SKAI Intelligence, a South Korean company, was named as a major contributor in this white paper jointly published by ABB Robotics and NVIDIA. ABB Robotics and NVIDIA published the white paper 'Industrial Physical AI in High-Precision Manufacturing' on the 20th.

Sky Intelligence announced that it participated as a Main Contributor alongside Deloitte and AsiaInfo, providing technical input regarding industrial digital twins, work-unit synthetic data, robot vision data, and Real2Sim2Real engineering loops.

The white paper suggests that the core evaluation criteria for physical AI implementation should be expanded from 'Model Accuracy' to 'Deployment Readiness'.

In the process of the high-precision manufacturing industry transitioning from traditional fixed automation to robot vision-based physical AI, robot system and AI computing rather than single model performanceIt is emphasized that establishing an engineering system that integrates testing, simulation, synthetic data, and field feedback is important.

In the white paper, ABB Robotics provides digital engineering capabilities based on RobotStudio, while NVIDIA provides a simulation and AI computing ecosystem based on Omniverse.

Deloitte contributed industry transformation strategies and ROI evaluation methodologies, while AsiaInfo contributed enterprise AI platforms and data governance perspectives.

Sky Intelligence constructed work-unit synthetic data centered on key variables of high-precision manufacturing sites, such as industrial tolerances, reflections, occlusions, positional changes, and assembly deviations.

In accordance with the 'validation moved forward' concept presented in the white paper, we focused our technological capabilities on enabling the assessment and improvement of risk factors—which typically only become apparent after the physical production line is established—at the digital engineering stage.

Real2Sim2Real is treated as a continuous correction mechanism rather than a one-time data conversion.

It is a structure where field feedback is reflected back into digital assets and utilized for the next stage of model training and deployment verification.

Morgan Mao, Global CEO of Sky Intelligence, stated, “The purpose of synthetic data capabilities is not simply to generate more data, but to bring low-frequency but high-risk failure conditions into the digital engineering phase to form traceable engineering loops.”

Meanwhile, Sky Intelligence and ABB Robotics signed a Cooperation Framework Agreement last June, linking synthetic data generation pipelines and physical AI We are jointly pursuing technology verification.

This white paper will be available on the official ABB Robotics website.