Jaden Schaefer

So the Wabi driver is essentially trained and validated inside of what is called Wabi World.

This is a closed loop simulation environment that automatically builds digital twins of real environments.

So it's, you know, simulating all of the sensors in real time.

It's generating edge case scenarios.

It's allowing the system to learn from its mistakes without human intervention.

So, you know, they kind of have this whole simulated world they've built and the AI is learning to drive inside of it.

But the simulated world is, you know, very similar to the real world, which we can do these types of things when we have, you know, technology like Google Maps and Street View and all this other stuff.

And the technology behind it, really, because I don't think Google's licensing that data, but you could go drive a car around and get real information and then use that to replicate these kind of digital worlds that it's used to simulate with.

So according to Yuritsan, this like approach that they're taking is going to make their system and it's going to allow it to reason about its surroundings more like a human driver and generalize from fewer examples than traditional AV systems.

Wabi has spent the past four

and a half years developing that technology.

And they've done it for highways and surface street tracking.

And it was 35 minutes.

And I was like, what the heck?

Why is it taking 35 minutes?

And it was because it wouldn't go on the freeway.

It was going on back roads.

And that's kind of where Waymo was at, at the time.

And maybe these things are changing.

Um,