# Securing the Belgian North Sea Shoreline from Small Boats with Docked Drones at PZ Westkust - Event: NestGen ’26 - Published: October 2, 2026 - Length: 31:51 - Watch: https://nestgen.org/library/watch/v2Oo2-d7Knc/ > The transcript of a recorded NestGen session, for reading or for use with an AI tool. It comes from the video's > captions; automatic captions can mishear names and terms. ## About this session Public safety teams securing coastal zones face a different response challenge from fixed-site operators. They need broader visibility across open terrain, faster verification of incidents, and a way to respond before threats move out of range. In this NestGen session, Belgium Police will share how docked drones support shoreline security against small-boat activity, and what public safety teams should think through when deploying drone response across exposed coastal environments. ## Transcript **[0:00]** Hello everyone. How are you guys doing? How have the sessions been so far? So, welcome to another exciting session and hope you guys are having a great day. So, today we will be discussing about how the Belgian police is securing the North Sea shoreline from small boats using dog drones, especially at the Politie Westkust, which is a which is a police department in Belgium. And uh we have I think a lot of people joining in from different parts of the world. So, yeah, please feel free to drop in uh where you're joining from in the chat section. So, we have two chat sections. One is a session and one is a public one. So, in the public one you can chat with everyone from all of the sessions. And yes, uh without any further ado, let me also tell you about some of the housekeeping rules. So, this session is getting recorded and you will be able to find this later on on our social media handles. And please feel free to grab uh **[1:05]** key valuables or key insights from this particular session. Uh take the screenshots and tag #nation on LinkedIn. We would love to uh you to share your thoughts and tell us how uh do you think this was inspiring for you. So, yeah, uh without any further ado, I would like to introduce you guys to Christian. So, Christian Dedrie again has been a prima facie of the Politie Westkust. He's the deputy chief of police uh working over the last 34 years building up as a drone enthusiast and now also being a part of the committee for rolling out drone as first responder as well as drone autonomy related use cases for policing. And he has been closely working on some very interesting proof of concepts for some of the small boats problem, which has been a big concern for Belgium as well as as European nations. So, without any further ado, over to you, Christian. Uh I would love to hand you over the stage. And as mentioned, please feel free to drop all of your questions, where you joining **[2:06]** from, in the meeting chat. And at the end, last 10 minutes, we'll try to take up some of the questions which you can put up in the Q&A tab. So, over to you, Christian. And yep. >> Rohan, thank you very much. I'd like to thank Nithin and the whole Fleetbase team, especially you, Rohan and Kanshi, for the invitation and for the help to get to this presentation. I'll now screen share it, so we can start the presentation. I hope this is visible for everybody. >> Yes, it's visible. >> Okay. What I'd like to share with you today is a problem we were confronted with after Belgian coast. It is a problem from immigration, illegal immigration, small boats problems, which means that illegals coming from all over the world are **[3:10]** coming to France to stay in camps over there and try to make the transit to the UK. To make the transit, they used to use climbing to trucks and take their lorries to get on boats and get to the UK. But the new method arose last years, where they used small boats, inflatable ribs, which they entered and then did the crossing to the UK, which makes it very dangerous for them also. It was partly, especially in France, but the last years, they also came to the Belgian coast. And we don't have any response to that on how to tackle the problem. We try to fight it on three, uh, fronts. When they leave the highways, we have infrared cameras and we try to stop them over there if we know with which car they come. We know where are the hot spots and where they they drop material and personnel. So, there we try also to work with patrolling officers. But, the last front really is in the dunes and on the beaches, uh, where they hide and **[4:13]** where they inflate their boats to get, uh, into the water. And there we don't have any means, uh, because at night it's bit dark over there. So, we don't see anything. So, we were looking for a solution, uh, on how to tackle that problem. And so, we came to an proof-of-concept with a drone-in-the-box solution, uh, and with the software of Threod Base. We, yeah, tested everything in the period of June and August of this year. Um, Like you see in the footage, this is a footage, uh, of, um, June last year where we arrived