Nvidia is accelerating its efforts in the Korean physical artificial intelligence (AI) market by expanding collaborations with domestic companies such as Samsung Electronics, LG Electronics, and Kakao Mobility. The focus is on applying Nvidia's accelerated computing and AI simulation technologies across semiconductor design and manufacturing processes, as well as in autonomous driving and robotics development.
On November 9, Nvidia unveiled the results and development plans of its ongoing physical AI collaborations with major domestic companies. Physical AI refers to technology that enables AI to understand and predict physical environments, linking this understanding to the actual operations of robots and autonomous vehicles.
Samsung Electronics is utilizing Nvidia's GPU acceleration technology in semiconductor material research, process development, and device design. Notably, in a case study applying Nvidia's AI physical simulation platform, PhysicsNeMo, chip thermal and mechanical analysis was reduced from approximately 18 hours using a 32-core CPU to just 41 seconds. However, this result is specific to a case that also incorporated AI-based material property predictions and computational grid simplifications.
Kakao Mobility is advancing the development of a Level 4 autonomous driving system using Nvidia's autonomous driving platform, Hyperion, and safety verification technology, Halos. The company plans to integrate its autonomous driving software with Kia's purpose-built vehicles (PBVs) and leverage operational experience accumulated in the Gangnam area of Seoul.
This collaboration is significant as it demonstrates the expanding application of Nvidia's AI technology beyond data centers into the design and verification processes of domestic manufacturing and the development of autonomous driving systems.
LG Electronics is enhancing robot learning and manufacturing environment simulation using Nvidia's robot AI model, Isaac GR00T. The two companies have demonstrated robot operation and tasks in real manufacturing environments using LG's CLOiD robots.
They have also established a development system that connects actual data collection, virtual data generation, AI model training, and performance verification. Utilizing Nvidia's Cosmos and Jetson AGX Orin, they plan to support robots in repeatedly learning various manufacturing tasks and improving performance.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.