 two projects here by the nerds from SeaWatch to save lives in the Mediterranean. The first is the SAR app to help getting resources to coordinate them, and Search Wing, which is a drone we use to help find refugees, and I'll try and give you an overview of the situation in the central Mediterranean. My name is Oben Neugebauer, and I'm a customer of SeaWatch. This is a SeaWatch organization that does sea rescues. We operate two ships in the Mediterranean. We help people survive this dangerous passage. We also try to political pressure in order to help improve the situation. How does the situation appear? The trip to Europe is difficult, particularly those people who don't have the right papers. People can go through the overland, over mountains, or over the much more dangerous Mediterranean trip. What this means, I've brought a couple numbers along. What does the number 16 mean? One sixteenth is the likelihood that you'll lose your life when trying an attempt to cross the Mediterranean. When you translate this to the room, if we'd all started, we would have lost 30 people. People think that people do this willingly, but why would they do this given the numbers? The numbers are from the Missing Migrants Project. That's a good source for these numbers. Over this year, the numbers have dropped down to one in 44. Not quite as dangerous, but it's still an unacceptable risk for us. People aren't victims of a natural disaster. These are a man-made disaster. The European Union had the capability to end this. They had the capability in their hands. But so long as there's no actual organized sea rescue mission, that's something we wanted to do. That's why we started this do-it-yourself search and rescue mission. This is Harold Hufner, a worker out of Brandenburg. We wanted to ask a question up front. In the entire time of the Berlin Wall, 183 people died. Outside the European borders in the last week, there's been 300 people who died. Here, for example, several families from Syria are going on a fully unacceptable old fisherman's boat because there's no other way to come to Europe. So they've built an old fisherman's boat and then rebuilt it into a relatively useful ship. This is a sea watch. It's not just about saving people, but also observe what's happening. This is what we've done from the beginning. We hacked the talk on this talk. So we continued and brought the ship into the Mediterranean. And we can see it here. That's the 12-mile zone. That's the radius where countries have influence on seawater. And our area of operation is the red-escaped one. And we're looking for people in need of rescue. And this is what we found in the 7th August of 2015. That was our key experience. And we came to the edge of our capabilities. We had 512 people. We had to help. And we were just out of our resources. We had a boat, which was just drifting on the sea. And there were people who were not drowning, but they weren't getting enough water. And that's why they were dying. And we kind of, we were taken aback that we just didn't find them over three days, even though they were in our area of observation. And so we kind of started a political action with the Spree boat and put 128 people from the Bundestag onto this boat to make them feel the politics on their own body. And then we decided to get a new ship that has medical facilities to treat dehydrated people. And we need aerial reconnaissance. It can't be that there's a ship drifting for three days and we can't see it. But there was a second key experience in the beginning. And that was our effort and the Igeis on the islands. And that was because lots of people have came together and helped the refugees from people who were professionally into that business up to hackers just helping. And the networks that evolved in the Igeis, they helped us in our operations in the Mediterranean. And four organizations in Germany followed this, this, our example. And these organizations also sent ships into the Mediterranean. And this can be a challenge. These 13 ships, which are in the Mediterranean now, must be coordinated. And then we can, we as techies can help there as well. And that's where our app comes into play. Even though we don't have this app in operation quite yet, we were able to save 20,000 people from drowning or dying, possibly dying in the Mediterranean. To show how this looks like, I'll show you a video. This is the situation in August. You see how the boat kind of starts to collapse because they're overloaded. And the, and the ground plates inside the boat just disintegrated the plastics and the, and the boat just starts to sink now. It was, it was a good fortune that we were on site with our ship and able to save the people. And we were able to give them, to give them help. But this can be very, very, this can end very differently. You can also see that we were together with the second organization. It was the organization, and they were there with their ship, Juventa. They took on board 400 people that day. We see the corporation general, or the coordination generally works. And here there's a new post where they again save 100 people from, from the Mediterranean. 