 I'm Flavio Desmone as Julian just said and I'm going to moderate this session for you. It's about digital twins. Actually I'm the director of strategy and innovation at Anita, which is a consultancy focused on digitalization of transport and the panel that I'm going to introduce is the panelist are actually my colleagues in two projects which are GAMS and S-DOOM. And as you can see we have well three panelists today so we have Silver Lamee and Matias Router and Gottfried Almer. I will introduce them one by one. So Matias please come to the stage. Thank you. Matias is the S-DOOM project coordinator. He is the head of the Institute Digital of Joanne Mbrezart which is Austria's second blood research and technology organization. He holds a PhD degree in computer vision. And then let me introduce Gottfried. Thank you. Gottfried has more than 13 years of experience as a service and project manager for ASPINAC, Austrian Operator of Highways. He is also currently involved in the digitalization program of the Austrian Ministry for Climate Action and Energy. And then we should also have Silver which needs a minute for the microphone and he will join us soon. So I will tell you a bit more about him just in the meanwhile. I will tell you a bit more about digital twins while I give him some time. Okay, no he's here. So Silver, thank you for coming. He's the GAMS project coordinator and he has been managing international projects focused on geometric mapping for Drusat for the last three years and he has more than 10 years of experience in image processing and engineering. So lights are not on me today but on them. So I will now let them speak about digital twins, specifically about on digital twins applied to infrastructure planning including data collection, data integration and data users. So Matias, do you mind giving us a little of overview on Astrum? Please can you show the second slide, not the following but the next one. Thank you. The next one. Yes, this one. Thanks. Okay, thank you. Yeah, thank you for the opportunity to speak on this stage about digital twins of road infrastructure. So Astrum is an EU funded project and together with the GAMS it forms like a multimillion dollar investment of collecting data from road infrastructure and much more important bridging the gap to the end users. So making this data, making this information available to those users who can make use of it, for example, to use roads more efficient to create, to do maintenance of roads more efficiently and also to do driving on our roads safer and greener. I'll talk a little about Astrum itself. So Astrum is a project which has the goal to do road condition monitoring. This is a classical survey application, right? So you use like a mobile map or camera or as we have done it, very lightweight camera system coupled with INS inertial measurement units and GNSS geolocalization and we use those systems to gather raw data in the forefront. Then we transfer the georeferenced image information to a central data platform where it is further processed. The processing is basically an AI-based processing which then translates image data to like detections of road damages. Those detections can be like cracks, potholes, ruts, large-scale geometric deformations of roads and so on. And this information is then georeferenced and collected over time into a central database. But this is not the end of processing. Within this database, of course, we also are collecting then time series of one and the same road damage over time. And this time series information then allows us to do a prediction. So based on what we know about road damages from the past, we can predict how an individual road damage will develop in the future and when actual road maintenance will be done in a most ideal fashion. And this information then we're distributing. So it's not helping the database but our main goal is also to distribute this information. And one channel of distribution is of course making the information available to road operators like today we have Gottfried Alma here as a representative of one road operator but also to the driver on the road. And here Austria, especially ASFINAC is on the forefront of deploying infrastructure to vehicle connectivity via ITS G5. So we have those broadcast channels where infrastructure can talk to modern cars. And we use this broadcast channel to give drivers immediate feedback on how the road is shaped before they are driving on top of it. And how can you make use of that? You can then direct those drivers. You can on the one hand recommend like a lane change. If you have heavy trucks, you can like recommend do a lane change here and do not drive over all on the same road damage all over again and make it worse and worse and worse. Or if you have exactly geolocalized vehicles like equipped with high accuracy Galileo receivers with authenticated positioning information, then you can also recommend in lane position changes. So drive just a few centimeters to the left or to the right and then you can avoid individual road damages. And with this, I'm used the road more efficiently and post-tone maintenance operations. So this in a short glance is the purpose of ASRIUM. That's great. Thank you very much. Well, Silber, do you mind giving us a quick presentation on GAMS project? And please can you show the previous slide on GAMS just before this? Yes. No, before the previous one. This one, thanks. Thank you. Thank you. Thank you for the opportunity to talk here also. So GAMS project is a European project with a consortium of seven companies and one university. The goal of the project is to develop a prototype of a mobile mapping system. And our motto is robot mapping for mapping because we want to develop maps for robots, I mean for autonomous vehicles, but we want to do it with an autonomous vehicle. So that's why it's robot mapping for robots. And to do that, we have a different technology in our course on GAMM, including autonomous vehicle, of course, but the different sensors that needs to be embedded on the vehicle to collect all the geodata. And by that, I include mobile mapping systems and all the different cameras, sensors, and also the GNSS, of course. We include a new GNSS receiver that includes the Galileo technology. They have new features, recently, a new service opening at Galileo GNSS. And we include these features. It includes a better positioning and precision system, and also a technology, an anti-spoofing technology to avoid hacking. Also, we have different technology, including the fusion of all these different sensors. It's