WEBVTT 00:00.000 --> 00:20.000 Thank you all for coming here, you're in for a ride because we're talking about maps and maps of the best. 00:20.000 --> 00:25.000 So welcome Petia from Humanitarian Opposite Map Team talking about drones. 00:26.000 --> 00:27.000 Thanks. 00:31.000 --> 00:35.000 Thank you everyone, thank you for joining this session from drones to data. 00:35.000 --> 00:41.000 So you hear a lot about mapping and you hear about humanitarian open street map teams. 00:41.000 --> 00:47.000 So I'll just start with a raise in terms of how many of you know or use open street map. 00:47.000 --> 00:50.000 I knew there would be a lot. I think it's the best conference. 00:50.000 --> 00:52.000 I go to a lot of events and it's less people. 00:52.000 --> 00:54.000 So thank you for joining today. 00:54.000 --> 00:57.000 So my name is Ilya Set, Petia Congalova. 00:57.000 --> 01:01.000 You can see my long job title, but as you can see it's not a developer. 01:01.000 --> 01:09.000 So you very much hear a lot about practical examples and some of the open source tools that we use. 01:09.000 --> 01:16.000 Yeah, I'm not a developer, but I'm somebody who's worked in international humanitarian development sector for quite some time. 01:17.000 --> 01:23.000 And I'm very passionate about the role of tech for good and open source tools that are really usable. 01:23.000 --> 01:30.000 And that's where I feel very passionate about what we do at hot, because I think it's a alliance with a lot of my mission. 01:30.000 --> 01:34.000 And it's my first time at post them. That's why I added this a little logo. 01:34.000 --> 01:42.000 So it's been overwhelming, but also really really exciting and really nice to see so many passionate people in one place. 01:43.000 --> 01:45.000 Thank you. 01:45.000 --> 01:55.000 What I'll try to do in the next 20 minutes is tell you a bit about the organization and talk about what we're saying this open mapping ecosystem for all. 01:55.000 --> 02:08.000 So what it means, what are the these examples and disaster response in humanitarian action and talking about the tools, which I'm sure you'll be excited to hear how you can get involved and hopefully time for questions. 02:09.000 --> 02:17.000 So hotter the humanitarian open street map team started 15 years ago, basically response to the earthquake in Haiti. 02:17.000 --> 02:23.000 So when the earthquake happened, there were no maps available for a lot of humanitarian respondents to reach. 02:23.000 --> 02:32.000 At that time, hot was the number of volunteers that developed at that time one of the tools, how the task manager you hear more in a bit. 02:32.000 --> 02:47.000 So we grew and I'll tell you about the team and the volunteers, but at the core of it is why we exist is very much as this mission statement you can say that we have local community source map data is accessible. 02:47.000 --> 02:56.000 So people who live in the regions which I'll talk about basically have can generate this open map data and that data is used for decision making. 02:56.000 --> 03:11.000 So that means sometimes local governments making decisions about the specific infrastructure, sometimes it's around like specific routes and sometimes it's about reaching people specifically in like poster quick and in disasters. 03:11.000 --> 03:19.000 What is the core of it? I'm not going to go longer in this slide, but often get off. What's the difference between hot and open street map? 03:19.000 --> 03:33.000 Obviously open street maps, you know, you're over where it's been existed for over 20 years and it's at the core. It's one of those core platforms of what we use and what we do. 03:33.000 --> 03:38.000 And this is the difference which is kind of in simple terms trying to explain that. 03:38.000 --> 03:46.000 A bit of the role of when we started like 15 years ago is strengthening the open street map communities and open data focusing on those regions. 03:46.000 --> 03:53.000 So you can see Asia Pacific, Eastern South and Africa, Western Northern Africa, Latin American, the Caribbean. 03:53.000 --> 04:05.000 So as our organization, we are an NGO, we have around 70 people, hopefully remote most of my colleagues are based in the regions and doing all the implementation of projects. 04:05.000 --> 04:13.000 And I'm today also joined by SAHA, our tech leaders at the front and she'll be able to answer all your very very very nice questions as well. 04:13.000 --> 04:17.000 So these are the places where we operate. 04:17.000 --> 04:23.000 So what do we mean by this open mapping and maybe sounds abstract, open mapping ecosystem for all? 04:23.000 --> 04:34.000 So one of the core things that we believe is that when we speak about disaster response or people having access to data is that they really need accessible technology. 04:34.000 --> 04:37.000 That's open and free for anyone to use. 04:37.000 --> 04:44.000 So it's a very kind of low barrier to access to people who maybe don't have GIS knowledge or have more detailed experience. 04:44.000 --> 04:47.000 And what we say in terms of like vulnerable communities. 