CyberWire Daily - AWS in Orbit: Monitoring critical road infrastructure at scale with Alteia and the World Bank. [T-Minus AWS in Orbit]

Episode Date: March 29, 2024

You can learn more about AWS in Orbit at space.n2k.com/aws. Baptiste Tripard is the Chief Marketing Officer at Alteia. Aiga Stokenberga is the Senior Transport Economist at the World Bank. We explore how Alteia and the World Bank are leveraging AWS's cloud, AI, and space capabilities to monitor critical road networks at scale to support large scale infrastructure investments. From road networks to bridges, they share real-world applications that are making a difference in emerging economies. AWS in Orbit is a podcast collaboration between N2K and AWS to offer listeners an in-depth look at the transformative intersection of cloud computing, space technologies, and generative AI. You can learn more about AWS in Orbit at space.n2k.com/aws. Remember to leave us a 5-star rating and review in your favorite podcast app. Miss an episode? Sign-up for our weekly intelligence roundup, Signals and Space, and you’ll never miss a beat. And be sure to follow T-Minus on LinkedIn and Instagram. Selected Reading AWS Aerospace and Satellite AWS re:Invent Alteia and the World Bank assess and enhance road infrastructure data quality at scale using AWS Audience Survey We want to hear from you! Please complete our short survey. It’ll help us get better and deliver you the most mission-critical space intel every day. Want to hear your company in the show? You too can reach the most influential leaders and operators in the industry. Here’s our media kit. Contact us at [e-mail] to request more info. Want to join us for an interview? Please send your pitch to [e-mail] and include your name, affiliation, and topic proposal. T-Minus is a production of N2K Networks, your source for strategic workforce intelligence. © 2023 N2K Networks, Inc. Learn more about your ad choices. Visit megaphone.fm/adchoices

Thank you for watching. Welcome to AWS In Orbit. I'm Maria Varmasis. We're working with AWS to bring you an in-depth look at the transformative intersection of cloud computing, space technologies, and generative AI. On AWS in Orbit, we're exploring not just what's possible, but what's meaningful in the realm of space and cloud innovation. We grapple with the complex challenges and unparalleled opportunities that arise when we use space to address pressing issues right here on Earth. Episode 3. Monitoring critical infrastructure at scale with Altea and the World Bank. Baptiste Tripard is the Chief Marketing Officer at Altea.

Iga Stogenberga is the Senior Transport Economist at the World Bank. We explore how Altea and the World Bank are leveraging AWS's cloud, AI, and space capabilities to monitor critical road networks at scale to support large-scale infrastructure investments. From road networks to bridges, they share real-world applications that are making a difference in emerging economies. My name is Ayagas Tokenmag. I'm a Senior Transport Economist at the World Bank.

I work on the Latin America region currently, but also support the global transport team on various knowledge products. I previously worked in other regions of the World Bank and the World Resources Institute and several other places on logistics, energy, transport, similar topics. Aiga, why don't we start with the problem the World Bank was working on and trying to solve, if you could set the stage for us there. Sure. I mean, so there's kind of the big picture problems we try to solve, which is reducing poverty and increasing shared prosperity. And now we have expanded our mission to also add on a livable planet. So the climate aspects are becoming much more central in everything we do in terms of both

resilience, which is much more relevant for many of the poorest countries we work on. And then in terms of decarbonization and transport is one of the main contributors to carbon emissions in many countries, and especially in Latin America, which is a very urbanized region. Transport is one of the main contributors. And so everything we work on has to be kind of through this climate lens, if you will. And so the World Bank works on two main types of work. One is investment lending, which is the bulk of what we do.

And infrastructure accounts for about a quarter of all of our lending. So both transport, but also energy and digital development. And so that means basically helping governments prepare projects, finance those projects, help implement them. This can range in transport. It can range from rural roads, interurban roads, ports, mass transit in cities, etc. And then the other type of work we do is really sort of like the knowledge creation and analytics. and analytics.

