What’s Next in the Autonomous Vehicle Space

As we wrap up our automotive podcast series, Autonomous Systems expert Bob Leigh weighs in on key market trends for Autonomous Vehicles. By way of an early preview, Bob also discusses what RTI has in store for visitors to our booth at CES 2020 in Las Vegas. Spoiler alert: It's all-new and epic.

In Episode 45 of The Connext Podcast:

[0:08] Important market challenges in the autonomous vehicle space
[4:17] What can we expect from the autonomous vehicle market in the imminent future?
[11:35] What keeps autonomous vehicle company executives up at night?
[13:06] Where does RTI come in at this point of autonomous vehicle market adoption?
[14:57] Can you tell us a little bit about some of the projects that RTI is involved in?
[18:34] Can you tell us about RTI’s presence at CES 2020 in Las Vegas?

Podcast Transcription:

Steven Onzo: Welcome back on the podcast, Bob.

Bob Leigh: Thanks.

[0:08] Important market challenges in the autonomous vehicle space

Steven Onzo: For listeners just tuning in or for those who haven't heard the last few podcasts, this is the final episode of a mini-series we've been doing, which focuses on IIoT and the automotive market. Bob, you and I spoke in the first episode on this series and kicked things off by discussing the technical side of autonomous vehicles.

Today I'd like to bookend the mini-series and discuss what's next for the automotive industry and what sort of things people can expect to see from the autonomous vehicle market in the future. So I think it's appropriate to start off by talking about some of the big market challenges in the autonomous vehicle space as of now. For example, consumer business models and supply chains.

Bob Leigh: Yeah, those are big topics. It really seems to be anyone's game at this point. We have companies, ride sharing companies, that we've seen that seem to require autonomous vehicles to be profitable, but I'm not sure that technology, frankly, is developing fast enough or as fast as everybody expects. So perhaps ride sharing isn't going to be as big as we thought it was maybe two or three years ago. We've got people exploring models for things like joining a club where you can get any car you want on the weekend as part of that club and it'll come pick you up and then you drive it out for the weekend, to other models where, what Tesla has recently suggested where you can share your car out or be part of a ride sharing program where the car comes and picks you up but you drive it.

So it's not really a Level 4 autonomous vehicle, which is driverless, but it's just going a few blocks from wherever it's parked to pick up a passenger. And who knows what's going to fall out of here. It's the kind of thing where it's really tied to how well the technology works to what business models are really going to execute. And the big unknown is how consumers are going to behave.

So, a lot of uncertainty in the business models that are out there and that are going to work. I think the challenge still is who's going to dominate, what's going to be the winning formula and all of these companies still scrambling to create IP first of all, because we can build value on IP, and that IP certainly extends to how do I build my supply chain for this? How do I get my costs down? How do I create a user base that is going to be committed to using our cars or using our transportation model and what am I going to charge for that? And all of that is still in development, I'd say.

Steven Onzo: On a side note, last time you had mentioned that some people think it's best to develop the best IP, can you dive into that just a little bit more?

Bob Leigh: Well, there's risks in all parts of this market and getting to market and being successful and what we mostly think of is either selling a car or selling a service and companies that do that. But that's not the only strategy you can take in trying to be successful in this market. And another strategy is, I'm going to develop some unique IP that no one else has, that we do better than no one else, and I'll either sell that IP or lease that IP out to other companies building autonomous vehicles and autonomous systems. And that perhaps has a lot lower risk because you don't have to actually build the market and build a customer base, and compete in the very competitive, who's going to drive who, where market and you're just focusing very narrowly on this one little piece of IP.

So for small companies with smart people, that may be a better approach because you can focus on the technology and do that really well as opposed to worrying about the whole supply chain issues and sales and marketing issues that go along with selling to the public.

[4:17] What can we expect from the autonomous vehicle market in the imminent future?

Steven Onzo: Last time we talked we said it was tough to say where autonomous vehicles were going to be in the future. What about, autonomous vehicles aside, what about the market? What can we expect from the autonomous vehicle market in the imminent future? You know, if we were to take a peak just over the hill a few years, what would that view look like or what do you think it would look like?

Bob Leigh: I think there's two markets that are, to me, very interesting right now. The first one is actually the trucking and mining market for autonomous vehicles. And this is interesting because we're seeing more and more deployments in this area and the need for autonomous technology is really driven by real use cases and real value that can be provided.

So, the need in both of these markets is to increase safety and autonomous vehicle technology can do that. And so I'm not talking necessarily about driverless trucks in this case, but things like lane keeping and lane detection and collision avoidance and following can help. If drivers are not paying attention then it can avoid a collision or can give them more information so they can better navigate long trips on the highway or urban areas or whatever the scenario is.

So that's one thing that's driving this technology, but perhaps pushing it even further is labor shortage in both of these markets. It's tough to hire drivers for trucking and for mining. And in the case of mining this equipment costs millions and millions of dollars and the operating costs are very high as well. So if you don't have a driver you have equipment that's not being used and you're losing a ton of revenue and you have capital that's not being properly utilized. So spending a few million dollars to get a technology installed and tested and running, and really like tens of millions or hundreds of millions, is actually probably a good return on your investment because you can do more with that equipment and you don't have to rely on a driver sitting in the cab all the time.