just too late. Um, we get a call out from a person hearing, uh, people on the beach carrying a boat. He called the police. We went out, but by the time we were over there, the boat was already in the water and we couldn't reach them anymore to stop them. So, we were, yeah, looking for a method, uh, on how to tackle that problem. Um, and **[5:16]** then came, uh, the rise of drone-in-the-box solutions. Um, the way we worked, uh, in the past before we did the POC, um, was there were people at the police station. They get into a police car. They drove out to the site where, uh, we got an alert. And then we launched the drone on the site and we tried to see, uh, what was happening. We were streaming our footage live to the uh, police station. Um so we could get the footage out. Um as you can see, it is a a process that takes people time um to get out there. So we want to make sure it went faster, more efficiently. Um that's why we did the drone in the box. So it's not another gadget that we would like to implement uh the drone in the box. No, we are really looking at something that would really help us make the **[6:22]** more efficiently and make the decision-making uh more rapidly. The drone in the box gave us a faster situational awareness. We got the live image that we could share um with the dispatching, with the patrolling agents um and we also could deploy our resources in a more targeted way. We could confirm by the drone if it is really uh illegal people in the dunes or if it's only a tourist walking his dogs. Um And we could scale it also. So we could uh make it move from one dog to three dogs to more dogs. This is a a dog like you can see uh with police uh signature. So it's really recognizable for other people that this is a police drone and not just any uh drone. So it does help us with getting uh calls from drones in the dunes. We know it's our **[7:26]** drone. How did we do the Pocket Attic? We uh flew more than uh 86 69 missions, more than 33 hours. Um we all completed them. I think as you can see on the the statics. What was the important things we took away with our talk is that we had 24/7 sensors, we had AI detection, and we had integration in our ways of working in and our software we already did. What was very important for us also is before we departed with the drone, we always have to check we have to check if the weather is okay, if there's no other traffic in nearby zones to fly so we can detect and avoid. Why is this? Because we are at the beach side, lots of wind over here. We are also **[8:27]** very near to a military airbase. We are also close to a civil airbase in Ostend. We have a helicopter training area here. So it's always important to know when we fly it is safe to fly and we are aware of what's flying around us and the other persons flying around are also aware that we are there present. How did we fly it and what did the life missions show us? We did missions during daytime and also during night time. One of those missions was also trying to see how fast we could respond to a situation and what was the gain for our police force. So as you can see on a certain day we got a report of a dangerous situation on the beach side where there were children in danger because they were in a deep pit. **[9:27]** We sent out a drone at that location. The drone automatically recognized the crowds size estimation so that is something we put in the AI software. Detect groups of people. And as you can see when we were flying out there at the location, he already gave us the alert that there was a group of people over there. Um when we arrived at scene, we could see it was a a specific known location for us. It was a child's beach where they sell uh self-made flowers uh for children. Uh we could directly zoom in and see that the dangerous situation uh was not that dangerous. It was only uh one of the yeah, let's say sellers who had made a big pit and probably one of the other sellers was jealous that his pit was bigger and he could sell more. That's why they called the police. But with this incident, we saved uh patrol two patrolling officers that that didn't have to go out, that didn't have to check the **[10:30]** uh situation. So with this uh we could easily get more efficient uh patrolling officers. Another example was um bad parked uh car. Somebody called to tow away the car uh to our police station because the car was uh in the wrong place. The drone also could capture this immediately. Uh after a couple of minutes, we could get onto the scene. We saw that the drone was parked um on a private part where we cannot tow away cars. So we didn't need to interfere. No patrolling uh officer was needed on scene. But we tried to help the people. We could see the number plate. In our database, we could connect the owner of that car. And like that, uh we could easily phone him and say he has to move his car. Um so you see from normal footage when AI comes in, the difference is from looking **[11:32]** to interpreting everything we have. We could detect things, we could