“We are very interested in how machines, without necessarily a lot of human intervention or minimal human intervention, can actually do this negotiation themselves to allocate who gets access to various spots within the center for loading and offloading.
In the maneuver, it is getting from the gate to that location safely and with respect to the limitations of any of the vehicles: their turning radius, their ability to back up and do particular kinds of maneuvers. It ensures that all of that happens efficiently and safely. That is our application that we’re working on now.
It’s this tight maneuver transaction, smart transaction between relatively autonomous trucks and other vehicles that ensures all the stuff gets moved as quickly as possible. It is subject to the urgency as dictated by the supply chain and by the ability of the individual entities that operate these machines to place a value on what it means to be first in line that day or to getting access to a specific loading or unloading zone.” -Dr. Karl Wunderlich
Full transcript
Francis Adanza:
Welcome to the Down the Freight podcast, where we sit down with transportation, logistics and supply chain subject matter experts to discuss digital transformation projects. I’m the host of the show, Francis Adanza, and it’s a pleasure to welcome Dr. Karl Wunderlich, Director, Surface Transportation Division for Noblis.
Karl, it’s great to have you. Can you please tell the listeners a little bit about yourself, your company and what you’re responsible for at Noblis?
Karl Wunderlich:
Sure. Thanks, Francis, and thanks for having me on the podcast. Yes, hi, everybody. My name is Karl Wunderlich. I do work for a company called Noblis. We’re located here in the Washington, D.C. metropolitan area. We may be a little bit different from other folks that Francis has had on his podcast in that our company is a not-for-profit, and we support the USDOT and other federal government agencies in a number of areas that relate to emerging technologies and how they apply to a variety of missions across civilian and non-civilian space, including, of course, the transportation and logistics.
My background is in applied mathematics. So I come from the University of Michigan originally. Well, a few years ago, I’m not sure you can tell how old I am on this podcast. But in any case, it’s been a few years since I left the University of Michigan. But I’ve spent my career in the area really trying to look at how emerging technologies can transform or take on some of the real wicked problems that we see in the service transportation infrastructure. And in that role, I have a couple of different hats I wear for Noblis and it includes being the director of our Autonomous Systems Research Center, which was recently set up. And then for the last 5 or 6 years, I’ve been the director of our Surface Transportation division supporting primarily the DOT.
Francis Adanza:
Awesome. Well, thanks for that overview, Karl. For those who are not familiar with Noblis, can you provide a little bit more of what your company does in regards to the transportation and logistics side of the business?
Karl Wunderlich:
Yes, absolutely. Because as I said again, Noblis is a bit of a different kind of company. And our engagement with transportation and logistics is fairly unique. So again, as a not-for-profit and supporting agencies, we don’t offer products, particularly to the market, and most of our support to these agencies because we are looking across the board. More like a think tank role, honestly, looking at how these technologies could potentially impact or transform things in transportation, making them more efficient, more productive, more cost and fuel efficient but also safer.
So in our role, we support an office called the USDOT Intelligent Transportation Systems Joint Program Office, which has a connection to several other global entities within the Department of Transportation specific to freight logistics planning as well as just general utilization of the surface transportation system. And so we look at how things even as simple, for example, if I go back to early in my career when it was an attempt to try and get probe data to identify hotspots and congestion along major interstates. That was a big goal in the ’90s to actually provide these kinds of things that estimate travel times accurately. And so we were at the forefront of trying to position the government in a way to facilitate the provision of that information collectively to the transportation network as well as the roadway operators like state DOTs who manage the interstates.
Francis Adanza:
Well, it definitely sounds like that think tank has been keeping you busy. More recently, is there a specific or current project that you’d like to share?
Karl Wunderlich:
Yes, absolutely. One of the areas of specific interest that we’ve been taking a look at is blockchain technology and what it could mean, honestly, to the transportation system. And so I think when I say blockchain, folks on they may be listening, they merely would be thinking of like, oh, cryptocurrency, cryptocurrency billionaires and bitcoin and all that kind of stuff. And that is true. That is one application. Cryptocurrency is the most well-known and most successful, honestly, application of the underlying blockchain or distributed ledger technology that runs it.
But it’s not the only one. Just to understand a little bit more about how blockchain could have a more even larger impact, honestly, on transportation. But just merely a reference to things like bitcoin, it may be useful just to think a little bit about what it is fundamentally, what distributed ledger and blockchain does. And honestly, it’s really a marriage of a couple of different technologies, including cryptography and sort of the shared notion of a distributed ledger that gives a lot of power outside of use cases that are just limited to things like cryptocurrencies and exchange of value and not using, for example, Fiat dollars or euro currency.
