Episode Transcript
[00:00:10] Speaker A: Welcome to the EU Energy Projects Podcast, a podcast series from Enlida and France focusing on the clean energy transition for the European Union and the EU Commission funded energy projects that will help us achieve it. My name is Aretid Daradimu. I. I am the editor of the EU Energy Projects Podcast and your host.
[00:00:35] Speaker B: Hello everyone. In this episode of the EU Energy Projects Podcast, I'm joined by Daniel Albuquerque, senior project Manager at the R D Center at edp, who is going to talk about the Talos project.
Welcome Daniel, and thank you for joining us today.
[00:00:55] Speaker C: Thank you so much, Jonathan. Thanks for having me and for the invitation. It's a pleasure.
[00:01:00] Speaker B: So please start then with a brief overview of the Talos project and its aims.
[00:01:08] Speaker C: So Talos addresses a very concrete problem that we have in the industry in the renewable energy sector, which is the fact that photovoltaic plants, so PV plants are scaling fast, becoming more remote, and therefore its operation is getting very complex and at the same time that the operation and maintenance costs are only increasing and of course that safety constraints and the workforce availability presents nowadays a real challenge in which I believe that we are in a very critical and turning point phase within the sector.
So Talus as a project is designed to operate in this real world and very challenging conditions that we have on field.
And our ultimate goal is to develop autonomous robotic solutions and digital solutions to perform operation and maintenance tasks in PV plants.
And we will be demonstrating our solutions in two, three different scenarios, as we call them.
We have the ground mounted PV or the conventional PV that we see across the fields throughout Europe.
Then we have floating pv, so that's a different sort of photovoltaics that are PV panels that are being installed above water bodies, in this case in water dams. And then we have also another case which is Agri PV in which we use agricultural fields to also produce energy through PV panels.
And we are also automating different operation and maintenance tasks, I will just call them O and M tasks, which are the monitoring part, the inspection, the cleaning, vegetation management, and for the agri PV case, we have also the crop management parts that are being automated. This is a three year project, we have a budget of ten and a half million and the project is about to end in September this year.
[00:03:18] Speaker B: Okay, we'll go into some more details about the individual use cases a bit later. But what have been the milestone achievements to date?
[00:03:28] Speaker C: I guess that so far we have demonstrated a lot of things already, but I think one of the biggest milestone that we have achieved is the open call that we had In Talus, in the total budget that I've mentioned, we had around 2 million euros that was used to support other parties. So European SMEs and startups that later in the middle of the project, they joined the project to develop other than other solutions than the ones that were initially envisioned in Talos. So this Open Call was seen as a way to broaden the scope of the project and a way to enhance the ecosystem of solutions that we were initially developing within the project.
This Open call, in our opinion, was very successful. We funded 12 companies, 12 different ideas and solutions that were quite challenging as well. And all of them were already fully demonstrated and fully validated and we are quite happy with the results already.
So that Open Call is already one of the biggest achievements that we got within Talos.
And it's complementing a set of different solutions that we already have within the project.
So we are talking about more than 30 different solutions in just one project.
So we are talking about digital tools, robotic solutions, and all of them create this ecosystem, as I usually call it, of different O and M solution that we developed at the same time that we were using, for instance, since the very beginning, co creation in our technologies, especially the ones that were initially in the scope of the project. And that's also one of the biggest achievements that we had. The fact that we could co create the technologies from the very beginning is something quite unique in our experience.
[00:05:42] Speaker B: And what was co creation in this context? I mean, normally it involves consumers, but I think in this case it was slightly different, wasn't it?
[00:05:53] Speaker C: Yes, you are totally correct. Within Talus we have like a big consortium. We are around 15 companies throughout Europe and we have this big opportunity within the project since we have the entire value chain of photovoltaics in the consortium itself. So it makes it very easy to co create since this becomes quite natural because we have the the technology developers in the project, but we have also several end users within the project. And those end users were engaged from the very beginning in the technology development phase. So they help steering the development towards the needs and the pains that these end users face in field. So that's something that we believe it's also unique and that's why we use this. Co creation from the very beginning was something natural in Talos.
[00:06:51] Speaker B: And what differences have emerged with the different use cases? I mean, there must have been different challenges and how have they fed through to the differences in technologies?
[00:07:03] Speaker C: That's a good one. Yeah. The project is very ambitious in its core because if we selected just one use Case or one of the PV scenarios that we had would be a good challenge already to justify a project like this.
And we were tackling different use cases in different PV scenarios so that I think it gives us more or less of a scale of what we intended to achieve and how impactful we want to be within this project.
But I'll give an example. One use case that has a drastic impact on the robotic solution selected depends a lot also on the PV scenario that you're in.
