OpenStreetMap in the Field

A brief overview of strategies and software

Overview

• OSM from 10,000 feet
• Two Field Strategies
• A buffet of apps

OpenStreetMap

The greatest dataset on earth?

• Extremely flexible data model
• Extensive “ecosystem” of applications, tools, etc
• History tracking like nothing else

OpenStreetMap is amazing. We use it everywhere we can, and it’s been a game-changer for us. In OSM, there are no layers or columns, there’s simply points and lines (and relations) and the attributes we apply to them.

A point representing a bench is in the same space as a detailed building footprint with all of its business information, address, etc.

For a data consumer, there is a well established pipeline of OSM data to things like locators, network datasets, vector tiles, etc., making it very easy to pull OSM data into other spaces. Esri even has a feature service that updates every 15 minutes, for core features like roads, buildings, etc.

Worth Noting

• Community standards: “verifiable”
• Data license constraints

See: Public Domain Map

I think “EPA” and I think of quantitative observations at points in time. OSM, though, is all about features which have some verifiable presence in the world. I do hope that there is enough overlap between how we at Kendall use OSM and how you might in your work.

I also recognize that federal agencies may have statutory limitations and cannot use ODbL-licensed data sources in their work. Being open-source, it’s actually not that hard to run your own, private OSM-like database, which you can license any way you want. We actually do this at Kendall for certain kinds of data that aren’t appropriate to add to OSM itself.

In fact, OSM-US put together the Public Domain Map for the express purpose of allowing federal agencies to to use it under a public domain license. Maybe worth a look?

Whether it’s OSM itself or a variant, any OSM editing tools can simply swap URLs and roll with it.

Field Strategy One: Asynchronous with Mapillary

The Scenario

• You have things you need to map out or check on in the field
• You have non-mapping staff regularly in the field already
• You’ve got something like a GoPro

The Process

1. Field staff walk the territory / path / etc with the GoPro
2. Captures are uploaded to Mapillary
3. Mapping staff can virtually “step through” and edit map

Mapillary: Post-Upload

Pretty simple, here’s us stepping through. Depending on the quality of the camera and what you’re doing, this can be really useful. I’m not a botanist, but I can spot specific species of plants even from this.

Mapillary: Object Detection

Mapillary does some processing of its own. Based on the camera metadata and the internal GPS, it uses structure-from-motion techniques to estimate positions of recognizable features, like signs, benches, and the like.

This video is showing the web based editor of OSM and the Mapillary overlay data. The imagery itself includes some of the “edge” lines that Mapillary detects.

Worth noting: everything Mapillary does is possible for you to do locally from your own videos / images. 360° imagery with enough coverage of an area can be used to create surprisingly accurate point clouds. Not something we do at kendall, but something I’ve messed around with myself.

Field Strategy Two: Live and In Person

OSM Go!

A major caveat to open-source: sometimes a project goes stale! Nobody has been actively maintaining this app, and you can see some frayed edges in the way things are labelled / displayed.

It’s still a useful tool, and could be a jumping-off point

Go Map!! (Apple)

Every Door

MapComplete

• Customizable presets
• Cross-platform (even browser)

Live Demo!

MapComplete

Questions?

Be in Touch!

josh@jcarlson.page

https://jcarlson.page

@jcarlson@mapstodon.space