SK Telecom is introducing “physical AI” that combines digital twins and robot training to ease difficulties in applying artificial intelligence at manufacturing sites due to their complexity and the lack of validation environments.

On May 11, SKT said through its newsroom, “We implement real processes in a virtual environment and apply the results of verifying various scenarios to actual sites.”

SKT’s digital twin platform is built on technology based on Nvidia’s Omniverse. Nvidia Omniverse is a platform that supports simulations in a 3D virtual space that precisely recreates real environments.

Based on this, SKT develops it into a form that can be applied to actual manufacturing processes by combining functions required by industry. It reflects data generated from different equipment and systems into a single virtual environment and verifies operating scenarios before process changes or equipment adoption through “What-if simulations” that compare results while changing various conditions.

The robot training platform trains robots based on the virtual environment of the digital twin platform. It repeatedly verifies and learns various working conditions and situations in the virtual environment. SKT stressed, “This allows us to secure robots’ response capabilities and stability before applying them at actual sites.”

SKT’s digital twin platform is being applied and expanded mainly for manufacturing customers. The robot training platform is preparing for full-scale expansion after a pilot stage.

Baek Seung-yoon (백승윤), head of SKT’s Data Intelligence Team, said, “Physical AI is key infrastructure for shifting manufacturing site operations to an AI-centered approach.” He added, “We will expand applications to various manufacturing industries such as semiconductors, automobiles and shipbuilding, and focus on creating practical value for AI at manufacturing sites.”