Welcome to NVIDIA's AI podcast, where we explore the expanding world of artificial intelligence with the people who are developing and applying AI to the problems and opportunities that cut across our lives.
And of course that means racing full size, all electric driverless cars as fast as they possibly can go.
Joining us on this segment is Justin Cook from RoboRace. a London-based company that is developing the technology and the race series that will pit one autonomous electric car against another.
Justin, welcome. Hello, how are you? You're in London at the moment, although I know this race series takes you all over the world, so welcome.
Thank you very much. So just to give you the background, RoboRace is... the first ever driverless electric racing championship. really being a platform where the most progressive companies in the world could contribute in an open AI platform.
And I think we feel. that the subject of driverless is very topical right now, but what we see is that the progress is quite slow and our real kind of main objective is to create an environment where we can accelerate those learnings.
We know that five, six, seven, eight of the world's biggest companies are all developing it because we're told but we don't know at what stage they're at i mean we do you don't actually you probably do the two of us are probably the only two that do Are you a gear head?
Are you a computer head? Are you all of the above?
How did you get involved in RoboRacer? How did that come to be?
Was it your background? Was it your interest?
Was it you lost your driver's license? I was born in 1981, and it was the ascension of Michael Jordan...
Mario Kart and motorsport in its like prime.
So I'm all of those problems all rolled into one.
And I've always been obsessed with technology.
So when we were kids, we had the first TV that had a VCR in one, so all in one. and we were always like, we knew there was better technology out there, but we couldn't get our hands on it, so we were always like,
When can we get the Atari? When can we get it?
I heard if you order it from Japan, you can get it.
So it's always been in my blood. It's something I love.
I work with a group of amazing people. Dennis is like an incredible founder and everyone who works in this company.
We don't really hire anyone that doesn't believe in the vision and where we're going, and I think that's the only way to build a successful company.
For the people at the company, give me the ratio of gear head to silicon head.
We're pretty good actually. So I think what we're really good at is we have a balance of people who've worked in motorsport.
So we have like a guy who works on the Rover race team who used to work with Ayrton Senna, which I just find incredible.
And he's got these unbelievable stories. about adjusting certain things in the brakes for Ayrton and knowing all these incredible details.
But enough that they question what they're doing today to know that there's room for disruption.
And then enough guys who don't know what the hell they're talking about that come from mobile phone software who are like, Why don't we just do it like this?
And everyone's like, uh, because no one does it that way.
And they're like, well, that's probably the reason why it's the most interesting point of what we're seeing.
And so I think with any disruption you create it by approaching it with a different mindset. and respect for the past.
If you try and break any of those two principles, you normally get someone who sort of You get fake disruption, I think.
So you get people going like, oh, this is the way to do it.
But you have to know the rules so you can break them, basically.
I think that's really the secret. Well, let me back up a little bit.
So RoboRace, the race series, like I drive an all-electric Fiat, and I refer to it fondly as the poor man's Tesla.
This is not what we're talking about. We're not talking about my car racing Teslas and Nissan Leafs and, you know, let's throw in the Chevy Bolt around a track or a course.
Yeah, so the car itself was designed by Daniel Simon, who worked on Tron and Oblivion.
So of course, it looks like some kind of far out incredible spaceship.
And really what we're seeing at the moment is, so we're in the development stage pre to us.
So we're launching the mobile world Congress, which is a scoop for you guys in Barcelona.
So that's coming up towards the end of March.
Monday, 27th of February. and we are planning to unveil the car and then after that we are going to take it and show people for the first time on a track somewhere.
And in the meantime, what we're using is a car which we call DevBot, which is basically a kind of down, dirty raw exciting you know kind of you know blacked out looking robot it's the it has a driver cabin so the race car will not have a cabin it's just completely flat looks like a rocket This car DevBot has a cabin and the reason why it has a cabin is we're able to take the learnings of a driver and put them in there and we're allowed to program the algorithms and so we get to see the comparison.
Right now what we have is the computer telling you like the tires were burning down too much on the right hand side or it wasn't able to take that racing line. but the computer can't tell you how it feels.
And so there's this magical moment where you have this overlap of the human and the machine working together to further technology. how does the computer feel like?
What is it saying when it's telling you, here's how I feel like, I'm scared.
Yeah, no, it can only give you data. It can say to you, like, you know, try to take that vector 19192, like, you know, what's the reason it failed and the guys who are smart enough to know are going to evaluate the data and say okay right it's because it dragged like this much or it was, you know, it needs to come down by two miles per hour as it exits that corner and they'll work it out.
But the computer's not going to put its hand up and say, you know what, it absolutely grated the back right-hand tire when you went around that corner because you were pushing me at speed. a driver would never take.
