Noob OSM editing using UAV imagery – Transcription
Perfect. Loading. Awesome. Thanks. Hey, everybody, I’m Jeff Ferrara with DigitalGlobe here to talk about OSM editing using UAV imagery. A lot of you guys know that DigitalGlobe is a satellite imagery company. I’m a drone nerd. I love drones. This presentation isn’t so much about satellite imagery, but more my passion about drones and getting the imagery I capture into OpenStreetMap. I’m not a programmer. This is not a technical talk. I hacked through this solution using tools I use on a daily basis and got to a solution that worked. A highlevel problem. I moved into a brand new development neighborhood around 2015 around the same time I discovered OpenStreetMap through a mapathon with Kevin Bullock at DigitalGlobe. He challenged me to get my drone data into OSM so that I could have better use of getting an Uber to my house, getting a pizza delivered and emergency services that was a result of out of date maps. He challenged me to get the drone up in the air and update OpenStreetMap. This is how the map that I discovered of the neighborhood looked in 2015 using 2012 imagery. The star is where my house was. There’s a huge gap of field there where there’s no development. But I was clearly living in that spot. It was a huge challenge. I needed to take action to get the services that I took for granted before. Here’s a little bit of background of my drone. This thing is out of date by now. I’m in negotiations with my wife to get a new one for Christmas. Crossing my fingers on that. This thing can be had for just a couple hundred bucks now. And the capabilities in using the software can get you guys pretty useful data. So to overcome kind of the limited transmission range of this out of date drone, it had a pretty cool followme feature. I took a bike ride having the drone follow me. Probably freaked out a bunch of my neighbors. But I took this data, I ran it through picks 4D to generate an orthomosaic and a DSM and a 3D mesh product. It’s really cool. I recommend you try it. First, I ran it through Picts4D, processed the video into stills and generated this orthomosaic and others. Since it didn’t have geolocation, I cheated and took the imagery and merged with the orthomosaic and turned it into a geo reference product that I could input into tileset on Mapbox. Load it into a style and use the leaflet API to bring it into OpenStreetMap. Then I was finally using my drone imagery to create vectors and add features in the neighborhood. Finally got it on the map. Here’s kind of the level of detail I have been able to distinguish and what I’m mapping currently. Features like manhole covers, solar panels, fire hydrants. Down to fine detail that would be otherwise superchallenging to do in satellite imagery. And finally, I’m on the map and I’m getting pizza delivered to my house. And my Uber is having no problem picking me up. So all the problems are solved thanks to all of these providers to help me get there. And yeah. That’s it. Thanks. [ Applause ]
