 Dwi'n cael ei wneud. First of all, I have a short link here to a little web app I made with Anvil as a way to share the slides with you, share any links, relevant links to the talk. Also, collect feedback and if anyone's interested in sort of the stuff I'm talking about, then they can just fill that out and they'll receive some contact from me. I'll bring this up at the end, but if you want to take a look now, you can get all the slides are being uploaded now, so the link should go to that. You can give me feedback at the end. My name is Ben Nuttall. I'm now the technical programme manager for the Raspberry Pi Foundation. We're based in Cambridge in the UK. I maintain a few open source libraries and projects, including GPIO0 and PiWheels. I write things on the internet and you can find my links on there to various things. I'm going to be talking about AstroPi, so this is the summary of the talk really. There are two Raspberry Pi's on the International Space Station. We let kids write Python code and we send it to space. I'm going to be telling you how we do that and why we do that. Sharing some cool photos from space and a nice time lapse and talk about how you can get involved. First of all, the background of the Raspberry Pi Foundation. A lot of people know us for the computer that we make. We were founded in 2008, four years before we had the product. That time was generally spent building that product and that was our idea of what we were going to launch with. We're all around educational charity based on computer science education and digital making. We've been making and selling Raspberry Pi computers since 2012, but we do a lot of other things. We merge with CodeClub and CodeDojo and we support them in growing a network of coding clubs and school clubs and youth clubs around the world. The sale of Raspberry Pi funds our educational mission, but our activities are not limited to Raspberry Pi, especially things like CodeClubs. We have sold 27 million units across the board from hobbyists, education users and through industry. Those sales fund what we do, so it's really great that we have that sustainable income to carry on as well as providing the computer that we wanted to make available to users. We run lots of youth programs, clubs and competitions and things like that. I'm going to talk about one of them called Astro Pi. In 2018, 10 years after we were founded, we were awarded a contract from the British government to create the National Centre for Computing Education as part of the Department for Education. It's really great that in just 10 years we've become the go-to people in the UK for this kind of thing. Astro Pi was developed in 2014 and it's a combination of the fact that ESA, the European Space Agency, they run education programs to get kids interested in space jobs in the future. Raspberry Pi Foundation, obviously, we run education programs, so we decided to collaborate on one. That year, Tim Peake was the first British astronaut for many years to go to the ISS. Tim was an incredible STEM education advocate and ambassador. We were interested in working with Tim and we were thinking, well, what can we do? What kind of program could we run with ESA? Maybe Tim can take a Raspberry Pi to space with him and we can run code that kids have written on that Pi and he can be involved. That's essentially what happened. We started to look at what we could do. The Raspberry Pi in 2014 looked like this. This is the Raspberry Pi 1 Model B+. It was the same sort of standard as the original Pi that came out. It's a 700 MHz on V6 and has a video called for TPU, half a gig of RAM. That's what we're working with. We developed a hat and add-on board for the Pi that gave us some interesting sensors. We thought that would be more interesting that we could run kind of science-y kind of experiment things in space. Temperature, humidity and pressure sensors and then inertial measurement units. Gyroscope, magnetometer, accelerometer and 8x8 RGB matrix. You can scroll text across it and do images and things like that. There's also a mini joystick. I like the Raspberry Pi. This is the thing that's for sale. You can buy this so anyone around the world can get access to the same sort of hardware. These are kind of fun to use in interesting projects at home as well. The Raspberry Pi Camera Module. We've got a version 2 that's a higher resolution than this, but this is what we had at the time. It's a 5 megapixel camera, but still really capable. There's a great Python library you can use called Pi Camera. We need to put it in a case. Put all this inside a case. It would have been nice to send it up in this case, but apparently it's not space-proof. This is what we ended up with. This is the Astro Pi unit in its entirety. We've got two of these made. They make the £25 Raspberry Pi look even cheaper. The case cost about £5,000. We've got a few buttons on there so they can be operated by the crew if they need to press a button to run a program or something like that. The little red thing over the joystick there, we'd actually made this case and the mini joystick from the tent out would stick through. At one point, Issa said to us, we don't really like having pointy metal bits on objects. Can you cover it with something? A couple of engineers in my office were running around looking for something. I actually had a Lenovo ThinkPad that had broken. It was sitting on the side and it opened it up and went, oh, we can use one of these and picked the little nipple off the keyboard and they had to buy another one, but mine went to space. How do you get it to space? It's basically a big pile of paperwork and a big rocket. Those are not to scale, but not far off. We sent that space. Tim went to space in a separate rocket and we both made it. We've got those two in action there. They use the bogan arms that they hold the laptops on the side of the station. They don't use tables. They use these arms to hold everything in place. We have two of them up there and they can move them to different modules if they want to do something different or have them point out the window. The camera is actually embedded in the