prioritize and we could activate and patrol and summarize afterwards what has to be done, where was it and what action was needed. This is something that was into the software from Fleet Base that we could actively use. But more to the to the real case of small boats, also in the software there's an automatic detection of if you see something it will automatically detect a boat, it will detect people. Um We also introduced the AI agent to tell us if they detect something of crowd gathering like I told you on a previous slide. And if that is gathered, the operator of the drone automatically gets a warning that there is crowd gathering. And he can immediately see then what's happening. So the drone operator doesn't have **[12:37]** to continuously look at all the footage of the drone or the drones in the dock. He can let the AI decide when to give an alert and at that moment he can look. So we are sure thanks to the AI we don't miss anything. Like on this footage you can see infrared footage of a beach side where they have a gathering of people. Um when you went up closer we can saw people sleeping in sleeping bags which is some of the things the tries migrants do before they take the boat. So we um took a closer look, sent a patrol. This time we were lucky it weren't illegal people, they're just students that were sleeping out after a heavy night. The AI also detects people like you see in this picture. It is something that we might have missed because it is so small and when you're driving over it, it is very difficult **[13:39]** sometimes to detect people in the dunes between the trees and the bushes. We can also see that there were tents built up at the beach. They could find campers which is something they always bring with them also. Also at night time the detection works. We could detect people on the dunes. We could detect the boats on the beach at night times. And from all that footage, we continuously get live streams. Those live streams links are coming in our dispatching room. And we could also share the live feed thanks to a QR code with the patrolling officers that we sent on scene. So they know in advance what to expect and where exactly to expect it. Before we can deploy all those things, there are a couple of conditions **[14:41]** that we have to agree on and that we have to make sure that everything is safe. We have the legal and the aviation issues where we have to coordinate our flight profile. We have to have emergency procedures. We have to have permits. And the B loss framework has to be in advance um made available. One of the issues we especially uh tangled with was the electronic conspicuity. To how can we see other traffics around us? Um We had to push also for alignment with the skies which is our national uh, aviation authority on how to fly in certain zones near to uh, airfields. Um, also because we are um, police uh, station of course police officers uh, uh cyber security and data has to be uh, very secured. Um, we have to have encryption, **[15:42]** we have to logging and data resistance. We have to be reviewed by our DPO, data protection officer. Um, also the privacy and the proportionality is something that we have to take in account um, the uh, minimum image processing has to be the retention periods of the the data we capture. We have to make up a DPA which very uh, transparent. Um, also uh, not to forget the operational management is very important, the human in the loop. We have to have to train him. Uh, we have to have an escalation matrix on how to work. And after each mission uh, we had to do an evaluation. With all those things in the picture um, we could bring this out to the public and also to our uh, political responsible people which makes sure that the trust arises when we are clear about the boundaries in which we work. Um, and that they're uh, very important to make it clear **[16:46]** that not only the possibilities uh, are there. We went from one uh, POC uh, in Koksijde. We want to expand it to more locations which can easily be um, what is important then you have to choose the right location. Uh, so we have a good coverage and also the privacy uh, check is in order. Um, so we have the time from the first image the number of useful missions, decision impact, and the privacy and security incidents all figured out, and then we can scale it to do more than just one dog. And for our police owner specifically, we would like to do three dogs on specific locations. We will have a real-time intelligence center from which we are operating the dogs. We will have one operator instead of a team going out. We have secure connection. We **[17:51]** have real real-time images, sensors 24/7 available. So, this is uh what we want to do. We also tested it with home-based uh goggles because normally we don't have 24/7 operators or drone pilots in our police station, but we have 24/7 drone pilot who is at home waiting to be called on. And to save some time, we experimented instead of letting him come to the police station, let him work from home out