엔비디아가 미국에서 17일부터 진행된 슈퍼컴퓨팅 2024 콘퍼런스에서 폭스콘(Foxconn)이 엔비디아 옴니버스를 통해 블랙웰 테스트와 생산성을 확대하고 있다고 21일 밝혔다.
Support for cost and energy reduction at Foxconn, Mexico, and Taiwan factories
Robot AI simulation with Omniverse → Deployment to Jetson-based robots
NVIDIA announced on the 21st at the Supercomputing 2024 conference held in the United States from the 17th that Foxconn is expanding Blackwell testing and productivity through NVIDIA Omniverse.
Foxconn, the world’s largest electronics manufacturer, is using NVIDIA Omniverse to meet demand from Blackwell, which is now fully operational. NVIDIA Omniverse, a platform for developing industrial AI simulation applications, is helping bring its manufacturing facilities in the U.S., Mexico, and Taiwan online as quickly as possible.
Foxconn uses NVIDIA Omniverse to virtually integrate its production facilities and equipment layouts, NVIDIA Isaac Sim for autonomous robot testing and simulation, and NVIDIA Metropolis for vision AI.
Omniverse helps industrial developers maximize efficiency by testing and optimizing on digital twins before deploying costly changes in real-world environments. Foxconn expects significant cost savings and annual reductions in strategic usage of more than 30 percent at its Mexico plants alone.
■ World's Largest Electronics Manufacturer's Omnibus and AI Utilization Plans Foxconn factory planners are using Omniverse and NVIDIA AI to build a physical AI-based robotic factory to meet demand.
Omniverse enables the creation of digital twins that virtually integrate production facility and equipment information from key industrial applications such as Siemens Teamcenter X and Autodesk Revit.
During the construction process, Foxconn teams use the Omniverse digital twin as a data source to communicate and verify the exact layout and placement of equipment. Omniverse’s virtual integration is said to save factory planners millions of dollars by reducing costly changes in real-world operations.
■ Omniverse Digital Twin, Implementing Robotics for Manufacturing Once the factory’s digital twin is built, the factory becomes a virtual training ground for Foxconn’s autonomous robots, including industrial manipulators and autonomous robots. Foxconn’s robot developers simulate, test, and validate AI robot models on NVIDIA Isaac Sim before deploying them on actual robots.
Foxconn can use Omniverse to simulate robot AI and then deploy it on NVIDIA Jetson-based autonomous robots. On the assembly line, the Isaac Manipulator library and AI models can be used to simulate automatic optical inspection, object identification, defect detection, and path planning.
Additionally, Omniverse enables production facility planners to test and optimize intelligent camera deployments before deploying them in real-world environments. This will help us understand the factory floor, support worker safety, and lay the foundation for a visual AI agent framework.
■ Foxconn plans to replicate precision lines worldwide Foxconn reportedly plans to use NVIDIA Omniverse and AI to replicate its precision production lines around the world, with the aim of rapidly deploying high-quality production equipment that meets unified standards to gain a competitive edge in the market.
Foxconn's Mexico plant is expected to begin production early next year, while its Taiwan plant is expected to begin production in December.