Autonomous systems expert Bob Leigh introduces RTI Connext 6, built to address the common challenges of autonomous systems. Learn what it takes to get an autonomous vehicle on the road and how the disruption of the transportation industry will change the future.

In Episode 28 of The Connext Podcast:

  • [3:34] Addressing the challenges of highly and fully autonomous systems
  • [5:16] Implementing the DDS standard into major platforms for autonomous systems
  • [7:00] How Connext 6 addresses safety within autonomous vehicles
  • [8:31] Leveraging data fusion with Connext 6 to provide autonomous vehicles with a coherent picture of the world
  • [12:08] The value of data-centricity

Related Content:

Podcast Transcription 

Steven Onzo: Hello everybody and thanks for joining us for a special episode of The Connext Podcast. Today we’re announcing the release of Connext 6, the first connectivity software for highly autonomous systems. We are joined by RTI's Autonomous System’s Market Development Manager, Bob Leigh.

Bob will discuss the current state of the market, and how Connext 6 can solve some of the industry's toughest challenges. Thanks for joining us, Bob.

Niheer Patel: Welcome to another episode of the Connext Podcast. Today we have some really special news, and with me here to share it is Bob Leigh, our Senior Director of Autonomous Systems.

Niheer Patel: Bob, thanks for joining.

Bob Leigh: Thanks Niheer.

Niheer Patel: Let's talk a little bit, first about what your role is within the company, and then if you expand, what is market development for RTI?

Bob Leigh: Sure. I do market development for Autonomous Systems, which the largest segment of that is certainly autonomous vehicles, that includes things like drones, and even hyperloops, and flying taxes and anything that's really mobile and fully autonomous.

Niheer Patel: The sky, anything?

Bob Leigh: Land, sea, and sky.

Niheer Patel: Nice. Underground too?

Bob Leigh: Underground as well. Eventually I'm sure there’ll be some space autonomous vehicles too, but we're not quite there yet.

Niheer Patel: Okay. Market development, what do you do for market development, what does market development for RTI mean?

Bob Leigh: Market development exists within the Products and Market Department which includes not only market development but technical marketing engineers that supports standards and reference, architectures and demos, and product management, and so our role as a department is to figure out what is the future, and how do we deliver that future, and then market development is really the outward-facing component of the department working actively with the market to understand customer requirements, where we are going, what standards are important. From day-to-day I work with customers, I work with partners, I work with our sales team, gathering that information trying to bring a cohesive story back to the internal component, or the internal parts of RTI and basically say, "This is what the market needs. This is what we should do." Then figure out with the rest of the company how we should go about doing that.

Niheer Patel: So you help the rest of RTI understand what's going on in the real world, what are the challenges we need to solve? The intelligent folks in the engineering team will help build back with folks like product management, technical marketing, and professional services, we've talked about them a couple times. Then bring it back to then tell the story to the world.

Bob Leigh: That's right. Yes, so another big part of my role is business development, even though it's not a traditional BD type of function. I do a lot of that trying to communicate with customers and then have them communicate with us, so we are all I guess pulling in the right direction, in the same direction in these new markets.

Niheer Patel: Okay. Then speaking of communications I know we have a big announcement today. Can you share that with the world and our customers?

Bob Leigh: Yes. Today RTI is announcing the launch of Connext 6, and this is a product suite that encompasses our DDS Connext Pro product, and well as our DDS Connext Micro product. There are a ton of very exciting new features that really built to address the needs to autonomous systems and specifically autonomous vehicles.

[3:34] Addressing the challenges of highly and fully autonomous systems

Niheer Patel: Great, thanks. We will be digging into the details of these features in future podcasts, but let's start with the challenges of autonomous systems. Can you enumerate these for us? What are the problems that need to be solved?

Bob Leigh: First it's good to understand our position within this market. Our focus is on Level 4 and Level 5 autonomous systems. If you know what that is, great, if you don't, it's the SAE standards to describe the level of automation within an autonomous vehicle, and it goes from zero with no automation up to Level 5, which is fully autonomous which means there is no steering wheel, and the car essentially can drive itself under any conditions that it's going to run into.

Level 4 is not quite as challenging in that the car will be highly autonomous and drive under certain conditions autonomously. The driver doesn’t have to pay attention, it may only work within a certain city, or under certain road conditions, but under those conditions, the driver doesn’t need to be fully attentive.

The difference between Level 3 and Level 4 is this need for driver attention, so Level 3 you always have to have the driver paying attention, and that's mostly what we have in the leading edge cars on the road today, but Level 4 is where you don't have to pay attention.

Bob Leigh: We are addressing the needs of that Level 4 and Level 5 system which need very complex sensor fusion, pulling a lot of data from different sources, merging it together, there's a lot of bandwidth required, very low latency required, and when you do it in a high-performance safety, secured system.

[5:16] Implementing the DDS standard into major platforms for autonomous systems

Niheer Patel: Providing all these capabilities and fitting it into the existing ecosystem as such, right, I know this is an emerging set of technologies, but there are some players, so how do we fit in with them? How do we take our technology and work with folks who are really heavily involved?

