# Detecting Landfill Fires Early with Autonomous Docked Drones for Polish Waste Management - Event: NestGen ’26 - Published: October 2, 2026 - Length: 30:11 - Watch: https://nestgen.org/library/watch/B7x4f8yNR4A/ > 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 Landfill operators face a fire risk that's hard to catch in time. Fires can smolder undetected within waste piles for hours before breaking the surface, and by the time smoke is visible, containment gets far harder and costlier. In this NestGen session, Megadron will share how autonomous docked drones support early fire detection at waste management sites, and what operators should think through when deploying drone monitoring across landfill and waste facilities. ## Transcript **[0:01]** All right. Hello and welcome folks to another session at NestGen. We are now starting with uh the email time zone. So I'm sure like there's a lot of new people. Uh hello and welcome to everybody who has recently joined in and if this is your first at um welcome. Uh we uh we are using the chat box here on the right side uh to also see um where you're joining in from. um and uh uh what kind of challenges you are facing today with dog drones. Of course, like we have Matias over here who is our speaker for this session. Uh and we will both be looking at um your questions, your comments throughout the session. Um now, uh speaking about uh specifically what our topic is today, uh here's a scenario that's worth sitting with. Uh think of a landfill fire. uh once it goes up deep into the waste pile, it can burn for weeks. It releases toxic smoke, it evacuates neighbor neighborhoods and it costs millions **[1:05]** of dollars. There is there has been historically no way of catching it early on except for now uh that with stop drones there are some uh early signs where you can catch a landfill fire very easily and quickly. Uh and that is exactly what uh Matias is here to share with us today. Uh Matias is from Mega Drone uh one of the leading enterprise drone integrators in Poland and they've uh partnered with a Polish waste management operator to build the country's first autonomous dog drone fire monitoring program at landfill sites. The stack uh is DJI do 3 custom zomal AI detection and 247 automated patrols. Um but I'm going to not I'm going to not give too many details uh because that's what the entire session is for. Uh speaking of like um what uh who our speaker is today, Matias has uh u uh specialized in photogometry and geoinformatics uh as a graduate and then he has worked in classical surveying at uh um shipyard **[2:09]** before joining me. He spent uh several years abroad at Murphy Geospatial who is now a part of Wulport uh running terrestrial laser scanning, surveying and drone data collection projects across Ireland and Germany. He's he's since returned to me drone where he leads the enterprise technology segment specializing in DJI Dog 3 where uh autonomous drone systems, LAR, 3D scanning and robotics uh especially for industrial applications. These are part of his day-to-day life. A very very warm welcome to you Matias and thank you so much for joining us at the NestGen stage. Uh without uh any further ado I will pass on the stage to you and let's dive right into it. >> Okay. Thank you so much uh for my introduction and let me share the screen. Can you see it? >> Yes, absolutely. >> Okay, perfect. Uh so again my name is Mat Kren and I work an enterprise UAV **[3:11]** specialist uh at Meadron in Poland. Uh today I would like to show you a practical use case of autonomous drones uh system in the waste management industry. Uh more specifically I will talk about how we use a autonomous docking system uh to monitor a waste management facility in OPA. uh and how this technology can help detect potential fires uh at the early stage. Uh so let's start with the location and the environment uh where the system is operating. Uh so our area of interest our project is located in OP uh in the southwestern part part of Poland. Uh OP is a medium uh sized city with around 125,000 inhabitants. Uh and it's located relatively close to the Czech border. Uh however, **[4:12]** the most important part uh of this project is not really city itself is just a scale of the operational area. Uh the waste management facility is located uh outside outside of densely populated city center and covers more than uh 400,000 square meters. So as you can imagine uh this is quite a large area to monitor manually. Uh this makes traditional fixed monitoring less flexible and this is where uh an autonomous drone starts to make a lot of more sense. Uh let's go to the case description. I have a small video on our YouTube channel. I think I can show you just the two minutes of this video with English subtitles. uh because the the whole video is I think too too long for for today for this session. So that just prepare **[5:39]** This is a small introduction from site in Opola. So uh what uh what is the actual problem we are trying to solve? Uh one of the risks at waste management facilities uh is uh heating and extremely in extremely cases fire uh during the biodegradation uh of biological waste. Uh heat is produced and g such as metan may also be generated. Uh so inside of a waste pile the temperature can become surprisingly high. And here is the most difficult part. A fire does not always start uh where we can immediately see it. It can develop underneath the surface somewhere inside the waste pile. So from the outside everything may look completely normal but the time smoke or flames become visible the situation can can already be much more serious. **[6:43]** Uh so the ear the ear we identify an abnormal