This also applies to trucking where a driver can maybe drive longer or do more with more resources. If you think about trucks following each other in a caravan, if one driver can drive the front truck and you have two trucks behind him, you can now ship two trucks behind him that are driving autonomously just following him, you now can do three times as much volume with the same labor and this can help solve some of the labor shortages. So that's a very interesting market that frankly, may get to maturity faster than the more exciting and sexier autonomous vehicle market for ride sharing and passenger cars.

So that's one to look at. The other one that is still really, really interesting because there's so many billions of dollars being invested is, what's going to happen in autonomous vehicles. And right now I think the focus is, let's get to Level 3 autonomy or let's get as close as we can to Level 3 autonomy and see how that works. And so if you recall Level 3 is when you can not pay attention until the car tells you that it's time to pay attention. Whereas Level 2 is you need to keep your eyes on the road all the time.

So if we get really close to Level 3 and then we eventually achieve Level 3, then we have a case for more value in that car and that you can take longer trips. And if you're fatigued it's not going to impact your driving ability. Perhaps you can take a break from focusing on the road. And perhaps that's going to make the roads generally safer. So that's where I think the focus is right now imminently. Level 4 and Level 5 where you're completely disengaged from driving seem to be still a long way away from mass market adoption and much longer than we thought. I've seen predictions that say in 10 years we're either going to have 10% penetration or 90% penetration, which is basically a way of saying we have no idea.

Steven Onzo: Right, that's a pretty huge gap. I actually read somewhere that some people are talking about skipping Level 3 and going straight to Level 4. Does that make sense? I mean it seems like you'd want to take it one step at a time but with emerging technology that's coming around, I think some people want to go straight to Level 4. Like you said, it is a race.

Bob Leigh: Yeah, so that's interesting because if we just look in detail on what the description is. So Level 3, the car should be able to transition to a driver with a comfortable transition time. So you have to be available. You still have to be there. You can't really fall asleep at the wheel for example, but you maybe don't have to be concentrating at all times.

When we talk about Level 4 then the car needs to drive on its own without human intervention under certain conditions. And so that's the critical piece. It may be actually easier to say this car is Level 4 within this geographic area or on this particular route, than it is to do a Level 3 car for a general purpose use. And there's an immense amount of value in doing that because that's now your ride sharing use case where the car only operates within Sunnyvale or within the Bay Area and it knows that area really well and we've covered lots of use cases. So that's a Level 4 car that can operate autonomously within that area. And that's what we're seeing a lot of testing for today.

That might be an easier problem to solve then, I'm going to sell you a car where you have Level 3 capability but I have no idea where you're going to go.

Steven Onzo: Right. So free range to go anywhere. That's going to take a lot of time to...

Bob Leigh: So there's inherently a lot of complexity and risks in Level 3 with not a huge benefit compared to say a very advanced Level 2 system, like a 2+ system, which has all the safety capabilities and is going to keep you safe, but you still need to pay attention.

Steven Onzo: So in a controlled environment you can eliminate ironing out all the details of pretty much driving wherever you want coming across any obstacle that the world gives you.

Bob Leigh: Right, and things like left turns. So within a city, all the red lights and the left turn lights operate the same way, and people tend to behave in a consistent way generally. There's a culture of how you drive. And if you go from Sunnyvale, say to Boston, well all of the learning that you had in Sunnyvale is not necessarily going to work in Boston because people drive differently and the lights look different and the way you navigate a left turn could be different. So there's a huge amount of local behavior that has to be built into these autonomous systems. And so if you can restrict that, and if you think about the highway driving use case, if you're talking about trucking for example, you could achieve Level 4 where you're going to go from this exit to that exit and the driver doesn't have to pay attention at all, that's a much simpler use case.

[11:35] What keeps autonomous vehicle company executives up at night?

Steven Onzo: So with all this being said, I can only imagine if I was working for a company and an executive at that, what would keep me up at night. So what I want to ask is what do you think keeps autonomous vehicle company executives up at night? Is it all of this plus more?

Bob Leigh: Well, it's when am I going to see return on these billions of dollars that I've invested? The car company is investing a ton of money, not only in autonomous vehicles but also in connected cars and they're investing way more money, for a longer period of time, than they ever have before. So this is a huge gamble that these billions will eventually pay off. And it's not a two year return or a three year return, we're talking like it seems like a decade return on these investments. In any normal market, if you were a public company that just would not fly. That's not the way you typically behave, to speculate on new technology that costs them many, many billions and is going to have a very long time to return.

So, I'm sure that's what keeps most executives up at night. Are we going to win or are we going to lose this gamble? And it's really all about the market dynamics and competition and how fast can we get to a point where we are actually seeing value and making some money on the investment.

[13:06] Where does RTI come in at this point of autonomous vehicle market adoption?

Steven Onzo: That makes a ton of sense. Well much like you've discussed, it's a pivotal moment in the success of the market. Where does RTI come in at this point of autonomous vehicle market adoption?

