Coming Out of Stealth to Drive AVs Forward

Yashad Kulkarni10min

Unknown Speaker

00:03

All right. Well,

Speaker 1

00:05

I'm, I'm very excited to welcome our next guest. This is Raquel artisan. She is the founder and CEO of a brand new company that just came out of stealth yesterday called Wahby. That's taking an AI first approach to solving self driving vehicle technology. quick background on Raquel, she is a full professor is kind of considered an AI pioneer, but also was leading the r&d efforts as a chief scientist over at Uber ATG. So and there's there's a lot of other things that she's been doing. But um, rock. Hello, thanks so much for joining us today. Thanks, good to be here today with you. Well, let's dig right into to this company. Um, and actually, to get the timeline down, you Uber ATG was acquired by Aurora. And that happened in December was announced in December, and now we are here like, you know, six months later, and you have this new startup? Did you start working on it? When did you start working on this company? And tell us a little bit about what you're actually trying to do here?

Speaker 2

01:16

Yeah, so I guess, you know, I left, you know, we're never going to be there were three months ago. And, you know, to start this new company, where we, with the idea of having, you know, different way of solving inventors or songwriting. And, you know, the, basically, this is a combination of, you know, my 20 years career in AI as well as more than 10 years in self driving. And, you know, thinking about using a new company, something that was always in my head. And the more that, you know, I was in the industry, the more that I started thinking that, you know, going away from the traditional approach and trying to have a diverse different view of sources, it was actually the way to go. So that's why I decided to do this company.

Speaker 1

02:09

Well, so you've been working on AI for a very long time, and then working to apply it towards self driving vehicles for some time as well. Why do you think not as Uber, but the rest of the industry hasn't white? unlocked it, we see really small commercial scale, way, Mo is one example. But it's pretty limited on it is growing. But why haven't we seen sort of bigger commercial scale deployments of, you know, Robo taxis, for example? What's the limiting factor there?

Speaker 2

02:43

Yeah, so we, we've seen that you're seeing meaningful progress right over, you know, the last 17 years, right, since the DARPA challenge. But as you, as you mentioned, there's the commercial deployment is, you know, very simple, and it's manipulation domains. And, you know, one of the reasons for this, and I think it's the utilization of what they call a traditional approach, that has not really utilize the full power of AI, in order to, you know, to server such a difficult task, such as driving, and, you know, realizing complex, you know, manual tuning. So this makes, you know, scaling of this technology, and particularly finding the long tails of the scenarios and things that might happen when you drive particularly difficult.

Speaker 1

03:33

I want to talk a little bit about the approach we're using, because it's a bit of a combination, and one of the one of what you use is deep nets, which you can, it's a little bit complicated, but maybe you can explain that briefly. But for me, my understanding is that deep nets, typically are used in a limited way. And then rules based AI is is generally how the approaches for self driving vehicle development. Um, and the problem is, is because using deep neural nets, there's the black box effect, which is basically not being able to verify or validate why the system did what it did. So, how are you solving for that?

Processing audio...