And so the problem that we face is usually in investment lending projects, it's the lack of data at the scale of a country that would allow us to advise the government in how to even prioritize those investments. Later on in project preparation, there's much more money going into actually collecting data, but at the stage of sort of just, what do we even need to do

to achieve the mission of producing poverty and increasing shared prosperity on a livable planet? Like what is, we really lack the data at the scale of a country on the quantity and the quality of the infrastructure assets that the country has. Thank you for setting the stage there. So now we have a good understanding

of what the problem is that we're trying to solve. And so it feels like a good time to bring in our second speaker, Baptiste, on how to go about doing that. Baptiste, why don't you introduce yourself? Hi, my name is Baptiste Trippard and I work at Altea as a CMO, Chief Marketing Officer. as a CMO, Chief Marketing Officer. And my background is in aerospace engineering, so everything related to aircraft design. That led me to design drones back in the 2010s, and now working in the field of software development and artificial intelligence. Thanks, Baptiste. So to start off, perhaps could you give us a bit of context on how you learned how aerial imagery could really change

how people do business? Back to the 2010s, I got involved in what's called kind of spatial computing now. As I said, when I started building commercial drones for different types of applications. And the idea was back then to start using small unmanned devices equipped with consumer cameras to take high resolution pictures from the sky. And it used to be and it still is highly valuable for many different types of applications ranging from agriculture to construction. But also, as you mentioned, some conservation projects.

One of the highlights of that time was to do a mission in Antarctica to take aerial images from penguin populations and understand from the sky how climate change had an impact on the way they all behave together and where they're going. And I think it's kind of fundamental for the topic we're discussing today. At that time, it really opened our eyes,

understand like the value of remote sensing for anybody really, from archaeologists, conservationists to larger organizations like the World Bank. larger organizations like the World Bank. And so what changed from 2010 to now is all the developments that happened in the field of geospatial and cloud technology. And what used to be available at the very narrow scale with the drone business is now available at a much larger scale.

Almost in near real time, you just click on a button on the AWS Open Data Registry and you access infinite amount of satellite imagery. And so you have kind of like that technology breakthrough, public initiatives like the Copernicus project in Europe to map with satellites the entire Earth every, I think it's every week, or even like private initiatives funded by investment companies

and fueled by private organizations like Planet or Satellogic that now launch in the sky satellites that are the size of a shoebox. So all of that kind of like connects to a point where data is much more available in real time and can be processed very easily thanks to AWS infrastructure. So you were speaking a bit about AWS, and I wanted to ask a bit about how AI also comes into the picture here and how Altea uses both AWS and AI. Well, and I'll start with that, you know, and Aiga highlighted it as well.

start with that, you know, and I highlighted it as well. Data or access to data is paving the way for what I would call a smarter and more sustainable future, right? So we have that very beautiful synergy between technology and the planet. But the problem is, you know, there is a notion of data out there. And this is really hard to transform that data into information, something that has a value for an organization. And because we have so much data available, it's not possible, or at least really difficult, both from an operational but also a financial standpoint,

to not automate some of those processes. And that's where I think the combination of artificial intelligence and cloud computing comes into play. AI is not a new technology. I mean, it was available already in the 70s. However, it consumes a lot of computing power. And it's really that combination that makes it so powerful today. And this is where Altea is standing.

And this is where Altea is standing. What we do is we allow organizations to produce real-time information from that ocean of data that is out there, leveraging both AI models and cloud technology. Excellent. Okay, I'm going to switch to Aiga now. So we've been speaking a bit at length about Batiste and Altea's solution. So I'd love to know how the World Bank,

kind of how you met essentially, and how you realized that there was a solution here that you both could work on. So in our case, there was a very specific ask from our clients. So there's sort of like a cascade of, you know, working together as being Altea's clients and then obviously the World Bank responds to governments. And in the case of Peru, which is one of the countries I work on, we are being asked to prepare essentially the next generation of rural roads projects or programs.

Peru has a long history of very successful rural roads investment programs that have had really significant impacts on rural poverty reduction, etc. But they're looking for a new way, a new sort of a season of roads projects that would be at an even bigger scale and that would have more robust prioritization criteria built in. And so this is sort of where the bold bang comes in because, you know, we don't just come with money. We're supposed to come with increasingly more knowledge, analytics. And for that, we really needed this at scale information on where are the roads and what are sort of what's

their conditions to be able to advise the government. At the same time, I was working in Mexico, which is, you know, another country where we lack that kind of data. And even though we don't have a project right now, the government is looking for technical assistance in helping to design an asset recycling strategy. So it's essentially advising how to use the road assets that the government manages at different levels. So it can be the federal or the state governments to generate revenues from those assets. I'm sure you're familiar with toll roads, but this is just one way that this can be done. But obviously for that to happen, you need to know what assets you're managing and even a country

the level of sophistication and income of Mexico honestly doesn't have that information at the scale needed and so we were basically looking for various solutions that we could think of and the World Bank