Bob Leigh: That's been the most important set of initiatives, is getting for me, and for RTI and this market is getting the DDS standard and then our product as part of the major platforms for autonomous vehicles. The first one we were in is ROS 2, and that's the Robotic Operating System which has been used in robotic systems for many, many years. As an open source project, ROS 2 is a redevelopment of that system to make it more suitable for commercial and other, I guess, more safety critical applications. One of the choices that that group made is to use DDS as the underlying middleware.

The DDS standard is part of ROS 2, and you actually can get ROS 2 bundled with RTI Connext as the DDS implementation. The other one which commercially is probably much bigger, is AUTOSAR, and we've been working for the last year to get the DDS standard introduced to the AUTOSAR adapted specification, and AUTOSAR today is used by more than 50% of all cars shipped, so it's a very important standard for interoperability and development of cars in general, and then AUTOSAR Adaptive is a new standard, and now it has DDS as being essential, we are setting up AUTOSAR to be able to deliver the functionality needed for these Level 4 and Level 5 autonomous systems.

[7:00] How Connext 6 addresses safety within autonomous vehicles

Niheer Patel: Okay. Then I just want to, just one more point and then maybe go back to what we were talking about before, and dig deeper into what the autonomous vehicle looks like. Then the next part is, great we’re in these open standards, we are working with the ecosystem but putting a car on the road could be dangerous, right? There's a lot we have to consider with human safety, the driver, passengers and then pedestrians, and property damage. Where does Connext 6 fit in with the idea of safety within the vehicle?

Bob Leigh: The thing that differentiates RTI and Connext 6, from pretty much every standard, and every other connectivity or communication platform out there, is our ability to address the non-functional requirements that you need permission in safety critical systems. Security is part of that, and then safety certification is part of that. We are able to provide a path to safety certification with our product, and Connext 6 is a big part of that. We have a version of our product that has security features and low latency features, and large data handling features that will be certifiable for ISO 26262 for the automotive market, and it also includes, as I mentioned the security features which is another important component of having a safe system.

You want to make sure it's not hackable, and you want to make sure that it works as intended, and safety and security both work together to make that happen.

[8:31] Leveraging data fusion with Connext 6 to provide autonomous vehicles with a coherent picture of the world

Niheer Patel: Great. With that backdrop, right, the safety and security backdrop, if we go back to data fusion, and we think about what kinds of data we're pulling in to a vehicle and what kind of processing we're doing and why, certainly autonomous vehicles need to be able to see where they're going, for example, so there's lidar, camera. Can you help describe, first maybe from a data flow perspective, the complexity that something like DDS would help solve, and specifically Connext 6 could help solve.

Bob Leigh: Yeah. The important thing to have within an autonomous vehicles is an accurate picture of the world around you, and that is probably the largest engineering challenge, perhaps AI driving and others are just as challenging, but getting the right data is key to success in getting highly and fully autonomous systems. If you don't notice a person crossing the road, and you can't properly identify, that's just not acceptable within this environment.

Bob Leigh: To do that, we have to have multiple sources of information, and most autonomous vehicles are pulling data from your traditional sensors, speed, how fast the car is going, what's the angle, what's the GPS location of the car, but they also have radar which is very good at detecting objects that are in front of you, and measuring how far they are away, and can work through things like fog and rain, and we have vision which is very good at recognizing what an object is. Then we have lidar which is very good at creating a detailed map of exactly where on the road you are and where other objects area.

Bob Leigh: But individually those sensors don't have enough information to recognize what is happening, and what sensor fusion does is take those different sensors, fuse them together into one coherent picture of what the world is. If there's a car in front of you the radar may be able to detect, but it doesn’t know if it's a car or a bike, or a truck. Then the vision can identify what the object is, probably classify it, and then your LIDAR can give you a very precise distance measurement and where other objects in your field that you, and those things all work together to give you this accurate picture.

Niheer Patel: Right, and I think one important thing to note is something like LIDAR will operate at a much higher rate than something like GPS or even radar. This vast, diversity of data flows has to be managed, and then mapped on top of each other.

Bob Leigh: Yeah. Absolutely. You’ve got data coming at much different rates, very different volumes, so a camera gives a very predictable amount of information, measured in pixels, at a fixed rate, whereas lidar will give you variable amounts of information depending on how many objects there are in your field of view.

You need something that can adjust and respond to this varying rates of data, and line up data from lidar on the front, to radar on the sides to your GPS sensor and put it all together, and traditionally, without DDS you're probably using multiple different protocols to do all of that, because some of them are very high speeds, some of them will be very low latency, some of them need really big bandwidth, but DDS standards, and RTI Connext, is the only sort of product and solution that can handle all of that under one standard.

Niheer Patel: This is a great transition, right. We've talked about how Connext 6 can bring this level of safety, just functionally from being able to manage all the different sensors in the data rates, on top of that a path to actual ISO 26262 certification, but there's this idea of ease of development and really being able to architect the system. If we go in and kind of take maybe an OEM or system integrator, or Tier 1 supplier perspective, what are the benefits that something like Connext now brings to those folks in the ecosystem?

