AEye Insights: Making Tracks in the Rail Industry
In this installment of the AEye Insights series, AEye’s VP of Business Development & Strategic Initiatives, Akram Benmbarek, sits down with AEye Advisory Board Member, global railway industry leader, and former Director of IBM’s Global Rail Innovation Center, Keith Dierkx, to discuss the state of innovation in the rail industry here in the U.S., major transformative trends and challenges, detection requirements for passenger and freight railways, and much more.
AB: Welcome to AEye Insights, where we talk industry trends with proven business leaders. Our guest today is Keith Dierkx, AEye Advisory Board Member, global rail industry leader, and former Director of IBM’s Global Rail Innovation Center. Keith, welcome and thanks for joining us.
KD: Glad to Akram. I’m really excited to speak with you and the audience today.
AB: Thank you. As VP of Strategic Initiatives at AEye, I have the pleasure of working closely with you on enabling applications that solve complex problems in the rail industry, and I’m lucky to be able to leverage your vast expertise in the space. Can you please tell our listeners a little bit about your background and how you became a rail expert?
KD: Yeah, I’d be glad to. It’s not as if I graduated from undergraduate at Berkeley and said, “I’m going to go into the rail industry.” I actually decided I was going to go into high tech and high tech sales, and it just so happened that when I had gotten through my training, my manager asked me if I would take over this small little account, and that account happened to be Southern Pacific Railroad, which was headquartered in downtown San Francisco. That was over 30 years ago, and then through an evolution in my career over time — and by the way, I’ve always thought, even in technology, that industry expertise and industry knowledge is the tip of the spear in terms of creating value for clients — and so as my career progressed, I left, went to an IoT company focusing on supply chain and transportation, became a CIO in the transportation industry, and then came back into rail, not only freight rail, which was my first love, but later on into passenger rail as well. And of course, they’re very different industries. One moves things, one moves people. And so, the requirements and the capabilities and the things that they need to do is obviously very different. But there are some things that are very much the same. One is innovation: railroads, although they looked like they were smokestack industries, there’s an amazing amount of technology that they use. When I was in Beijing, as you mentioned, I opened a Global Rail Innovation Center in China. Of course, they have the largest network of high-speed rail. Most recently, they made some breakthrough announcements on “maglev” (magnetic levitation) technology. And of course, many people have heard — and although it’s not rail in the traditional sense of wheels on steel — even Hyperloop is an extension of the same type of technology. So, the rail industry has always been moving forward, looking at serving their customers in a much better fashion.
AB: We talked about innovation a couple of times, and innovations in the rail industry are not something that we hear about in Silicon Valley. Of course, in Japan, Beijing and China a lot is happening. Except for the Hyperloop, I don’t hear much here. The reality is that if you go back to history, you’ll find that the rail industry had something to do with the creation of this massive innovation hub in the Bay Area. After all, Leland Stanford, the founder of Stanford University, was a rail industry tycoon and he was the CEO of Central Pacific Railroad that later became Southern Pacific. So, Keith, fast forward 150 years, what is the state of innovation in the rail industry here in the U.S. and what are the major transformative trends or challenges that we’re solving with technology here in the U.S. and in the rail industry?
KD: Yeah, that’s a great question, but since you got to share a story, I’d like to share two quick stories. You mentioned high tech in the Bay Area — Leland Stanford, the senator from California, and also the namesake for Stanford University. But of course, railroads were a big part of the whole westward expansion of the United States under Manifest Destiny. And many people don’t realize that time zones in the US were actually created by the rail industry so that they could have consistent timetables. So, when we think of East Coast, Central, Mountain, and West Coast times, that was all around making sense for those towns and cities that grew up as we expanded westward to have a logic to the timetabling system.
The second one, which is probably familiar to everyone who’s listening in, is we’re very familiar with the telecom provider Sprint. They were recently acquired by T-Mobile, but Sprint actually stands for Southern Pacific Railroad Internal Network Telephony. And because they have the right of way, that’s the land on which the rail track is laid, they laid a communication network over that same track and later on spun that off in the late 70s, early 80s into Sprint, the telecommunications and mobile service provider.
So, in terms of that innovation you talk about, it’s been happening on a constant basis, obviously, throughout that 150 years. I think the thing you think about most today, though, is this idea of Industry 4.0 or digital transformation. And I know that we’re going to get into it in a little while. But I would say that innovation today, when you think about managing very complex rail networks in the US, that might be freight over 50, 60, 70 thousand miles of track, the network optimization, the tools that you need to use for that are highly complex. When you think about European railways and timetables down to the minute with high speed rail, regular commuter traffic and then even freight trains on a mixed track, the complexities that are there hidden from the user, that is the passenger or the shipper that’s moving those goods, but the industry has always been on the forefront of innovation. It’s just not a B2C company. It’s a B2B company that’s vital to moving freight and people in the United States and around the world. And that’s why you see also as development takes place in the Third World, a lot of that development is focused on rail as well.