back that points out the back. We ran a few different competitions. The first one was just to let's see what we can do and let people come up with their own idea and do something. The whole purpose was for kids to get their code running space. The last few years we've run two distinct missions. One called Mission Zero, one called Mission Space Lab. We've been running these annually since 2015. The first year Tim was the astro pi astronaut. He was the ambassador. We've had a different one each year from a different represented country. Tomapus gave from France and Alex from Germany and Fabio from Italy. They've been the figurehead for the programme during their time. Mission Zero is actually just done in a web browser. It's a really low barrier to entry, really simple to get started and enter. There's a website called Trinket which is Python in the browser. They made an emulator for the sense hat. When you run the same Python code that runs on a Raspberry Pi to control the sense hat, it runs your code on the LEDs there. You can also use sliders and things to emulate the temperature going up and down and things like that. There are about 5,000 or so entries for that this year because it's really easy to get involved and it's open to anyone from the ESA countries. You get 30 seconds of run time, but you submit your code through Trinket, comes to us, we test it and they get run in space so everybody gets their slot. It's just kind of hello world. You can write messages, you can do pixel art. You can use conditionals and things, kind of basic stuff, but you have access to the sensors but not the camera. You can't log any data, there's nothing taken back and you can't use the camera. So it's just, as I say, just hello world. So really, really good for younger kids. You get a certificate at the end that has a map on the back showing you where the ISS was when it ran your code. The grade line is the path of where it was going and that little red bit in the middle is where it was during the 30 seconds that your code ran. And then Mission Space Lab is kind of a much more involved programme, sort of takes the whole academic year to do, but what you end up with is a three hour run time on the ISS. You come up with an idea for a science experiment, something you want to do and that's you submit that as part of the idea phase. You come up with the idea and you submit it to us, we analyse it and judge your idea based on its sort of feasibility, scientific merits and things like that. And then the next phase, the successful ideas, we send them a kit so they get a Raspberry Pi and a sense hat and a camera. So they have the same equipment that's in space and then they have some time, a few months to develop their code, write everything up. We give them an image that's kind of replicant of what's in space with the same libraries installed and things. We kind of choose two certain libraries we think would be useful and let them use those. And then they submit their code, we test it, we analyse their code and check that they're sort of still doing what they originally said, that kind of thing. There's two different themes to the programme. There's one called Life on Earth, which is about studying life on earth from space. So looking at that one, you're looking out the window, you can take pictures of the earth. And Life in Space is on the other Astra Pi unit. It's inside the station and you're reading the centres and things and thinking about what life is like as an astronaut. So looking at the setting and the environment in the space station. And then they run their code, they get their results back so if they've logged any data, if they've taken any photos, they get those back. Then they have some time to write a report on their findings. They can do further analysis, so they can write some more code to do some more graphs and things like that. And they write their report, they send their report in, we judge the reports and then we have some overall winners. And they all get certificates showing where their programme ran over three hours. So they kind of get pretty much two whole laps of the earth in that time. So you can see the path of where that's been. So it's kind of a look of the draw, if you're looking to look at cities or look at land or something like that, then look of the draw whether you might spend a lot of your time over the ocean or something like that. There's also night time phases, so when the ISS is on the other side of the earth from the sun, everything goes black. But in three hours you've got a good chance of hitting something interesting. So a lot of missions, space lab ideas, so things like dissecting crew presence with sensors. So kids thinking, what have I got available? I've got temperature sensor, I've got humidity sensor, we've got pressure, we can tell the orientation and we can tell acceleration and things like that. And they can kind of check the conditions and they can decide, well what can we use, could it be that if the humidity sensor rises, suddenly if there's a spike in the humidity sensor readings, an astronaut has drifted past. And it's about getting them to think about what they can use with the limitations they have. They can do, they can log data and just kind of see what happens. They can analyse that back on earth and look at any patterns or see if they can spot anything interesting that took place. If they're taking pictures out of the window, they can do time lapses looking down at the earth. And a lot of the ideas they come up are kind of based around global issues like climate change. We have a few teams looking at wildfires, see if they can spot evidence of wildfires in the photos. Things like forest depletion and looking at the greenery of looking down at the continents and colour analysis, that kind of thing. And also things like shrinkage of lakes. They look at archived photos from ESA and from NASA, compare them to any ones that they took and kind of relocate them and do comparisons. It's kind of really interesting that they have these kind of really, really big ideas that they want to think about. So libraries that we kind of install, so obviously