Requirements, Communication Protocols and the IIoT

In Episode 3 of The Connext Podcast:

• [1:20] Social media for machines! What social media and DDS have in common, plus a great example from medical robotics [3:16]
• [6:40] About addressing the challenging communications requirements for telerobotics, especially in space (rovers on Mars and the ISS
• [13:00] How Connext DDS can help you solve your system challenges, especially with respect to integration, communications and security challenges
• [16:10] Why people are choosing RTI and Connext DDS for their next-gen systems.
• [18:00] What a system is and why autonomy is so key
• [19:00] Connected cars, vehicle-to-vehicle comms, sensor systems and the evolution of these systems into larger, smarter systems of systems with ad-hoc P2P networks
• [24:45] The transition from PhD student in computer science to CEO, to Product Director, to VP of Sales - who is breaking records!
• [28:28] Insights on RTI’s record growth

Related Content:

• [Customer Snapshot] European Space Agency: Telerobotics and Haptics Laboratory
• [Customer Snapshot] RMC MicroLab: Minimally Invasive Robotic Surgery with the DLR MiroSurge
• [Product Page] Connext DDS Professional

Podcast Transcription:

Lacey Trebaol: Hi, and welcome to episode three of The Connext Podcast. I'm Lacey Trebaol and I'm here today with Niheer Patel, and we're going to be speaking with Dr. Edwin de Jong, the VP of sales at RTI. Edwin brings with him more than 20 years of experience in the software industry. He is one of the pioneers in the development of large scale real-time publish/subscribe middleware that is now revolutionizing the industrial internet of things. Now Edwin is spearheading the incredible business growth that we're seeing here at RTI.

In today's podcast, we're going to cover a variety of topics. We'll be touching on what social media and DDS have in common, and why a point solution isn't the right thing for the IIOT. We'll also be talking about autonomous cars, vehicle-to-vehicle communication+s, sensor systems, and how the evolution of these systems into larger, smarter systems of systems with ad hoc peer-to-peer networks is kind of the next thing.

We'll wrap up today's interview with a discussion about Edwin's transition from PhD student to CEO to product director, and then to the VP of sales. It's quite the conversation. We hope you enjoy this episode.

Niheer Patel: Edwin, thanks for joining us today.

Edwin de Jong: Sure, you're welcome.

Niheer Patel: Let's start out with something simple. What is DDS?

[1:20] Social media for machines! What social media and DDS have in common, plus a great example from medical robotics [3:16]

Edwin de Jong: Great question. So the way I've been explaining DDS recently is basically it's social media for machines. So using DDS, you build basically a society of machines, devices, and whole systems that work together to achieve a common purpose. So we as people, we use social media all the time. You can use Facebook, Twitter, Instagram, all the others you can share with your friends and your followers where you are, what you're doing, what you're up to. You can say, "Hey, I'm here at Starbucks having a great time. Anybody care to join me?"

Or friends can like that, say something about it, or they can actually say, "Hey, I would like to join you." So through social media it's more than just sharing your status and what you're doing. You can ask actually also coordinated activities with your friends.

So our software enables something similar, but for machines. So devices, machines, and systems, they can post their status. They can invite others to participate in a coordinated activity. Now the difference with social media for people, like you and me, is that for machines, these interactions happen at the speed of physics, which is very, very different.

Let me give you an example. So our software is being used for example in a robotic system for heart surgery, and this system actually performs surgery on a beating heart. Quite impressive when you think about it.

Niheer Patel: Yeah, that's insane.

Edwin de Jong: Right, because today when a surgeon performs heart surgery, they need to stop your heart from beating first because forming the procedure on a beating heart would be quite dangerous. And so to stop the heart from beating, they put you on life support and then they perform the procedure. And that is very evasive and still quite risky as well. So with this system, and it's of course still in early stages of development, but with this system you can perform the surgery on a beating heart. And the way it works is the surgeon uses a display that shows the patient's heart as a static surface, so they use image processing to display a static surface. And then the surgeon uses haptic feedback controls to perform the surgery. So as they make the incision, they actually feel the resistance build up in the controls.

Lacey Trebaol: Joysticks.