temperature the more time the fac the facility has to react. Let's go to the typical solution. Of course be honest with you a thermal monitoring itself is nothing new. Uh a traditional solution would be install fixed thermal cameras around the facility and B so far uh this solution has some more important advantages. Uh it can operate continuously. It can uh automatically detect temperature changes and it doesn't not require a drone pilot. Uh but there is one major limitation. of course is very expensive and the camera stays in one place so it can only see the area that was selected during installation at a waste management facility. This can become a problem uh because waste waste piles move their their **[7:49]** size changes and operation area can be recognized. Uh so our solution is a solution with autonomous docked uh drones. Uh so instead of placing many sensor around the facility, what if we could move the sensor to wherever we need it? And this is basically the idea behind the autonomous drone system. The drone can automatically leave the dock station >> and follow a predefinite mission, collect thermal and visual data, uh return to the dock, recharge and prepare for the next flight. And if the operational situation changes, we don't have to physically move the infrastructure. We can simply modify the flight route or the inspection area. Another important uh advantage is localization. **[8:50]** If the drone detect something unusual, we don't only receive information that there is a problem. Uh we can also get a image thermal data and the geographical localization for the event. Of course, it's not perfect system. Autonomous operations still have to comply with avation regulations, uh, weather conditions and can affect operations and personnel need proper training. But compared with traditional fixed monitoring, the system gives us much greater flexibility. So I think this is my favorite part uh of my uh project that and that we have additional benefits for using autonomous drones. So if we already have a drone permanently available on the site, why should we use it only for one task? **[9:51]** Uh exactly the same platform can perform security patrols and help detect unauthorized access. It can inspect infrastructure and provide regular visual documentation. We we can also use photoggramometry by by measure the volume of material stockpiles. Uh and also to uh and also because we repeat missions over exactly the same areas, we can compare data over time and monitor changes in the terrain. Uh so this is the photo from our site in the OP. As as I mentioned before, one side many times. So we have two predefinite patrol routes uh and the drone performed them every hour and the altitude for approximately 70 m. **[10:53]** uh obscale a audience is active during uh the mission which is particularly important uh because we also have to consider the the local environment and birth activity for thermal monitoring and automatic alarm threshold has been configured at 200 uh degrees C. So when we uh so when usually high temperature is detected the system can immediately uh highlight the event instead instead of waiting for someone of manually analyze all the data. Uh once a week we also performed a volutric uh missions over the sand stockpile. This allows us to monitor how the amount of material change over time. Uh and during the night of course the drone continuously uh **[11:53]** regular patrols when the facility uh has significantly fewer people on site. So one installation is supporting thermal monitoring uh security and surveying at the same time. Uh yes and here uh I have a small uh description of how was before and how is today with automatic smart and safer drone solution. Uh before the installation uh I I want to quickly uh tell you that monitoring was mainly uh based on manual operations. Uh typically this mean around four to six patrol routes per day, limited nighttime monitoring and manual stockpile measurements. Today the system can perform up to 48 automated patrol routes per day. So in terms of monitoring frequency, we moved from several inspections a day to almost continuously coverage. The stockpile **[12:56]** [clears throat] sorry survey is also performed um every week. Uh so which give us a 50 52 regular measurements per year instead of occasional manual measurements. Uh and perhaps even more importantly every mission creates a digital record. So when we we have geo reference images, thermal information and a history of previous uh missions uh we can just back to this information every time if if we need it. Uh this is one of the biggest advantages uh of automatization more frequ uh frequent data uh collection. Yeah, more frequent data collection uh but without increasing manual workload at the same time and **[13:56]** let's go to also very important part uh of whole project AI sol AI solutions uh up to this point uh I have mainly talked about automatization but automated flying is not only part of the system. Uh the next step is uh to automate the analyszis of the data as well. So this is where artificial intelligent become particularly useful AI solution. Here also I have a small video unfortunately not from the site but from our practical use uh with this system. As you can see on on the uh short video uh the system can detect a high temperature. Uh we just create a special mission that up to 100° **[14:58]** we should have a information about uh temperature. So drone can collect a huge amount of data and if somebody still has to manually review every single image every single mission we have automated the flight but we haven't fully automated the workflow this is why uh intelligent software