[12:08] The value of data-centricity

Bob Leigh: We probably need four or five hours to go through those, just even at a high level.

Bob Leigh: But the key, high-level concept is Connext is a data-centric architecture, which means you develop your applications based on data interfaces, and to do that you need to define a common data model and that data model, essentially represents the real world. What that means when you get down into an architecture's developing, is that you're developing algorithms and applications that use data that are representative of the real world, and really have nothing to do with application developed by Jim down the hall who is doing this, completely independent of that.

Niheer Patel: Right.

Bob Leigh: It's just we're working with data that represents the real world, and that means we can develop our applications, make them work really well for that particular component, and we don't have to concern ourselves with what other components of the system are doing, what kind of architecture they have, or what kind of design choices are made in developing those particular applications.

Bob Leigh: When we're going from the state where we don't know how to build these systems, and we're doing lots of research, if we're cobbling things together, and lots of it runs on laptops, to more of a proof of concept application where we need to rip up things, and change things, and improve things to production development. The data-centric nature allows us to move our IP along that chain without redeveloping and re-coding it, and allows us to incrementally add and change and modify different applications without rewriting the whole thing from scratch every time.

This is extremely powerful in allowing IP to be reused and developed, and persists through lifecycle, but also making the development process much more efficient, and you can have much tighter feedback loops from one step to the next, because you're all speaking the same language, which is a language about data, and what the data is.

Niheer Patel: Not only providing value to being able to architect the vehicle in the first place, but now enabling efficiencies in the development lifecycle, and deployment lifecycle of the vehicle?

Bob Leigh: Yeah, and we're seeing a huge demand for that within the industry because traditionally, you develop some proof of concepts in the research phase, and you say, okay, I'm going to improve my torque performance, and I'm going to write some code to do that, but then when you go to production, you basically say, "Okay, this is how it should be done, I'm going to rewrite everything, because now I need to do it for production standards for different platforms and different language."

Bob Leigh: But when we look at autonomous vehicles, software is so important, and so key to the value proposition, and then so complex, you can't do that with every new model that's developed, and you can't do that with every algorithm, it's just too expensive, and you don't have the nimbleness to respond to needs in the market. The industry has to move this way, and they know this to compete with some of the electrical vehicles, startups and other autonomous vehicle companies.

Bob Leigh: They're just learning how to do that, and DDS certainly helps with that, because it's suitable both for the research prototyping phase, and it can get you into a production system.

Niheer Patel: All right. DDS and particularly Connext 6-

Bob Leigh: Connext 6, yes.

Niheer Patel: … and as an extension the rest of RTI as well, I'll just throw that in there. Great! Thanks Bob. We're winding down on time, but I want to give you another opportunity. Is there anything else that you want our listeners, or customers, prospects, anyone working on autonomous vehicles, or autonomous systems, in general, to know?

Bob Leigh: Well, this is a challenging time for an industry that's being disruptive … disrupted, and is doing the disrupting, and we don't really know what the future will look like, and it's not just about how will autonomous vehicles get to market, and what they look like and what they will do. It's more about: What's the business model going to be? How many cars are going be sold, who is going to buy them, and how are they going to pay for them?

Everything about this, the transportation industry in general is being disrupted, but what we see is that the common thread in all of these disruptions and changes is the DDS standard, and is RTI Connext. We're part of ROS, we're part of AUTOSAR, and we have many transportation companies that include hyperloops and flying taxis, and electric autonomous vehicles, using DDS as part of their architecture, so because of this data-centric nature of DDS and because Connext 6 can support all these different platforms, and different standards, and different industry consortia, when you choose RTI you're really protected against future disruption and changes within the industry, and whether you go to AUTOSAR or go to ROS, or develop something brand new, we can still support that.

Niheer Patel: Definitely not only getting competitive advantage but really hedging your future, the future of your business, the future of your deployed systems. Then the last question I'll ask, Bob, if folks want to reach out to RTI and learn some more, how should we best be enable them? Where should they go?

Bob Leigh: Well, the website is a great place to start, if you got RTI.com, you can look for Autonomous Systems under our Industry menu. We have a press release out today on Connext 6 which has a ton of information about the features, and then I'd invite you to contact your local account team, and they’ll be happy to give you a demo, and tell you more.

Niheer Patel: Cool! Thanks, Bob, really appreciate it. Alright, thanks everyone!

Steven Onzo: Thanks again, Bob, for that terrific interview, and thanks to all the listeners for tuning into another episode of the Connext Podcast. Next week we continue our Connext 6 Podcast Series, and learn what's new in Connext DDS Micro. If you have suggestions of feedback on this or other episodes, please contact us at podcast@RTI.com. Thanks, and have a good day!

INSIDE THE PLAYBOOK

Get the latest updates and insights from the RTI newsletter

Subscribe to the Newsletter