And so in this case really what blockchain is, the simplest way to understand it, maybe the one I’ve used to try to explain it. It’s like a big shared spreadsheet and which everybody in the community that uses that spreadsheet to record transactions can monitor and look at. So it’s a complete ledger. Every entry is a transaction between one or more parties that acts as sort of a smart contract between those 2 entities. When the work is done and is complete, then value is exchanged. Sometimes it’s value, but there can be other kinds of tokens as well that relate to essentially the trustworthiness of the entities. Essentially, like you get a report, for example, an Airbnb host or something, right? The same kind of tokens can be used to not only exchange value but also rate the reliability of that other entity as a partner in the exchange.
So these kinds of spreadsheet-style capabilities are all excellent. But the problem is how can you trust if that’s secure? And so this is where the cryptographic component comes in. Because it’s dangerous, I suppose, to have a large distributed ledger that records high-value transactions that include exchange of value or ratings about trust. If it can’t be actually reliably trusted itself to be an honest record of what happened and if it could easily be spoofed, then there’s a problem. So the notion that cryptographically is that we use hashing techniques and other cryptographic methods to ensure that all of the information that’s encoded in the distributed ledger is actually irrefutable.
So the point is that if anybody goes in and changes any entry in our jointly shared spreadsheet, it’s a mental model here so it’s not really spreadsheets, then that causes essentially everything in the entire spreadsheet to change value. So anytime somebody tinkers with one small part of our ledger, that instantly changes essentially everything in it. So the problem that we’re trying to solve here is that when one person makes one change and everybody has multiple copies, I guess that’s what it said, it’s distributed ledgers. We all have our own copies of it. If one ledger looks different than the other ledgers, the shared ledger, then those ones that are different are overwritten. So the past transactions are inherently preserved going forward.
And as long as there is actual attention paid to that ledger, as long as the folks who are using it are paying attention, it’s fairly straightforward and very secure to keep those transactions going forward. This is essentially the basic technology in a simplified way that we use to look at high-value cryptocurrency transactions using bitcoin. But this fundamental underlying capability could be about any kind of relationship, any kind of exchange, any kind of agreement between humans, humans and machines or machines.
Francis Adanza:
Thank you. That was super helpful. It was like a quick crash course on blockchain. So as you think about blockchain and its potential use cases, what was the problem that you were trying to solve as it pertains to logistics and transportation?
Karl Wunderlich:
Yeah, right. So actually, there are a lot of things that folks can use or think about using blockchain to do related to the transportation system and the logistics supply chain. And the most prevalent use of this relates to chain of custody. So for example, the goods that are being moved within the supply chain are being handed off from one entity to another as a transaction, right? And so those transactions can be embedded into the distributed ledger the same way. And so there are a fair number of relatively mature emerging but relatively mature capabilities to do this. And there are some great use cases where companies have used these sort of smart logistics planning, smart contract capabilities to improve the management of their own supply chain and the chain of custody for high-value cargo. So that’s really a great area.
But the area that we are specifically researching here at Noblis relates to machine-to-machine and machine-to-human transactions related to collective maneuver planning. So in this case, autonomous or highly autonomous vehicles using a blockchain construct to enable tactical maneuver planning in and around areas where there may be contention for space. So this could be as simple as an intersection on the roadway or as complicated potentially as any as managing access and movement within or around a distribution center shared by multiple vendors or utilized by multiple vendors who are needing to get in essentially to particular loading locations or other aspects of logistics and supply distribution center.
So in these cases, we are very interested in how can machines without necessarily a lot of human intervention or minimal human intervention actually do this negotiation themselves to allocate who gets access to various fronts within the center for loading and unloading but also in the maneuver of getting from the gate essentially to that location safely. And with respect to the limitations of any of the vehicles, whether turning radius or the ability to back up and do particular kinds of maneuvers, ensuring that all of that happens efficiently and safely, that is really our application that we’re working on now. It’s this tight maneuver transaction, smart transaction between relatively autonomous trucks and other vehicles to ensure that all the stuff gets moved as quickly as possible but is subject to the urgency that’s dictated by supply chain and by the ability of the individual entities that operate these machines to place a value on what it means to be first in line that day or to getting access to a specific loading their unloading zone.
Francis Adanza:
Wow, that’s pretty amazing when you think about the development of smart cities and then all the autonomous vehicles that are running through it. As you work with clients on developing this solution, what were some of the interesting things that you’ve learned in the process?
