BMW uses the platform for data sharing, wormholing, virtual robot training, and even robot fleet management.

When NVIDIA announced its Omniverse Enterprise platform last month, it was immediately clear that the technology’s real-world potential is huge.

Here’s the TL;DR: Omniverse Enterprise enables teams spread across the globe to create and share photorealistic 3D worlds that obey the laws of physics. It also acts as a connector between the various 3D software packages used on any project, making it easy for designers, architects, engineers, and other creators to collaborate on the same 3D world—and do it in real time.

During the initial showcase, NVIDIA touted BMW as one of its biggest Omniverse users. But they were short on details to explain exactly how the carmaker is using the tool in its 31 factories worldwide.

Good news for us, because BMW has shared some specific ways they use Omniverse Enterprise for efficiency gains, including a massive 30% gain in planning efficiency. Here are the highlights.

Data sharing

BMW uses the platform to share data among team members. Omniverse combines point cloud data, models from Autodesk Revit, and Dassault’s CATIA.

How does it improve efficiency?
BMW gains back time by eliminating problems related to data transfer, file management, and tedious conversions between different formats.


Factory design is a virtual process. To verify the design, a company needs to perform hands-on testing.

BMW does this testing by having an employee put on a motion-capture suit, and having them “wormhole in” to part of the virtual factory to test it out. As they perform a task, a second worker makes adjustments to the set up in real-time. In short, they tailor the setup to the human worker in order to ensure good ergonomics and optimize the process.

How does it improve efficiency?
This process enables virtual testing, so no physical building is required. It also enables remote testing for employees anywhere in the world. This, in turn, greatly increases the speed of iteration by tightening up the feedback loop between design and verification.

Training robots

BMW is also making extensive use of NVIDIA Omniverse’s Isaac robotics platform to generate synthetic training data for autonomous robot assistants.

Isaac uses synthetic data generation and domain randomization to create “an infinite permutation of photorealistic objects, textures, orientations, and lighting conditions.” BMW feeds these simulations to their robots to train them for detection, segmentation, and depth perception.

How does it improve efficiency?
This minimizes the need for real-world robot training and speeds up the process of robotic bootstrapping. This, in turn, means robots can be deployed faster, and start moving materials around the factory to assist human workers.

Fleet management

The Omniverse includes NVIDIA Fleet Command, a toolset for humans to monitor and control robot operations. The worker can assign missions and track progress on a photorealistic digital twin of the real-world factory. If the robot needs help, the operator can connect, solve the problem, and then return the robot to autonomous control—all without leaving their station.

How does this improve efficiency?
The tool makes it easier to scale autonomous robots, while offering an easy mechanism for addressing autonomy problems.

For more about how BMW uses Omniverse, click through here.