So one of our use cases is PV panel cleaning, and that's different. It's a totally different challenge. If we're talking cleaning on ground mounted PV or if we consider the floating pv, it's a totally different solution that, that needs to be designed for the same use case. So since we, since we changed this PV scenario, the solution needs to be totally different.
And I'll give this specific example that our solution for the floating pv, it's something quite small in the sense of robotics. It's like an 80 kilogram robot that is more or less a large size autonomous vacuum cleaner that nowadays is very common in our households, like a ROMBO or something.
And this robot can be put on top of PV panels.
So for the floating pv, it's the perfect scenario. It's a robot that you put on top of the PV panels and they just go across the entire plant without much of a problem. And that has been working in a quite good way and we are quite happy with the solution.
But for the ground mounted PV which is like a utility scale power plant, we had to engineer a totally different machine.
We are talking about a cleaner that is a huge robot, a 12 ton machine, so quite big with very large brushes as well. So they could clean the PV panels and with the big water tank, so the cleaning is even more efficient.
So for the same use case, just clear example of the robotic solution needs to be totally different depending on the scenario that we are in.
[00:09:38] Speaker B: And obviously AI is an important feature of that. And how's that set up? I mean, presumably there has to be some sort of mapping which would be unique for each installation.
[00:09:52] Speaker C: Yes, that's correct. And even for the different robots, they also have some artificial intelligence associated with it.
But I think this will depend a lot on the use case. But I would say that in general terms we are talking about three different layers of AI within talus.
The first layer is associated with the data that we got, especially from the scada. So the monitoring systems of these PV power plants, we are getting real time SCADA data from the Power plants. And we have to have these AI agents to start digesting all this data and try to correct some of problems that we might face when we are collecting this sort of data.
And this AI layer also applies to the digital twins. For each pilot set that we have, we have a specific digital twin that was built so we could mimic the plant operation and try to anticipate problems on the different components of these PV parks. So we are talking about the panels themselves, but also the inverters and all those components. So we have this first layer of AI, then we have a second layer of AI that it's running on edge on the robots themselves. But for data interpretation, in this case, in the drones that are inspecting these PV plants, or even the robot that we use for crop management in monitoring, we are having this AI on hedge that is helping identifying, for instance, PV panel defects right away in real time, but also for the crop part, helping estimate the crop health and yield also in real time. So that's the second layer that we have in Talus, the third. So the last but not least, I would say that this third is it needs to be associated with the capacity of these robots to navigate autonomously.
And this per se, it's like a colossal feature because it's really hard to achieve. And we just need to just to give ourselves a sense of, of how difficult this can be.
We just need to think about Tesla, right? Tesla in the past few years is always, at least they were calling full self driving for a feature that is not doing full self driving. And they were always saying that these full self driving capabilities would be implemented in all their cars for the next semester or for the next year. But they are saying that in the last decade and they have not yet achieved that full autonomous navigation.
So that's just to give a sense of scale of how difficult this can be, especially in a power plant where we have different risks also associated with just the operation of this kind of assets.
So yeah, there are also other different challenges like the fact that these sites are very characterized by being remote places with very poor Internet connections. So having an autonomous robot operating there without a proper connection, it might be difficult even to justify it and to operate it safely. And these sites are also very featureless in the sense that all rows of panels are the same. So it's very difficult for the robot to understand in which row it's in. So it's really challenging. It's a true nightmare to work autonomously in these conditions. But yeah, we are trying our best and I think we are having very good results at the same time.
[00:13:50] Speaker B: And then what still needs to be done in the remaining months of the project.
[00:13:55] Speaker C: So so far most of the technologies that we are intended to develop, they were already pre validated. And I say pre validated in the sense that they were validated, but it just as a standalone solution.
And now during May and June, so the next two months we'll be demonstrating the whole ecosystem working together at the same time. So we'll be having demonstration activities in each of these PV scenarios. So in the conventional pv, floating PV and agri pv where all will be demonstrated together. So we are talking about a digital platform that I call it the digital brain of Talos, in which we can orchestrate and orchestrate our robotic fleet and see all the data that is coming from these PV plants in real time. At the same time, this digital brain has a recommendation engine that is fed by this data and it tries to plan missions ahead of time, taking into account, for instance, return on investment schemes, etc. And at the same time that we are executing all these robotic missions together with all the tech developers. So this is something that we want to achieve for the next two months and then until the end of the project, we'll be focusing on finishing everything. But the focus will be more on the techno, economic and environmental analysis of the project. So in every solution that we developed, we'll have these analysis being made just to clarify the pros and cons in each solution and then complement it also with specific exploitation strategies to each robot and to each digital solution.
So the partner could further mature the technologies after the project and have something that would be market ready.
[00:16:02] Speaker B: How do these solutions differ from others that are in the market?