And so this is the part where it's really interesting is the computer starts taking a racing line that a human would never take.
But just like if you've seen the way DeepMind do the Pac-Man, It does a movement some way significantly before the puck's coming down a bit like Wayne Gretzky.
It's going to where it knows it's going to end up.
It's all about prediction. So that's something I wanted to get into, that if you have a human and a computer and a robot essentially driving this car, how are the driving styles structured? starting to show how they're different how are they emerging as like we are two different things or maybe like they're similar in this way and different than others Yeah, I mean, we're seeing a lot of interesting patterns.
We're seeing very, very many different things.
So actually, another scoop for you is in Buenos Aires, we're going to Argentina with two DevBots. on the track at the same time.
They'll be going plus 150 kilometers an hour.
All the stuff that has been shown publicly has been like 11, 20, 30 miles per hour. with the exception of Audi's Bobby.
We're pretty proud of that. What we're going to see is really I guess we're going to try to show as much as possible of what we call showing the invisible, so what the car sees and hears and understands as it's making decisions.
But there's no rhyme or reason because everything that we know is based on things that we think are correct, right?
So a human does something because it's always been done that way. and now and again someone comes along and betters it, but the computer is going to look at every single potential aspect and work out a path that has never been done before, and that is the root of all scientific development.
And we'll start to see how that changes.
And I think what's crazy and what's interesting is that how, you know, you mentioned Pac-Man or in the Go tournament. humans learned how to play Go better.
There's the potential, it seems like, where we can learn how to drive better at the end of the day by learning from robots.
Absolutely. I think we're at a really fascinating point.
We work with the guys at DeepMind, we know them well.
One of the things which is fascinating for me about what Mustafa and his team are doing is that when they won the GO Championship, when they were asking the Chinese player what had happened, He said that he had changed his behavior because the computer was playing in a way that he had never, ever been taught to play.
In a race series, are we gonna see robot cars racing with different styles?
Are we gonna be able to apply? Is it, you know, because there's always the sort of movie trope of the French driver versus the Australian versus the Spanish driver.
Look, I think there's some really interesting stuff here.
One of them is that the racing formats are not going to be what you know.
So think more like Wipeout, F-Zero. The inspiration has come very much I think drone racing is doing some great stuff.
I think the inspiration has come much more from Of course, our start point is entertainment.
Our primary in terms of the cars is safety because obviously we want this to go to the roads and want people to understand it.
It's much more useful if I put the car in an environment where it has to break down from 200 to zero, and if it crosses over a line that's marked on the floor or a laser line that's cut across, they've killed a person.
So if there's a robotic dog that they have to avoid, A, it's a supreme learning for the road, and b it's much more entertaining than 10 cars going around in a loop because i think i think this is not the case because we tested there is an assumption that if 10 cars all have the same car, which is, just to clarify, everyone has the same hardware, so it's a true championship of software, a real intelligence battle. if everyone has the same car, can the front car ever be beaten?
We've seen that it can, is the truth, because basically you've got to come up with a racing line that's never been done before and you've got to push the absolute limit at speed where perhaps the other car would break down.
But I think that the real the real magic is going to come from, you know, all the best scenes in James Bond or when the cars are racing through traffic.
Right. It's not about two cars like racing on an open road, like out burning each other.
So I think what we're looking at is how we incorporate obstacles, how we incorporate real life behavior.
Again, trying to benefit cities, maybe there'll be some electric trucks in there on the track. you know we're working on lots of different scenarios but whatever's coming, it's going to be super exciting, I can guarantee you that.
You've no doubt ridden in the development vehicle.
I mean, what's it like? Is it... Like any other car?
No, it's not like any other car. The noise is extraordinary.
The feeling is extraordinary. The inability for me as a human being to not want to take control is the most overwhelming thing.
We put Lucas Degrassi in there who drives for Audi and Le Mans.
He's like a world leading driver. His comment when he got out of the car was, it was literally like, whoa, you know, like, excuse my language, you know, like, he was like, he's like, you know, I drive 200,
80 miles per hour. What's absolutely crazy is you're trusting the car to be in total control.
I'm not scared of that speed. but all of a sudden someone else is controlling me at that speed.
And my instinct is like, I want to, like, I want to, you know, like we're control freaks.
We're human beings. Like that's what we do.
And he was like, you really feel the speed when a car's going 275.
You know, you're a passenger. And yet, I think probably not too far into it, maybe you don't relax because this is a different thing.
But I mean, with autonomous cars, I certainly didn't, but I think the professional racing.
There's a big difference between a road car.
I've been in a Tesla at like 50, 60 miles an hour, and I felt like I was in cruise control.
I was delighted. When you're going 200 miles per hour around a corner, on a racing line, which is basically like, you know, the sharpest possible angle because it's trying to beat the ultimate, you know, lap time, you know, that's probably, that's probably slightly different.