innovative as we include the statistical algorithm, of course, to fuse all these sensors and all this information, but also we include new information, like positioning inside camera images and also automator information to fuse all the information to have a better positioning system. This is the GAMM project. And we are looking now for a replacement for our autonomous vehicle company, because they just went bankruptcy recently. So to finish our project, we are searching recently to replace this partner so we can finish the project, because the autonomous vehicle is one important part of our project. Of course. Thank you. Thank you very much. Go for it quickly. Can you give us two, three minutes introduction about your point of view with ASPINAC on S2? Thank you. Can you show the following slides? Not this one, the next one, this, and the next one. Thanks. Thank you. I'm from Austria's main operator, main role operator ASPINAC, and Digital Twins is a topic which has come upon us. It's becoming very prominent. Actually, it's found us instead of us finding it. And the project Asriom is a key project to lead us along the way there. We've been using, in our minds, Digital Twins for a long time now, only we call them differently. For instance, GIS applications, where you can see the map of Austria here, where we have all sorts of data categories which can be displayed, which you can see on the left side. This is accessible to all ASPINAC people. And all sorts of information is digitized and is presented. So, next slide, please. If we zoom in here into one of these apps, could we have the next slide, please? Thank you. Then you can see, for instance, one section of the highway where there is, in this case, the inclination, the steepness of the highway is displayed. So, you can see orange is on the bottom, that's 3%, and then it gets red because it's already 4%. And if we click another box on the left side, next slide, please, then on the same highway stretch, you get displayed all the metal signs. So, now, the next slide, please. We are proceeding from this type of static information to the next steps with ASRIO. So, the first of challenge we have here is that we have to make this whole thing real-time. So, like in the example before, if you have variable message signs, variable metal signs, that you could get the information within seconds in this digitized format. The other challenge we have is that to fill data into a data system like that, you have to connect nationwide technologies and processes for the whole thing to work. So, this is what we're doing with ASRIO. As shown in this project, ASRIO could relay EGNSSRTK information, which is this correction of positioning so that it can get well under one meter of position accuracy and feed this data into an infrastructure system which could provide this to the vehicles. And in this way, build applications to further planning we have to get to a better planning even and maybe to output safety-related messages because of some road damages ahead for vehicles. To illustrate this, and next slide, please. I want to show you how digital twins fit in here. So, this is what we did in ASRIO. We have this system right on the top. It's a nationwide system which gives you the correct correction data for any point in Austria. And this was fed into our ASFINAC CITS into this, again, nationwide system which provides traffic data to vehicles passing by. And with this, into a prototype digital twin, we output a few messages to see if the whole thing works. But you can see that nationwide systems are here connected on a bilateral basis where we actually want to get to. And last slide, please. If we could have the last slide, please. Thank you. We would actually like to replace the system with an ordered system where you have all of inputs going into their appropriate layers and all of the applications which use this data, they take it from specialized interfaces, standardized interfaces, and can use it for applications. So, this is our mindset where we want to go to and what we're using ASRIO for. Thank you. Thank you very much. So, Matthias, so, I mean, we know that ASRIO requires many components to work together in an integrated system. And what we are curious to know is what are the main buttons next to the system that you found and how you solved it or tried to solve the issues? Yeah, thanks for the question. So, within ASRIO, we already tackled a wide range of different disciplines because we needed to start with, of course, with roadside image capture and geo-referencing. Okay, this is a survey, survey application. Then with data transfer to a central data platform of the image data. Then with machine learning techniques to process and create road damage data. Then with distribution techniques to road operators. Then with connectivity, infrastructure to vehicle connectivity. And then, of course, also with autonomous driving functions because there needs to be some vehicle who reacts on incoming road damage data and drives accordingly. So, this baseline was already very wide. ASRIO as a project will close by the end of this year. And, of course, it comes so far within a project. The next stage will be a rollout phase of the whole system with multiple scanning units and also multiple, like, end customers. And as the main obstacles I see here, of course, now we have seen on the one hand, it's, of course, the agglomeration of all this data within a central database and keeping the database consistent over time, especially when you have then multiple sources of data in the end. So, we are not just restricted to our own camera system which can do this. We can use basically every scanning system which creates stereoscopic image data of road surfaces. And based on this information, we see, like, establishing compatibility, of course, adapting the machine learning techniques, but also data transfer is still as one big obstacle because we need to bring this information from roadside vehicles in very short time into our central data platform. Thank you. Thank you very much. It's interesting to see how to deal with many different stakeholders and how to integrate data as best as you can. Silver, I mean, what have been the biggest challenges for you in GAMS that you had to deal with and how you dealt with them and what you feel was the main thing to solve and how we solved that? Thanks. Yes, we have many challenges in GAMS project. I would name two challenges. Right now, we have, of course, the fact that we want to use and to exploit GNSS Galileo, new features with improved accuracy and hand spoofing