things like Pie Camera and the Sense Hat Library, there's various alternatives, but we've been using FM for the ephemeraet data. So finding out where the ISS is at any given time. And then things like Reverse TO Coder to look at that to the nearest city or something. They can log that data and kind of keep that, or they can look it up later because the path of the ISS is something you can look up retrospectively. Things like GDAL, TO Spatial, obviously Numpy, SciPy, Pandas and stuff like that. We also have TensorFlow, OpenCV, Scikit Learn, Scikit Image, and we do get kids using libraries like that. We've had people train machine learning libraries to spot things like wildfires, like I said, and deploy those. It's really interesting to see how advanced some of these programs can get. So the process for how Mission Space Lab works, we send a recipe, an image that we've created, installing all these libraries and all the kind of modifications and things and all the tooling that we need to launch their programs. We do all that work, we send it to ESA. They do security hardening and sort of checking things over. So we are actually on the ISS LAN, so we have lots of rules to obey. And we test, we get the image running in flight, and we look to see if there's any issues with connectivity and stuff like that. So we do a bit of Earth-to-Space debugging, which is quite fun. We actually had an issue this year where we were using a Wi-Fi dongle, because when they put it in the window, they want to be able to transfer the files as they're recorded back to a Dropbox server on the station, and then they can download it from there. We'd been testing on a Wi-Fi dongle with the exact same product ID and the same packaging and the same serial numbers and stuff like that. But it turned out to be an entirely different chip on the inside, so the kernel drivers that we built for the Wi-Fi driver just turned out to be completely incompatible. So we had to try and find the right one with the right USB IDs and remake those kernel modules and do a separate payload to send the new Wi-Fi drivers. So we have lots of fun things like that to deal with. So we test the Astra Pi in-flight, we resolve any issues, and then a later stage, once the kids code has been come in and we've judged it, we then do a payload where we send the student code in a zip file, we extract that and give these operatives all the instructions to what they need to do to run that. The student code then runs and we're logging what's going on the whole time and we transfer any output from that and the logs to the ISS Dropbox server. We're then able to download files to Earth and then ease to send them to us and we distribute them to the teams. But we also, when we're receiving those files, we actually check through all the sys logs and we find some errors in the code that weren't caught in testing. We get the chance to rerun some of those if we're lucky. So we literally have things like kids who've divided by a value and if that value is from the accelerometer, then it's recording the G-force. Of course, in our testing, that didn't come up, but you send it to space and they're dividing by zero. That's always fun as well. So some of the, a couple of the winners from this year, they're really interesting. So this is a team called the Fire Watchers from Portugal. So they were one of the teams that I said that was detecting evidence of wildfires. They used open image databases, things like that, they could get from NASA and they're comparing images and pulling interesting information out. And using the, because it's an infrared camera that's looking out the window, there's extra information they get from the near-infrared so they can analyse vegetation and plant life and do a lot of maths and stuff on that too, to get some information out. And they end up with some really cool, really cool looking pictures as well. Another team from Poland called the Happy Pie, they were analysing photosynthesis and the observability of underwater life. They've got some really interesting kind of fractal images on the sort of mountainous landscapes and things like that and did some, again did some post-analysis on their images, which is really cool. The winners all get together and have a webinar with a former astronaut and ask them questions about things, which is really interesting for them. Here are some photos that were taken in the last couple of years so I'll just skip through some of these, some really interesting sort of cloud stuff and those sort of landscapes and coastlines. That's kind of the sun, a really odd angle, kind of hitting across some, so I can't tell if that's land or cloud. You've got really interesting sort of hurricaney looking stuff and big ice, patches of ice. This is a selfie I took. It's kind of cool. And there are, these three are, this is actually from this year, there was a team that managed to capture this and there are three images of the, it was the Soyuz module docking with the ISS and it was just really, really lucky timing that they managed to capture it. They're looking down at Earth and the Soyuz is sort of slowly reaching and coming towards the ISS. It's just really, really lucky. And so another thing we're considering is to add a third mission. So there's something we're probably going to be piloting in the next run. Something that sits between mission zero, the really easy low barrier to entry competition, a mission space lab which kind of takes up your whole time for the year, something with less time commitment than that, varying levels of difficulty perhaps and something that doesn't require ISS involvement. So what we did was we logged a load of data this year. We ran some extra experiments, logged data, we took photos, so we've got kind of a good bank of images that we can use and we're looking to do kind of small periodic challenges that people could drop in and out of and do rather than having to commit to doing the whole project for the whole year. So this is something we're considering this year because working with the ISS is really, really cool and it makes it really interesting but it's a lot of work to make