Bob Leigh: So we start working with our customers, and we're selling to companies that are developing the technology and integrating the technology into systems. And we come generally at two phases but fairly early on in the process. One, when they're doing the very first steps at developing an architecture, a proof of concept for how the vehicle will be built. And that usually starts with a couple of years of research and proof of concept validation and testing and then an architecture decision is made and it is to adopt Connext DDS and then you've gotten many years of development before you get to market.

The other point that we come in is where companies have perhaps started building their architecture, building focused on their IP and they have a bunch of IP. They're built on their own in-house architecture or more likely built on something like ROS. Then they've decided, "Okay, we've proven out the technology, now we're going to start development for a production system." And they will transition from a platform that was perhaps only suitable for research to Connext DDS because now they get the safety certification options and a higher performance and more reliability and they can build in robustness and reliability into their architectures as well. So that would be the next inflection point. They still have many years before we get to actual production deployments but it comes much later in the company's development cycle.

[14:57] Can you tell us a little bit about some of the projects that RTI is involved in?

Steven Onzo: This is a good segue to talk about some projects that RTI is actually involved in. Can you talk a little bit about that?

Bob Leigh: Yeah, I can talk about them in general. This market is very secretive. Many of our customers feel working with us is a competitive advantage so they don't want to talk about their architecture. They don't want to talk about their use of Connext DDS before they go public and they're nearly ready to launch their product. But we're in more than 40 autonomous vehicle programs that are on a production track. More than half of them, around 25 or so, are things like passenger cars and trucking and things that you'll see on the road. They range from ride sharing to electric vehicle startups to traditional OEMs that are building the next generation of their in-car architecture. We're in the very early stages of some and at others they're getting ready to launch their product within the next year or two.

What we almost always are working with our customers on are the really hard sensor fusion path planning architectures within the car. So these are running on high powered processors and they're taking in a lot of data and things like our quality of service parameters and the ability to build in robustness and the ability to handle different types of data with the same framework and the same API's, is really why they're using us for that use case. I like to call sensor fusions the pointy end of the stick, and that's where we usually enter into a customer.

Then we also have projects doing teleoperations. There's a lot of cool capabilities that we offer in our product to help with, connecting real-time over unreliable networks. Then more recently we're being built into the lower level safety domain of the autonomous vehicle. The part of the car that's doing things like actuation and steering and some of the data that's critical to actually make the car move and function.

So, there's a wide gamut of what we're doing, but usually it's centered around, I've got a lot of data I need to move it quickly, I need to move it reliably and I need to be very flexible in how I build and deploy my system because IP's evolving, platforms evolving, and our use cases are evolving.

Steven Onzo: Do we see customers who start off by using Connext for one thing, say it's for sensor fusion or for the safety aspect of it, and end up using more features than they thought they would over time?

Bob Leigh: Certainly, and that's typical. So that's why sensor fusion is that pointy end, that's the problem that they're trying to solve. That's the hard problem they say they need-

Steven Onzo: It's the foot in the door.

Bob Leigh: ...to solve that first. So we're the ones that are best able to solve that problem for them. Once they've made that decision, then now it doesn't make any sense to bring in other protocols and other technologies when you've already got one that can handle just about all of your use cases. So, it goes from there to say, "Okay, we can use it for teleoperations. We can use it for some of the control." Sometimes we'd say, "Okay, well we even use it for providing data to the infotainment system because we can rely on DDS security mechanisms to isolate those systems." Then it evolves from there. I would say no two customers are the same and what else they use us for, but 90% of them use us for the sensor fusion and really hard autonomous functionality.

[18:34] Can you tell us about RTI’s presence at CES 2020 in Las Vegas?

Steven Onzo: Well, finally, I know that RTI will be at CES for the first time this coming January. Can you tell our listeners what they should expect to see from RTI there?

Bob Leigh: Yeah, that's pretty exciting. We're going to be at CES for the first time with a booth. What's probably more exciting than the booth, as wonderful as that is, we'll have a few customers announcing their use of DDS, some of them that could be going to market within a year or two. Others that may be a little early on in the process, but they'll be the first end use customers using DDS out of all the ones that we're using that are ready to announce publicly details about their architecture and why they are using DDS. So that's really exciting for us.

We're also going to showcase all the different integrations and platforms that we support. So when you choose Connext, you're really choosing a platform that'll run from the ECU to the cloud, support of multiple different hardware systems and can integrate with things like ROS and AUTOSAR and MATLAB and Simulink and other platforms as well. So we'll be showcasing how Connext can be that that backbone you can rely on to really support your entire infrastructure needs. That's the first time we'll be telling that complete story as well. So that's pretty exciting and then we expect to have things like panels in our booth and a few other announcements with both partners and customers. So, there should be a lot going on.

Steven Onzo: If you'd like to find out more you can always visit our website on the events page. You can find out more about CES and RTI's presence there. Well, Bob, thanks again for joining us. It's always great having you on and we look forward to having you back on soon.

Bob Leigh: Okay.

Steven Onzo: Thank you.

Bob Leigh: Thank you.

Autonomous Systems

Customers

Products

Products

Services

Support

Developers

Resources

Company