my engineer colleagues are very familiar with the traditional ways of surveying road assets. You literally drive with a car with a GoPro camera and another type of camera and you record the roads. And as you can imagine, it takes about a century if you're trying to cover the road network of Mexico or Peru. if you're trying to cover the road network of Mexico or Peru. And so basically we started looking around to try to find a solution that is leveraging satellite imagery, but doesn't cost what I've seen these kinds of projects cost in other regions where I've worked, where if you're thinking of leveraging high-resolution imagery,

this can easily spiral out of control in terms of, you know, the cost. And we were trying to find a solution that, you know, we could apply not only in Peru and Mexico, but also potentially sort of develop a blueprint that can, you know, the methodological blueprint that could serve other countries' needs as well and could be potentially scaled to even, you know, the world, all the clients where we work in. So I had some support from the global transport team in the World Bank, which is more sort of the knowledge generation hub and typically supports these sort of global knowledge initiatives

and basically created a coalition of people who were really looking for the solution. And we were able to... We were connected with Altea through the chief economist's office at the World Bank who had also been thinking about this topic more from the analytical and knowledge creation perspective because they are always looking for great data

and comparable data across countries to be able to do statistical analyses, how transport connectivity affect incomes or poverty, et cetera. And so it was kind of this lucky confluence of needs. And in the end, we were able to develop something that worked. But in the beginning, were able to develop something that worked. But in the beginning, I wouldn't say that we, you know, we were

buying a solution that was ready. We were sort of ready to work on the solution with Altea, knowing what they had in mind. It sounded very promising. And I think that that's what happened. It was sort of working together on the solution for a couple of months. Excellent. So from your perspective, how did the project go? I mean, I think the initial, my role was really to sort of define the objectives, what we really wanted to get out of it in the end, and broadly the kind of indicators we were looking for for the road networks. for for the road networks um and uh even though my background is not in road engineering as i still i think my role was expected to be a bit more on that as well like how does the world bank prepare projects what is the you know like should we be looking at road condition at the scale of

10 meters or is 100 meters the sort of um normal typical project scale that the World Bank would finance. So there are more of these practical questions that I could help answer. Also, one thing that I was really trying to help with was the finding of what's called ground truth data, which is essentially like data we know to be true underground and try to literally scrape the entire World Bank network to try to find teams that actually have that kind of data and try to help as an input for the models that Altea was developing. And then Altea did the actual work, of course, in terms of training the models and to see what makes sense and what would be sufficient detail for us to be able to use those results. Baptiste, from your perspective and from the LTA perspective, how did it go?

Yeah, so what we've tried to do, and I kind of explained it a little bit, but I'm going to try to do it from a more technical perspective, is work backwards through the problem, right? So what were the key constraints that the World Bank had in terms of project deployment? And in that case, it was access macroeconomic KPIs while minimizing the cost of such analysis. So it all comes down to finding the right balance between results granularly and costs of data.

We focused on building a solution that aggregates lots of heterogeneous databases that have one thing in common, you know, they're all freely accessible. We've especially leveraged low-resolution satellite data that we actually uploaded from the AWS Open Registry. And what we've done is aggregate all of that information together, contextualize all that information together, and process it together to build the required metrics. and process it together to build the required metrics. Actually, it's kind of as if we've artificially increased the resolution of the low-res satellite data by stacking so many layers of information to it.

And at the end, we've obtained the right amount of data for IGAS needs and the pre-project analysis. It does not solve all of the problems. I mean, you cannot really have a high level of detail with such data sets, but it was sufficient for the needs of the project. And, you know, we see often in the AI space people that try to use a sledgehammer to crack a nut.

I mean, honestly, in that case, the whole point was about being rational in terms of data cost, as I said, but also minimizing the data processing cost. So we've been working a lot on building algorithms that would only go and take the right amount of data because obviously, you know, energy consumption, computing costs are also very important in such projects.

So we've tried to optimize that as well. And to do that, we've leveraged our software platform called ETHER. It's kind of like what I would call an operating system for vision AI, so visual data and artificial intelligence. And all of the building blocks that are built within that platform were leveraged to do such analysis at scale. I mean, when we talk about a country-level assessment, it's between 5 and 10 terabytes of data for each project, so it's pretty massive.