Formula One, in particular, at least the automotive folks claim that they're helping develop the cars for all of us, the consumers.
I think ABS, things like that came out of racing.
But this sounds like a much more broad application of this racing.
It's not just to go fast, it's to go fast. to run a car far more safely and to do the kinds of things that you described.
Look, I don't want to say anything bad about Formula One.
Some people are saying that as a format that it's become monotonous and obviously they've changed ownership and all that kind of stuff.
At the end of the day, they are responsible for a huge amount of innovation in vehicles.
What I would say is that I think that people are looking for something different.
I think that this generation that's coming through is certainly looking to a greener future, to an electric future.
And so I think that we're not just driverless, we're electric, which I believe is very important.
In the case of a partner like Michelin, Michelin had come to us and saying, we want to develop our autonomous tire and use Roborace as a platform to test that.
We're not asking them to make a tire just for us because that tire has no use to anyone.
What they're doing is they're improving their road tire technology that's going to be rolled out to consumer. valuable platform.
What's not useful is when a team goes and spends 400 million in a wind tunnel and then basically that goes nowhere and the team goes bust two or three years later, which is what we're trying to avoid.
That's really the main reason why we've gone after, I would say, a solid hardware. so people are not spending time tampering or changing stuff because no one's going to have that car, but they are all going to have software.
There's Waymo, there's Google, there's an NVIDIA.
You know, there's Apple. This is the true battle is software.
So we're making a competitive platform for people to start to understand that and see that.
From your vantage point and doing the development that you guys have done already and what has yet to be done, what's really hard about this?
The racetrack is a controlled environment, so that'll happen maybe sooner rather than later, but how soon do you see these vehicles Not the super cool looking Tron one, but autonomous vehicles on the road.
So my personal opinion is this, is that they only work when the total environment is all autonomous.
And we're quite a way off of that because you're not just going to get rid of all of the cars.
So I think you're going to see zoned areas or you're going to see progressive cities like Sweden or Mazda in Abu Dhabi, North City, like various other cities around the world. where you have full control, full autonomy, full robotic cars, and then they work like a dream.
And I think once people start to see the benefits of that, no traffic, no accidents, zero emissions, it's going to become inevitable.
And I think that's basically the model that we have to move towards.
Where we're going to have problems is when you've got like the Tesla guy who's been told, you know, poor Tesla, I feel so sorry for them.
You know, the guy sitting there reading his iPad, it says clearly in the instructions, also partly you need to keep your wheels on. hand on the wheel the whole time and he'd keep the feet on the pedals and he goes and reads an iPad and he gets killed by a combine harvester.
I mean, you know, and everyone says, oh, it has a tele...
What do you want to do? It's got instructions.
If people don't follow the rules, then you have problems, and that's why a control end-to-end environment is the optimum because the computer It's margin for errors like 0.9999, God knows what.
Humans are responsible for 97 percent of accidents on our roads every single day.
Imagine the number of lives that would be saved.
Imagine the number of minor incidents that wouldn't happen.
Imagine what that means for the insurance companies, for commerce, traffic control by digital mapping.
It's an amazing possibility and we have to get there and we need to get there as soon as possible and that's really what we're trying to do.
You say that this will change cities. How do you imagine that it changes?
I mean, you talk about less traffic, et cetera, but what are the kind of broader changes about how cities might be designed or... redeveloped for that matter, et cetera.
Yeah, so look, I think, you know, I was having a really interesting conversation with some guys who've been working on Singapore and Saudi Arabia recently.
And you know, the truth is, It's challenging to just rip up all the cities, but my personal view is I imagine that if you think about all your main roads, you'll keep your main roads, but all your side streets, there's no reason why they can't be gardens.
There's no reason why cars are not in banks of 20 cars like the bicycle schemes they have in cities. and you take one when you need one.
I think the life cycle of a car is something like 4% of the time it's in use and the other 96 it's in your driveway or outside your house parked.
And there's definitely a value in that. You only need to look at the way Uber's changed our behavior in terms of using taxis and riding when you need them.
The physical need to own a car is definitely disappearing.
So I think decluttering of cities is something that I think will happen naturally, which is a beautiful thing.
We are trying to create a place where people go, whoa, that's the benchmark for autonomous.
Let's all do that. And in doing so... It's almost like it's a selfless mission.
It will drive everyone forward. It will accelerate the entire. entire opportunity and the goal.
And that's Dennis's vision. And we're all living by that.
Justin Cook, RoboRace. Thank you so much.
Take care, man. PostScript here. If you're listening to this podcast on iTunes, Google Play Music, or whatever your favorite podcast player is, do us a favor and leave us a review.
Thanks for listening.