Edwin de Jong: Yeah, with joystick-like controls. So for the surgeon, it's pretty close to how they do this today, except they're not manipulating the instruments directly. And then these haptic feedback controls, they actually control a robotic system consisting of three robot arms, and those arms perform the actual surgery on the beating heart, and of course compensating for the beating of the heart in real-time, so the instruments keep a constant distance from the surface of the heart. They make an incision so they move closer. And of course this is a closed feedback control loop because the heart, it's not beating at a constant rate. It changes slightly, so you need to sense that very closely, how you need to position the robot arms and everything.

So to do that, we run a three kilohertz real-time feedback loop between the haptic feedback controls and the robot arms. So that's 3,000 updates per second. It's not just per second because those updates need to be spaced very equally across that one second. It's not like you can send 3,000 updates in the first half-second and then nothing in the second half. That would be disastrous. So it needs to be very equally spaced.

So going back to social media for machines, now you can probably see how this works, right. It's like the haptic feedback controls, the two of them, and the three robot arms are really three almost independent autonomous systems. They're sharing their state, the controls, basically, what their position is, where the robot arms need to move. The robot arms basically receive the information, they control their movement, they send status information back. Is everything still operating in according to how things should work so there's nothing going wrong? And they share all the information 3,000 times per second.

And of course an important difference there is social media, like Facebook and Twitter, they can easily keep up with 3,000 updates per second aggregate across all users, actually, orders of magnitude more than that. But you as a single user can't send 3,000 updates per second to Facebook or to Twitter, they would lock you out, right. That's like a denial of service.

Niheer Patel: Or let alone to every one of my friends. Yeah, 3,000 a second just ...

Edwin de Jong: Yeah, you can't do that. And actually, you can't go through a server at all just because of the latency that that would take. So the way that this is set up is all these components share the information peer-to-peer. So it's really like ... You have like in this case five, and actually six if you also count the display, machines or devices that are, relatively speaking, autonomous. And they are coordinating their activities, so to speak, through social media. And it's this social media that our software provides.

[6:40] About addressing the challenging communications requirements for telerobotics, especially in space (rovers on Mars and the ISS

So that's one example. Now to make things a little bit more interesting, there's another example, quite different. So NASA is using our software to control robots, like rovers that drive around on Earth from the international space station. Why would they want to do that? Well, really what they want to do is control ...

Lacey Trebaol: Why not?

Edwin de Jong: Yeah. I know, just because ...

Niheer Patel: Sign me up.

Edwin de Jong: Yeah, really what they want to do is control robots driving around on Mars from orbit, right. And this is an experimental setup.

Lacey Trebaol: Telerobotics setup.

Edwin de Jong: It's telerobotics, exactly. That's exactly what it is. And especially the link, communication link from the International Space Station down to the robots and back up, is very challenging. There's a high latency, multiple seconds. If you send something down, it takes multiple seconds for the robots to receive it, and same thing when you go back up. It's also very challenging link in the sense that the bandwidth is limited, you have intermittent connectivity. The link isn't always there. So very challenging environment.

And here they use DDS to basically communicate between the International Space Station and the robots on Mars, which can be more than one. So again, you see as an example the International Space Station is one participant, the robots each are participants in the system, and through social media, so to speak, they share where they are, what they're doing, and they coordinate their activities.

Now what's interesting here is that when you look under the hood, it's the same communication protocol being used as in the surgical robot. I mean, you think about it, that's quite impressive because keep in mind the surgical robot, 3,000 updates per second, the Mars rovers from International Space Station, a two second latency link ... Very, very different environment, and yet the same communication protocol is used in those two very different environments. And with DDS, this is possible because of the quality of service tuning that you can do in this case on the reliability protocol. So I don't want to go into the details of how that is done right now. That's maybe a conversation for another time, but just keep in mind that really this is the only communication protocol as part of DDS that can be used in these two very different environments.

Now why is that even important? Because why not use just point solutions, very different point solutions, in those different environments? This is where the industrial internet comes in because what we see is that those very different environments are now more and more being integrated into one system. Not that we're going to do heart surgery from the International Space Station down on Earth anytime soon.

