Free your mind: Digital twins aren’t just for buildings

What if a tech company presented a virtual keynote event where the on-stage CEO was digital and nobody knew it? That’s exactly what happened this past April and we just found out about it. At Nvidia’s GTC event, CEO Jensen Huang was represented by a digital twin, created with Nvidia’s own Omniverse technology. 

More importantly, as it relates to the IoT, however, was how this technology can be used. We’ve gone from creating virtual “people” to digital buildings and even roadways. And as this technology continues to improve, it will be utilized to virtualize practically anything.

Nvidia spilled the beans on its technical achievement in a video last week. Frankly, as a student of Computer Science, I find it fascinating. (Sorry to borrow from Mr. Spock, but it’s true.)

I highly recommend watching the full video because it shows how far we’ve come in this space in such a short time. And it illustrates the huge cost savings that can be achieved when it comes to creating digital twins.

I don’t mean the virtual CEO example, of course. I’m sure Jensen is still well compensated when it’s him or a digital replica of him at a keynote event. Take a different use case such as the BMW production factory Nvidia created using its Ominverse solution.

To build this production line in the real world would cost hundreds of millions of dollars before a single vehicle could be built. There would certainly be construction challenges and mistakes to be rectified, adding to the cost and effort. Inefficient processes might be found but not easily addressed due to physical plant design constraints. And BMW wouldn’t know the exact results of its vehicle production until the line was used to create test cars.

Here’s where the digital twin concept comes out way ahead. By working with Nvidia, BMW was able to create a virtual production line modeled upon the physical plans. And it’s a true working model, complete with all of the robotics used to build cars, all of the parts and materials. There are even virtual people at various stations.

It’s worth noting that companies other than Nvidia are seeing the light when it comes to providing digital twins for buildings and enterprises. Protocol noted earlier this week that Unity, creatory of a leading 3D game engine, sees demand for these too.

Digital twins of physical structures aren’t that new, of course. So when Nvidia showed off its virtual highway it really piqued my attention. The company created a digital twin of an actual driving area in Germany. Why? Because it’s safer to train the AI of self-driving cars in a real-life simulation than it is on real roads. 

Think about that: The vast majority of learning used by the next generation of vehicles may take place in a sort of virtual classroom that mimics actual physical spaces. Sure, the self-driving AI models may not learn 100% of what they need to know on digital roads. But if you can cut down the learning model time before a car even hits the road, that’s a big time savings. And it’s much safer too. Stacey wrote about that last year. 

The concept is sound. By attaching physical sensors such as cameras, radar, accelerometers and the like to a real self-driving system, you can provide inputs from a digital twin of the highway. It’s a bridge between the physical thing, a car in this case, and the simulated digital roads, which are replicated to exacting specifications of real locations. 

Aside from being fascinated by the technology behind Nvidia’s Ominverse, this presentation opened my eyes to a world of digital possibilities that I never realized. 

Maybe one day I can try out digital tires on my car, driving virtual twins of the roads around my house to pick the ones with the best performance or most comfort. Perhaps I can test different robot vacuums in a digital twin of my house and based on the specifications of each can see which does the best job at cleaning my floors. Or maybe I’ll outfit a digital twin of my home with virtual solar panels and use real weather input to see how much energy I can create with various configurations and models of those panels.

I’m just taking a stab at what might be soon possible here, of course. But one thing I do know is that digital twins aren’t just for buildings anymore.