2016 has been the most deadly year in, in our history. This is our colleague Martin, who has recovered it, died, baby. All in all, 5,000 people or above 5,000 people have died in the Mediterranean or on the borders of the European Union. If you want to solve this problem, it's not going to be, it's not going to be that easy. And the European Union has the capability of doing this. There is not even, there's only one European piece of legislation. And that's from 2001. And that says that the trans, the companies that are transferring the people are doing the, are doing the registration. But this is politically not quite possible yet. Therefore, the European Union is not likely to, to open up legal ways of immigration. And so we're thinking, we're wondering what we can do. And wondering what we can, or we can help. So we're back with aerial reconnaissance. We started a try. We bought a ultra lightweight aircraft. And we flew it down to the, to Tunis. And we had a few very, very productive. We had a very productive few days. And we actually found lots of boats. But the official, the officials down there reject, reject, or retracted our, our allowance to travel through their airspace. And that just shows that it's not, it's not able to do this. And it's politically not wanted that there's a civil eye watching what happens. And the cost of Tunis. And therefore, this hasn't worked until now. So this is our method of reconnaissance or primary method of reconnaissance. And you can picture that if it's a small boat on the horizon, it's practically impossible to see it. The other thing is that there's the Libyan Coast Guard is pushing us back. And we are, that we are violating different rights of the Libyan of the Libyan sea space. And therefore they're intervening in our rescue missions. And this is something which happens. And this is very, very important to document these kind of incidents. And there's another problem that we have. We are, we are, the other problem we have is that we're visible for far distance. And it's good on the one hand that they can find us and they can come towards us and that we can help them. But the problem is that we have, that we can rescue them, but we are quickly at the end of our capabilities. And we can kind of know which one is a higher priority to rescue or which is the other. Therefore, we said we need help. And that's what we need. Quite pragmatically, we have two projects. One is search wing. We have a drone, a model aircraft, which is equipped with cameras, and which is helping us to find the boats in time. And the other way of, the other project is the SAR app, which helps us coordinate the ships and gathers all the information needed to assess the situation. That's why I'm passing on to my colleague. So we had the situation when we started what, what working with Seawatch, that just was in the last year, around 12 organizations started working to try to help and rescue refugees from the Mediterranean. So we are now in need to coordinate which ship is going to rescue which boat of refugees. So we had the idea for this app to coordinate this. So we had this idea for this coordination app where we had one central structure where incidents could be reported to. And then after a boat has been reported, we could coordinate which rescue ship would be allocated to this boat. So it's a coordination system for all the rescue boats. I registered within. So also, and then also, also rescue, also boats in need of rescue are also reported to this app. So all the, all the NGOs see where all the rescue boats are going and also where the refugee boats and also the coast rescue has access to it. So everyone knows what's going on at the same time. So here we see how this looks like. So this is a map view and we also have a view for where there are the cases. On the left, there are different kinds of filters. On the left, there is like a system where you can see how urgent a situation a boat is in. And also you have an option where you can see the rescue boats and see which one is online or which one is currently not working. And you can also see if a ship entered in a new case of rescue. And here's a case of a few. So here you can see all the cases which are currently reported. And well, that's basically about it. So until now we have been working on this on our own the two of us. So we have started a prototype which is working currently, but we are in need of help. And we are hoping for your help. We are hoping you can help us with the security concept. We have a centralized system, which might be good to decentralize this. And also we are looking for geo-hackers. And people are knowledgeable about this. And like any other ideas which could help us to make this app better and faster. Hello, everyone. Ben Tuenstein. A thousand thanks to CWATCH. It's an incredible work that they're doing. Lots of applause. And the lack of responsibility of the European Union and the government's error is a massive shame. We've known about this CWATCH project. Lots of refugees from Turkey to Lesbos attempted to make the trip. And that's where the idea came that we can help with technical things such as a quadcopter. But this wasn't really enough for the right range. And so then we thought about what are some other things we could do that could help. How could we watch over a little bit of the coastal sea so it could actually help someone before they... so we could save them before they drown. That was the whole idea of a drone watch. We wanted to make something that actually helped. And that's where