Lacey Trebaol: But maybe we could?

Edwin de Jong: Maybe we could. The only way around, actually ... Might be a use case, right, if something bad happens in the International Space Station, you can perform a medical procedure from Earth, but in everyday life, it is very well possible that you have let's say a surgeon over here at Stanford, a patient in Miami, and now I have the haptic feedback controls and my display here in Stanford where the surgeon is. I have my surgical robot set up in Miami. And now I have to work across a very different type of networking connectivity than when I have everything available in the same hospital, in the same room maybe even.

Lacey Trebaol: Physically connected, right? Yeah.

Edwin de Jong: Physically connected. Exactly. So through quality of service tuning, you can make these different setups work. And that is what we're going to see more and more driven by the industrial internet. Another example, right, is patient data being transmitted from an ambulance. Ask the patient who's in transit ... I don't know if you've experienced yourself or maybe you've seen it in the movies when an ambulance arrives at the ER. It's a very chaotic scene, right. Who's the patient? What's the condition? What do we need to do?

Niheer Patel: A lot of information has to be passed back and forth. If there's ...

Lacey Trebaol: In a very high-stress environment.

Niheer Patel: Yeah. Yeah.

Edwin de Jong: Exactly. And then they need to prep maybe the OR, or whatever needs to happen. So in this case, the ambulance transmits the patient data in real-time, so that when the ambulance arrives, they know who the patient is, what their condition is, what they need to do, the OR or whatever is already set up. So it's a very different solution to the problem.

And there you see that it's maybe not the space to Earth link, but the ambulance, of course, is transmitting data over LTE. The data is being processed in the hospital where you have your standard high speed WiFi or ethernet cables even, and basically these different environments are now part of one and the same system.

And yeah, that's the type of environment that DDS provides. It's social media for machines, that basically runs across any type of connectivity that you can imagine.

Niheer Patel: So high latency, low latency environments, life-critical to really cool scientific type of work, all sorts of networks involved, all sorts of communication protocols involved, or I should say communication mediums, whether it's LTE or WiFi or a satellite link or a link from the space station down to a rover. So DDS is really used everywhere ...

Lacey Trebaol: To get data ...

Niheer Patel: And can be used ...

Lacey Trebaol: ... Where you need to have it, when it needs to be there to perform the various actions on the other end of that.

Edwin de Jong: Exactly right.

Niheer Patel: That's insane, to know that it's in so many things. And it's like, just the medical example, the benefits to the doctor performing the surgery, right ... Getting the still heart and to the patient ...

Lacey Trebaol: Not having to put a person on bypass is huge.

Niheer Patel: Right.

Edwin de Jong: Right. That is huge.

Niheer Patel: That's amazing. The benefits there are huge.

Lacey Trebaol: The risk reduction that that technology will bring into that surgical suite is sort of ... It's unlike anything I think they've ever had. The idea that you no longer have to put a patient through that is crazy.

Edwin de Jong: For heart surgery, right.

Lacey Trebaol: Yeah. Anything you can do to make that safer ...

Niheer Patel: Right.

Lacey Trebaol: We support.

Niheer Patel: You did a great job explaining to us what DDS is, and we're here at RTI, so it'd be great to know what does RTI do?

[13:00] How Connext DDS can help you solve your system challenges, especially with respect to integration, communications and security challenges

Edwin de Jong: We provide Connext DDS, which is basically the industry leading DDS implementation available from RTI. It's not just an implementation of the DDS standard, it's really a whole tool suite for developing DDS applications. Essentially you can say it's the most complete connectivity framework for industrial internet applications. We already touched on that briefly in our previous conversation. It runs across close to 100 different platforms, across high-speed networks as well as WiFi, radio, satellite links, and all of those in one system if you want. It comes with all the tools that help you fine-tune the deployment of DDS applications to some of those specific networking environments. It also comes with a whole suite of infrastructure services, for example routing service bridges DDS data between different network domains, different networking technologies, like between an ethernet network in a factory and LTE out to devices out there in the field. We already talked about the ambulance, for example, where you will basically have to bridge the patient data as it is transmitted in real-time over LTE to basically ethernet in the hospital networks. Our routing service allows you to do that bridging with zero programming, just some configuration and it does it for you.