04:47.000 --> 04:55.000 So they're able to contribute to the map and also use the map once it's generated. 04:56.000 --> 05:04.000 Free simple steps in terms of explaining this like mapping process and that also then relates to the tools that I'll talk about. 05:04.000 --> 05:11.000 So for those who've been in the geospatial truck crew probably noticed very well, but the first step is very much imagery. 05:11.000 --> 05:15.000 Like if you have no imagery, you're not able to create a map. 05:15.000 --> 05:18.000 So that's normally satellite or drawing imagery. 05:18.000 --> 05:26.000 The second step is the digitalization or a lot of it is now that manually with a lot of volunteers of drawing this lines. 05:26.000 --> 05:30.000 Polygons of where the buildings are where their roads or waterways. 05:30.000 --> 05:36.000 And the third and very kind of simplified terms component is actually really critical, which is the local knowledge. 05:36.000 --> 05:39.000 We can also see, okay, this is the building, but what is it? 05:39.000 --> 05:47.000 So if we're here in Belgium and this is maybe a place in Kenya, we wouldn't know what that building is. 05:47.000 --> 05:58.000 So having the tools where local communities can undone knowledge is that a hospital is that a local sense or maybe it's a critical component. 05:58.000 --> 06:09.000 So what that looks like, like those three steps and this is what we've been focusing is this end-to-end mapping journey. 06:09.000 --> 06:17.000 So what we think of it is what if local communities have no, there's no imagery. 06:17.000 --> 06:22.000 So from nothing to actually being able to create a map that's being used. 06:22.000 --> 06:25.000 So this is the journey which I'll talk you through. 06:25.000 --> 06:30.000 The first step is basically the imagery, satellite or drawn imagery. 06:30.000 --> 06:44.000 Then moving into remote mapping using the tools AI assistant mapping, adding the local knowledge and then having tools to go and to kind of down being able to download or use the map. 06:44.000 --> 06:53.000 So before I jump into water, the tools that are related to each part of what I've just explained is end-to-end mapping solution. 06:53.000 --> 07:00.000 I wanted to talk some examples when I say local communities can have the tools so that they're used. 07:00.000 --> 07:04.000 So we have tried this journey, so hopefully you stay with me. 07:04.000 --> 07:19.000 Some examples where you can see in Bali, Nepal, Argentina and Sierra Leone, where we've done this whole process of starting from the aerial image of the way to generating a map. 07:19.000 --> 07:26.000 So I'll just give you two examples and a lot more that you can also read on the website. 07:26.000 --> 07:32.000 So this one is in, you can see here the map is pre-town and Sierra Leone. 07:32.000 --> 07:37.000 So what we were doing there is supporting communities in informal settlement mapping. 07:37.000 --> 07:42.000 So working directly with some local organizations, some dwellers international, 07:42.000 --> 07:55.000 basically working with them to collect the strong imagery to the field mapping and generate the maps that are then used in that cases to have insights into risk hazards, 07:55.000 --> 08:01.000 but also what I mean by services in these cases was also disability access. 08:01.000 --> 08:08.000 Another one which I wanted to play a short video, but I don't think that's going to work, so you can look it later in the slides. 08:08.000 --> 08:20.000 There is an example of the Balinese disaster management agency where this whole process was followed to basically map the volcano evacuation routes. 08:20.000 --> 08:28.000 And actually sound, you can talk to him later with also part of working locally with the disaster aid and so apologies. 08:28.000 --> 08:29.000 You can watch that. 08:29.000 --> 08:33.000 I wanted to just give you a bit more local context into that video. 08:38.000 --> 08:48.000 Just refresh. 08:48.000 --> 09:05.000 Hey, so the tools, so hopefully you stayed with me, I'm like, okay, how do we start in this journey of coming from, you know, no data to actually having a map. 09:05.000 --> 09:11.000 So these are the tools that we have developed and are maintaining at the moment. 09:11.000 --> 09:15.000 There are just different stages of development and you can also take a look at them. 09:15.000 --> 09:19.000 So what is the gap that we are filling with the technology development? 09:19.000 --> 09:22.000 So the first one, as I said, is you need imagery, right? 09:22.000 --> 09:28.000 Satellite imagery is very expensive and it's also in a lot of areas where we are working with. 09:28.000 --> 09:32.000 It doesn't have the resolution that is needed for mapping. 09:32.000 --> 09:36.000 So one of the tools that we've developed is called the drone task manager. 09:36.000 --> 09:42.000 It's basically allowing people to create local area imagery. 09:42.000 --> 09:52.000 And it also, the main difference to maybe either the other software that's available is that it allows you to split the flight plan into different tasks. 