And I'll finish by that. I mean, we highly rely on AWS infrastructure to do that. And again, none of this would have been possible like 10 years ago. And several components are necessary for that kind of project. So accessing all the data sets with the open data registry, leveraging also Amazon Elastic Kubernetes Service and the Elastic Computing for the orchestration and the analysis of all the data. And this is really important

because there is an orchestration of different tasks between the pre-processing of the data, the launch of the AI models, and then, as I was mentioning, the ground-truthing with data coming from the ground. So it's a lot of heterogeneous data that just navigate together but to achieve the result that we've had.

Let's talk about that result. So Aiga, I'll go to you about the impact. Could you walk us through a bit about what the results were? Just to preview the kind of World Bank, the timelines that we worked with at the World Bank are a little bit longer maybe than the private sector. So at this point, what we're really working on is sharing these results with the government, trying to make sure that the road agencies really understand how this was generated. It's not a black box.

We're trying to kind of open the hood as much as we can, explain the process, even though it's obviously a highly specialized type of analysis. But that's kind of the first step, is to really deliver these results to our client, which is the Peruvian transport authorities. And at this stage, we're really trying to narrow down the project. And once that will be done, we will be able to use those results in economic analysis. We've already done a lot of that

at the country scale. So using the results that were produced by Altea to overlay them with climate risk factors to see, you know know where there are overlaps between poor condition roads and high climate risk so that's like a first flag for us to to know that this could be a you know this should be an area we consider for the investment uh we use the results to um assign a much more you know assign much more realistic travel speeds on each of those roads. And once we have those assumptions on speeds layered on top, we can use this data to basically conduct accessibility analyses to local markets, to schools, etc. And that's another way or another further analytical results

that we can use to really define the investment project. I mean, that process is quite long in the World Bank. And I think in any multilateral development institution, the project preparation will be like a year. So we're somewhere in the middle of that. And then, you know, we will be able to build the road in, let's say, three or four years. And we're hoping for a scale of, you know, several hundred, if not thousand kilometers.

But I mean, this is not something we've, you know, we'll be doing for the first time, you know, in terms of the actual works and expected impacts. So in some other countries where I worked, I mean, we really see the impact on the ground in terms of, you know, improved school completion rates, especially for girls, reduced maternal mortality because women can now access clinics and not miss medical appointments. We did quite intense, intensive analysis of these kinds of impacts in the poorest countries where we work, so in Haiti and Bukhanafosso in my previous region.

And it's really incredible what a road that can be functional all year round and not be washed out by rains, what it means for connecting people to opportunities and keep kids in schools, etc. That's incredible just to think of the journey the data has taken and what an incredible real world impact it's having. So often when we talk about data that's coming from, that's aerial or from space, the impact seems a bit abstract. But this is very, very concrete and has quite an incredible impact. So what an amazing story.

Baptiste, I'm just curious, from your point of view also, can you imagine in other ways how what you've been doing, how that will have other real-world impacts? Or do you have any anecdotes that you wanted to add to what Ayaga just shared? Yeah, I mean, and let's go back for a second to what we generate. You know, what we create as an output of the workflow

we've been talking about is a digital road network model where each pixel represents, you know, truly what's going on on the ground. And we assess also the condition, so if the road is paved or unpaved, and the quality of the road, like ranging from poor to very good. So now, you know, a country can have access at very low cost

to a digital record of their own network. The analysis can be done over the course of the years to understand, you know, how it changes. And this is actually pretty rich information, especially in the context of climate change, to understand how your infrastructure is resilient to all the pressure that it receives from the elements. And I think that would be the first thing, you know.

Now you have a model that you can leverage to either understand how it changes over time by conducting the same analysis frequently or by running simulations where you digitally create an environment and see how your model is supposed to react to it. And all of that is really interesting. We've seen in many cases, you know,

people used to manage their infrastructure by being very reactive to things. And this is now a way to become proactive, kind of like building something to prevent an event before it actually happens. And that path is necessary if we want to build infrastructure that are more resilient

and more sustainable. So this is very hopeful in terms of technology enablement. And what I think is very important for everybody that runs such AI project, it's always layered to my opinion. So starting with something that already provides value

like that road network that we create and then add new layers of information, simulations or computing that will go more in depth and help different fields of work like transport safety, fields of work like transport safety, basically emergency response, or building bridges to avoid pressure of climate events. So this is how we see the future, basically building and adding more tools to that initial base that we've been building digitally. I wanted to make sure as we sort of start to sum up, if there was anything you wanted to add that maybe we hadn't discussed yet, or if you wanted to offer any advice or maybe even a reflection on the project that you have undertaken and any things that you have come away with. Any reflections you might want to share? Sure. I mean, I think Bajis kind of talked about some of this idea that this project is not like a static project. It can be dynamic over time as new questions emerge or new questions for us as the World Bank, you know, like the

