Mapillary

Exploring the Evolution and Impact of Mapillary with Ed from Meta. 

Topics include Ed's journey with Mapillary, the process of uploading and utilizing street-level imagery, and the integration with OpenStreetMap.

Ed talks about the challenges of mapping with various devices, the role of community contributions, and future potentials in mapping technology, such as using neural radiance fields (NeRFs) for creating immersive 3D scenes.

The episode provides insights into how Mapillary is advancing geospatial data collection and usage.

00:00 Introduction to the Map Scaping Podcast 00:57

Meet Ed: Product Manager at Meta 02:09

Ed's Journey with Mapillary 03:59

What is Mapillary? 07:00

The Evolution of 360 Cameras 09:20

Uploading Imagery to Mapillary 14:10

Building a 3D Model of the World 19:10

Meta's Use of Map Data 21:24

The Importance of Community in Mapping 24:15

The Importance of Authoritative Data 24:49

Meta's Contributions to Open Source Geo World 25:27

Real-World Applications: Vietnam's B Group 28:16

Innovative Mapping in Detroit 31:38

Future of Mapping: Lidar and Beyond 32:20

Exploring Neural Radiance Fields (NeRFs) 35:40

Challenges and Innovations in Mapping Technology 45:25

Community Contributions and Future Directions 46:37

Closing Remarks and Contact Information

 

Previous episodes that you might find interesting

https://mapscaping.com/podcast/scaling-map-data-generation-using-computer-vision/

https://mapscaping.com/podcast/the-rapid-editor/

https://mapscaping.com/podcast/overture-maps-and-the-daylight-distribution/

 

 

 