we thought we wanted to engage ourselves. And that's where we've done really happily to save people. So we consider what we needed to do. So the problem with the plane overwatch is pretty hard due to border issues and legal issues. And here's an example of one NGO drone on the far right. But this is an example of one cost like a half million euros. But that's not something we can do. Maybe if we get 10 euros from 150,000 people, but that's probably not possible. How can we get... we really wondered how can we get this as cheaply as possible. Something that was okay to lose. So we thought what are some of the scenarios, how to be used. So the first scenario, all I wanted to see if there's something beyond the horizon. That was the first scenario. Is there a boat beyond the horizon that needed to be saved potentially? So due to... because sometimes radar limitations couldn't get much more than 15 kilometers away. Could we send something perhaps an hour or two away? Just to be able to see when you've spotted something at a distance with radar. Can we see is a sailboat or some sort of fisherman boat or something? And we thought that drone would also be the sort of the second scenario. Could we see to see something and identify it after radar detection? So the idea was that we create this small drone. We just throw it in the area. We thrown into the air and we were remotely controlled. And we thought it would look a little difference along the coast of Lesbos. And we thought we would also... there could use IR cameras, find people in the water. And then the... that was our first scenario, but then the Turkey made the deal with the EU, which meant that all of a sudden that wasn't the high-risk transit zone anymore. So that's why we thought for the smaller one was more ideal for a ship launch and ship landing. Because it's a little harder to land on a ship, so we needed the small one. And then we... this thought was that we could also really increase the radius that the seawatch could see. So that it would still come back to the ship. And so you can see... this was our recommendation for the for the longer-range second scenario for maintaining an overwatch around the ship to increase the range. And so we had a bit of a learning curve. So we started out this way. We built... Yes, we found out that some people were doing long-distance FPV. And they built in a camera into the aircraft. And then they just put a lot of batteries into the aircraft. And they came like... had ranges of 200 kilometers. And they posted very boring videos on the internet. And then in those videos, 50 percent of the time, you just see water. So for like 200 euros, you can get the model and the engines and electronics. Yeah, so we started sticking stuff together and learned, well, this is not that trivial, as we thought. And so it was a bit... it's a bit more difficult. Ben, you had a model pilot because I'm not one. Yeah, we met people as we did... as we did a presentation on that. And he said, yeah, of course I can help because I build these things and can fly them for a long time. So it's... because if you can't fly them, it's difficult to test them. So we started our first experiment. And yeah, this is our learn curve, pretty literally. This was our first endurance test. And we wanted to see how long it can fly. And we said, okay, we'll just let it run circles above our airfield until the aircraft said it's gonna... and it's gonna be empty. So we stand there, it was raining, and the aircraft was doing its circles. And at one point it just... at the 51th turn, it just decided to do something different. And then it nearly... and it came down just like just over the... over the rims of the forest. So yeah. And then it came down there. So we had a bit more learning curve. And we noticed that Eitelkeit and Posertum just doesn't help with such a project. And we wanted to like do a low flight skimming and want to do like a bit of posing. And so yeah, we noticed that there was a conflict in the software. And yeah, there was a bug in the software that... Yeah, part of the software in this specific situation, this... the software just did something unexpected and went into... went into a tree. And for every one of these errors we have... we have a... we have a taped... taped the plane for that. So we have multiple videos to show you. Yeah, we supposedly had a video. I pressed on play. Help. Oh, they're all gone. It's also part of the learning curve. But yeah. This is an important part of a learning curve. This is a... This is the first curve it is flying completely autonomously. The cameraman was standing under the first curve. So the question was if we can do that again... This is just don't move. It'll work. Just don't get out of the way of it. And... Yeah, it worked. It came back and landed right in front of our feet. So we do have like kind of tasks. And this is the second one. It was like land in the net. And it just didn't. This is the second attempt. And it didn't. And the ship is normally moving on top of that. So there's nothing that can go wrong. What could possibly go wrong? So, hey, it's just Styrofoam. So you can glue it together. So now you can see the small aircraft in action. And yeah, it actually at one point it just starts working. So you actually are possible. You're capable of landing inside a net. Another problem is to