NC AI said on May 28 it formed a consortium with Hyundai Rotem and was selected as the final project operator for a national R&D project ordered by the Agency for Defense Development (ADD) on an integrated simulator and modular robot system based on Physical artificial intelligence (AI).

The project is a large government initiative to integrate control of multiple types of unmanned robots and build a digital twin-based simulator and modular robot hardware. NC AI will lead overall development of a “world model,” a core technology needed to implement a next-generation robot foundation model (RFM).

A world model is a Physical AI technology that lets robots simulate physical laws and environmental changes in the real world and convert them into training data. It is seen as a way to address the “Sim-to-Real Gap,” in which robots trained in virtual environments malfunction in real settings.

NC AI in March unveiled world foundation model (WFM) technology that delivers comparable performance to global state-of-the-art models using only 25 percent of the graphics processing unit (GPU) resources. Instead of the existing approach of generating video and then reasoning with a vision-language model (VLM), it applied an architecture that directly derives robot actions in “latent space,” the stage just before video generation. The company explained it recorded a comparable success rate to Nvidia’s robot AI model “Cosmos” based on 18 core tasks.

NC AI CEO Yeon-su Lee (이연수) said, “We have joined a national project that will be responsible for national security together with Hyundai Rotem.” Lee said, “Based on a lightweight world model, we will develop a Physical AI simulator that connects the virtual and the real and take the lead in accelerating the defense AI transformation (AX).”