Episode Transcript

(00:05):
Welcome to another episode of the Map Scaping podcast.
My name is Daniel and this is a podcast for the Geospatial community.
My guest on the show today is Ed. 4 00:00:16,117.323394982 --> 00:00:23,767.323394982 Ed is a project manager at Meta, a fantastic person, and today on the podcast we are going to be talking about Map Hillery. 5 00:00:28,582.323394982 --> 00:00:39,682.323394982 So I've previous previously published an episode about Ma Pilary and I'll link to that in the show notes and, and also include some other links in the show notes today to other relevant episodes that you might find interesting. 6 00:00:39,682.323394982 --> 00:00:49,972.323394982 I, I would be remiss if I didn't mention that this episode was recorded some months ago, but due to some technical issues and some major life up upheavals. 7 00:00:50,482.323394982 --> 00:00:51,382.323394982 It is only being published. 8 00:00:51,382.323394982 --> 00:00:51,652.323394982 Now. 9 00:00:51,802.323394982 --> 00:00:55,132.323394982 That being said, it's a fantastic lesson and I really hope you enjoy it. 10 00:00:57,897.323394982 --> 00:00:59,427.323394982 Hi, ed, welcome to the podcast. 11 00:00:59,547.323394982 --> 00:01:01,47.323394982 This is absolutely great to have you here. 12 00:01:01,47.323394982 --> 00:01:05,547.323394982 We've talked a bunch, but we've never recorded the conversations, and this is what we're doing today. 13 00:01:05,877.323394982 --> 00:01:08,487.323394982 Today I really want to have an update on map. 14 00:01:09,297.323394982 --> 00:01:15,627.323394982 I know you have a ton of, uh, history with, with this product, with this service, and that's what I'm keen to chat about. 15 00:01:15,957.323394982 --> 00:01:17,487.323394982 You're a product manager at Meta. 16 00:01:18,87.323394982 --> 00:01:24,567.323394982 Why don't we start there? Why don't you just give us a brief overview of who you are, what you do at Meta, and then your relationship to papillary. 17 00:01:25,797.323394982 --> 00:01:27,147.323394982 Sure Daniel, so hi. 18 00:01:27,207.323394982 --> 00:01:28,497.323394982 Thanks for having me on the podcast. 19 00:01:29,157.323394982 --> 00:01:31,227.323394982 So yeah, I'm currently a product manager at Meta. 20 00:01:31,227.323394982 --> 00:01:32,907.323394982 This is, uh, a new role for me. 21 00:01:32,907.323394982 --> 00:01:37,167.323394982 I've, I guess, had a history of working with a community around MAP blurry. 22 00:01:37,227.323394982 --> 00:01:47,247.32339498 Um, and then as maper became a part of meta four years ago, I started working on a lot of our strategy around open street map and community building. 23 00:01:47,637.32339498 --> 00:01:49,732.32339498 Given that Meta uses a lot of open map data. 24 00:01:50,397.32339498 --> 00:01:54,957.32339498 And so that was kind of my, my foray into product management and then started thinking about tooling. 25 00:01:55,227.32339498 --> 00:02:00,987.32339498 And so right now I'm working on a few different things as product manager at Meta, but Maps, tooling is one of them. 26 00:02:01,107.32339498 --> 00:02:08,7.32339498 Um, maybe we talk about some of the other stuff, but, uh, MAP has been something that I've been working in for around 10 years now. 27 00:02:09,177.32339498 --> 00:02:13,377.32339498 Could you just tell me how you got involved? What was the story there? Sure. 28 00:02:13,467.32339498 --> 00:02:16,947.32339498 So I was working or actually studying in Sweden. 29 00:02:16,947.32339498 --> 00:02:19,117.32339498 I moved to Sweden in 2000 and. 30 00:02:19,992.32339498 --> 00:02:22,902.32339498 14 and I was studying something completely unrelated. 31 00:02:23,322.32339498 --> 00:02:32,892.32339498 I was really interested in foreign tech companies in China, and as I was doing my masters, I, I needed a job and I was looking at interesting companies in Sweden. 32 00:02:32,892.32339498 --> 00:02:45,102.32339498 I know there were a lot of like incredible startups in southern Sweden and then across the water in Copenhagen and I found MAP and just emailed them pretty much almost to the day, 10 years ago. 33 00:02:45,537.32339498 --> 00:02:48,147.32339498 Just said like, Hey, I, I love what you guys are doing. 34 00:02:48,147.32339498 --> 00:02:55,257.32339498 Like this idea of building a, a database of the world using street level imagery and using phones to map the world. 35 00:02:55,257.32339498 --> 00:03:01,857.32339498 And yeah, just emailed them, asked if I could find a role there, and luckily they got back to me soon after. 36 00:03:01,857.32339498 --> 00:03:06,897.32339498 And yeah, that was, that was the very beginning of a really incredible journey with Mapley. 37 00:03:07,737.32339498 --> 00:03:09,417.32339498 Yeah, that, that's, that's really cool. 38 00:03:09,477.32339498 --> 00:03:14,967.32339498 So just, just to clarify, you had no sort of prior interest in, in mapping, uh, background in geospatial. 39 00:03:14,967.32339498 --> 00:03:15,327.32339498 You just saw this. 40 00:03:16,557.32339498 --> 00:03:18,732.32339498 Yeah, the the startup and thought, wow, that is awesome. 41 00:03:19,812.32339498 --> 00:03:26,502.32339498 Yeah, I had no background in mapping other than just loving it since I was a kid, like looking at atlases and maps and so that was the seed. 42 00:03:26,652.32339498 --> 00:03:33,972.32339498 And so when I was looking at interesting startups in Sweden and saw one that was related to maps, it like immediately jumped to the top of my list. 43 00:03:34,557.32339498 --> 00:03:36,987.32339498 And already like it was very early days of Ba Boy there. 44 00:03:36,987.32339498 --> 00:03:43,737.32339498 I think they had six people working on the, on the team, but there was already like an incredible community swell around it. 45 00:03:43,827.32339498 --> 00:03:53,127.32339498 Images were coming in from around the world and, and that was a really intriguing thing that I think I resonated with that I know a lot of other future employees and also people contributing, resonated. 46 00:03:54,402.32339498 --> 00:03:55,872.32339498 And I can see that now with the numbers. 47 00:03:55,872.32339498 --> 00:03:58,32.32339498 I think maybe we should dive into those a little bit later on. 48 00:03:58,422.32339498 --> 00:03:59,892.32339498 Maybe we've jumped in a little bit deep. 49 00:03:59,952.32339498 --> 00:04:03,912.32339498 There'll be some people listening to this that don't know what papillary is. 50 00:04:04,452.32339498 --> 00:04:09,237.32339498 Could, could you give us an overview? What was it when you started and what is it now? Sure. 51 00:04:09,477.32339498 --> 00:04:09,687.32339498 Yeah. 52 00:04:09,687.32339498 --> 00:04:16,797.32339498 So I, what's funny is I, in preparation for this, went back to your 2019 episode with YA, Eric, and listened to that and I thought, oh. 53 00:04:17,352.32339498 --> 00:04:25,212.32339498 This might be a bit outta date now, and I, I listened to it and it's like there was hardly anything there that I thought wasn't relevant today. 54 00:04:25,212.32339498 --> 00:04:32,202.32339498 And even if you go back to 2013 when map blur started, I think there's still a lot of relevance in the original idea. 55 00:04:32,202.32339498 --> 00:04:38,472.32339498 And so the idea of map ary is that, obviously map data is hard, and collecting that data is hard. 56 00:04:39,132.32339498 --> 00:04:44,562.32339498 But in 2013, all of a sudden we had these smartphones in our pockets that had GPS. 57 00:04:44,712.32339498 --> 00:04:46,152.32339498 The cameras were getting a lot better. 58 00:04:46,932.32339498 --> 00:04:55,32.32339498 And there was computer vision technology, there was improvements obviously in, in cloud services where you could scale a platform like this really quickly. 59 00:04:55,182.32339498 --> 00:05:08,982.32339498 And so the idea of MAP is to build a platform which people could contribute to easily, to with cameras and, and the smartphones to build a street level imagery platform and then use computer vision to derive map data from that imagery. 60 00:05:09,162.32339498 --> 00:05:12,612.32339498 And so we, we can talk a bit more about that and, and how that's useful. 61 00:05:13,47.32339498 --> 00:05:23,37.32339498 But it was this convergence of cameras and smartphones and GPS and, and the ability to scale a platform around that to get map data in all different places around the world. 62 00:05:23,877.32339498 --> 00:05:32,397.32339498 What happens to that map data? What, what is it used for and who gets access to it? So the map data is used for all sorts of different things. 63 00:05:32,397.32339498 --> 00:05:35,247.32339498 It's pretty much the majority of people contributing to. 64 00:05:36,102.32339498 --> 00:05:40,302.32339498 Map are doing so because they wanna understand the world around them in some way. 65 00:05:40,422.32339498 --> 00:05:49,452.32339498 They're building a map or they're updating a map, and that can be anything from asset management that cities are doing, keeping track of where their traffic signs are located. 66 00:05:50,532.32339498 --> 00:06:04,932.32339498 We've had a lot of use cases around the humanitarian sector and uh, identifying buildings that might be susceptible to flooding or working out which parts of the city have good cycling infrastructure and which parts of the city have bad cycling infrastructure. 67 00:06:05,472.32339498 --> 00:06:18,162.32339498 There's a variety of different use cases, and in the past, mapware used to make the data, uh, available for OpenStreetMap under the ODBR license, and then we would charge for commercial use. 68 00:06:18,192.32339498 --> 00:06:27,642.32339498 And one of the good things is when we joined Meta in 2020, we're able to open up a lot of the data and make it available for anyone who's interested in using it. 69 00:06:27,642.32339498 --> 00:06:32,652.32339498 And, and so there's, it's much easier now for a company or a researcher or anyone with a. 70 00:06:33,222.32339498 --> 00:06:34,452.32339498 Interest to use the imagery. 71 00:06:34,812.32339498 --> 00:06:39,447.32339498 And so we're seeing a, a kind of a resurgence of all these exciting use cases, which we, we can talk about. 72 00:06:40,462.32339498 --> 00:06:40,632.32339498 Yeah. 73 00:06:40,662.32339498 --> 00:06:50,292.32339498 Yeah, I, I'd really like to talk about that later on, but at the, the moment, so we understand that MAP is a, a platform where people can upload imagery to it. 74 00:06:50,847.32339498 --> 00:06:52,977.32339498 It says in my notes any device. 75 00:06:53,277.32339498 --> 00:06:56,307.32339498 Um, but mostly we've talked about, about cell phones at the moment. 76 00:06:56,307.32339498 --> 00:07:09,282.32339498 What other devices can we use to, to upload imagery to papillary? One of the big changes that I think has probably occurred since 2019 when you spoke to young Eric is just how many more? 360. 