climate aspects of, you know, like this is not something that we were working on essentially when I started the bank 15 years ago. And, you know, it's possible that in 10 years we'll be working on new things and we will need to add new lenses of prioritization to our project. So I think I agree that, you know, this is sort of the basis for which we can start, with which we can start and layer new things on top. And we're already thinking about expanding

this methodology in some other countries. So, you know, adding, again, leveraging free data to be able to say something about the road safety characteristics of roads, you know, not so much exactly where accidents actually happen, but can we predict where they could be happening based on the road features that we can extract from free satellite data? So that's just one example.

Another thing, I mean, just in terms of the kind of value that we gain from this project, so it's, you know, obviously it's going to be a really important input for us in project preparation. It also allowed us to save a lot of time and money in comparison to, you know, the other things we were considering, like these traditional surveys or even the more innovative solutions like drone surveys or obviously high resolution satellite imagery would have been completely unaffordable. In the case of Mexico, it also really, it was like a matter of just feasibility of surveying those roads. You know, parts of the country you cannot really travel to due to safety reasons.

And, you know, in many other countries, this is even more so, you know, that the remote assessment is really the only feasible way of doing this. So it's been, you know, a very kind of from all aspects, like a very high value proposition for us. And we're really in the process of disseminating this work to other teams. You know, many of my colleagues are really interested in using this in their countries. Yeah, in terms of the value, it's that.

It's the time, it's the money, it's the safety. It's the ability to also not only use this data selfishly for our project, but also share it with these more analytical knowledge teams that are, you know, this is super highly valuable for any kind of economic analysis to be able to, like Matisse said, over time track the impact of things. You know, for instance, once we have the road built or improved, you know, we can then do this analysis again a couple of years later. And we can, you know, scientifically measure what the impacts have been on various things we actually care about. You know, like these things I mentioned before, like school completion rates or poverty or the extent to which farmers now not only produce for their own consumption, but also sell at markets. You know, there's many different things we can measure. And I think that the fact that we could come up with a methodology

that is really affordable, that means that we can really do this over time. And as a tracking tool, some teams have asked me if there's a possibility of using this also to kind of even monitor the implementation progress of projects, which is sometimes also difficult, especially in the countries where there's like an ongoing civil war and the work sites are really hard to reach. And so, I mean, I think the possibilities are quite open in terms of what this could be. Yeah, I could absolutely see that. Many of the things you just mentioned, I'm blowing my mind a little bit at the thought of doing all those things and much more easily than in the past. Baptiste, I know you already provided really beautiful thoughts on the future.

I just wanted to give you that opportunity to reflect. Well, so one thing that I feel is going to happen is AWS is going to continue to innovate. We are going to continue to innovate. We are going to continue to innovate. Satellite image providers are going to continue to innovate. And, you know, we will be given the opportunity to work with more high-resolution data within the next month, years.

And what's fascinating is the level of granularity of the analysis that we'll be able to make at scale is going to probably be multiplied by a factor of 10 within the next couple of years. or 10 within the next couple of years. And I feel like we're only grasping the surface of all the things that we're going to be able to extract from remote sensing data. And that's actually very hopeful, especially when it's tied to development project

or conservation projects. Because we, as people from Earth, we will need that level of granularity to really get a sense of the course of our actions. And in that specific case, technology will be able to provide that. So I think that kind of project also paved the way to that type of innovation. And I hope it will inspire more tech companies to follow that route because, and this is also important, you know, there is a way to be financially profitable, leverage AI and have a positive impact. So I think

it's important that people realize that out there. And that's it for AWS In Orbit, Episode 3, Monitoring Critical Infrastructure at Scale with Altea and the World Bank. A special thanks to Iga Stogenberger and Baptiste Dupard for joining us today. For additional resources from this episode and for more episodes in the AWS InOrbit series, head on over to space.n2k.com slash AWS. This episode was produced by Alice Carruth and powered by AWS. Our AWS producer is Laura Barber. Mixing by Elliot Peltzman and Trey Hester, with original music and sound design by Elliot Peltzman. Our executive producer is Brandon Karpf.

And I'm Maria Varmasis. Thanks for listening. Thank you.