with the goggles, which also worked perfectly. What are the benefits if we use dogs instead of normal operations? We can see that the coastline for our coast is 16 16 km km. If we fly the normal way, we drive by car along the coast road, it takes about **[18:54]** nearly 2 hours to tackle the whole coast. If we then look at what gives us the benefit, if you see the round trips with the classical drone and a normal shift of 8 hours, we can pass over some hotspots six to seven times, and the average between uh drone passing over a specific location or hotspot will be around 1 hour. If we take it out to the dock and we place three docks along our coast side, we will pass over a hotspot around 18 to 19 times and the the time in between two passages will be 26 minutes, which is much better if we have people in the dunes trying to inflate a boat and trying to take that boat onto the sea. It's a really, really better timing for that. Also, if we look at the the cost, the personnel cost that the gain **[19:57]** normal costs uh weight will be uh for one inspector, one shift, we gain around 570 euros per shift, um which means if you get 40 shifts, we already break even with the investment of one dock. We want to expand this to the whole Belgian coast to secure the whole coast, so which means we would deploy around 10 to 12 docks, um which can be used in all the different police zones around the Belgian coast. But can only be used for um illegal immigration and small boats, but can also be used for other activities uh beyond that. We can guide them from one rock or we can let each police zone uh yeah, operate their own dock. We can also push uh the rest together in the rock. We can put out a fixed-wing drone to cover the whole area and indicate where a dock should go. We can also still send **[21:03]** out a normal patrolling car with an non-dock drone. But, the future would be to mesh everything together, not only the dogs, but also the CCTV cameras, the ANPR cameras, and the body cams from all our police officers. That would be the future to mesh this all in in one fusion. >> [gasps] >> Also very interesting was the the post data search we could do with the forensic search. We could look in our footage and search for everything we would like, cars, people, Also live incidents is something we used from Fluid Base, which let us work very closely together on an incident. We could share footage, have the same mapping, and that's led us to a better operational cooperation between the drone pilot, the officer in the real-time intelligence **[22:05]** center, and the patrolling officers on the scene. We got some publicity of newspapers, of course, for this. And um I would wrap up my presentation now because my time is over. And if there any people do have any questions, please let me know. >> So much, Christian, for that interesting presentation. And I think one key point that I can take into notice, trust arises when the boundaries are as clear as the possibilities. That was fantastic statement that you gave. And I think that's how uh public safety departments and policing, let's say, use cases, our policing authorities can utilize this technology. So, yeah, we do have a lot of questions. What I will do is I will just take them one by one. So, let me take up this first question, and it's very interesting. How has the alerting sequence been since the program got live? **[23:06]** Did the manned guards completely remove any specific metric you measured for this to prove the POC? That's the question. >> Could you >> Yeah. >> Could you show the the question, please, Rohan? So, again. >> Yeah, let me try that. Okay. Yeah, I think you can see it now. So, how has the alerting sequence been since the program got live? Did the manned guards being completely removed or any specific metric you measured for this to prove the POC as successful? >> Yeah, okay. Um the alerting sequence um is something we at the small boats departures are weather driven, which means only with good weather times they depart. So, we let our dog fly all the time. As yeah, **[24:12]** as a sort of patrolling virtual patrolling officer in the sky. And he detects then automatically something has happened in the dunes. We detected lots of things, Derek, but in the period we tested, we didn't detect any migrants. We detected lots of people like you saw, people walking with dogs, yeah, guys having a party into the dunes, all those kind of things happened. Did they remove manned guards completely? No, no, of course not. We we we work together. But at the moment we get a detection, the operator can look what he has detected, which means that we only sent out a team or a patrolling officer at the moment it is really necessary. So, not all the detections get a response from a patrolling officer. Uh anything specific matrix for this to prove the park? Yeah, we we measure times, response times. Uh we measured **[25:15]** accuracy of the AI, the height of the flight, when we can see something good, how fast do we fly to get a good measurement. Yeah, we