77 00:07:10,167.32339498 --> 00:07:10,587.32339498 Cameras. 78 00:07:10,587.32339498 --> 00:07:11,7.32339498 There are. 79 00:07:12,27.32339498 --> 00:07:24,237.32339498 And so if you go back to when I joined Mapware in 2015, I remember we would get these weird looking 360 cameras in the office, and some of them were Kickstarter campaigns and they were clunky. 80 00:07:24,237.32339498 --> 00:07:28,527.32339498 The software was fairly bad, the hardware was a varying quality. 81 00:07:28,587.32339498 --> 00:07:33,327.32339498 They were expensive, but these were kind of the pioneers in 360 cameras. 82 00:07:33,327.32339498 --> 00:07:37,917.32339498 And then since that time, I think in large part, due to companies like. 83 00:07:38,637.32339498 --> 00:07:40,707.32339498 Pro and, uh, Insta 360. 84 00:07:41,397.32339498 --> 00:07:46,317.32339498 There have been more a resurgence of good 360 cameras that can be used for mapping. 85 00:07:46,347.32339498 --> 00:07:51,687.32339498 And I think their primary use case was actually people taking reels and, and nice ski videos. 86 00:07:51,687.32339498 --> 00:07:52,827.32339498 Nice surfing videos. 87 00:07:53,247.32339498 --> 00:07:56,277.32339498 But we've been able to benefit from that, particularly the ones with GPS. 88 00:07:56,277.32339498 --> 00:08:06,957.32339498 So I mentioned smartphones earlier, but particularly over the last five years, we've seen much more of the imagery on our platform coming from cameras like the GoPro Max, which is just a very easy to use action camera. 89 00:08:07,632.32339498 --> 00:08:13,602.32339498 And then you have all the way through to like the professional grade cameras, the Trimble cameras, the mosaic cameras. 90 00:08:13,602.32339498 --> 00:08:22,692.32339498 Um, there are a bunch of names that may or may not mean something to the people listening, but these are the kind of cameras that collect very high resolution imagery. 91 00:08:22,782.32339498 --> 00:08:35,532.32339498 Um, they're quite expensive, but they're used by cities that are maybe doing asset management at a very high frequency and, and want to be able to zoom in and look at an address or they need that, that crisp imagery. 92 00:08:35,982.32339498 --> 00:08:37,62.32339498 So we have the full range. 93 00:08:37,857.32339498 --> 00:08:44,577.32339498 So my, my guess is with the higher resolution cameras, you can extract more details from those images and therefore derive more data from them. 94 00:08:45,912.32339498 --> 00:08:46,812.32339498 Yeah, precisely. 95 00:08:46,872.32339498 --> 00:08:58,212.32339498 Um, you can imagine if, if someone is, so we have our own computer vision algorithms running on the imagery, but then there's also people that are using map imagery and, and writing their own models to derive. 96 00:08:58,932.32339498 --> 00:08:59,652.32339498 Extra things. 97 00:08:59,652.32339498 --> 00:09:16,122.32339498 And so there's people that have been doing, uh, optical character recognition where they can read text in an image and then, you know, discern what kind of, uh, place am I looking at? What does the restaurant name say? And if you have a grainy image, like a lower quality image there, it's hard to read text. 98 00:09:16,122.32339498 --> 00:09:19,512.32339498 So that's one example where you, you really do wanna hire resolution image. 99 00:09:20,392.32339498 --> 00:09:30,792.32339498 And what does the upload process look like? How do I get these images into papillary? So for your smartphone, you are going out into the world using our app. 100 00:09:30,822.32339498 --> 00:09:37,812.32339498 Maybe you're mounting it to your windshield, or maybe you're walking down the sidewalk and capturing sidewalk imagery, but that is fairly easy. 101 00:09:37,812.32339498 --> 00:09:39,192.32339498 You, you, you've captured imagery. 102 00:09:39,192.32339498 --> 00:09:44,52.32339498 It captures imagery, uh, automatically as you go based on the distance that you've traveled. 103 00:09:44,202.32339498 --> 00:09:47,442.32339498 You go home and then as soon as you connect it to wifi, it starts uploading. 104 00:09:48,792.32339498 --> 00:10:03,672.32339498 For the action cameras, like a GoPro Max, which is a popular 360 cam, you can just take your SD card out, put it in your computer, drag the files to a desktop, uploading software that we have called the Mapware Uploader, and yeah, you just drag and drop it. 105 00:10:03,672.32339498 --> 00:10:12,912.32339498 And one of the improvements we've had over the last few years is we've added support for a bunch of different camera types, 360 cameras to make it that drag and drop process easier. 106 00:10:12,912.32339498 --> 00:10:17,22.32339498 So we handle the GoPro Max, we handle Insta 360. 107 00:10:17,487.32339498 --> 00:10:23,817.32339498 And there's a, a file format called cam, which is kind of a common 360 file format that we also support. 108 00:10:24,87.32339498 --> 00:10:28,317.32339498 So it's, it's much easier than it used to be to upload a lot of this 360 imagery. 109 00:10:29,487.32339498 --> 00:10:29,667.32339498 Yeah. 110 00:10:29,667.32339498 --> 00:10:33,897.32339498 Can, can I tell you a little bit about my own experience with the platform? Yes, please do. 111 00:10:34,137.32339498 --> 00:10:36,237.32339498 You can tell us the good and the bad and the ugly. 112 00:10:36,792.32339498 --> 00:10:40,137.32339498 so I got really interested in this, uh, a while ago now. 113 00:10:40,137.32339498 --> 00:10:44,697.32339498 I was like, oh, how can I get involved with this? I think I, I looked at using my phone. 114 00:10:44,697.32339498 --> 00:10:46,287.32339498 I was like, ah, that's not really the answer. 115 00:10:46,287.32339498 --> 00:10:49,227.32339498 I want something I can melt on my head when I bike around and contribute. 116 00:10:49,707.32339498 --> 00:10:55,287.32339498 It seemed like a really, sort of low friction way to contribute to the, to this awesome project. 117 00:10:55,692.32339498 --> 00:11:04,782.32339498 So I posted on LinkedIn, Hey, does anyone have any experience with this? What kind of 360 camera should I buy? Obviously, I'd read your blog posts and said, and I could see that you support all these cameras. 118 00:11:05,112.32339498 --> 00:11:16,2.32339498 And this guy, you might know him, Justin Meyers replied and say, Hey, a wee while ago, meta gave me this camera on the promise that when I was finished with it, I should give it to someone else who, who was gonna continue contributing. 119 00:11:16,542.32339498 --> 00:11:24,972.32339498 So he sent it to me, I put it on my helmet, started biking around my, my local town here, mapping away, I could see that there, there was a lot of places that weren't covered. 120 00:11:26,202.32339498 --> 00:11:29,412.32339498 And it was pretty amazing how easy it was, to be perfectly honest. 121 00:11:29,412.32339498 --> 00:11:37,722.32339498 That upload that you talk about is, is, is magical because it's literally just drag and drop the, the file into the folder and off it goes. 122 00:11:38,142.32339498 --> 00:11:38,862.32339498 And it was pretty cool. 123 00:11:38,952.32339498 --> 00:11:40,962.32339498 The whole experience was, was really seamless. 124 00:11:41,592.32339498 --> 00:11:47,742.32339498 And I can say that because I tried also contributing to, to other projects and it was painful. 125 00:11:47,772.32339498 --> 00:11:50,802.32339498 Like it was, it was, it was painful in comparison, so. 126 00:11:51,777.32339498 --> 00:11:52,737.32339498 Congratulations on that. 127 00:11:52,737.32339498 --> 00:11:57,777.32339498 It's a, it's a really easy way to contribute and I think you've made it really accessible for a lot of people. 128 00:11:59,367.32339498 --> 00:12:00,987.32339498 Huge credit to the team on that. 129 00:12:01,17.32339498 --> 00:12:14,817.32339498 'cause there's a bit of work that they put into, so there's a series of pythons scripts called Mapware tools, which is the underlying logic of how do you take this imagery or video and process it and, and how do you understand all these different file formats. 130 00:12:14,817.32339498 --> 00:12:19,917.32339498 And so they did a lot of that work and then incorporate it into that desktop uploading software that you used. 131 00:12:20,562.32339498 --> 00:12:24,312.32339498 It's really important because that's what makes people want to go out and do it. 132 00:12:24,312.32339498 --> 00:12:24,642.32339498 Again. 133 00:12:24,642.32339498 --> 00:12:30,912.32339498 If you had to spend your whole weekend writing scripts to process the imagery, you probably wouldn't do it very often. 134 00:12:30,912.32339498 --> 00:12:35,22.32339498 And so, yeah, we really wanna make it as easy as possible to contribute. 135 00:12:35,502.32339498 --> 00:12:38,712.32339498 And it's also good that just the cameras enable that too. 136 00:12:38,712.32339498 --> 00:12:41,592.32339498 That much smaller, they're less expensive. 137 00:12:41,742.32339498 --> 00:12:46,422.32339498 So thank you to Justin for sending it to you and getting us some coverage in beautiful New Zealand. 138 00:12:47,172.32339498 --> 00:12:47,322.32339498 Yeah. 139 00:12:47,322.32339498 --> 00:12:49,122.32339498 I, I really appreciate him doing that as well. 140 00:12:49,122.32339498 --> 00:12:51,762.32339498 And I'm, I promise to, to send it on when, when I'm finished. 141 00:12:52,212.32339498 --> 00:12:54,762.32339498 I also wanna say that up until now I've been talking about imagery. 142 00:12:55,377.32339498 --> 00:12:59,577.32339498 And it's important to understand that it's not just images that we're talking about, we're also talking about video. 143 00:13:00,27.32339498 --> 00:13:13,947.32339498 That surprised me a little bit because it seemed like, I guess, an extra pain in the ass on your side in terms of the processing, but it makes it a lot easier for people like me because I don't have to think about the frequency at, at which I take images. 144 00:13:14,247.32339498 --> 00:13:20,307.32339498 You know, I don't have to play around with like a, some sort of distance calculation every three seconds, every two seconds or, or whatever. 145 00:13:20,607.32339498 --> 00:13:23,97.32339498 I can just turn it on, go and turn it off. 146 00:13:23,457.32339498 --> 00:13:24,837.32339498 That was, I was quite surprised about that. 147 00:13:25,917.32339498 --> 00:13:33,267.32339498 Yeah, there were definitely challenges with some other cameras in the past because they had these intervals, so a lot of them had some sort of time-lapse mode. 148 00:13:33,267.32339498 --> 00:13:37,977.32339498 And then take a photo every five seconds or every two seconds and five seconds is a lot. 149 00:13:38,37.32339498 --> 00:13:42,717.32339498 If you're driving at a hundred miles or a hundred kilometers an hour, 60 miles an hour. 150 00:13:43,482.32339498 --> 00:13:46,242.32339498 There's a large gap and you might miss things like traffic signs. 