find the net. But that's something different. But the net landing isn't carved in stone. And there's other... And we're open for suggestions how to catch a drone with a minimum speed on a moving target. And there's the concept of... There's a... There has a trail... It trails away on a rope and then the rope just is caught by a rod. So the minimal velocity for these things is about 50 meters per second. So if we're landing against the wind, so our relative speed, if we're flying into the wind, if we're landing into the wind is slower, but we're having double turbulence is created by the wind hitting the ship. And then we're in a turbulence spot behind the ship. And therefore we have to fly with a higher speed. And yeah, on this point I want to say a friendly hello to the Etihad Zurich. Which unfortunately doesn't have a time for us because they need to start their commercial production. And the Etihad Zurich can build tail sitter, which then essentially hovers hanging on the propellers and land them on its tail. And we saw and we found it really cool. And it has a lot to do with a lot of partial differential equations, which is pretty complicated to solve. But the proof is made that it's possible. They're not saying how and they're saying that they would quite like to help us. But we would quite like to know how they're doing it and therefore... Of course, you can solve it easily. You take a copter software and a plane software. We have the idea of combining a copter with an aircraft. We're switching the software from an aircraft mode into a copter mode when the aircraft is in a vertical climb. And actually it is done that way and even though it's a bit strange. But the Etihad Zurich did it in one software solution and we would quite like to know how they did that. So help us appreciate it. So why don't you land on water and you're... And then we have to pick up the aircraft. And the simple solution is we have to let out a boat and then pick it up. And worst case scenario, we are actually going to do that. One of the questions we had was how to get the thing waterproof. And because saltwater is a... How to get a thing waterproof. And we found rubber that you can encapsulate the electronics in to get the electronics watertight. But this is plan C, the last option. Because if the drone like... If the drone like flies into the net and rebounds and lines in the water, we want to pick up the drone and hope that it still works. The idea was that we also... We wanted to include a Zatmodem but the Zatmodem was more expensive than the entire drone. So one day of aircraft operation costs like 700 euros. So technically for that money we can throw away like three aircraft. A proper aircraft throwing it away and that's like... That was the aircraft launch and the autonomous aircraft which then we had to take control of. So that was a flight into the woods quite literally. Always these trees that randomly appear in your flight path. Modern technology of presentation. That's how it looks like. The video was the 23rd December. And yeah, this morning the aircraft still looked exactly like that. And Styrofoam, you can still glue it together. Yeah, we actually wanted to show something. We want to show you which kind of decisions we did. We took while building it. And there's different software stacks that operate those. There are different hardware solutions for our use case. Because a lot of the technology for the drones is export control because it comes from the US. Because it's theoretically a microcontroller with just a standard bunch of sensors. Which is then wrapped in a bit of software. But the US obviously classifies this as a weapon. So on all these small sensors you can buy for your Arduino's on breakout boards. And we're momentarily using the Pixhawk. It actually has this problem that you can have to leave your name and a position of a number of documents. So we thought, yeah, we're real hackers. We're going to do it on Linux. So we're going to use a Raspberry Pi or something similar. And we're going to use that and put a real-time kernel on top of that. And then let the Arduino Pilot software run inside that. At the moment, we can let it run on this system. And because we don't only want to have the claims, we want to show you the video. And yeah, obviously it seems like it doesn't work. From a vendor in Spain, there's a small board. It's designed as a shield for the Raspberry Zero. And the Raspberry costs like five quid as a small board. And there's an ARM CPU on it. And it has two USB ports. And for example, this board, you have a sensor shield where people do quadrocopter hacking with. And there's a gyroscope on it. There's a pressure sensor on it that you can attach a GPS module to it. And it's in there also. They're also shipping it with a special Linux distribution, which you have to register to download. Essentially, it's used for copters. And they've got two videos from their many, where they're using it for aircraft. If we, yeah, we essentially notice that if we compile ourself at sex faults and the signaling on the CPU just has latencies we can control. And everything comes in. Everything starts jittering, even though we didn't do anything. So we're not as far as that yet. We're sure that this is doable. So we have thought of a few challenges that we want to talk to you about. And we think