything in consumer electronics manufacturing and until now, this level of accuracy and fidelity wasn’t possible in simulation and digital twins,” Shi said.
The technology is also being used by WORKR, a California-based robotics automation company that provides robotic manufacturing systems. WORKR is developing AI-driven robotic solutions using ABB robotics hardware and synthetic data generated through NVIDIA Omniverse simulation tools. The company plans to demonstrate systems built on the technology at the NVIDIA GTC conference in San Jose, presenting robotic systems designed to be deployed without requiring users to program the robots directly.
Ken Macken, chief executive and founder of WORKR, said the collaboration could expand access to advanced automation for smaller manufacturers. “This collaboration is about making industrial AI deployable today,” Macken said. ABB Robotics said the technology will be made available to the company’s installed base of approximately 60,000 RobotStudio users once RobotStudio HyperReality is released.