We’re going to talk a little bit more later about Industry 4.0 and digital transformation. And that transformation refers not only to technologies, but business models and also people transformation, and I’ll talk a little bit about that as well.
AB: Having had the chance over the last year or so to interact with a few rail companies and system integrators that are driving innovation in the rail industry, I noticed when it comes to autonomy or safety applications, some railway customers are inquiring about 200 meter detection and others need 600 meter detection. Can you shed some light over the detection requirements between passenger and freight railways? I believe that kind of distancing detection requirements depends on speed and maybe some other factors.
KD: Yeah, absolutely, that range would matter. I mean, to a large degree, the use case is going to determine the scan with the pattern that best fits that. So, you’re absolutely right. And we’re going to talk a little bit more about one of the real values that I see. I didn’t touch on it when you asked the innovation question. But, now’s a great time when we talk about AEye and your software defined products, that 200 meter and 600 meter might be both on a high-speed train. In other words, when I’m at full speed and I’m blasting across the countryside at 400+ kilometers per hour, I want that 600 meter range so that I can be aware of what’s in front of me and have that presence in awareness. And then when I’m coming into the station, I might want to have 200 meter because oftentimes you’ll see and it’s true, but it’s not safe, you’ll see passengers crossing over the platform from one side of the tracks to the other side. So, the fact is, in freight rail, it takes a longer time to stop a train because of the weight of that train, and so you might want to continue that on the longer-range scan pattern of 600 meters.
But at the same time, you might make a decision that throughout that journey you might want to vary that time. So, again, it really depends on the use case. And, I’m going to talk a little bit about digital corridors and digitizing operations, but you really want to scan with the use case that you get to.
The other thing that to me is really powerful about AEye is this idea of intelligent LiDAR — the ability to define through software-definable capabilities based upon the use case in the patterns that you want to have, and automate that process. And I think that’s a really powerful tool. And when you think about what railways want — they want the ability to detect things, they want the ability to acquire them, they want the ability to classify them, and then they want actionable information to be able to do that. And I think that’s one of the things that is really going to power software-definable, intelligent LiDAR in the rail industry is the ability to mimic a business process.
AB: Ok, thank you for the plug and for mentioning the software definability and talking about its capabilities and its use and benefits to the audience. Maybe I should step back and talk a little bit more about what you meant by the software-definability. iDAR is the first solid-state intelligent LiDAR in the world, and what intelligence means is that it has edge processing capabilities that allows you to customize the scanning patterns that Keith talked about and detection configurations to use cases. And to your point, Keith, yes, for a train you can move from detecting an object or a vehicle or pedestrian at 600 meters to switching the scan pattern to see in high definition small objects in the near range. So, based on these capabilities, Keith, that allow you to customize detection parameters to an application, can you think of other possibilities in the rail industry where iDAR can enable the industry and the professionals working in it outside maybe safety or autonomy, outside detection when you’re moving to detect a vehicle or pedestrian, are there any applications that you can think of?
KD: We are in the rail industry and we’re moving a train on track, so let me take you on a little bit of a journey: We’ve got a train that starts and it’s going down the track, and I might want to inspect the track in front of me. I want to inspect the vegetation that ties the ballast. If I happen to be in an electrified passenger train, I might want to check the catenary wires overhead. So, as I go down the track, and many of these things are fixed objects, I can, through software-definability, match that scan pattern to the geolocation point that I am on the track, being able to do that. I might also know because of safety regulations where there’s a work gang working on the parallel track next to me. So, to the point you just made earlier around the 200 meter or more localized scan pattern, I might want to look at geo-fencing, so to be aware of a group of workers that are maintaining the track as I go on my journey. Next up, I’m going to come across a bridge. And today, of course, in many cases, it’s a very potentially dangerous thing to inspect a bridge. Obviously, they do it with safety in mind. And so, you would inspect the front of the bridge, the sides of the bridge, the underneath of the bridge.