there's reasonable that it's doable on a real Linux real-time operating system. So we think it makes sense to develop our own sensor board to put on a Linux PC. Or maybe we could also find a very tiny processor to put on the sensor board. We don't really care which way around. So this would be one of the challenges. And the second core challenge, well, a third core challenge, apart from landing into this net. And also like how to find the, like for the net, how to find the center of this net. How do we calculate it? So if we take the edges of the net and like then give it to the drone. And it should be following this coordinates with all the waves and stuff. And it's landing in the net in the end. So we are still working on this. And our search and probably most interesting challenge is the how to transmit the data from the drone to the ship, actually. So that's a, the watch can actually use it. So we have tried to transmit video with an analog system and with infrared. And maybe then there is someone sitting in front of a monitor for two hours where nothing is happening, which is completely empty, hoping for something to just come along quite quickly, which you might, which might have been someone you should have to rescue. So yeah, that doesn't really work. So for the small one, we are, we think that might be working. But for the big one, we have thought about something, something different. It should be flying in quite a high altitude. And it's again about being really cheap. And standard technologies, standard transmission technology. Well, we thought that would be Wi-Fi. So that, that would be cheap for like quite little money. So we are in international waters. So no one should, should be able to do any harm to a Wi-Fi transmission. So on the ship, we were thinking about having a telco antenna, which we can put into a circle. So with small raspies and transmission, and we're doing software diversity and doing a Wi-Fi broadcast. So anyone who has done anything about Wi-Fi broadcast, like not only know what it is, but has seriously worked with it, we could really need some help because we are still having quite a lot of difficulties with this. So if you just blast that data you did, well, the data is still coming out on the other side, but there is a lot of errors coming out. So we need a better protocol, which works good for this. We want to fly in a quite high altitude and take pictures of the sea. Take pictures of the sea and on the sea watch, just a couple of minutes later, someone can look at these pictures and analyze if they can see anything on this picture. So this is an area we need, where we need help. So it doesn't really have to be Wi-Fi, but it has to be something, it can be something else if you have a better idea. So it shouldn't be more than maybe 15 years of hardware, because otherwise we can't use it. It might be better, but we want off-the-shelf hardware, which is really cheap. And the next problem is to take pictures from above, and we need high-resolution consumer cameras, which have 10 to 12 megapixels. And just a short while ago, that was very easy, but right now we can't find any consumer camera, which I can control from Linux remotely and get the picture in the end on my computer. The most similar thing we found was a Sony camera, which only has the interface of Wi-Fi. So this is developed to be paired with a smartphone, and this is quite complicated. If you have a very strong Wi-Fi next to it, this isn't going to work. So we have this antenna from the recipe to pull a string around it. We don't know, it's kind of an interface. So we have this camera, if anyone feels capable to work with this camera, which might not need connection via Wi-Fi, that would be super cool. You can hack on it. It doesn't matter if you break it, if it helps to develop something that works, that's fine. Do we have another challenge? Well, maybe a ground station. The ground station is not really a challenge. Maybe a few words about it. So all the software you have for the ground station is just crap. The best thing we found is Pison. So just talk Mufflink. Mufflink is awesome. That's a protocol we can talk to the drones. That's why the ground station is used to talk to the drones. With this pseudo shell, we can do more than the ground station can do. We want to build our own ground station with this shell. What we want to do is we get a next Y coordinate from the radar, we dial it into our system, we just throw it off and it goes and it does stuff. So we do have another video. We can show you how that looks from above. The camera is just attached with tape. That's how it looks from our standard test flight. This is the perspective. We can see details on the water surface. That's how it should look. This is about 300 meters. So these are upcoming monestones. We have now 100 kilometers of range with this with one of the batteries. But there are two in there. So we are thinking that we can double the range. So we're thinking that we're using 100 kilometers as operational range and the 100 kilometers as reserve for like wind and other stuff. So we want to develop software to help all of these organizations that want to save lives and improve the life of humans in need, which is wholesomely used for these purposes. So we're working on a lot at a license which then prohibits the use for the software in