151 00:13:46,242.32339498 --> 00:13:58,722.32339498 So that's where video's really helpful 'cause we can make a determination about what frame rate is useful to extract and then that just happens, uh, in the processing and you don't have to worry about it. 152 00:13:58,902.32339498 --> 00:14:02,862.32339498 So video's been something that we've started to support a lot more for that reason. 153 00:14:03,707.32339498 --> 00:14:06,147.32339498 So we talked a little bit about the, the data that you extract. 154 00:14:06,147.32339498 --> 00:14:11,652.32339498 Are you also building like a a 3D model of the world in the background? Yes, yes we are. 155 00:14:11,652.32339498 --> 00:14:16,182.32339498 So, ya, Eric spoke a bit about this, I think when he spoke to you last time and that's, that's still true. 156 00:14:16,182.32339498 --> 00:14:31,542.32339498 So once you upload the images around New Zealand, like every image, we process it and the first thing we do is blow faces and license plates and then delete that original image so that we don't have images that might have, uh, might be privacy compromised. 157 00:14:32,187.32339498 --> 00:14:45,687.32339498 And then what we do is we build a 3D model using a technology called Structure from Motion, which looks for different points across multiple images, which we, uh, using, um, computer vision can identify as the same. 158 00:14:45,987.32339498 --> 00:14:55,167.32339498 So maybe the corner of a building, we know it's the same corner of a building across multiple images and we can rebuild the world in, in 3D, uh, as a point cloud. 159 00:14:55,527.32339498 --> 00:14:57,987.32339498 And then once we have that point cloud, we can start to. 160 00:14:58,617.32339498 --> 00:15:03,837.32339498 Use semantics and, and, and our understanding of an image and try to overlay that onto that point. 161 00:15:03,837.32339498 --> 00:15:05,367.32339498 Cloud to position objects. 162 00:15:05,367.32339498 --> 00:15:13,467.32339498 So let's say we have an image and, and using semantics, we think we've identified a trash can or a crosswalk. 163 00:15:14,67.32339498 --> 00:15:21,387.32339498 We can then try and position that feature in that 3D point cloud and give you a estimated latitude and longitude of the object. 164 00:15:21,507.32339498 --> 00:15:25,557.32339498 And that's why a lot of people contribute to map is to be able to have that. 165 00:15:26,127.32339498 --> 00:15:30,837.32339498 3D processing take place behind the scenes, but then extract the map data out of it, not just the imagery. 166 00:15:31,902.32339498 --> 00:15:36,587.32339498 That must be a huge task given the different senses that you support. 167 00:15:37,737.32339498 --> 00:15:46,257.32339498 It's both a benefit and a challenge of map blur is there are so many different sensors, so many different cameras being used, so many different GPS thresholds. 168 00:15:46,287.32339498 --> 00:15:54,27.32339498 And so even, even internally sometimes, like there are conversations about what is the quality of map blur and it's hard to give a. 169 00:15:54,597.32339498 --> 00:15:55,257.32339498 Straight answer. 170 00:15:55,257.32339498 --> 00:16:06,987.32339498 I get this, I, I get asked this at conferences and in some places we have $50 Android smartphones contributing and that can be useful 'cause we're still, we're still getting photos that show what the world looks like. 171 00:16:07,587.32339498 --> 00:16:16,122.32339498 But then we can have places like Detroit, which have this incredible setup that're capturing pretty much every day of the year with a very expensive Trimble camera. 172 00:16:16,902.32339498 --> 00:16:25,212.32339498 And so we have the full range of quality, but the team then has to make sense of this and build a point cloud from different sources and different lighting conditions. 173 00:16:25,212.32339498 --> 00:16:29,502.32339498 And I, I think that's probably one of the most technically challenging aspects of malow. 174 00:16:30,97.32339498 --> 00:16:39,672.32339498 Do you specify that in the data that you derive? From, you know, from the imagery, from the video, like the sense that it was captured from and the estimated accuracy of that object. 175 00:16:40,872.323395 --> 00:16:55,692.323395 Yeah, the metadata of features generally contains the image it was detected in, and from that you can then discern which camera was used to contribute that image and, and so if you are using our API or doing research. 176 00:16:56,622.323395 --> 00:17:05,202.323395 And we, we see there's a lot people filter by particular users that they know contribute good quality imagery or by particular camera types that they know contribute to high quality imagery. 177 00:17:05,832.323395 --> 00:17:06,12.323395 That. 178 00:17:06,12.323395 --> 00:17:07,2.323395 That makes a lot of sense. 179 00:17:07,332.323395 --> 00:17:14,952.323395 And just can you gimme a quick overview of the kinds of features you are extracting? Yeah, it's been pretty consistent over the last few years. 180 00:17:14,952.323395 --> 00:17:24,672.323395 We we're looking at how we can grow the amount of features, but right now it's things like crosswalk, traffic signs, trash cans, bicycle parking. 181 00:17:25,452.323395 --> 00:17:29,82.323395 Streetlights utility poles, so, so those are all the map features. 182 00:17:29,82.323395 --> 00:17:32,922.323395 We're actually giving you an estimated lat long in space. 183 00:17:33,782.323395 --> 00:17:34,62.323395 Wow. 184 00:17:35,352.323395 --> 00:17:40,272.323395 And all of those are being put into OpenStreetMap and available through the API. 185 00:17:40,632.323395 --> 00:17:44,617.323395 Is that correct? There's a few ways, so you can access them through our API. 186 00:17:45,102.323395 --> 00:17:48,192.323395 You can access them via our web interface. 187 00:17:48,192.323395 --> 00:17:53,982.323395 We have a little toggle where you can turn on map data, zoom into an area, and then download. 188 00:17:54,627.323395 --> 00:17:57,57.323395 The map data, uh, that you've selected and filtered by. 189 00:17:57,657.323395 --> 00:18:01,707.323395 And then an open street map in rapid editor and some of the other editors. 190 00:18:01,707.323395 --> 00:18:07,317.323395 There's a layer that you can toggle that has our vector tiles with different map features. 191 00:18:07,557.323395 --> 00:18:09,537.323395 So it really just depends on what problem you're solving. 192 00:18:09,537.323395 --> 00:18:13,947.323395 But let's say you wanted to get them into open street map, maybe your mapping all the crosswalks. 193 00:18:14,397.323395 --> 00:18:21,57.323395 You could go into rapid editor, toggle the maple feature layer, see where all the crosswalks are, and then. 194 00:18:21,522.323395 --> 00:18:25,572.323395 Just zoom into an area and review the image, make sure that it's actually there. 195 00:18:25,722.323395 --> 00:18:27,192.323395 It hasn't been a false positive. 196 00:18:27,192.323395 --> 00:18:31,902.323395 And then you can add that to open Streete map so they're not automatically going into open Streete map. 197 00:18:31,902.323395 --> 00:18:37,212.323395 'cause Open Street map is very much about human curated data, but they're available for open streete map. 198 00:18:37,932.323395 --> 00:18:39,852.323395 As soon as I ask that question, it's like, oh, of course. 199 00:18:39,852.323395 --> 00:18:41,652.323395 They're not automatically going into Open Streete map. 200 00:18:41,652.323395 --> 00:18:46,32.323395 There needs to be a human in the loop and it makes a lot of sense to run 'em through Rapid Editor. 201 00:18:46,602.323395 --> 00:18:55,212.323395 And for for the listeners, if you are interested in, in the Rapid Editor, which you should be, 'cause it's incredible, I'll put a link in the show notes to an episode that we've already published around that. 202 00:18:56,172.323395 --> 00:18:59,742.323395 Was it 2020 that map joined Meta? Yes. 203 00:18:59,742.323395 --> 00:19:00,507.323395 June of 2020. 204 00:19:00,942.323395 --> 00:19:01,302.323395 Yeah. 205 00:19:01,482.323395 --> 00:19:18,777.323395 And what, what, what's the, like, how does it fit in with, with what, uh, meta is doing? How does Maps and papillary sort of enhance their organization? Yeah, so at Meta we are using maps and location in a variety of different ways, and we we're not the only company doing this. 206 00:19:18,777.323395 --> 00:19:25,137.323395 A lot of companies now, particularly as smartphone applications became more common, you know, 50, 10, 15 years ago. 207 00:19:25,137.323395 --> 00:19:30,42.323395 People need to answer the question of like, where am I? What's around me? There was kind of things, and that's just. 208 00:19:31,467.323395 --> 00:19:33,237.323395 Proliferated a whole bunch of use cases. 209 00:19:33,237.323395 --> 00:19:37,347.323395 And so for us, we have things like Facebook recommendations. 210 00:19:37,347.323395 --> 00:19:46,707.323395 You know, what, what do people recommend around me or on Instagram, maybe you are typing in the name of a restaurant and you wanna see a map rather than just, uh, just photos. 211 00:19:46,707.323395 --> 00:19:47,727.323395 You wanna see the map appear. 212 00:19:47,727.323395 --> 00:19:51,627.323395 And then, and maybe use that map as an interface to start exploring what's around you. 213 00:19:52,557.323395 --> 00:19:58,77.323395 Uh, we have wearable devices like rebound meta glasses, which are available now. 214 00:19:58,902.323395 --> 00:20:03,162.323395 There's the AI assistant and people wanna ask questions about what's around them. 215 00:20:03,162.323395 --> 00:20:14,772.323395 And so we have a, in a variety of surfaces within meta, where people need to know what's around me, where am I? And MAP data is an essential part of answering that question. 216 00:20:14,772.323395 --> 00:20:17,532.323395 So MapR is just one source to answer that question. 217 00:20:17,532.323395 --> 00:20:24,642.323395 Meta is, is very much involved in using and contributing to open data sources, though, to answer those questions. 218 00:20:24,642.323395 --> 00:20:27,432.323395 And so to that end, we have open street map, which. 219 00:20:28,137.323395 --> 00:20:31,407.323395 Meta has editing software like Rapid that we mentioned earlier. 220 00:20:32,7.323395 --> 00:20:36,387.323395 We make donations to the foundation, the Open Streete Map Foundation, O-S-M-U-S. 221 00:20:36,387.323395 --> 00:20:41,487.323395 But then we also are part of Over Chi Maps Foundation, which is like a multi-company collaboration. 222 00:20:41,487.323395 --> 00:20:46,917.323395 And so MapR is just one of these data sources that's feeding in and, and making better map data. 223 00:20:47,307.323395 --> 00:20:52,767.323395 And, and we really wanna do this in the open 'cause we think that we know the maps are incredibly hard. 224 00:20:52,797.323395 --> 00:20:55,977.323395 The world is changing very rapidly, particularly around things like. 225 00:20:56,457.323395 --> 00:20:57,687.323395 Places, points of interest. 