There’s a couple of ways I can do that. I can do that by the train going down the track and I can scan all of the things that are available. And I can use both the camera and the software-definability of this to target specific Regions of Interest, being able to do that within the Field-of-View. Now, underneath the bridge, however, I couldn’t access that by the train. So maybe I make a decision that I’m going to use a drone and I’m going to put AEye’s capability on that drone so that rather than slinging an individual and a worker underneath to inspect, I can now use intelligent LiDAR as a way to take that drone, bring it under the bridge, inspect for corrosion, wear, maintenance, anything that needs to get done. So now my train continues down the track and I’m going to the tunnels. And so, I can now use the intelligent LiDAR within the tunnels. I can pick my Regions of Interest, Fields-of-View. And by the way, those can be changed because maybe when I went through the tunnel three weeks ago, there was an area — and I always like to use the idea of a cavity at the dentist where the dentist says you don’t have a cavity right now on this tooth, but we’re going to watch it — and so maybe three weeks ago when my train went through that tunnel, we’re watching a particular Region of Interest, and now I will be able to inspect that as I take that train through that area again to ensure that either identifying work that needs to be done or work that doesn’t need to be done. Now, remember, as I was traveling fast, 600 meter range, I’m approaching a station, I want to inspect the tracks in front of me. I want to look at the platforms. Any of those Regions of Interest. Those can change as I get closer. But, within the station itself, I might want to track passenger flows. I might want to have a LiDAR pointing outwards from the center of the station, but maybe pointing inwards from the ends of the station. Are the stations too crowded? Are the platforms presenting a dangerous situation? Can I integrate this with a business process to be able to set off a series of alerts or attentions to management or these customer service people being able to do that? So now I’m stopped at the station, but I have to stop because we’ve got maintenance being done or inspection of the track. And there was recently autonomous trains inspection, and there have always been dedicated cars that do ultrasound or LiDAR, but now they’re automating this through the evolution of the technology into form factors and power consumption and the ability to integrate cameras and LiDAR and again, being able to define based upon where you are on the track, what your Regions and Interests and Fields-of-View would be.
These trains are now running as part of operating trends rather than dedicated trains themselves, which always cause challenges because you can’t run an operating train behind one of these due to the speed with which they run. And so now we’re starting to see more and more of these autonomous trains starting to be adopted. And to me, the software-definable nature of this — and integrating the business process, geospatial locations from GPS devices — really allows, and again, I think the automating of the scanning technology — the detect, the acquire, the classifying and then the actionable information that I talked about earlier right into the business process. Now my train’s going down the track again, and this is an area where maybe there’s a lot of vegetation and of course, LiDAR can see through the canopy and so I can look to see if there’s anything that I should particularly be aware of. Maybe if there’s a grade crossing coming up, and if I’m a high-speed train, I have the ability to stop at a shorter distance. So, I might actually look at 600, 500, 400 meters of range to scan when I’m going to the crossing.
But maybe I could do less. It doesn’t really matter from a safety perspective for a freight train because of all the tonnage that they have. They would obviously want the longest range to be able to do that because the time it takes to stop the train is much more difficult in a shorter amount of time to be able to do that.
Now, I’d like to talk about two other things, because you had mentioned some use cases.
There was an initiative a number of years ago, I believe it was 2008, it was the Rail Safety Act, but it talks about Positive Train Control, and what Positive Train Control is, is the ability to remotely slow or stop a train. Now, in order to do so, you have to know what the locomotives are, the number of cars on that train. But equally, if not more importantly, you also need to know what the terrain looks like. So how do I map all of that digital corridor that I talked about earlier: the ballast, the ties, obstructions on the side, the rail itself, the grade, vegetation, anything that might affect the algorithms that need to be developed for the braking of the train under the Positive Train Control Initiative, being able to do that.
The ability and the form factor today is that you could easily put this technology back on these operating trains, not just on those autonomous track inspection cars to inspect the track, but theoretically on the front of the train as well. And there were a couple of use cases that came out of China when we were there that we tried with cameras years ago that just weren’t possible with the technology, and that is using LiDAR on the front of a high-speed train and then communicating that data back to the train behind it and then scanning that and looking for any variations, comparisons for safety reasons.
And then, of course, finally we’ve got construction. I mentioned earlier when we talked about the growth of rail, as we look at high speed rail, the UK is doing high speed rail too — Asia, Africa — a lot of the developing countries are all looking at rail. They have a lot of passenger rail today, but it’s slow speed. But as they look at building out new networks, the ability to use LiDAR and intelligent LiDAR and scan patterns, all defined by software to match the patterns, to me, is a very powerful tool as we move forward.
AB: Thank you, Keith. What really caught my attention is that now you see, in the ability to change scan patterns, the possibility to address several applications and use cases in one. And the most interesting part of it is that you can do all this on revenue train. Not only can you use it for safety or maybe even autonomy, but I can use the sensor or the system, the iDAR system, because, of course, we’re talking about both LiDAR and computer vision, I can use it for autonomy, safety, infrastructure inspection — both the track and the top, maybe the wires, the catenaries — and also do roadside mapping.
This is really fascinating. It looks like there is a lot of work ahead. I want to bring up the fact that iDAR has both LiDAR and the camera bore-sighted so you get RGB and XYZ data, which, to your point, comes in really handy in these applications.
This has been really a great wealth of information. It looks like there are a lot of opportunities for us and a lot of work ahead. Really appreciate you joining us, Keith. Your expertise is really invaluable and it’s great to catch up with you today.
KD: I just want to thank you as well. I’m really excited about these technologies, and because, built into it is this ability to match through the software-defined capability to a business process, I think it leads to a much quicker adoption as well. So, the best of luck to AEye and I wanted to thank you for giving me the opportunity to talk about a subject I love a lot, which is the rail industry.
AB: It’s my pleasure. And thanks again, Keith. To all of you out there, thank you for joining us on AEye Insights, and we’ll see you next time.