Korea: NVIDIA Accelerates Physical AI in Industry

Key Takeaways

1Samsung announces simulation accelerations of up to approximately 1,600 times and chemical calculations up to 5 times faster with NVIDIA tools

2Kakao Mobility plans a Level 4 autonomous driving system, relying on Hyperion, DRIVE AGX Thor, and Halos, starting with Kia vehicles

3LG and NVIDIA have factory-validated robotic tasks with the CLOiD robot and a reusable data and simulation loop

Why it matters : These collaborations illustrate the transition of physical AI from the lab to production across various Korean industrial sectors.

At Samsung, chemical calculations and engineering simulations are accelerating significantly with NVIDIA's libraries and platforms. Kakao Mobility is preparing a Level 4 autonomous driving system using Hyperion, DRIVE AGX Thor, and Halos. LG Electronics and NVIDIA have demonstrated robotic tasks on the CLOiD robot in a factory setting, utilizing a data and simulation loop designed for repetition.

Samsung Measures Massive Accelerations in Simulation and CAD

According to NVIDIA, an example of AI homogenization at Samsung has reduced simulation time from about 18 hours on CPU to 41 seconds on GPU with a coarse grid, resulting in an acceleration factor of approximately 1,600. In the field of materials research, Samsung's sMOL platform has achieved certain quantum chemical calculations up to five times faster end-to-end thanks to cuEST. Samsung is also developing thermal and mechanical analysis chains at the chip scale using PhysicsNeMo. Upstream, the company is working with cuISS and cuDSS to accelerate sparse linear systems and is expanding GPU usage to process and optical simulations. Finally, Samsung is exploring cuEquivariance to speed up tensor operations used in training and inference of atomic AI models.

Kakao Mobility Prepares Level 4 with Hyperion, Thor, and Halos

Kakao Mobility plans to develop a Level 4 autonomous driving system leveraging NVIDIA's computing and safety technologies. The project combines Kakao Mobility's proprietary driving software and its operational experience in Gangnam, Seoul, with NVIDIA's vehicle computing platform. The roadmap includes utilizing the NVIDIA Hyperion reference platform, equipped with DRIVE AGX Thor, as well as NVIDIA Halos to ensure safety during development and validation phases. The initial development will focus on vehicles specifically designed by Kia, which will incorporate Kakao Mobility's software and NVIDIA technologies, aiming to enable the emergence of future Level 4 mobility services in Korea.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

LG and NVIDIA Validate Industrial Gestures on the CLOiD Robot

NVIDIA and LG Electronics have demonstrated manipulation and robotic manufacturing tasks on the CLOiD robot in a factory environment, relying on NVIDIA Isaac GR00T. The partners established a data loop connecting real-world collection, augmentation via NVIDIA Cosmos, training of GR00T models, embedded computing with Jetson AGX Thor, and evaluation through simulation, to create reusable development and validation flows. Teams are using this setup to refine a repeatable process for training, evaluation, and improvement of robots intended for production.

Agent Tools and a Rising Korean Ecosystem

NVIDIA specifies that its Agent Toolkit would be capable of linking simulation tools to AI models to assist tomorrow's engineering agents in performing analyses and comparing different design options. The extensive collaboration with Samsung encompasses materials research, process development, and device engineering. More broadly, Korean industries are implementing accelerated computing, artificial intelligence, and simulation in semiconductor manufacturing, robotics, and autonomous vehicles, aiming to transition physical AI from the development stage to production.