226 00:20:58,347.323395 --> 00:21:01,527.323395 I've seen all sorts of stats, but sometimes they're changing it. 227 00:21:02,67.323395 --> 00:21:11,307.323395 I, I've seen stats that they say change 30% on average each year, so that a place, you know, if you have a hundred places, 30 of them won't be there anymore by the end of the year. 228 00:21:11,577.323395 --> 00:21:22,707.323395 I've seen other stats that put it higher, but the bottom line is they're changing rapidly, and so we need things like street level imagery to be able to get a sense of what's on the ground, has it changed, and make sure our maps are accurate. 229 00:21:23,947.323395 --> 00:21:24,87.323395 Yeah. 230 00:21:24,87.323395 --> 00:21:28,827.323395 And this, um, your approach to collecting it by building a community, I think is really interesting. 231 00:21:28,887.323395 --> 00:21:30,717.323395 Not always easy. 232 00:21:30,717.323395 --> 00:21:34,797.323395 I would imagine a communities that they, they must be difficult to sort of foster nurture. 233 00:21:36,627.323395 --> 00:21:48,687.323395 I think every company in maps is trending this way though, that the traditional approach of just having some authoritative source decide is, is not the only way there. 234 00:21:48,687.323395 --> 00:21:49,802.323395 There's still importance for that. 235 00:21:50,922.323395 --> 00:21:56,232.323395 I think most companies that are serious in mapping are trying to take a multi-pronged approach. 236 00:21:56,232.323395 --> 00:22:01,542.323395 So there are cases where you want high degrees of confidence in the data, like administrative boundaries. 237 00:22:01,542.323395 --> 00:22:09,252.323395 You, you really wanna be careful with when you're taking that data, particularly national borders, but then even county boundaries. 238 00:22:09,252.323395 --> 00:22:17,712.323395 Maybe the government's a good, good source there, but there are other things where that crowdsource component is, is really valuable and places is an example of that. 239 00:22:18,717.323395 --> 00:22:25,137.323395 Google Maps has done a really good job there of building that feedback mechanism to get all their users who are looking at the map daily. 240 00:22:25,317.323395 --> 00:22:37,827.323395 Also then giving feedback back to them about which places are out of date, and so whether it's Google Maps or Apple Maps or no meta and yeah, our involvement in OSM and o ture crowdsourcing is a, is a key part. 241 00:22:37,827.323395 --> 00:22:39,327.323395 But yeah, you're right, it's challenging. 242 00:22:39,327.323395 --> 00:22:41,997.323395 It can lead to issues around quality. 243 00:22:41,997.323395 --> 00:22:45,867.323395 It can lead to issues around vandalism sometimes. 244 00:22:45,867.323395 --> 00:22:46,977.323395 That's something that we deal with a lot. 245 00:22:47,922.323395 --> 00:22:58,32.323395 But it, it, it's definitely like a very minor, minor price to pay for the brilliance that you get from just having the variety of people and, and the diversity of people. 246 00:22:58,32.323395 --> 00:23:01,272.323395 And that's something that I think is missed when you just have one authoritative source. 247 00:23:01,272.323395 --> 00:23:05,82.323395 You miss the fact that I, I see two bikes on the wall behind you. 248 00:23:05,892.323395 --> 00:23:06,882.323395 You are interested in cycling. 249 00:23:06,882.323395 --> 00:23:13,692.323395 So someone like you might have an incentive to improve the map data about cycling in your area, and then someone else is maybe a truck driver. 250 00:23:13,692.323395 --> 00:23:16,527.323395 And this happens a lot in some communities where the truck driver cares about. 251 00:23:17,352.323395 --> 00:23:23,352.323395 Where can the truck go down? And then they can update the map and make sure that truck access restrictions are, are on the map. 252 00:23:23,352.323395 --> 00:23:30,792.323395 And so that diversity of knowledge and opinion is really what makes the map as incredible as, as open streete map and over chair is today. 253 00:23:32,52.323395 --> 00:23:35,142.323395 Yeah, I think, um, well, if we had more time, it'd be worth diving. 254 00:23:35,427.323395 --> 00:23:42,777.323395 Into, to overture a little bit more because I, I find it fascinating, but I wanna mention, I wanna sort of hit on this idea you talked about of authoritative data. 255 00:23:43,107.323395 --> 00:23:55,947.323395 And it's really interesting to think of authoritative data as maybe something that is crowdsourced but has been through this extra layer of, um, I don't know, uh, ground truthing via maybe the rapid edit, for example. 256 00:23:55,947.323395 --> 00:23:57,147.323395 So a human in the loop. 257 00:23:57,777.323395 --> 00:24:03,237.323395 And I think depending on where you are, I think authoritative data is, is sourced quite, quite differently. 258 00:24:04,242.323395 --> 00:24:04,842.323395 Yeah, absolutely. 259 00:24:04,842.323395 --> 00:24:06,882.323395 That, that's a, I think a great point. 260 00:24:06,882.323395 --> 00:24:11,262.323395 Like to put an authoritative label on something doesn't necessarily mean it's good. 261 00:24:11,262.323395 --> 00:24:12,882.323395 It's exactly what you said. 262 00:24:12,882.323395 --> 00:24:15,942.323395 It's where is it coming from? What was the method used to collect it. 263 00:24:15,972.323395 --> 00:24:25,272.323395 There, there's terrible, authoritative data and then there's excellent authoritative data, and, and so knowing a bit more about the, the origin of that data is really important. 264 00:24:25,332.323395 --> 00:24:28,92.323395 And so rapid has definitely been useful for that. 265 00:24:28,92.323395 --> 00:24:29,82.323395 We have things like. 266 00:24:29,712.323395 --> 00:24:47,997.323395 Addresses that cities contribute or buildings building data that cities contribute and a user can go into rapid review it, make sure that it's actually correct, and maybe 99% of it is, but by going through that process, they can remove the 1% that's maybe a false positive or is just out of date because the data was sourced a year ago. 267 00:24:49,462.323395 --> 00:24:52,672.323395 So we, we've talked a little bit about the uses internally within Meta. 268 00:24:52,672.323395 --> 00:24:58,342.323395 We've talked about how, in, in my opinion, meta's a, a really good citizen of the, uh, open source geo world. 269 00:24:58,342.323395 --> 00:25:02,272.323395 I mean, you make a lot of information, data available, you build a lot of tools. 270 00:25:02,632.323395 --> 00:25:08,122.323395 You foster a lot of communities through the work that you do and your, your sponsorship, which I personally really appreciate. 271 00:25:08,872.323395 --> 00:25:14,272.323395 Is there anyone else using this stuff? Like are there any sort of cities, councils, larger organizations that, uh. 272 00:25:15,372.323395 --> 00:25:25,847.323395 Doing the thing that we're talking about, that they're investing in, I don't know, using 360 cameras to drive around their, their local, whatever community and, and monitor the things and collect data in this way. 273 00:25:27,177.323395 --> 00:25:33,747.323395 Yeah, the, I'll give you two contrasting, well, not contrasting, but two different stories, different parts of the world, different methods. 274 00:25:33,807.323395 --> 00:25:43,527.323395 One is a company called B Group and they have, I don't know where they sit in Vietnam, but they're one of the growing ride sharing companies in in Vietnam. 275 00:25:43,527.323395 --> 00:25:43,972.323395 There are others, but. 276 00:25:45,57.323395 --> 00:25:51,507.323395 They're a ride sharing company in Vietnam, and you can imagine a ride sharing company really needs to make sure that people get from A to B in an efficient way. 277 00:25:51,837.323395 --> 00:25:55,647.323395 They're also delivering packages, so they want packages to get from A to B in an efficient way. 278 00:25:56,97.323395 --> 00:25:57,717.323395 They need to know what the end point is. 279 00:25:57,717.323395 --> 00:26:03,567.323395 So if the user's putting in a a place, they need it to be the right place and they hope to have that business in their database. 280 00:26:03,567.323395 --> 00:26:07,947.323395 And Vietnam is one of the fastest growing economies in the world. 281 00:26:07,947.323395 --> 00:26:11,577.323395 It's changing even more rapidly than some of the stats I mentioned earlier. 282 00:26:12,297.323395 --> 00:26:12,957.323395 And so they. 283 00:26:13,617.323395 --> 00:26:16,137.323395 Have had to get into the mapping business as a ride sharing company. 284 00:26:16,137.323395 --> 00:26:19,467.323395 They've had to figure out maps, and you've seen this with Uber, you've seen this with Grab. 285 00:26:19,467.323395 --> 00:26:28,107.323395 This is just like a natural progression for a lot of companies is that they realize they, they have the ability to make the map better, and that's a core part of their use case for their customer. 286 00:26:28,197.323395 --> 00:26:37,17.323395 And so they've put 360 cameras on a few of their rider, and if you just look at Malar and her Chi Minh City in Hanoi, it's incredible. 287 00:26:37,17.323395 --> 00:26:39,747.323395 Like they've captured almost all of the city. 288 00:26:40,167.323395 --> 00:26:59,277.323395 And they're doing things like working out where can cars not go down, but scooters can because a lot of their customers are on scooters or maybe a car can't go down this road between these hours, but it can go down that road between these hours and so that they need to get all the information about this and street level imagery a great source. 289 00:26:59,817.323395 --> 00:27:08,307.323395 They're also using it for that reason I mentioned before, which is optical character recognition and looking at which businesses are there and trying to position the. 290 00:27:08,982.323395 --> 00:27:15,522.323395 The building entrance more accurately and, and a whole bunch of stuff around places and, and categorization of places. 291 00:27:15,522.323395 --> 00:27:19,242.323395 So I really love that one because it's in an exciting part of the world. 292 00:27:19,242.323395 --> 00:27:36,792.323395 They've been very, they're using a GoPro Max camera that anyone can access, and I think they're proving that you can do this in a very scalable way and relatively inexpensive way, whilst also contributing back, because anyone else who's interested in Vietnam can then go and use that same debtor. 293 00:27:37,302.323395 --> 00:27:45,492.323395 I think that's, that's the truly interesting he thing for me here is that there's a clear return on investment in terms of business for them at solving a business problem. 