It can also perform transformation of data on the fly. Maybe you don't want to send all the patient data. Some of it maybe remains in the ambulance, some of it you want to transmit to the hospital. So you can make those transformations on the data if you want.

Lacey Trebaol: Some of it you might want anonymized.

Edwin de Jong: Right, for example. Maybe you don't want to stick the patient's data or whatever you're using on every data item that you send out. It includes persistent service that maintains system-critical configuration data on disk, so if you need to reboot the system or parts of the system, you can do that. It also includes a record and replay capability that helps you build easily. Things like simulation, training, offline analysis applications. So in essence, I would say RTI Connext is the most complete and also the most robust and proven DDS implementation out there.

Niheer Patel: Why do our customers come to RTI? I mean, it sounds like Connext itself is a really good reason to come to RTI to solve their distributed system problems. Maybe you could touch on ...

Lacey Trebaol: But DDS is an open standard, he also mentioned. We have one implementation.

Niheer Patel: Right.

Lacey Trebaol: We have competitors out there who have other implementations. And why are they choosing RTI Connext DDS?

Niheer Patel: Over ... Yeah.

Lacey Trebaol: Yeah.

[16:10] Why people are choosing RTI and Connext DDS for their next-gen systems.

Edwin de Jong: Good question. I think the answer to that is actually simple. It's trust. Like you said, DDS is standard, but really it's a set of APIs and protocols, and that's it. And how these get implemented differs vastly across different DDS implementations. And our customers, they all know that choosing the right connectivity framework for your next generation system or product line is very important. It's going to be the critical backbone of your next generation systems.

So customers choose to work with partners they know won't let them fail. And we at RTI, we have a proven track record here. We're using well over a thousand different system designs. Not only is our software proven in those system designs, but we've also built a wealth of systems architecture expertise through that. We've been involved in nearly all of these thousand designs. So we can use the expertise that we gained throughout all that in helping our customers define the right systems architecture, and basically we can help them avoid making design mistakes that otherwise will be very expensive and risky to fix later. And nobody else is in this position to help like we are.

Lacey Trebaol: And we like helping. We have a great services team. And our support is actually like win-win. Our customers, their systems are deployed. One of the common things, I've seen thank you notes come back in relating to from our customers is for the amazing support they actually received during the process, and all the help.

Niheer Patel: And even the engineering teams, right, they get involved almost directly with customers, especially these big projects. So it's the entire company working together with our customers to make them successful.

Lacey Trebaol: To make them successful.

[18:00] What a system is and why autonomy is so key

Edwin de Jong: Absolutely. Basically we provide a connectivity framework. So usually our customers tend to try and connect things. That's oversimplifying things maybe quite a bit, but what mostly all of our customers have in common is that they're building systems that integrate sensors with processing and control nodes and actuators. So what we help solve there is a systems integration problem between sensors, processing nodes, control nodes, and actuators. And what we see is that increasingly these systems have higher and higher levels of autonomy. You can think of a system really as a set of highly autonomous devices and other system components that coordinate their activities through our social media software for machines, right, as we talked about earlier.

Think about autonomous cars, for example. We're going to change the auto industry forever. So when you think about the car itself, the autonomous car, there's all kinds of sensors on there, like radar, LIDAR, cameras, and we have the processing and control nodes that collect all the data from data correlation, apply deep learning and all those things. And then in the end after the car basically builds and maintains this real-time picture model of what's going on in the environment, it sends control commands to the various actuators on the car, which are of course the steering system, braking system, acceleration, all those things.

[19:00] Connected cars, vehicle-to-vehicle comms, sensor systems and the evolution of these systems into larger, smarter systems of systems with ad-hoc P2P networks

So that's one example where our software is then being used to connect all those components on the car. And that's where it starts, on the car. But now of course we're also talking about connected car where your car is now being connected to cloud-based services. So you get the connection from the car to the cloud.

