Nvidia Is Pushing the Limits in Autonomous Driving

Autonomous driving is an engineering feat even with respect to small cars. But is the tech safe enough to scale up to trailer trucks?

In this video from "The Virtual Opportunities Show" recorded on Jan. 4, Fool analyst Asit Sharma and Fool contributors Demitri Kalogeropoulos and Jose Najarro discuss a few challenges involved in the new partnership between Nvidia (NVDA -3.87%) and TuSimple (TSP) aimed at producing self-driving freight trucks.

Jose Najarro: TuSimple. For those not familiar with TuSimple, they are a publicly traded company and they focus in creating autonomous self-driving trucks, which I think is pretty impressive. Trucks are such a huge machine, that you need to have so many sensors, because obviously a car can really cause a lot of damage, but a truck, self-driving, if that goes circuit can definitely cause some insane damage. They partnered up with Nvidia, chips to work on autonomous computing, to work with their autonomous driving. For those not familiar with Nvidia, they are semiconductor company that makes chips or processors and accelerators. The reason you need these accelerators is because when you're doing autonomous driving, you really need to be able to compute all this information, all these sensors that you're seeing at real time. Hey, there's a car in the left for me, hey, there's a car in the right of me, hey, I'm going x amount of speed in this time, speed zone. I need to be able to read all this information in a fast matter, and to be able to do that, you need to have a high performance product, and Nvidia is working with that. I think it's pretty interesting that, these big trucks. I've done a few research of TuSimple. They've done testing in the past. I believe they are in the Southern part of the United States.

I believe it's around Texas, Arizona. I forget what actual states they're working in, but they have already done a few self-driving trucks, where they go from one warehouse to another. At the moment when they do that, there's obviously someone in the truck to make sure to takeover if something bad happens. But they've done fully autonomous to some extent trips within certain warehouses, which I think is pretty impressive. For them to partner up with one of the leaders in this market, I believe that's good news for both TuSimple, for Nvidia, where it shows that, hey our hardware is definitely needed. But just overall for the market, it shows that autonomous driving, even though, I want to say two years ago, we probably thought this was something we weren't going to see within the next 10 years. I feel as we continue to see more articles like this, this is something that we might even see as soon as maybe five years from now. That's the topic I wanted to talk a little bit about.

Asit Sharma: Jose, something that I've been curious about in the autonomous driving industry, are you familiar at all with the role that insurance has to play in this industry? It's something that from time-to-time keeps coming into my mind as a barrier. If the technology is slight nascent and maybe we will see the first bona fide fleets of autonomous commercial vehicles maybe in the next five years, as you say, what does the insurance industry need to do to make sure that there's some acceptable loss ratio for them? Though it seems like an expensive thing to ensure at the beginning of the industry. Demitri also, if you have any thoughts on this or have come across any articles that would be interesting?

Najarro: I haven't come across any articles about it, and that's a great question, something I'd never thought about. But the first thing that comes to mind is, with anything like this, the most important thing is data. I feel like a lot of insurance companies have already been collecting that data. I feel like insurance companies that hey, if we're able to track your driving movements, if you don't break too often, if you don't do so, we give you better rates. I don't know which insurance company has been pretty sure. I've seen in some insurance companies do that. I feel like some of these insurance companies are going to start working with some of these autonomous vehicle companies and say, "Hey, we're going to need to put our sensors or whatever into your vehicle, so we can get that information, and now with that information at hand, we'll be able to price accordingly this type of how much we would charge you for your premium and for whatever goes from there.

Sharma: That makes a lot of sense. Then I followed some insurtechs that are using this technology telemetry to adjust that, gather the data and then use it to adjust how they're pricing their insurance, what the term should be, what again, an acceptable loss ratios for them. Any thoughts? Demitri?

Kalogeropoulos: Yeah, that's exciting. I can't wait for autonomous driving to be all over the place. I spend a lot of time on the roads, I bike on the roads and people they're like, aren't you terrified of a machine, like driving round you in that situation. But I prefer to have little machines involved than just regular drivers that I think. What the other interesting thing about this, when I first was getting into this industry, I just assumed that the way this would work is that we'd have sensors in other cars and they'd be communicating with each other or on the road, we'd have the smart systems that had to communicate with each other, but it looks like obviously that's not the way it's happening. We're creating these super processing, all inclusive little robot machines that can just handle all this processing on their own in real-time, and that's amazing. I don't think that was possible, but that could be way I dropped out of engineering school and became a business major [laughs] because I would witnessing that happen.

Najarro: This is interesting.

Sharma: Go ahead.

Najarro: I was going to say I definitely think that's where it would go in the future. But I think to start off, it's weird to start off, you need to have this big bulk because not all the cars out there are going to have the sensors at once. But I do believe as it becomes more normal, all these cars are going to be able to communicate, so you are going to be able to reduce the cost because you're not going to need all the sensors. Because now you will be able to share that information with another vehicle. I can foresee that in the future and I know we just talked about predictions. But in the future I do believe the costs and we won't need all these sensors and all these products in one car because there will be more of a shared style road. But I believe that's far off in the future from now.