is so important AI can help detect objects identify the thermal outormalize compare images from different missions and generate automatic alerts. Instead of showing the operation operator hundreds of images and asking, can you find something unusual, the system can point directly to the areas that may require attention. The final decision still belongs to the operator of course but AI can significantly reduce the **[15:58]** amount of information that has to be reviewed manually. In other words, uh I can say that from collecting data to extracting useful information from the data. Uh AI solutions. Uh again uh here we have a small screenshot from our WhatsApp group with with our drone uh that we of course in our Polish language that we find uh four four people uh in area where we shouldn't find the people uh and here on the right side we have a a picture that uh our AI solution can find how many tracks we have on the site. So for example during a security patrol the system can recognize as I told you people uh or vehicles in specific areas uh during repetitive missions it can compare the current situation with the previous **[17:04]** uh data and highlight important changes. Imagine that a vehicle appears in an area where vehicles normally shouldn't be there or does the gometric of stock price change unexpectedly. Instead of expecting an operator to notice every detail, the system can flag the event and ask the operator to take a closer look. And I think in the direction in which autonomous drone operator are developing the goal is not simply to make the drone fly without a pilot. The goal is to create a system that collects data, understands what is happening and and helps people react faster. Um I think I can share uh with you a link to our our video uh on the uh beginning of **[18:05]** the meeting. Uh there is like a 10 minutes of uh the presentation of whole site and uh for the last site uh slide I want to just tell you thank you. uh in our case in OPE our system supports as I told you thermal monitoring security patrols uh voluometric measurements and regular documentation of the facility. So for me the most important is the the most important is this idea and this idea is very simple. Early detection give uh gives people more time to react and in an environment where a small thermal anomaly can potentially develop into a serious accident. Uh that additional time can be extremely valuable. Thank you so much for your attention and again my name is Mat from Madron and I will be happy to answer any questions I think on I can share the **[19:09]** the link from the v video if you if you need it. Matias you can also uh uh stop your uh screen sharing um okay no problem as we uh so the audience does have a lot of questions I think that's a sign of like a great presentation and I think like um uh lantern monitoring and thermal nominate detection is specifically like I think your session is one of the most unique sessions today uh in the agenda uh so we do have a lot of questions which I will pop up one by one um on to the goldcast screen itself self. Uh so let's start with the first one over here. Somebody has asked, "Landfills change shape constantly as waste is added." Uh that is correct. Uh does the drone's flight path need to be updated regularly or does it adapt to the changing terrain? How does it work? >> Uh sorry. Uh where is Wait a second. Probably I have some problems with the internet. >> Are you able to see the question on my screen? On the on your screen? **[20:10]** Uh yeah. Oh okay. Land uh landfields change shape constantly. We waste uh Okay. Okay. Okay. I I I I see it. >> So the question essentially is of course like with new stock piles getting added. The landfills will constantly evolve as a site. They will change shape. They will change their form factor. So how do you manage your mission planning accordingly? >> Okay. So uh yeah also I saw uh questions. So yeah the site is quite big and uh puff needed to be updated regularly especially during for example summer where we have a different uh areas to u uh to the most important area more important areas. Sometimes you have it uh in a summer time and different part of area is more important during uh winter time. So yeah, of course we uh we have to update it regularly. **[21:13]** >> Understood. Uh I think that kind answers the uh question. Uh it's a continuously evolving mission planning. Let's go to the second one here. So uh somebody has asked beyond fire detection that you just spoke about are you using doc drones for anything else on the site? So like monitoring waste levels or tracking vehicle movements >> as I I think as I mentioned before in the uh >> my project. Yeah we with AI solutions we have a different tasks so we can uh check how many vehicles we have on the site how many people we have on the different areas. So yeah, of course we we have >> So is it a part of your like daily patrols like what kind of what what different kinds of reports? >> No, the most important part is like measuring temperature of course during the summer summer but also as I mentioned during the winter we have a different tasks because of course temperature is is uh less **[22:15]** important during the winter. So uh during the winter for example more important is a security uh uh security areas uh just checking uh and tracking vehicles of course. So yeah in the different parts uh of the air we have a different tasks and as I mentioned before of course we can tracking vehicles checking stock parts everything >> correct I think that makes a lot of difference in terms of like how much ROI are you getting out of the hardware setup. Um now moving on to uh the next