294 00:27:45,492.323395 --> 00:27:50,412.323395 But at the same time, you know, they're making things better for everybody else that wants to use the the map as well. 295 00:27:50,682.323395 --> 00:27:51,597.323395 And it's interesting to hear you say. 296 00:27:52,437.323395 --> 00:27:59,517.323395 Like all these ride sharing, you know, apps and businesses and you know, multiple other organizations, they're in the mapping business as well. 297 00:27:59,637.323395 --> 00:28:06,597.323395 Because as soon as we, the devices told us where we were or are in the world, we had to figure out where everything else is in the world. 298 00:28:06,597.323395 --> 00:28:08,67.323395 Otherwise we have no context. 299 00:28:08,457.323395 --> 00:28:11,212.323395 And when things are changing, you know, it's difficult to keep up. 300 00:28:12,462.323395 --> 00:28:14,172.323395 Yeah, it's, it is very difficult to keep up. 301 00:28:14,172.323395 --> 00:28:16,212.323395 And, and so I mentioned two use cases. 302 00:28:16,212.323395 --> 00:28:19,962.323395 The other one is Detroit, the city of Detroit. 303 00:28:19,962.323395 --> 00:28:32,532.323395 And so they're a bit of a different use case in that they have an obligation to their citizens to maintain assets, things like traffic signs and, and knowing which parts of the city are suffering from blight. 304 00:28:33,612.323395 --> 00:28:36,762.323395 And they had this GIS officer called Dexter who. 305 00:28:37,182.323395 --> 00:28:38,262.323395 Came up with the idea again. 306 00:28:38,262.323395 --> 00:28:45,822.323395 It started starting with, I think a GoPro Max or something similar and just driving around with that and, and proving that he could collect useful data for the city. 307 00:28:46,962.323395 --> 00:28:52,362.323395 And he won a Mayor's innovation grant, which I think was acknowledgement that, hey, that like, this is a really good idea. 308 00:28:52,392.323395 --> 00:28:53,652.323395 You're collecting useful data. 309 00:28:54,852.323395 --> 00:29:05,292.323395 He, long story short, he used that to get a much better camera and then that turned into like getting more sponsorship from the city and, and General Motors, which is obviously the big company in Detroit. 310 00:29:05,757.323395 --> 00:29:12,927.323395 And so fast forward to today, they have a Trimble, can't remember the exact model, but a very good Trimble camera. 311 00:29:13,377.323395 --> 00:29:15,657.323395 They have Lidar on, on top of a vehicle. 312 00:29:15,657.323395 --> 00:29:20,637.323395 They have a dedicated vehicle, two drivers, and they're driving for most of the year, collecting data. 313 00:29:20,817.323395 --> 00:29:29,487.323395 And their philosophy was like, in the past we used to pay companies to do this, and they'll do it maybe once every few years, would have to wait ages for the data. 314 00:29:30,342.323395 --> 00:29:41,772.323395 But they realized like we have people going around all the time in a city solving various problems, fixing things like let's have a dedicated vehicle that's always capturing multiple things, make that data available and open. 315 00:29:41,772.323395 --> 00:29:45,522.323395 And they've been able to do things like reduce insurance rates. 316 00:29:45,732.323395 --> 00:29:49,392.323395 'cause insurance is obviously based on whichever statistics they have. 317 00:29:49,392.323395 --> 00:29:52,872.323395 And in some places they thought there were less people living than there actually were. 318 00:29:53,547.323395 --> 00:30:00,807.323395 Being able to like update census records based on the imagery collection and, and prove that the population was higher than it originally thought. 319 00:30:01,107.323395 --> 00:30:10,707.323395 That's just another example of like really being quite innovative around, uh, the city's resources and making them go much further with, uh, the dedicated camera. 320 00:30:11,127.323395 --> 00:30:13,737.323395 It's a really interesting idea actually, because it just occurred to me that. 321 00:30:14,367.323395 --> 00:30:19,77.323395 Yeah, so they need to collect the, the base data, if you will, but somebody else is doing the processing. 322 00:30:19,377.323395 --> 00:30:25,647.323395 Somebody else is doing the heavy lifting in terms of determining is this a rubbish bin or is this a, a street sign kind of thing. 323 00:30:26,127.323395 --> 00:30:31,947.323395 And then somebody else is also doing the heavy lifting in terms of distributing that information, making it freely available to people. 324 00:30:32,157.323395 --> 00:30:38,187.323395 Because if it was, if they had to do all of this internally, my guess is they would have to pay someone to collect the data. 325 00:30:38,487.323395 --> 00:30:41,817.323395 The burden of processing it would be on them, and then they'd have to host it and update it. 326 00:30:42,627.323395 --> 00:30:45,482.323395 Somewhere themselves, and this is kind of done for them. 327 00:30:46,827.323395 --> 00:30:52,827.323395 Yeah, and precisely that like, and I guess that's why it takes it, it would often happen much less frequently than it did. 328 00:30:52,827.323395 --> 00:30:57,777.323395 It would be more expensive, and they've got a really good writeup about their, their story. 329 00:30:57,777.323395 --> 00:31:03,847.323395 But I really like their philosophy as a city, and I think it's something that every city should aspire to, but it's. 330 00:31:04,842.323395 --> 00:31:17,442.323395 Really trying to be smart with taxpayer funds and saying like, in the past we used to go out and do multiple surveys, right? You do a lidar survey, you do a street level imagery survey, maybe you do another survey to, to like manually counter the traffic signs. 331 00:31:17,922.323395 --> 00:31:24,552.323395 Why don't we just do that in one drive? We can collect a lot of that information in one drive and, and so that's what they're doing now. 332 00:31:24,552.323395 --> 00:31:33,762.323395 And yeah, anything a city can do to kind of use taxpayer funds more efficiently and, and put more money towards the services and service delivery is, is a really good thing. 333 00:31:34,647.323395 --> 00:31:37,392.323395 Yeah, it, it's hard to disagree on that one. 334 00:31:38,22.323395 --> 00:31:39,42.323395 You mentioned Lidar. 335 00:31:39,192.323395 --> 00:31:40,2.323395 There any plans for. 336 00:31:42,177.323395 --> 00:31:46,882.323395 To sort of support the collection of lidar, 'cause we see Lidar creeping into devices now as well. 337 00:31:48,102.323395 --> 00:31:52,62.323395 Yeah, there's not, it's a question that comes up a lot and it's a good question. 338 00:31:52,122.323395 --> 00:31:55,482.323395 It would be nice and we could definitely do a lot more if we did support it. 339 00:31:55,602.323395 --> 00:32:02,202.323395 I don't think we have it on a roadmap or like the capacity to support lidar, but there's so much amazing there been Lidar out there now and. 340 00:32:02,847.323395 --> 00:32:03,147.323395 Yeah. 341 00:32:03,177.323395 --> 00:32:06,177.323395 Never say never, but it's something that's, yeah. 342 00:32:06,177.323395 --> 00:32:07,407.323395 We don't have on our roadmap right now. 343 00:32:07,987.323395 --> 00:32:12,792.323395 So speaking of roadmaps, let's move on to the sort of future gazing section of the podcast. 344 00:32:12,822.323395 --> 00:32:15,72.323395 Where are we going from here? We're not doing lidar. 345 00:32:16,267.323395 --> 00:32:16,752.323395 Big, big, disappointing. 346 00:32:17,502.323395 --> 00:32:18,882.323395 A sigh goes through the audience. 347 00:32:18,882.323395 --> 00:32:36,792.323395 What? What are we doing? One of the really cool things from earlier in the year, one of the blog posts that I think was widely received was one that Peter Conta wrote about nerves and nerves of these Stanford neural radiance fields, which again, might not mean much to the audience, but this is a. 348 00:32:37,362.323395 --> 00:32:42,822.323395 A way to create highly immersive 3D scenes from images as input. 349 00:32:43,752.323395 --> 00:32:58,302.323395 So we mentioned the 3D point cloud before, that's just like the, these isolated points in space, which, uh, you, you can apply textures to them, but they're not great if you're trying to get immersed into that scene and, and kind of feel like you're really there. 350 00:32:58,302.323395 --> 00:33:02,172.323395 Whereas a Nerf is really good for that, that visual experience. 351 00:33:02,172.323395 --> 00:33:05,112.323395 And so it's using computer vision to be able to. 352 00:33:05,712.323395 --> 00:33:12,702.323395 Not just turn images you see into a 3D scene, but it can also create what they call novel synthesis, new views. 353 00:33:13,122.323395 --> 00:33:18,72.323395 So imagine you're walking around a statue, you're taking thousands of photos of that statue. 354 00:33:18,72.323395 --> 00:33:21,132.323395 There might be bits and pieces of that statue that you missed. 355 00:33:21,762.323395 --> 00:33:31,992.323395 It can start to fill that in using AI to kind of guess what's there, but also really like making sure that the lighting looks consistent across the whole scene. 356 00:33:32,682.323395 --> 00:33:42,672.323395 And I think that's really exciting because obviously Meta is, is all about allowing people to, uh, experience the metaverse and go into a scene remotely wearing a headset. 357 00:33:42,672.323395 --> 00:33:54,762.323395 So I, I think we'll be doing more exploration around that and being able to teleport to, you know, Australia where I'm from, and, and go to the Great Ocean Road or the Sydney Harbor Bridge and explore it with a headset. 358 00:33:55,2.323395 --> 00:33:57,612.323395 That's one thing that I think we're really excited about. 359 00:33:58,452.323395 --> 00:34:01,782.323395 Also, there's a lot we can do around, uh, the data as well. 360 00:34:01,782.323395 --> 00:34:13,47.323395 But maybe I'll stop there on, on Nerfs in case you had any, any No, I think, I think that's fascinating and I often think that if this future of virtual reality, augmented reality is gonna be realized, the, the. 361 00:34:13,302.323395 --> 00:34:14,622.323395 Promise is gonna be realized. 362 00:34:14,622.323395 --> 00:34:22,122.323395 I think we're gonna need a lot of amazing base data, you know, to be able to project some of these things onto it or maybe help people navigate it. 363 00:34:22,542.323395 --> 00:34:24,462.323395 We're, we're gonna need some sort of realism over it. 364 00:34:24,462.323395 --> 00:34:30,162.323395 And I think I see projects like, uh, like this really contributing to that, you know, at, at a global scale. 365 00:34:30,282.323395 --> 00:34:31,542.323395 So I think that it's fascinating. 366 00:34:32,337.323395 --> 00:34:35,917.323395 There's an interesting trend now, like O obviously you've had the success of. 367 00:34:36,537.323395 --> 00:34:41,607.323395 Projects like Minecraft and or, or games like Minecraft and Roblox and people creating new worlds. 368 00:34:41,607.323395 --> 00:34:47,127.323395 And we have that in Horizon Worlds, which is a meta application on, on Quest devices. 