00:10:22 · Dmitri Dolgov, Aditya Agarwal
[00:10:22] S0: Of a, you know, living room experience, like Yeah. You know, fully stretched out, which you can't quite do in the in the eye base. [00:10:28] S2: San Francisco as a city kinda goes through its ups and downs. And the city is obviously kind of had a hard time during COVID. It's doing better today. One But of the things I often wonder about is that it's pretty amazing that Waymo kind of first got its urban kind of like setup in San Francisco. So perhaps, like, tell me, was that a conscious... Like, is it just a question of proximity that you happen to be in San Francisco because, you know, the the offices are here? Or is there anything particular about being in this city? Or was it, like, easier to do in terms of regulation and so on? How come it was San Francisco first? [00:11:01] S0: I would say yes to all of the above Okay. Except maybe, you know, the part about regulation. But it was... And it's a it's a great market. Okay. So if you look at the, you know, commercial potential, it is massive. It's, you know, one of the top cities. It's true in The US and, yeah, actually worldwide. Number one. Number two, it's... There's a lot of complexity when it comes to the core test of driving. Sure. So, you know, we wanted to learn and improve the core capabilities of the Waymo Driver as fast as possible, and this was a great place [00:11:35] S2: to be. So being in this space would allow you to kind of feel confident. Right? [00:11:38] S0: Yeah. And you did mention that it is, you know, close to home, and that that is a very important [00:11:43] S2: Makes a
00:11:43 · Aditya Agarwal, Dmitri Dolgov
[00:11:43] S2: Makes a [00:11:43] S0: big difference. You want, like... Yeah. You want, you know, your team to use their product, right, to be excited about it, to experience any limitations of it. That that matters tremendously. [00:11:52] S2: Yeah. Maybe last question for this segment. I'm curious, Dmitri, as you are in a Waymo, are you constantly noticing all of the ways that it could... You know, it is either going right or perhaps not going as right as you hoped? Like, are you kind of constantly in a critic mode as you are kind of, like, riding these cars? Not on [00:12:10] S0: the I PACE anymore. Okay. Honestly, I tend to... I I I used to. I, you know [00:12:14] S2: So you're already living in the future, basically. [00:12:16] S0: Yeah. That's my more of a... When when I PACE, I just use it in my space. I work. I'm on a phone call. I I... They made me... I'm relaxing and, like, I would have, you know, my eyes closed and be, you know, taking a quick nap or meditating. In the all high, especially in the early days as we're bringing it up, everything... I I pay a lot of attention to every little detail from the driving behaviors to the, you know, driving comfort to any issues with the vehicle platform itself. I see. So that... Yeah. My mind naturally goes there on the new thing.
00:12:16 · Dmitri Dolgov, Aditya Agarwal
[00:12:16] S0: Yeah. That's my more of a... When when I PACE, I just use it in my space. I work. I'm on a phone call. I I... They made me... I'm relaxing and, like, I would have, you know, my eyes closed and be, you know, taking a quick nap or meditating. In the all high, especially in the early days as we're bringing it up, everything... I I pay a lot of attention to every little detail from the driving behaviors to the, you know, driving comfort to any issues with the vehicle platform itself. I see. So that... Yeah. My mind naturally goes there on the new thing. [00:12:47] S2: Yeah. Very cool. One of the things I have difficulty wrapping my head around when it comes to tackling self driving is what was the first thing that you kind of... When the team first came together, what is the first artifact that you were trying to produce? Because it feels so different than getting a web app up and running or getting your first mobile app. This is what we were all building in 2009. Let's get the first mobile app working and so on. So what was the first almost artifact?
00:13:20 · Dmitri Dolgov
[00:13:20] S0: So when we started in 2009, it was not about building a product. Our first order of business was just to understand the problem better. So in service of that, when we started, and honestly, the question at that time was, should we even try? Is the time appropriate, or should we just go back and come back in a decade or so? So in service of understanding the problem, when we started this project in 2009, we created a couple of milestones for ourselves with that goal of exploring the space. And one challenge that we created for ourselves was to drive 100,000 miles autonomously, which at the time was, you know, orders of magnitude more than anyone has, you know, ever done. And the second one, which actually turned out to be more difficult, was to drive 10 routes all over the Bay Area, and each one was a 100 miles long. And we had to do each one from beginning to end with no human intervention. There was a human behind the wheel, but the challenge was to complete one. And in doing that, you end up building every part of the system. Right? So just the crazy, amazing, magical days of an early startup, right? Calibrating and putting new hardware in the car, you're building perception, planning, controlling all of the tools and GUIs. Everybody is doing everything, but you're learning at a crazy rate. So that was, you know, first couple of years we succeeded, and then we said, okay. Now there's, you know, there's, you know, a there there, so let's think about what the product could be. And we kinda went back to building more of the fundamentals and laying the foundation for that what became eventually Waymo and the product you see today.
00:14:58 · Ruchi Sanghvi, Dmitri Dolgov
[00:14:58] S1: So if you were to explain Waymo, like how one would explain the networking stack or something that we all understand, like how would you explain the Waymo end to end system? [00:15:09] S0: So there's a couple of things. End to end has, you know, specific meaning in the field, I can, you know, speak to that. The way I would describe, you know, the, of course, you know, on the car, it's a system that goes, from sensors. We use three main sensing modalities, cameras, sliders, and radars. That information gets processed and kind of encoded into a more compressed representation. And then there's a generative part, which, you know, on the car is the, you know, the the part that makes decisions and generates driving behaviors. You know, that's the the decoder or the planning part. All of that, you know, the the the that's running in the car that we call the Waymo Driver, that's just one part of the system. Really, when we think about the technical challenge that we're solving, there's three main pillars. We call them the driver, the simulator, and the critic. And it starts with us with building kind of large, high capacity off board models. They have kind of three teachers for each one of those tasks. And then they get distilled into their students. And the driver acts, the simulator generates features, often in synthetic environment, and the critic judges and evaluates. Right? So then all of them kinda have to work together in order for this ecosystem to move forward.
00:15:09 · Dmitri Dolgov, Ruchi Sanghvi
[00:15:09] S0: So there's a couple of things. End to end has, you know, specific meaning in the field, I can, you know, speak to that. The way I would describe, you know, the, of course, you know, on the car, it's a system that goes, from sensors. We use three main sensing modalities, cameras, sliders, and radars. That information gets processed and kind of encoded into a more compressed representation. And then there's a generative part, which, you know, on the car is the, you know, the the part that makes decisions and generates driving behaviors. You know, that's the the decoder or the planning part. All of that, you know, the the the that's running in the car that we call the Waymo Driver, that's just one part of the system. Really, when we think about the technical challenge that we're solving, there's three main pillars. We call them the driver, the simulator, and the critic. And it starts with us with building kind of large, high capacity off board models. They have kind of three teachers for each one of those tasks. And then they get distilled into their students. And the driver acts, the simulator generates features, often in synthetic environment, and the critic judges and evaluates. Right? So then all of them kinda have to work together in order for this ecosystem to move forward. [00:16:37] S1: Do they all run on the car itself? Like, does the simulator need to run on the car, or can it run on the cloud?