[00:16:08] Speaker C: Most of the solutions that are currently in the market, they are not autonomous. So as I said before, this is where the difficulty remains.
And I think that's precisely where Thales shines, because we are achieving that with a lot of good partners.
They have been challenged by these constraints of real operations and they have engineered some good robots that could perform these missions completely autonomous. And that's something that we don't have currently in the market. So we are quite happy with this as well.
And the fact that we have a common platform that can orchestrate so many different robots with just one solution. So we are talking about more than 30 solutions that can be orchestrated by this platform. This is something that it's novel in the sense that it's not seen anywhere else. I believe every tech developer has its own platform, so sometimes this is very they are working in silos. This is very common. And and we are hoping to have this interoperability between all these technology providers. And I think that is paying off quite well.
[00:17:31] Speaker B: And can the technologies be adapted for other use cases in the energy sector?
[00:17:38] Speaker C: Yes, we believe so and I think we have some good evidence that might be the case, to be honest.
One specific example are the drones. So we have drones for inspecting these PV panels, but drones per se, they are very agnostic to the use case normally.
And for instance, the drones that are being developed for Talus for this photovoltaic use case in the energy sector, they have also been used for instance in the wind sector. The wind energy sector, but also in the military, for military purposes.
So they are very versatile. That's one of the examples.
The second example is one that we have in which a robot that was initially designed for a very specific use case was in the middle of the project repurposed to another one. So we added another feature. So another use case for the same robot. So it's a multi purpose robot that ended up been quite innovative. That's something, as far as we are concerned, unique in the market.
In this specific case, I'm being very vague, I know, but unfortunately I cannot reveal more information about the robot due to intellectual property rights. But I hope that at the end of the project we can elaborate more on this specific case.
[00:19:10] Speaker B: And you mentioned commercialization. And what are the plans to commercialize any of the technologies beyond the end of the project?
[00:19:20] Speaker C: Yeah, those plans are not yet finished, we are still working on them. But I would say that we have two different ways of potentiating the project results because in the project we have different partners, ones that are more research focused and another ones that are more profit focused, so to say. So for the research centers, I would say that their idea is to continue to be very disruptive while guaranteeing that the robotic platform is agnostic enough to be applied to other use cases. So that's more on the research area, not very related to commercialization.
Besides the fact that these disruptive technologies can also hint the market on the way they should follow and on the solutions that they could also pursue if they want to commercialize it for the safety that we have within the consortium. So the tech developers, I believe that the targeted solutions, the target solutions, are in a very good position to be commercialized in the future.
This is just a matter of now deploying these systems, having more maturity in their solutions.
But I would say that they are very close to markets.
[00:20:44] Speaker B: And do you see the technologies being provided as a service rather than, for example, an individual utility or renewable company owning a particular robot.
[00:21:01] Speaker C: That's a very good question. I believe both models are valid.
And for instance, even within the consortium we have this dilemma in the sense that EDP is the coordinator, but it's also one of the end users. But it's also true that RESH there is an O and M provider. So O and M is again operation and maintenance provider.
They are also one of the possible end users of the project, but it would be different. In fact, I think both scenarios are possible if it was like these solutions to be a service, it can be provided by rez. And I think they are very interested and willing to take these results to improve their offer to their clients. So I think they are very willing to do that. For edp, I think this is a matter of optionality. Even if it is a service or if we acquire our own, this is just a matter of optionality. To be honest,
[00:22:03] Speaker B: that was one specific aspect. And in general, how did the end users anticipate benefiting from their participation in the project?
[00:22:13] Speaker C: I think from the energy sector, my last reply was, I think it was referred to all the energy sector end users that we have in the consortium, both from the vision of the O and M provider, but also from the utility.
Of course, for us we want to test as many options as possible, especially on this novel sector of robotics. And how could these robotics enhance our general operation of the assets?
But for the agri food sector, I would say that we have two other natural end users, Wageningen Research and also certh. They work a lot on the agronomist side of things within Talos.
And I believe that for them it's very useful to understand this nexus between energy and food production, because this is far from being trivial.
But it is also true that they are understanding better how these robots can be utilized, for instance, in crop management in the future.
And we have another spin off from a university in Edencour that has shown very good results in predicting crop field using a sensor.
So this could be a good indicator on how further exploration could start and collaboration between these entities in the future.
[00:23:50] Speaker B: Okay, good. Well, thank you very much Daniel for your comments and discussion and we look forward to following to the end of the project.
[00:23:58] Speaker C: Thank you, Jonathan. Thank you.
[00:24:02] Speaker A: You've been listening to the EU Energy Projects podcast, a podcast brought to you by Enlit and France. You can find us on Spotify, Apple and the Enlit World website.
Just hit subscribe and you can access our other episodes too. I'm Aretita Radimo. Thank you for joining us.