question over here. Uh, for those of us managing waste sites in other countries, did Mega Drone need a special regulatory appro approval for autonomous flights over a landfill or did standard Polish drone regulations cover it? >> Uh, this question is quite hard because we have a specific regulation in Poland. Uh, our our government uh changed something. Uh, but it's hard to describe with uh **[23:19]** just on this session. This topic is so long. Erh and be honest in Europe you have a different regulations that in our different parts of the world. So >> yeah, we have some regulations but I think we don't have enough enough time to uh talking about this uh unfortunately. Uh but yeah, it's very in one word in one in one word I can tell you that it's very hard to prepare this because this is autonomous flights and in with our government it's hard to uh to with this regulations. I understand completely and I think like we as you mentioned like we may need a different session alto together to talk about just BV loss and uh uh in uh uh your neighborhood in your region. I think uh that this fact itself kind of put uh puts mega drone on like um the leadership map in drone autonomy in Poland **[24:21]** and uh of course like not only uh like being able to fly a drone is not uh like it's just one part of the puzzle. Uh but securing B loss approvals I think I'm sure uh it it must have been a very huge process. So >> I think be honest with you this is the biggest part of creating [laughter] the whole project. Yeah. >> Right. Right. So I'm sure like other enterprises who are joining us today especially from Poland should be super inspired to work with me drone in the future and work uh and hopefully share insights as well. Uh I have another question over here. Uh false positives on a landfill seem like uh seem likely given that decomposition generates heat naturally. Uh, how does the system disting distinguish between normal normal activity and an actual fire risk? That's a good question. >> Uh, yeah. Yeah, true. Uh, the biggest problem uh during the summer uh we have with glass because of course sun **[25:23]** can >> heat up glass. >> Exactly. And this is the biggest problem. But with drone we can find this temperature relatively faster. And be honest, thank you. Thanks for uh draw our drone on our system on this area. We found a free uh specific problems uh with fire and we we we could uh just delete the whole problem uh because we found it faster uh faster uh thermal activities. So um yeah this this system uh actually works uh and helps us to f find fire risks. >> So essentially is it like there is a certain like degree of temperature that the threshold if the threshold is crossed only then it's a fire risk. >> Exactly. >> Okay. Understood. And the drone uh drone's thermal sensors are pretty good at **[26:27]** capturing the right temperature. >> Mhm. Mhm. >> That's why you say that like it's pretty reliable in that way. Makes sense. Uh I think that answers the question. And I think I have one last question from the audience. Um landfills produce methane and other gases. Is the drone carrying a thermal sensor only or is it also does it also have gas about about gas detecting to detecting we needed a spec specific uh part of the drone. This is this is like uh detecting gas uh system but this is not exactly with uh matrix for 4D. So we have to find from different company for example this system and uh we can integrate with uh with docu and with matric. >> Uh I completely agree and I think like thank you so much Matias for your **[27:29]** transparent answers. I think uh this session has been really really valuable for our audience and uh guys whoever has posted these questions I hope your questions were answered. I think somebody has posted another question here. The changing site >> um in the chat uh >> right I will just try and present this on the screen. Yeah, >> this is also sorry because this is also interesting part that we can adapt to drone uh different systems as I mentioned for example uh the gas sensor but also some lights we can adapt uh some different different parts of the drone we can adapt to the system >> right uh makes sense so I think uh what I What was the question? It's actually somebody from Skyok reaching out to you uh that they are developing a critical **[28:30]** infrastructure protection using fully autonomous drones. Uh you may want to So actually I was just about to drop Matias's uh LinkedIn uh over here in the chat box. Anyone can feel free to like take this conversation offline, connect with him um and of course like speak about the session at NestGen. Uh I'm just dropping it here >> I think. Give me just a second. I can uh drop here a link to our video because it's also quite interesting. >> And we have English subtitles as I mentioned before. >> So give me just a second. Uh here I can drop. Okay. This is the link uh to our uh YouTube channel and to our video. >> All right. I think that's perfect. Uh so yes coming back to uh what we were talking about uh thank you so much Matias for uh your transparency sharing all the insights with us. I think this has been an incredible distinctive session uh today **[29:31]** at Nesten and everyone please feel free to connect with Matias over LinkedIn and um be sure to check out the video as well. Uh all right I think we are now running out of time so Matias I will take your leave now. Thank you so much for joining us today. >> Thank you so much. Have a great day guys. >> Yes. Okay. We will be publishing this session on NestGen YouTube as well and we'll notify everyone once the session is live. So you can always come back to it. >> Okay. Thank you. >> Bye. Bye.