Karl Wunderlich:
Right. So what we learned is that there’s a lot of use cases out there where machines and people have to share space, right? So when you think about the supply chain, logistics and freight movement in general, you can think of all the places where you actually need or adjust that space to conduct the transaction or undertake the transaction of moving the materials, right? And that’s not always straightforward. In fact, there’s some very interesting use cases going on right now as we’ve been starting to talk with folks about this in a more serious light related to just curb space management.
So for example, in a place like Manhattan, it’s very difficult to load and unload because when you talk about a place with space at the premium, Manhattan is a great example. And so the notion that vehicles have to block traffic is accepted essentially as a way of life there in order for resupply to occur. But there are ways, when one truck does it, it may not be such a big deal. When you have multiple trucks all vying for positions at the same time, then there could be both safety concerns as well as just accuracy in terms of just accuracy of delivery and efficiency concerns. So in these cases, there are essentially ways of auctioning off the space, this curb space in a very intelligent way to allow the unloading and loading to be conducted safely and also with a minimum impact to surrounding traffic.
All of that is essentially started in a ledger. And it’s just an agreement among, for example, the city and the vehicles that use that space, that curb space to do this. Honestly, the curb space management, it goes beyond just for truck loading, unloading. It also relates to passenger pickup as well. So if you can take that same multiple use of curb space and then use a mechanism like a distributed ledger, a blockchain technology to secure it, you can have people making reservations, acquiring that space, understanding and communicating to the rest of the community about the availability of that space in a way that makes it safer and more efficient and more predictable for everybody trying to use the space.
Francis Adanza:
Fantastic. So where are you in the development of this concept? And when do you think this will reach mainstream adoption and will start seeing autonomous vehicles in smart facilities and/or cities come to life?
Karl Wunderlich:
Right. So Francis, that’s a great point. Our job at Noblis, we are asked to think out 5, 10, 20 years. But parts of what I’ve been talking about are really more near term and other parts really are further out. So in terms of the planning for detailed maneuvers among autonomous vehicles or highly autonomous vehicles, I think this is still a little bit further away because we have yet to really get to the point where these vehicles are utterly devoid of human control. So drivers are still supervising how those machines and those vehicles operate. And that’s okay for now. I think that’s appropriate at the level of our technological capability and the offerings on the market to do so. But we are increasingly moving towards additional assist.
And one of the areas, honestly, where automation can come into play even earlier than somewhat general driving around, let’s say, in New York City where, for example, that would be a tough ask for an autonomous vehicle, given the pedestrians and the fact that there’s multiple lanes and that there’s cross traffic and a lot of other stuff. It’s a very difficult environment for an automated vehicle to really do well.
But one other thing I’d say is that in and around an area like a distribution center, then you’ve got an example where vehicles are moving relatively slowly. It’s a controlled space. It could be pretty good markings in that area to delineate where traffic could go or should go to channelize things. In any case, this sort of low speed relatively controlled environment makes a pretty good test bed for these kinds of tactical maneuver planning. And so I think in this case, the combination of blockchain, which is really as a technology ready to go, there’s nothing really too complicated that needs to be done in that arena. That technology is mature and ready for application. But integrating that with the machines and the autonomous vehicles that maneuver in these spaces, the distribution center is a great place to start.
So this is an example where I think even without necessarily autonomous machines or even just managing human-driven machines and the reservation for those allocated space could begin slowly. And then as the machines themselves, the vehicles become more and more automated, it sort of builds up. So right now, with just the access of any interest in autonomy or autonomous vehicles, you could have a blockchain solution managing access to some of these shared distribution center spaces, getting more efficient, more safer interactions, more predictable access to resources within the center that are maybe limited. I think that could happen now. So really tomorrow is a time where we could be using these kinds of technologies to manage the access, and I think there has been some movement in that area.
But as autonomy comes along, which I think is maybe for this kind of low speed autonomy at 3 to 7-year growth rate, then I think we’re talking about a pretty steady transition from what was an important thing for people to do, something that machines can handle on their own. And in the end, the humans will just really be managing the strategy and less the tactics of maneuver right in these low speed areas. So I feel like in this case, the specific use cases that we’re working on, by the time it’s not terribly far away from being able to implement something that is not only useful but grows into something relatively futuristic rapidly.