military or non-human applications. We have one video, we still remains. I went to show that it was fun. This is from our workshop. Sadly this is without background so I think that it has a nice waltz in the background. But until now it was terribly fun even with the challenges we faced. We learned really much things doing this project and we're asking for your support, help us. Help us develop this software into something usable for Seawatch and other organizations to help people and yeah, thank you. Downstairs we have in the Hack Center we do have close to the bar, we have an assembly. We're happy if you come along and talk to us, talk to Seawatch. Give us cool hardware or show us the partners in the direction of it. We can think of situations that we can 3D print them and that 3D printer models actually fly and we're open for everything. All of ideas that you have. Also what the Grand Station is concerned we want to have. Help us make the code nicer that doesn't suck full all the time. And yeah, we want to hang the aircraft under the ceiling so you see where we are so yeah. So we still have 10 minutes for questions. Please stand by the microphones. Yes, question. Two points of information for you. Have you thought of Lora for the transmit, for the transmit receive? We can for on that long range we can transmit not video data but like asynchronously for the long range and also can you use deep learning for recognition of boats on seawater. So essentially we probably want to do that on the aircraft. Therefore we need enough CPU power on the aircraft to do the recognition of the of the boats. But the first priority is video transmission. To your first comment, we haven't heard of it, we haven't heard of that technology. And as we already said we have the requirement that the hardware is not allowed to be. We do have bandwidth requirements and we have a normal a distance between the ship and aircraft of three to your 40 kilometers. And we want to transmit two megabytes per minute. If it's possible with that and if it's if it's possible for less than 20 euros and let us talk about it. If you can determine what is just water and what is not. Let's come back to our milestones. In March we want to put our first pilot into into practice. Because there are humans dying out there so we have to get going. We're going to do something that works now and it can get better but so we think Wi-Fi is the best course of action for this immediate time. Question from the internet? You said you need a He said you need a Alternate form of So Use this nylon strips to entangle the propellers and then use that to do that. But essentially this is just a net. I've also got a question for the net landing. So you said you want to calculate the point the center of the net over GPS. Have you thought about doing that optically? We thought of letting an IR We thought of an IR LED blinking and just track that. We thought of mounting cameras on the ship and then just It's possible of controlling the aircraft pretty well over this math uplink. So we're going to use then We can use that on the ship to target the aircraft into the net. So thanks for all your effort for the app and with the drones. Thank you for that effort. You have very you have very hard requirements and a very tough time plan. So do you have a website where you have a website with the expertise you need? We've got our website answer. We've got our website search wing.org It's hashtag search wing. Next thing we want so we're just waiting for the Congress to finish and then we want to carry together what still need to be done. So come to our website and you can see what we published there. Next question. For me a quite important question the altitude you want to fly at So at the height of one kilometers you have to take care of anti-collision with other aircraft. And there's a And our solution for that problem is we're just going to we're just going to broadcast the position of the aircraft to other organizations and then this is something we can just and what we essentially we don't need this transmitter then and we're flying over international waters and Essentially It's not we're in international airspace so So just So we're over international waters and we we're not we're not thinking of collision anti-collision systems quite yet because it's also birds that are they're heavier than those. I've you heard of project Marvin from to Berlin yes good So I don't think this is actually an actual working working product or am I wrong about that and the responses actually it started in 2013 and the and the the scenario they're working on is very similar and maybe you can use some of their algorithms that they're using but things have changed a lot so perhaps this doesn't apply we're actually more interested on actual capabilities in particularly with an underline they could possibly do this and the last question I've worked with digital photography for a few years and then there's a lot of different digital digital camera firmwares exist and there's some that are open that we could use and there is something called CEDK which unfortunately doesn't work with most modern cameras there are some cameras for some software for reflex cameras but we don't have that lots of applause thank you and thanks again for listening please give feedback on the translation to hashtag c3t that's