this kind of happen. So anything that we can do to run more activities without that involvement would be really good. So, as I say, we logged 24 hours worth of data so you know where the ISS was all the time and we logged all the sensors and took a time lapse. This is actually one of the images that came out of it. So that's something we're going to be piloting and kind of putting out challenges so that people can try if we give them something to think of, a bit like kind of project oiler stuff but based on data analysis of this data. So how can you help? So come and talk to me today if this is interesting to you or you just want to know more. Share your ideas with me for these challenges and that kind of thing and share your interest. Anyone in this room could mentor a young person or group when the challenges launch. We launch everything in September so Mission Zero will open. If you've got kids you can let them go to Trinket and do Mission Zero and submit their code and they'll get their certificate towards the end of the year. But you can mentor a group doing Mission Space Lab so actually help them if they're doing more advanced stuff. Having access to a professional Python programmer would be really advantageous and be really rewarding for you as well. And a good way to actually do that in a sort of formal way is to volunteer at one of the code clubs or coded dojos. If you look at those websites you'll be able to find more about how you can get involved locally and you can always contact me if you want to give you some pointers. If you've got kids yourself in school you can talk to their teachers about AstroPy tell them about it and offer your support as well and watch astropy.org and Raspberry Pi and AstroPy on Twitter for general updates. And again the link to the slides the link's feedback and interest so if you're interested in following me following up with you about this if you want to know more about volunteering or getting involved with AstroPy you can fill out the form and contact me through there. So I'll just leave this video running. I've cut out the nighttime parts so you just see the transition there but this is a video from three hours one of our three hour runs on the ISS. But I'm ready to take questions while this is running. Thank you Ben. We have five minutes for questions. Please go to a microphone near you. Thank you for an interesting talk. What's the kind of breakdown geographical breakdown of the countries of the children that are entering this competition? Is there a high proportion from any particular country and also from the analysis of these experiments that would have found anything particularly particularly interesting from analysing their data? I'm trying to think what I know about the country breakdown I don't really have it to hand but I know there's obviously a big participant a big number of participants in the UK because we're based in the UK and a lot of the stuff that we do is in English and is local to us it's easier for us to reach people but we do, ESA do operate in various countries as well they do a lot of outreach I can't really think off to my head there is a good spread across the ESA countries across Europe I can't quite think of what the distribution is like. On the second question there's lots of interesting things if we look through the reports and we get we've got something like 50 reports or something this year there's interesting things that people have found in things like if they were looking at forest depletion and lake depletion and things like that there have been noticeable discoveries that they've found there's lots of things like that Do you publish the code at all? We don't we sort of put that down to the the mentors to whether they want to do that Any more questions? What sort of connectivity is there? You mentioned a drop box so what kind of connectivity is there up and down between the space station and earth and how is it managed presumably you can't just add any old thing in there is that strongly, strictly policed as well? Yeah, so the there's a a company that you used to work with that do all the operations, the technical operations and things they do pretty much have any time access so when these experiments were running they were running for several days and we were getting daily updates we were getting syslogs and stuff like that and we were analysing what was going on so we did have that having that drop box made it a lot easier and that's why it's it's easy for them to reach that but they do have a solid I don't know what kind of speeds they're getting or anything like that but it's and the downlink of the files there's several gigabytes at a time and we're getting that quite quickly More like a personal curiosity has there been any project that turned the camera out there instead of towards the earth that interprets what's happening beyond the earth in outer space? So they don't have the opportunity to do that, they're just the asherpies placed in the window pointing at earth so they don't have the opportunity to look at a different angle unfortunately it's quite difficult to organise things like that so we don't have a lot of flexibility we just have to get what we're given but it would be really interesting if we have been talking about that and see if we can capture the curvature and things like that We have them for one more question Just out of curiosity, so thanks for a nice talk first How does the data compare to the data available in the public domain so your next task is data analysis task essentially so I'm just curious if GITS could grab something similar or is it some unique data set that you have? I mean I can't really say that there's any difference because it's it's just data any collected data is going to have differences but there's no distinct observable Do you get the data from ISS or any other space programs? Is there such data from sensors and cameras available in public domains? Sensors on the station or? I have no idea I'm just curious I'm not really sure I mean if there's sensor data available it will be interesting to compare it I haven't done that myself Thank you You're welcome