The next step after that is vehicle-to-vehicle communication, V2V. So right now ... It's not exactly right now, but in a few years, right, when we'll see more autonomous cars out there soon ... Yeah, in a few years. What's going to happen is those cars have the sensors to detect what other things in their environment, like other cars and of course pedestrians and bikers and everything, are doing in the environment. They need to sense that. That's why you have all those sensors.

But why do that if the car in front of you can just tell you that it's going to apply the brake, right? So rather than sensing that, the car in front just tells you. So now you get this ad hoc network that is basically formed and maintained as cars drive on the freeway or in the roads in the city, and they communicate directly to tell what they're going to do, and then others can respond to it. And of course you get a much quicker and also much, much better quality response once you've set it up like that, of course that's going to be ...

Lacey Trebaol: It's like an iRobot, right? In the movie when they're driving through the tunnels and all the cars communicate with each other, so they can go really fast and do ... Yeah.

Niheer Patel: Fast, but reducing accidents to zero accidents.

Lacey Trebaol: Reducing errors and accidents. Yeah.

Edwin de Jong: Yup, yup. Exactly. And then from there you go to connecting between the car and roadside systems, right. Why detect that the traffic light is red when the traffic light can just tell your car, "Hey, stop." So we'll get direct communication between those. And then from there you take all the data that you're gathering and basically move that in real-time to the smart city. And now the smart city can start to control the flow of traffic throughout, especially the busy areas in town, in an optimized way.

So you go from this simple systems architecture on the car where we connect the components there, to basically a whole global network that spans entire cities to optimize the flow of control. And it's these types of basically long-term connectivity roadmaps, right, because this is not something that's going to get created overnight. It's a whole connectivity roadmap that we help our customers solve, and this is just an example of that, but a very typical one where it's not just connectivity in one system. It basically spreads and then it gets bigger and bigger until you talk about global connectivity, and then all the optimizations you can do with all the data that you collect, and also basically create a much safer environment at the same time.

Niheer Patel: The autonomous vehicle story's near and dear to me because we also have a product for safety with vehicles and autonomous vehicles, ADAS, all those components ... Safety is really critical. We're talking about lives.

Lacey Trebaol: Connext DDS Cert.

Niheer Patel: So yes, we have Connext DDS Cert, which is a safety certified product that we can leverage in these safety-critical applications. And of course you throw in security to prevent against any kind of hacking for those autonomous vehicles, now you have this great set of products and frameworks to build out your connected vehicle story and protect it and protect the people around it.

Lacey Trebaol: And also to future-proof your system, which is kind of the point that you made that starting at the car is the step of this map for them, but as they go forward, the need to be able to really scale it out. We have the products to actually accommodate the scaling out of a system like that.

Edwin de Jong: Absolutely true, yeah, because that's one of the advantages that you have with DDS and particularly our implementation, is that you don't need to think upfront what data you want to share, what information you want to share, and who you're going to share it with. It's really like with social media, those connections can change over time, and what information is there you can share with anyone anywhere. Of course, subject to security policies ... Have that in place in well, but you have the flexibility to let that data flow the way that you want at certain points in time as the system evolves.

Lacey Trebaol: Yeah, I remember when I was a customer before I came to work here. One of the things that a mentor told me was that when you design a system, you have to design it knowing it's the one truth. The system will change and evolve. It's a dynamic thing. And so the best thing you can do as a system engineer was to approach it with that in mind and say, "How do I build the ability to accommodate change into my system and do it gracefully?" So that you're not breaking things and having to rebuild and recompile code even. It's built to accommodate change. And when I was there, this was right before we became a customer, that was why we chose Connext DDS, was because it enabled that, it let us build systems that could gracefully accommodate change. And change was inevitable.

Edwin de Jong: Yeah, absolutely.

[24:45] The transition from PhD student in computer science to CEO, to Product Director, to VP of Sales - who is breaking records!

Niheer Patel: Maybe we can switch gears a little bit. One thing we kept from our audience is that you were at one time a customer.

Lacey Trebaol: In a previous life.

Niheer Patel: In a previous life. So it'd be great to hear about your experience as a customer in ...

Lacey Trebaol: You went from being a customer to working in products and markets, and now you are the VP of sales.