Francis Adanza:
That makes complete sense. I mean, you touched a little bit on how adoption could pick up a little bit quicker in, say, a smart facility or a yard that a shipper or a retailer might have in terms of a controlled environment. Are there any things that you’re doing with like the DOT that would prevent the private sector from being able to adopt these things much sooner within their own controlled environment?
Karl Wunderlich:
Right. It’s a great question. And I think when we talk about regular roadway intersection interstate space, we’re talking about relatively high stakes environment with severe safety implications and a shared infrastructure that can have anybody on it, from a highly automated truck driven by a professional with years of experience, say, supported by a whole range of automation and driver-assist technologies on the same roadway as a Model T, an antique vehicle from 100 years ago, still out there, right? They’re open access on the interstate or any of the other roadways as well as pedestrians and cyclists and everybody else.
So we have a general purpose infrastructure and a huge variety of folks and machines operating in that system. It’s a very difficult environment to get it right all the time because some of these outlier use cases can be very dangerous. So for example, when the artificial intelligence in some of these autonomous machines encounter something it hasn’t seen before, it can’t predict exactly what the reaction may be. So in order to make that system very safe, the required testing is super high. Now in contrast, the controlled environment of the distribution center is an example of a space where access is restricted. There is only a certain collection of folks who are allowed to get in.
Access is something that you have to earn, honestly, to get in. You can’t just drive up in your car and say, I’d like to get in here and drive around or walk around. Again, whoever is in and around these locations, it could be an intermodal facility, a distribution center, you name it, the access is by definition restrictive. And the space itself can be designed to support autonomy even as it’s being sort of built up in its capabilities. So it’s repeatable. It’s consistent. It’s a managed environment. There are a certain subset of folks who are essentially trained and understand the rules of engagement in that environment.
And that’s important for the human drivers but also for the autonomous machines that have a closed set of finite expectations. Because it limits the number of cases where we would be concerned about things like safety, where moving at relatively low speed also makes things a little bit more amenable to early adoption because the consequences of failure in this case tend to be lower than if a vehicle is going 60 miles an hour on the open road. So in these cases where the speeds are low, in fact, more importantly, access is restricted and the environment controlled. It’s an example where we can get started. From that environment, we might be able to take a lot of the lessons learned and get the machines trained up so they can take on the more challenging general purpose environment on the roadway system.
Francis Adanza:
Got it. Well, Karl, this is a fantastic conversation and lots of insight. Really appreciate your time. So what’s the next for you and/or this project?
Karl Wunderlich:
Right. So again, we are a little bit different than maybe other folks that you have on the podcast, Francis. We are a research and scientific and advisory company sustained by the government. But we are doing some very interesting research. And folks are interested in seeing some of these concepts played out, not on the roadway with full size trucks and vehicles. That’s still, as I pointed out, a little bit further off. But we’d like to see some of our autonomous rowers and drones doing some of this collective maneuver planning that I talked about where the machines are negotiating with each other without any human intervention, agreeing to a maneuver plan and executing and then using a blockchain to not only record the transaction as a part of the maneuver plan but also identifying which one of these machines fail to perform, for example, a maneuver and how that is incorporated into a different kind of interaction for that machine in following maneuvers.
Essentially, machines that can’t be trusted to follow directions have to penalize the next interaction or transaction by having to reserve more space to essentially accommodate for their inability to maneuver in prior cases, right? So this is a self-organizing system with feedback that says, those who are good, trustworthy machines in the system that perform reliably but actually get benefits from negotiation for space and future interactions because they’ve proven themselves to, in this many cases, use just less space, right? And less leeway means more space for other folks, other machines in the system.
Although, like I said, we have not done this with full size trucks. I don’t want to get anybody too excited. So you’re not going to see some 18-wheelers maneuver around each other. That’s not the video you’re going to see. You’re going to see some smaller rovers in our lab environment doing this kind of activity. I encourage you folks, anybody out there, to check that out and that can be looked up at our website. It’s noblis.org/orchestrated-autonomy. I’ll do it again, so noblis.org/orchestrated-autonomy. So check that out. Hopefully, I’ve verbally given you enough information. I guess you could also just Google search us on Noblis orchestrated autonomy. That’s also a really good way to find our resources and what we’re doing in the research center.
Francis Adanza:
Definitely. Well, I’ll be sure to keep you and Noblis on my radar and track the progress of this project as well as other initiatives that you’re working on. Again, thank you so much for joining today’s episode.
Karl Wunderlich:
Awesome. Thank you, Francis. It’s been a pleasure to chat with you, and I look forward to hearing more from all around the industry in upcoming podcasts.
Francis Adanza:
Thank you.