strategy. It's really important for us. It's the key for our project and it's a new service that is difficult to exploit. We are the first to exploit this service. It's complicated, but we are succeeding on it. And also, the goal of the project, GAMS, is to build an HD map, of course, for public infrastructures or for private car companies developing autonomous vehicles. And the task of building the map is really complicated. We have to use artificial intelligence to extract information and to classify information like road marking, any kind of road marking or everything, every object that we can identify on the road. This is a task that we do using artificial intelligence. And after that, we have to generate a map that can be used by users. So we have to identify which standard as an HD map is really interesting for our users and convert to the standard. This kind of information. This is also a big challenge for us. Thank you. Thank you. It's very interesting to see your point of view on the challenges for GAMS. Goffred, I have a question for you about your opinion as a road operator about the actual importance and the benefits of the HD maps for road operators. So how do you think they can actually bring a benefit to you as a Svinag? Yeah, thank you. Well, mapping, of course, is something important for road operators. And what we've been seeing all these years is that one project follows the next and parts of it are taken along, other parts somehow get lost after it's been implemented. So with this HD map in a digital twin approach, we're hoping to get the data of these specified projects into a central space where it is not lost, where other people can build things, can connect data which we can use for planning or for warning vehicles on the road. Okay. Thank you. Matias, what have been the main challenges that made a feature for S-Trum? With challenges, you mean during the rollout? Yes. Yeah, so now we have mainly, so with Investor in Project, we have partners from five nations of Europe and from those nations, of course, we could get good support in how they do road operations. And even there, we learned, okay, this is not harmonized over Europe. So every country and even every country, every contractor in an individual country has different rules on how to maintain his roads and how to, like, do maintenance and do operations and how to interact with the drivers they're driving on. And following that road, we had many discussions then with additional road operators, and we have seen that if we are to envision like a Europe-wide rollout of such a system, like to integrate also surveying information from, which is gathered in different manners from all over Europe, but then also to provide the correct information for each road operator, we need at least a minimum standard of harmonization of those individual, say, road damage catalogs and way of operation. And of course, then on top of it, we need individualization. We don't want to change road maintenance because of our digital solution, but the digital solution has to adapt to individual needs. So therefore, also the machine learning techniques and the output, the desired output, the standardized road damage information needs to be individualized. And this is one, still one big challenge then ahead on the way to a broader rollout. Yeah, I agree. There is still a lot to do in terms of regulations and how to deal with these data, common to many different countries with different regulations over Europe. So it's a problem, definitely. Silver, I was curious to know, you know, I mean, data collection, of course, can be challenging sometimes because of different locations. For example, when you have to collect data inside tunnels or when there are roadworks and GPS or sensors may not always work. So I wanted to know how did you manage to actually collect data also in these difficult situations, in these difficult locations? Yes. Yes, it's part of the project to be able to map, to operate our mobile mapping and to build our map on every location possible with sufficient accuracy for our users. And by that, I mean less than five centimeters precision. So for dead rocketing areas like tunnels where GNSS information is not available, we count on our innovative multi-sensor fusion algorithms that will embed any kind of new data like automator information, also the vehicle activation and also camera images embedded on the vehicle. So we can have a lot more information to fuse and to analyze. So we can have our good positioning in between two signals of good GNSS data is received. Okay. Thank you very much. We have three minutes left. So a last question for you, Gaufré. It's actually about the integration of data on a U.S. road operator. So basically, how will you integrate these HD maps into your application portfolio as a road operator? Well, I can make that short. That's what I should try to show in the slides, that we already have a data system with many layers, many faceted layers, and we're just going to integrate that into this system. And then step by step, then we'll try to get this real-time, and that's how it will proceed. It's a natural process. So you don't see it as a complexity of the process. The complexity lies in the data sources to get nationwide systems to connect in real-time to your own digital twin. Okay. So maybe we have one minute or two minutes for a question from the public. If there is any questions, you can read to ask. No questions? Okay. Fine. Okay. So I may have another question for you, Gaufré, actually, about the benefits that you think that the users of road damage data can bring to Asfinac for the long term. Well, of course, you have to see as a maintainer that you're constantly maintaining your roads. And this, there's a lot of money inside of this business. And if you can shorten the periods you need to repair a road work, it's simply money you win. We are also trying, of course, to warn in the safety issue, which is one of the highest issues we have to warn vehicles if there really is some damage which we haven't been able to ward off yet from traffic. So, of course, we always have the safety issue, but you really have to think about the millions of euros that go into the maintenance of roads, and you have to try and optimize that all the time. So it's obvious that a system like that will simply bring us to another level. Thank you. Well, I'm glad I heard positive feedback about these projects and how to improve through digital teams, the maintenance of roads, and to, I mean, I'm looking forward to see the progress of these. Thank you all.