Edwin de Jong: Right, yeah.

Lacey Trebaol: So tell us a little bit more about that.

Edwin de Jong: Yeah, exactly. All of this is part of that story, sort of my life story. I could say it's definitely been an interesting journey. It never was in my plans to be a VP of sales one day, I can honestly say that.

Lacey Trebaol: No, when you were getting your PhD that wasn't ...

Edwin de Jong: Not really. I was a ... I still am, I guess, a computer scientist originally with a particular interest in distributed systems architecture. My PhD thesis was on parallel and distributed programming. My research has always been focused on what we call today data-centric architecture. I didn't know it at the time, but I've basically been working on DDS years before it became DDS as a standard. I actually read into RTI for the first time at the OMG, the Object Management Group, and together we worked on the first draft of the DDS standard.

Around that same time, you already mentioned there was a customer previously ... So around that same time, I actually started my own company in the Netherlands. That's where I'm from, that explains the accent. And we provided a technology very much related to what is now DDS. Back then what my company did was create an integration between RTI's NDDS at the time, which was more of a messaging middleware. It didn't have all the data-centric features yet. And we integrated that with a relational in-memory database. And through that integration, we actually drove many of the concepts that are now part of the DDS standard. So yeah, I was a customer of RTI in the early days or exactly around the time when we started working on the DDS standard.

And then, by the way, the company that we started back in the Netherlands got acquired by RTI, so that's how I ended up at RTI. At the time of that acquisition, I started actually in R&D team at RTI, and then it was a couple of years moved into product management, product marketing, which I did for five years. And now I'm in sales and when I think about it, really it doesn't matter much where I am. What I really like to do is evangelizing the DDS concepts, the architectural thinking behind it. That really is the fun part of the job, I feel. And you can really tell because when you're presenting to a group of smart engineers, there's no quicker way to put them to sleep than by giving them a product pitch.

But the beauty of our job is that we actually don't pitch our product that much. We mostly talk about system architecture. It's an entirely new way of designing large-scale distributed applications, the next generation for the industrial internet. And seeding that architectural thinking behind these systems is really what I love to do, whether I'm in R&D or marketing or sales, it doesn't matter. It's what I've been doing all this time and what I still do today and what I still enjoy doing. And the nice thing is with the industrial internet now, it seems more relevant than ever.

[28:28] Insights on RTI’s record growth

Lacey Trebaol: Well, you do it very well because as some people might know we recently had a press release that went out, and it talked about the fact that we had record growth and IIOT market traction this past year. So under your leadership there was more than a 40% growth in sales, which is not too shabby.

Edwin de Jong: Right, yeah.

Lacey Trebaol: That must be pretty exciting.

Edwin de Jong: That's truly amazing. I mean, we hit over 30% growth in 2015, then over 40% in 2016, and looking back at our first quarter this year, we again established I think it was around 48% growth relative to the first quarter last year. And we basically doubled the company in just over two years. Last year we were involved in 140 new system designs that are now using RTI, and as I mentioned, we're in over a thousand unique designs in total. So we added another 140 just last year. And we've really become the recognized leader in autonomous cars, intelligent transportation. We talked about that already a little bit. What we didn't talk about is that RTI is also a key leader in the transition to green energy. So now we're actually in production programs in major hydro and wind generators. And we talked a little bit about this, with RTI's help, tomorrow's medical devices will work much better together and improve medical care.

And last year has really been a turning point for us. I can say we feel truly honored to be trusted by so many customers by so many systems that we're designed into, and really it's our software that makes those systems run. And we're very happy that we're helping to define a safer, healthier, and I would say generally happier future for everybody.

Lacey Trebaol: That's what we all want.

Niheer Patel: Just to be happy.

Lacey Trebaol: Yeah.

Niheer Patel: I think that's a worthwhile goal.

Edwin de Jong: Sums it up.

Lacey Trebaol: Thanks for listening to this episode of The Connext Podcast. We hope you enjoyed it. If you have any questions or suggestions for future interviews, please be sure to hit us up over on social media, and you can also reach out to us at podcast@rti.com. Thanks, and have a great day.