we tested everything. >> Perfect. Sounds good question. So, I think the another question is regarding BVLOS. Did the approval process involve different authorities than a commercial BVLOS application and was it faster or harder? >> Um well, we are we are state operators, which make it a bit easier than a commercial BVLOS application. Here in Belgium, BVLOS applications are very restricted for commercial use. Um for our specific case, it was granted. We had different conversations with um national aviation authority, specifically the military here, because we have an in Koksijde military air base, **[26:18]** which you have very good contact. But, for each mission, we had to get an approval also. So, we had to call in to the military to say, "Can we fly now?" And we also get the approval from them. So, we have the approval was faster than the commercial ones because we are a state operator. It was also specifically about yeah, non-populated areas in the dunes everything. And when we fly, we fly above the sea. So, yeah, that was um Everything was granted. >> Thank you so much. So, we have another question. How were the KPIs measured for this park? Any key insights versus the traditional policing that you guys did? >> Yeah, what were we um We know what we were looking for. Um so, we have a very specific targeted in this **[27:22]** park. Uh it was detecting uh people in dunes and on the beaches. Um So, it had to be uh it had to be working. Um And uh key insights, yeah, um It it gains a lot of efficiency when you use this uh instead of flying out with a team of two men and one drone and following uh the drone. When you get uh the dock, you get the the better efficiency because uh it detects automatically. Um it flies automatically. So, you you don't need two guys. You need one operator. Um So, we gain a lot of efficiency. We gain yeah uh time. We gain um money with that. Um And the response is uh much better uh because you can see it before you send out somebody. **[28:22]** >> Perfect. Thank you so much. I think we have two questions on the similar lines, so I will just take one of them. How have you guys as a police adopted DJI dock? Is that [clears throat] allowed in the EU? >> Yeah, um DJI docks are allowed uh in the EU. Um Of course, um it is for um yeah. We use it for the normal operations. Let's say if we have special operations, we won't use DJI, but for this normal operations, um we we we can use it. There's no problem for that, yeah. >> Perfect. I think there's one question for me. Who is the provider of all these detections like crowd, people, and dunes, campers? Is it FlytBase? So, yes, like uh detections happen using the Verity goes AI layer, which is agentic AI layer inside the FlytBase platform, and you can use it in a on-premise or a cloud **[29:24]** version. So, let me take up this. I think this is also interesting. Did you struggle with the wind and weather on the coast? Do drones often not fly? Question mark. >> Um, well, we have lots of wind here, of course. So, uh but it still uh it still worked until I think 15 m/s, so we have some strong winds, and it still worked. The only thing you see and sometimes I had to get a hold of my heart in the landing procedure. If you see the battery dropping down and he keeps hovering half a meter above the box, yeah. Sometimes it's we didn't have any issues. It each time it was able to land. So, and it take off. So, it was nice. Nice to see how it works and how good the hardware was. >> Perfect, Christian. I think we have come up to the end of the presentation and **[30:26]** all the questions. So, yeah, I think thank you so much, Christian, for this wonderful presentation and indeed showing how you were able to prove this concept of using dog drones across the Belgium coastline, and I hope that you we are able to scale this across the complete Belgian coastline. So guys, thank you so much again for tuning in and for your patient hearing. We have a lot of more exciting sessions lined up. So please stay tuned. Head over to the agenda tab so that you can track all the exciting sessions which are there for the following meeting. So yeah, thank you so much Claire. Thank you so much I think Matt Ashton, right? Matt also joined. A lot of people joined in actually from different parts of the world. So thank you so much everyone. Any closing thoughts Christian before we close the meeting? Anything that you would love to put up as a closure? >> No, I'm really looking forward to expanding this digital along the whole Belgian coast >> [clears throat] >> to tackle the problem. **[31:27]** >> Yeah, I think for me the key statement has been trust arises when the boundaries are as clear as the possibilities. That was actually [snorts] the reality of how you need to proceed ahead with the technology keeping in mind all the integrity of safety in place. So yeah, thank you so much guys and looking forward to staying in touch. Please stay tuned and as I mentioned you can head over to the agenda tab. Have a great day ahead everyone. Bye-bye.