369 00:34:47,277.323395 --> 00:34:56,547.323395 But people also want to go to places they know and love and whether it's their home or their, uh, their favorite, you know, landmark in their city. 370 00:34:56,547.323395 --> 00:35:02,967.323395 And even at Mapley, like early on there were people in France and other places who were really like, passionate about preserving. 371 00:35:03,552.323395 --> 00:35:06,792.323395 The places in their city that they cared about and taking a lot of images. 372 00:35:07,182.323395 --> 00:35:18,192.323395 And so Nerf nerfs is a way for them to not just have the images of that place, but then also a, a, a model that people can then explore through various mediums and be much more immersed in that scene. 373 00:35:18,742.323395 --> 00:35:18,982.323395 Yeah. 374 00:35:18,982.323395 --> 00:35:22,792.323395 And I think when, when people visit places, they know they want it to be realistic. 375 00:35:22,852.323395 --> 00:35:34,342.323395 They don't want, you know, if I know my neighborhood well, I don't wanna visit this in a virtual world and have buildings missing and, and gaps or, I, I don't wanna see a, a vertical wall sort of leaning at a 45 degree angle. 376 00:35:34,342.323395 --> 00:35:35,422.323395 I want it to be accurate. 377 00:35:35,422.323395 --> 00:35:39,52.323395 And I can see this is, you know, this is part of what you're building in the background. 378 00:35:39,372.323395 --> 00:35:40,317.323395 Yeah, exactly. 379 00:35:40,822.323395 --> 00:35:44,32.323395 Are we gonna see map supporting, uh. 380 00:35:44,427.323395 --> 00:35:47,182.323395 Ray Band glasses as a, you know, collection device. 381 00:35:48,27.323395 --> 00:35:50,397.323395 Yeah, we, we've tried it, a few of us have tried it. 382 00:35:51,177.323395 --> 00:35:55,527.323395 It's definitely like something that we're trying to figure out. 383 00:35:55,527.323395 --> 00:35:57,627.323395 So there are a few limitations. 384 00:35:57,627.323395 --> 00:36:04,857.323395 Like one, originally the rebound meta glasses had I think a 32nd average duration of capture. 385 00:36:04,857.323395 --> 00:36:07,497.323395 And then that's been extended, I think it's about three minutes now. 386 00:36:07,797.323395 --> 00:36:08,247.323395 So that's. 387 00:36:08,832.323395 --> 00:36:13,212.323395 Useful, but like it quickly drains through the battery if you have them on for much longer than that. 388 00:36:13,212.323395 --> 00:36:14,832.323395 So they limit the capture time. 389 00:36:15,642.323395 --> 00:36:18,12.323395 You also need to rely on your phone's GPS. 390 00:36:18,252.323395 --> 00:36:28,782.323395 So we need to like work on the software there to make sure that we're syncing the video frames from the wearable to, to your phone, um, and utilizing that GPS. 391 00:36:29,382.323395 --> 00:36:36,807.323395 And then there's also the issue of when you're wearing glasses, you turn your head right, like we, we can't expect you to just keep a straight head as you walk down the street. 392 00:36:37,557.323395 --> 00:36:50,67.323395 And, and for map blurry, it's much better for that 3D reconstruction I mentioned when there's a consistent angle relative to the direction of travel, so you're looking straight ahead and continuing straight ahead, or you're looking to the side and keeping your gaze fixed to the side. 393 00:36:50,367.323395 --> 00:36:52,587.323395 Very hard to do that with Glasses 360. 394 00:36:52,587.323395 --> 00:36:56,97.323395 It's less of an issue because you've got the full perspective around you. 395 00:36:56,522.323395 --> 00:36:59,247.323395 But I think it would be good for like limited bursts. 396 00:36:59,247.323395 --> 00:37:05,877.323395 Like I, I would love to see it where if you wanted to use your ray bands to quickly, like walk down the street and, and maybe take a 32nd. 397 00:37:06,597.323395 --> 00:37:15,372.323395 Video of something that interests you, that you could do that and then go back and edit it later in open Streete map or map, or you could derive interesting features and, and put them onto the map. 398 00:37:16,182.323395 --> 00:37:21,102.323395 I could see it for, for scanning things like, I wanna remember this, the sculpture, this, this, whatever. 399 00:37:21,432.323395 --> 00:37:25,962.323395 And just, you know, being able to walk around, like kind of, kind of scan it. 400 00:37:25,962.323395 --> 00:37:27,132.323395 I think that would be really interesting. 401 00:37:27,432.323395 --> 00:37:41,952.323395 Also, wonder if this would open up, I, I realize you talked about positioning being a problem, you know, a challenge, maybe I should say that, but I wonder if it would open up mapping, um, indoors as well if you had something that was sort of less conspicuous than a, than a massive 3D camera. 402 00:37:42,762.323395 --> 00:37:43,572.323395 360 cameras. 403 00:37:43,627.323395 --> 00:37:46,87.323395 Yeah, mapping indoors is like a whole other challenge. 404 00:37:46,147.323395 --> 00:37:47,617.323395 I loved your podcast. 405 00:37:47,677.323395 --> 00:37:50,317.323395 Uh, it was with it a long way, um, from mapped in. 406 00:37:50,707.323395 --> 00:37:53,827.323395 He, he's great in like hearing the challenges there, I think. 407 00:37:54,582.323395 --> 00:37:59,592.323395 Indoors, you lose GPS and then you have to rely on other localization technologies. 408 00:37:59,592.323395 --> 00:38:01,722.323395 But glasses could be one of the inputs there. 409 00:38:01,812.323395 --> 00:38:08,862.323395 Definitely like having, having more imagery indoors, particularly people that are wearing transition lenses and are always wearing their glasses. 410 00:38:08,862.323395 --> 00:38:11,262.323395 I do suspect we'll see a lot more of that. 411 00:38:11,982.323395 --> 00:38:22,387.323395 I think like a lot of things in maps and location, generally, the more inputs the better, right? Like you have, it's very hard to build Acura Maps with just one input, and so you try to take. 412 00:38:23,397.323395 --> 00:38:24,267.323395 Different sources. 413 00:38:24,267.323395 --> 00:38:33,177.323395 Maybe the base map could come from a service like mapped in where you know the floor plan and then you have meta glasses to kind of fill it in and know where you are within that floor plan. 414 00:38:35,187.323395 --> 00:38:37,287.323395 we, you just mentioned different sources there. 415 00:38:37,287.323395 --> 00:38:38,637.323395 We talked about RayBan glasses. 416 00:38:38,697.323395 --> 00:38:45,417.323395 The other sort of obvious source to me, I dunno if RayBan glasses is, is that obvious? But like to me it is would be drones. 417 00:38:46,377.323395 --> 00:38:47,277.323395 And so my question is. 418 00:38:48,132.323395 --> 00:38:58,152.323395 Could you see map moving away from this idea of street level to just sort of imagery in general? I think partially, yeah. 419 00:38:58,152.323395 --> 00:39:01,242.323395 We have been really encouraging more sidewalks. 420 00:39:02,7.323395 --> 00:39:03,957.323395 Capture more like footpaths. 421 00:39:03,957.323395 --> 00:39:19,617.323395 I think going beyond where we started, which was all about roads and navigation, now a lot of our navigation focuses is how does a, how does a human on two legs or in a wheelchair or outside of a road, navigate from A to B? And we've encouraged more of that. 422 00:39:19,617.323395 --> 00:39:22,287.323395 In terms of drones, we don't encourage it. 423 00:39:22,917.323395 --> 00:39:26,157.323395 We go through various phases where we discourage it more than others, but I think. 424 00:39:27,57.323395 --> 00:39:33,567.323395 You kind of have to build a different platform for drones 'cause you need to start to filter in altitude and have a, have a sense of how you're dealing with that. 425 00:39:33,987.323395 --> 00:39:47,577.323395 That 3D model becomes more challenging because you maybe have a ton of camera views that are taken at ground level or, or slightly above ground level with if you're taking it from a car and then all of a sudden you have something like a hundred meters up. 426 00:39:47,577.323395 --> 00:39:49,227.323395 So that can, that can be challenging. 427 00:39:49,317.323395 --> 00:39:52,977.323395 The algorithms need to like be retrained to, to know how to connect those. 428 00:39:53,397.323395 --> 00:39:54,147.323395 But we've had. 429 00:39:54,792.323395 --> 00:39:59,922.323395 Quite a bit of success with people uploading drone imagery at, at a lower level. 430 00:40:00,642.323395 --> 00:40:05,322.323395 So you can imagine going down a European street, right? Uh, you lived in Copenhagen. 431 00:40:05,322.323395 --> 00:40:10,152.323395 If you're going down a a downtown area in Copenhagen, there's probably a lot of bicycles parked on the side. 432 00:40:10,152.323395 --> 00:40:11,652.323395 Maybe cars parked on the side. 433 00:40:12,612.323395 --> 00:40:20,112.323395 If you are just walking along or even driving in the road, you might not be able to see things on the side because of those cars and bicycles. 434 00:40:20,817.323395 --> 00:40:25,677.323395 But if you have a drone that's flying like higher, you can actually get a pretty good perspective. 435 00:40:26,187.323395 --> 00:40:29,937.323395 Probably not legal in a lot of cities, but it is an interesting thing. 436 00:40:30,327.323395 --> 00:40:33,477.323395 Maybe that's what they're doing in New Jersey at the moment. 437 00:40:33,582.323395 --> 00:40:33,822.323395 Yeah. 438 00:40:35,412.323395 --> 00:40:50,352.323395 What, what about, um, when we think about the future, uh, is there an object in the imagery that you'd like to be able to detect that you can't detect today? Or, you know, do you have things on the roadmap? Oh, we really want to get, uh, streetlights or, I don't know, we are gonna start reading signs ourselves kind of thing. 439 00:40:51,887.323395 --> 00:40:59,62.323395 I think places is the big one where if you look at Open Streett Map or Google Maps or Overture Maps or, or like any MAP database. 440 00:40:59,892.323395 --> 00:41:02,892.323395 Places is the really hard one to keep up to date. 441 00:41:02,982.323395 --> 00:41:21,192.323395 And Maper, I think, can play a better role there, discerning what is the name of this place, what type of business is it? Um, where is the entrance located? So that's probably the number one impact that Maper could have right now is, is answering those questions better. 442 00:41:21,192.323395 --> 00:41:27,462.323395 And historically, having front facing imagery in a car made it challenging to do that because. 443 00:41:28,62.323395 --> 00:41:30,762.323395 Maybe only part of the name is visible. 444 00:41:31,212.323395 --> 00:41:34,722.323395 Maybe the resolution is not high enough to do some of the things we mentioned before. 445 00:41:35,202.323395 --> 00:41:42,12.323395 But now that there's more 360 imagery, I think it's becoming more feasible to be able to, to do some of the things like building entrance and name detection. 