09:52.000 --> 09:53.000 Why? 09:53.000 --> 10:01.000 So in the example, for instance, in Sierra Leone, you had a lot of local community members that were trained to do that and they can split. 10:01.000 --> 10:09.000 So you have many people that can collect the drone imagery and then that imagery is uploaded, processed and stored. 10:09.000 --> 10:19.000 And then it goes into open area map, sure, maybe you're familiar with that, which is basically a platform where you can search, share that open imagery. 10:19.000 --> 10:29.000 So in the project in Sierra Leone, when this imagery was collected, it's also plotted into an open platform that's then used by others as well. 10:29.000 --> 10:39.000 So what happens next? You have the imagery in order to generate, we use again the example in Sierra Leone, those informal settlements, it's the remote mapping part. 10:39.000 --> 10:51.000 So if you're familiar with mapping, maybe it has the tools that you're aware of or if you're new, I would encourage you to check it out because that's your way of contributing to mapping or volunteering. 10:51.000 --> 10:59.000 So it uses that imagery and then the main difference is it creates a project for a certain area that splits it into tasks. 10:59.000 --> 11:15.000 So we have you can see here nearly half 1500,000 volunteers, not active all at the same time, but all of them select tasks and they do manually the mapping, which is whether it's buildings, roads, waterways, depends on the projects. 11:15.000 --> 11:29.000 So this is part of the digitalization, so we have the imagery and then the next step is adding the information related to, like whether it's building on roads or what's specific to the project. 11:29.000 --> 11:44.000 Another thing that we've been exploring of the last few years, it's called Fair, it's not fair for the principles where we've used the similar abbreviation, and like the RNA starts more for resilience, responsibility to the communities we work in. 11:44.000 --> 12:03.000 You might ask why, you know, it's a basically a service to assist mappers, and the main difference to anything else that's available globally is that a lot of like global models say models are first of all not open, but they're also not trained on the data sets and the communities where we work. 12:03.000 --> 12:20.000 If you use that model in a remote village in Kenya, it's not going to support you a lot. So with Fair, the main thing is richly to have open models to train them locally with the communities who are there, who can share the feedback directly in that specific area. 12:20.000 --> 12:25.000 And then support the mapping effort, which was also something I was trying to see earlier. 12:26.000 --> 12:40.000 And if you're still with me, so we had the imagery, we talked about this kind of digitalization step, and then the last part, remember that slide was okay, the local knowledge, right, if that's a building, then what, what does it mean? 12:40.000 --> 12:53.000 It's the additional information that's added, and that's another to all the field task manager that we have, and all there are so many open mapping, and especially field mapping applications. 12:53.000 --> 13:08.000 The key difference with this one is that it supports the efforts to coordination. So what I mentioned, a bit in the drone part, as well in the task manager is split areas into tasks that allowed coordination, which is really important, 13:08.000 --> 13:24.000 like in disaster response in areas of trying to coordinate and collect data much faster. And yeah, it's used, it's a standalone mobile and when application that uses ODK and Q field, I'm sure you probably joined some sessions about Q field as well. 13:24.000 --> 13:53.000 And another thing that's another tool in that ecosystem and that field mapping that was actually developed very much with the user's user focus is a lot of communities that we work with, they have no audio space and knowledge, and they're like, they don't maybe not considering or like doing the mapping that is explained earlier, and they very much said like, oh, a lot of in disaster communication happens to exist in apps, right, whether it's WhatsApp or signal. 13:53.000 --> 14:19.000 So this is when we develop something called chat map, which basically, it was used and you see one of the apps in response to Hurricane Melissa in Jamaica, so it's a group of people where they just share their location and a photo or a video and then easily this information can then be extracted and visualized directly. 14:19.000 --> 14:34.000 So we talked about the imagery, the digitalization, we talked about like the local data collection, and then in terms of like this end-to-end journey is like using the data. 14:34.000 --> 14:43.000 For instance, the data that goes by the task manager in top and street map, the export to maybe some of you have heard of it. 14:43.000 --> 15:02.000 It's another tool we maintain and it's you can download open street map data, but also some of the ways in which we also connect, especially with a lot of humanitarian organizations that need the data, the humanitarian data exchange, it pulls a lot of open street map data. 15:02.000 --> 15:08.000 There are different data sets there, but open street map is one of the main ones. 15:08.000 --> 15:16.000 And when we think about like the visualizations and the maps, again, there are different open source tools. 15:16.000 --> 15:28.000 We made an instance of you map, which is actually open street map, France is developed, but it's something that our community really needed and visualizing quickly like data that's been collected and that's used. 15:28.000 --> 15:39.000 Maybe that's a bit blurry, but that's again the response for Hurricane Melissa and Jamaica, and you can see kind of photos of before and during the storm. 15:39.000 --> 15:52.000 And I'll have a few, I think, two more slides, but I wanted to again bring back that the open source tool that I spoke about, the key focus of them is really accessibility. 15:52.000 --> 16:01.000 And being able to have access to generate a map that's community-owned and then it's used. 16:01.000 --> 16:08.000 It's in disaster response, like there are so many different projects now in climate action in health and others. 16:08.000 --> 16:14.000 And through these tools, this can happen at the moment. 16:14.000 --> 16:24.000 Okay, so the people before I finish, which I think is a critical part, is everything that I talked about is the huge amount of effort for us. 16:24.000 --> 16:42.000 I wanted to say huge thanks to the tech team, tech and data team, you know, for an NGO, it's a big tech team, but I would say for the number of tools that's like we're maintaining is actually quite a small team, but I want to say huge thanks to the whole team and all their effort. 16:43.000 --> 16:47.000 And also to all the volunteers and contributors. 16:47.000 --> 16:59.000 When I speak about like volunteers, especially mapping is like all the people that contribute to the map, but also the people that contribute to the software, like the open source contributors. 16:59.000 --> 17:07.000 And here is like the cure called is just more recent blog that we share the stories of some of our open source contributors that. 17:07.000 --> 17:22.000 Yeah, I want to say huge thanks to thanks to them and to all the volunteers, because that's especially with open mapping, nothing will be possible without the volunteers and the contributors. 17:22.000 --> 17:37.000 And I hope I'll have to time, but I wanted to end on ways to get involved or things that you can check, so obviously the website, if you want to read more about different examples and use cases, you know, so read about the tools in there. 17:37.000 --> 17:47.000 Of course, GitHub, I'm sure you've been asked 10,000 times right in this confidence about contributing, but we always welcome contributions on GitHub. 17:47.000 --> 17:57.000 And all the projects that I've mentioned, you see the repositories, their current current volunteer projects and in ways to get involved. 17:57.000 --> 18:05.000 We also run different working groups, so they're focusing on kind of some of them are more broad like community, there's governance and others. 18:05.000 --> 18:16.000 One that I lead is our tech and innovation, it's an open space, so if you're interested in joining with sometimes do like testing on some of the tools or discussions, but it's an open space for the team. 18:16.000 --> 18:19.000 Community to get together once a month. 18:19.000 --> 18:26.000 And yeah, it's like an email and I want to say thank you and hopefully this time for questions. 18:27.000 --> 18:47.000 So, thank you, bit, bit, yeah, there's definitely a time for questions, so I already see a fence, so please. 18:47.000 --> 19:00.000 Thank you very much for the talk. Can the tools, et cetera, you've been using, you've been talking about be used for sort of individual research projects in anthropology or archaeology or similar. 19:00.000 --> 19:07.000 Not, not new results to be released publicly, but not kind of public agency engagement. 19:07.000 --> 19:26.000 So the tools are open and they can be used for, yeah, for any any purpose, it depends on yeah, I think which tool you refer to, but obviously what we do is support like communities with like training specifically in disaster response, but they're open to I know if some I have my colleague for. 19:26.000 --> 19:34.000 More technical questions, but yeah, they are and if you do let us know if you have any questions, it's yeah, they're open for anyone to use. 19:34.000 --> 19:45.000 Thanks for a great talk. I'm curious about this tool field TM, I think it was called it seemed very similar to street complete. 19:45.000 --> 19:52.000 Could you talk a bit about the differences or if there could be some some cross collaboration there, maybe. 19:52.000 --> 19:55.000 Here's the developer. 19:55.000 --> 19:59.000 So yeah, I've been working on the field TM for a while. 19:59.000 --> 20:07.000 Street complete is very much a more of an individual experience right, you go out with your mobile and you collect data as an individual. 