446 00:41:42,657.323395 --> 00:41:48,507.323395 Can you gimme a quick run through of what needs to happen in order to be able to identify a place? So we talked about reading signs. 447 00:41:49,122.323395 --> 00:42:09,927.323395 What do you need to know? Like what, and when you think about if you had all the imagery you needed, what bits of information do you need to extract from that in order to be able to say, okay, this place to be able to build up this object? Yeah, the, so the first one is reading the name of the place, and that can be high 'cause there's all these different types of typography and there's different languages and you need to discern. 448 00:42:10,947.323395 --> 00:42:22,227.323395 What is the relevant, like part of the name here? 'cause it might be, um, you know, such and such name and then Thai restaurant, but you're taking the Thai restaurant party or you're taking the name and you're taking both. 449 00:42:23,337.323395 --> 00:42:25,617.323395 Then there's the, the entrance itself. 450 00:42:25,647.323395 --> 00:42:32,187.323395 So some of the larger buildings, and I think this is something that the ride sharing companies have to challenge a lot with is. 451 00:42:32,637.323395 --> 00:42:35,487.323395 A larger building might have a pickup and drop off area. 452 00:42:35,577.323395 --> 00:42:38,127.323395 It might have a separate area for delivery people. 453 00:42:38,127.323395 --> 00:42:40,467.323395 It might have a separate entrance for the customers. 454 00:42:41,397.323395 --> 00:42:46,647.323395 Think of a large shopping mall as a good example that might have these three different types of entrances. 455 00:42:47,67.323395 --> 00:42:51,747.323395 Discerning what the relevant entrance is, is is a bit of a challenge there. 456 00:42:52,167.323395 --> 00:42:55,557.323395 And then, yeah, positioning the, the building itself. 457 00:42:55,557.323395 --> 00:43:00,747.323395 So like, which you've got the building entrance, but which polygon is it relating to? Like what's the 2D. 458 00:43:01,347.323395 --> 00:43:04,767.323395 What does the 2D map look like? Doing conflation. 459 00:43:04,857.323395 --> 00:43:12,657.323395 Maybe there's already an existing building there, like, or an existing place, um, understanding that or maybe there's a different place there. 460 00:43:12,657.323395 --> 00:43:23,7.323395 Has it changed name or have we just identified an additional place that's opened up to it? So I think those are some of the challenges that come up, uh, that, that make place detection quite challenging. 461 00:43:23,517.323395 --> 00:43:29,247.323395 Yeah, having to talk to the good people at Foursquare as well, so they work a lot on this and it's, it's not easy. 462 00:43:29,487.323395 --> 00:43:32,517.323395 And you mentioned the rate of change before as well, like 30%. 463 00:43:33,507.323395 --> 00:43:34,77.323395 That's tough. 464 00:43:34,107.323395 --> 00:43:36,927.323395 That's, that's a real, that's a real tough, tough problem to solve. 465 00:43:37,647.323395 --> 00:43:44,217.323395 In my notes here, I can see something about, like, I can see some sort of references to the API to the developer community. 466 00:43:44,592.323395 --> 00:44:08,892.323395 And I, I, I wanted to ask, and this might be, this might be a really naive question, is there a world where you start to open the, you know, the API up to normal humans, if you will, via something like a, um, by putting an LLM in front of it and letting people ask questions of the data as opposed to approaching it like the, the typical structured database where it's an SQL query that you are, that you're firing off. 467 00:44:10,142.323395 --> 00:44:11,462.323395 Yeah, I like that question. 468 00:44:11,822.323395 --> 00:44:13,682.323395 There's definitely ways we could make it easy. 469 00:44:13,682.323395 --> 00:44:18,482.323395 I don't think it's something that we have on our roadmap, but it's something that anyone could build. 470 00:44:18,602.323395 --> 00:44:25,352.323395 Like I would encourage you with a lot of map blurry the data, the API to, to go and take it and, and build stuff with it. 471 00:44:25,352.323395 --> 00:44:26,627.323395 You could build a business around it. 472 00:44:26,627.323395 --> 00:44:33,782.323395 And people do, they build stuff on top of MAP and, and we highly encourage that because it makes the ecosystem healthier. 473 00:44:34,352.323395 --> 00:44:34,982.323395 We have. 474 00:44:35,862.323395 --> 00:44:37,332.323395 Not an LLM but a Python. 475 00:44:37,332.323395 --> 00:44:42,882.323395 SDK for example, that some students built fantastic students is part of their major league hacking exercise. 476 00:44:42,882.323395 --> 00:44:58,602.323395 And they helped us make the API just a bit easier for people to use so that if you're familiar with Python and you wanna download data and you don't have to worry about map tiles and, and some of the way the API is structured right now, you can just give it a bounding box, say give me traffic signs and download that. 477 00:44:58,602.323395 --> 00:45:00,702.323395 But I can see a next step of that being. 478 00:45:01,422.323395 --> 00:45:08,952.323395 More of a a natural language interface like you've mentioned, and it'd be really cool if someone built it, but it probably won't be us anytime soon. 479 00:45:09,777.323395 --> 00:45:10,137.323395 Fair enough. 480 00:45:10,137.323395 --> 00:45:13,707.323395 It sounds like you've got your hands full with all the other things that you're building and supporting. 481 00:45:14,757.323395 --> 00:45:16,317.323395 I feel like we're coming to the end of the conversation. 482 00:45:16,317.323395 --> 00:45:17,352.323395 We've covered a lot of ground. 483 00:45:18,117.323395 --> 00:45:25,512.323395 Is there anything you wanna leave the listeners with? Uh, is there anything that I haven't asked you about that you're, that you'd like to mention before we, before we round things off? Yeah. 484 00:45:25,517.323395 --> 00:45:30,402.323395 Map blurry is what it is because of all the people contributing to it over the last. 485 00:45:30,762.323395 --> 00:45:31,212.323395 10 years. 486 00:45:31,212.323395 --> 00:45:36,312.323395 So we had a, actually last year was I think the official 10 year anniversary. 487 00:45:36,312.323395 --> 00:45:43,842.323395 We had a nice celebration in southern Sweden around that and really wanted to thank everybody who's contributed over the years and taught us a lot. 488 00:45:43,842.323395 --> 00:45:49,442.323395 I feel I'm at a stage now where I'm learning more from other people who are using MAP blur because they find new ways to do things. 489 00:45:49,442.323395 --> 00:45:51,572.323395 And so I want to thank everybody. 490 00:45:51,572.323395 --> 00:45:53,402.323395 Like the City of Detroit is one example. 491 00:45:53,492.323395 --> 00:45:57,837.323395 B Group is another of like pushing the, the boundaries of what you can do with street level imagery. 492 00:45:58,517.323395 --> 00:46:02,357.323395 And as I mentioned just before, like I encourage people to, to use the data. 493 00:46:02,357.323395 --> 00:46:07,7.323395 It's, it's much more open than a lot of other street level imagery platforms. 494 00:46:07,397.323395 --> 00:46:10,577.323395 You can look at the licensing and the terms and, and see what you can do with it. 495 00:46:10,637.323395 --> 00:46:14,357.323395 Um, if you have questions about how to use it for commercial use, our lines are always open. 496 00:46:14,357.323395 --> 00:46:16,787.323395 And as I said, we encourage that as well. 497 00:46:17,267.323395 --> 00:46:19,727.323395 So we would love feedback. 498 00:46:19,907.323395 --> 00:46:26,627.323395 We'd love to know what would you like to see for us to make it better And we can't promise everything, but you know, the team is always. 499 00:46:27,632.323395 --> 00:46:34,412.323395 Learning from our users, and that's been a big part of how we've improved over the last 10 years, is taking that feedback and incorporating it back in. 500 00:46:34,592.323395 --> 00:46:36,662.323395 Making upload easier is one example. 501 00:46:37,112.323395 --> 00:46:50,312.323395 Is there any particular channel you'd like people to give you feedback on? Like should they tweet at you? Should they write to you on, on LinkedIn? Is there a I know form someone can fill out? Where, where can they go if they wanna give you some feedback? Yeah, I've never heard of tweeting. 502 00:46:50,392.323395 --> 00:46:51,782.323395 I'm not, I'm not sure what that is. 503 00:46:51,812.323395 --> 00:46:59,552.323395 Um, but you can find us on threads, uh, which is, which is a great, uh, we, we have a ary handle there on threads. 504 00:47:00,2.323395 --> 00:47:02,282.323395 Uh, we have Zendesk, which is, um, uh. 505 00:47:02,687.323395 --> 00:47:05,897.323395 Yeah, you can send us an email [e-mail], 506 00:47:06,17.323395 --> 00:47:12,677.323395 which is just where we, uh, keep track of all the emails that's coming in, and, and then you can probably find us on LinkedIn as well. 507 00:47:12,677.323395 --> 00:47:17,327.323395 We're active there, so whatever place you you prefer, you're welcome to contact us. 508 00:47:18,152.323395 --> 00:47:18,872.323395 Appreciate it. 509 00:47:18,872.323395 --> 00:47:19,592.323395 Thanks very much, ed. 510 00:47:19,592.323395 --> 00:47:20,732.323395 I really appreciate your time. 511 00:47:20,762.323395 --> 00:47:23,552.323395 Uh, appreciate all the work that you've done over the years and. 512 00:47:24,677.323395 --> 00:47:25,877.323395 I'm really grateful for meta. 513 00:47:25,877.323395 --> 00:47:37,127.323395 I think, like I said earlier, I think they've been a really great citizen of the open source geospatial world and um, yeah, they contribute a lot and they do a lot to support, uh, a bunch of different people and a bunch of different organizations and projects. 514 00:47:37,127.323395 --> 00:47:38,867.323395 So much appreciated from my side. 515 00:47:39,167.323395 --> 00:47:39,857.323395 Thanks very much. 516 00:47:39,887.323395 --> 00:47:46,202.323395 I will put links to everything I can in the show notes to help people find and discover more about you and, um, I hope I see you around. 517 00:47:47,462.323395 --> 00:47:47,912.323395 Thank you. 518 00:47:47,912.323395 --> 00:47:51,932.323395 Maps are much more fun as a collaborative project and they're much better as a result. 519 00:47:51,932.323395 --> 00:47:57,152.323395 So that's why we are building maps in collaboration through Map Larry and Overture and Open Streete Map. 520 00:47:57,152.323395 --> 00:47:59,942.323395 So thanks for letting me talk about it on your show. 521 00:48:00,347.323395 --> 00:48:00,887.323395 No worries. 522 00:48:01,397.323395 --> 00:48:02,57.323395 See you around Ed. 523 00:48:02,147.323395 --> 00:48:02,567.323395 Cheers. 524 00:48:03,122.323395 --> 00:48:03,662.323395 See you soon. 525 00:48:03,962.323395 --> 00:48:04,172.323395 Bye.