20:07.000 --> 20:15.000 The whole point of field TM is to coordinate field mapping, so it's using existing tools like Q field, ODK. 20:16.000 --> 20:24.000 We take a area that you need to map, we subdivide it into field mapable chunks, so trying to avoid like, 20:24.000 --> 20:30.000 traversable non-troversible linear features like roads and rivers and so on. 20:30.000 --> 20:44.000 And then each task is then kind of sent to the underlying tool, so that could be street complete if you know we make an integration there, or Q field ODK, like the underlying kind of survey data collection up. 20:44.000 --> 20:47.000 Sorry, the answer is. 20:47.000 --> 20:53.000 The only thing I'll add is here, some explain the difference, but in our community people use street complete organic maps, you know, 20:53.000 --> 20:57.000 there's a lot of like mobile applications for adding all the street map. 20:57.000 --> 21:00.000 That's not coordinated, but it's. 21:00.000 --> 21:03.000 Quick question here. 21:03.000 --> 21:13.000 So do you use or do you plan to use like different sensors to like, or what are you using, what kind of data you're using right now? 21:13.000 --> 21:22.000 Is it optical imagery, so RGB, classical RGB, radar, radar, radar, maybe a combination of those? 21:22.000 --> 21:26.000 Yeah, it's, it's primarily optical imagery for now. 21:26.000 --> 21:31.000 I think open area map is pretty much like predominantly optical. 21:31.000 --> 21:42.000 Obviously more unlike the, the fair kind of the AI stuff, there's been some combinations of, you know, maybe inside data, maybe we could use thermal infrared, 21:42.000 --> 21:45.000 or other, and particularly with drones as well. 21:45.000 --> 21:51.000 This is all very much more like research, kind of rather than production tools. 21:51.000 --> 21:58.000 So it's all optical for now. 21:58.000 --> 22:00.000 Hello. 22:00.000 --> 22:04.000 Thank you for the great talk. Just relevant to the description as well. 22:04.000 --> 22:12.000 But is it any drone, and don't can be used or like commercial or custom, maybe custom better? 22:12.000 --> 22:13.000 Yeah. 22:13.000 --> 22:15.000 Yes, a good question. 22:15.000 --> 22:26.000 So currently the drones that we recommend using are two specific drones that DJI many for pro and the DJI many five pro. 22:26.000 --> 22:29.000 We do support many of it DJI models as well. 22:29.000 --> 22:33.000 Lots of the workflow is currently built around DJI drones. 22:33.000 --> 22:48.000 We do, we've recently added support for Argypilot, more acute ground controls, like more open source DIY drones that you may build, or, you know, where you load your own software. 22:48.000 --> 23:02.000 We're targeting less that kind of market, right? Because this is more making things accessible for, you know, people in developing countries and tends to be easier to get hold of, you know, commercial. 23:02.000 --> 23:05.000 So it's changing over time. 23:05.000 --> 23:12.000 The main issue here is around APIs that are available, you know, and SDKs, I should say sorry. 23:12.000 --> 23:21.000 So most major manufacturers these days don't release, you know, easy ways for you to integrate with their drones. 23:21.000 --> 23:29.000 So it will change over time, but currently we also support the Potensic atom one, as a very cheap lightweight drone. 23:29.000 --> 23:31.000 And it costs about $300. 23:31.000 --> 23:35.000 So that's the cheapest drone that we support with some caveats. 23:35.000 --> 23:37.000 But this will change. 23:40.000 --> 23:44.000 Yes. Thank you for the presentation and for the amazing work. 23:44.000 --> 23:57.000 I was wondering, how do you address the data protection issues, which may stem from the aerial images, like photography cars or people, how do you do it? 23:57.000 --> 24:01.000 Yeah. So I guess it's more based on best practice. 24:01.000 --> 24:07.000 We tend to fly out shoots between like 120 meters. 24:07.000 --> 24:16.000 That would generate with the sensors used on the drones imagery that you can't identify people's faces at least. 24:16.000 --> 24:20.000 Obviously, you can identify, you know, buildings, that's the whole purpose of it. 24:20.000 --> 24:25.000 You can identify other features on the map, but, yeah, and personally identifyable stuff. 24:25.000 --> 24:28.000 It tends to be, it's not high enough resolution for that. 24:28.000 --> 24:42.000 If it was, then, yeah, we would definitely implement some, like, reduction of the resolution to a point where you can't identify before it's published at least. 24:42.000 --> 24:50.000 How do you go from road, road, road images to geolocated dr. Masek? 24:50.000 --> 24:55.000 Oh, we're just up there. So we use a tool called open drone map. 24:55.000 --> 25:08.000 So we're on very good terms with the team, open drone map, and that's kind of the predominant open source tool that we use to take, you know, thousands of raw image images, 25:08.000 --> 25:13.000 we have references them, and then Mosek's them enjoy single author Mosek. 25:13.000 --> 25:24.000 We also get 3D products as well, so we'll have point clouds, and we can make digital terrain models and various other things from that. 25:24.000 --> 25:28.000 Okay, this is all time we have. Thank you, Bertie and Sam. 25:28.000 --> 25:34.000 Thank you very much. 25:38.000 --> 25:43.000 Thank you very much.