Auto-Tag-Suggestion App for more accessibility data in OSM (A Wheelchair/Rollstuhlfahrer-Routing Revival)

General talk
, ,
1

Hello everyone, I’m an Internet of Things student with a CS background and I have an idea for my masters thesis, that needs some refining and help.

My plan would be to create a mobile app, that uses the phones sensors like gyro, acceleration and tilt, to track and make out if there is a curb or steep incline to overcome on the wheelchair users path. Also heatmaps would prove useful to filter out, if someone with a disability constantly goes a different route than regular users. I’m skeptical on how to turn that into reality though, especially the second part.

Ideally the app would track the users route via GPS, figure out through sensor data is something abnormal is happening or the surface is too rough at a certain part and suggest multiple different entries via the OSM API to tag the paths with keys such as:

  • smoothness=*
  • footway:surface=*
  • kerb=*
  • barrier=step
  • incline=up/down (above 6%, see DIN EN 17210)

I have read through the OSM Automated Edits code of conduct, that’s why I deviated from my original plan to fully automate the entries, and rather rely on user confirmation first. Also I’m still unsure on how to deal with duplicates or rather how to check for them correctly, after recording a route, where such tags already exist.

Please don’t take this the wrong way, I’m just a bit lost and I would love to hear your thoughts on this concept, as well as how to realistically start and what to keep in mind when creating such an app. I haven’t found anything like it yet, except wheelmap.org (which uses OSM), but it focuses more on locations/amenities, rather than the actual paths to get there.

I also would like to add a map displayed in the app with routing navigation, avoiding certain key/value pairs. or just by the “wheelchair=*” tag. In the past I did display nodes on a Map using Leaflet, but I never implemented something that advanced. I found https://wiki.openstreetmap.org/wiki/DE:Rollstuhlfahrer-Routing which follows the same idea minus the sensor-assisted suggestions, but unfortunately none of the links are active and I couldn’t find anything here in the forum.

I would be very happy about any input regarding this topic and I wish you guys a happy new year!

2

I’d suggest that before actually uploading anything to OSM you have a go at measuring things (several times, with different people) and then look at what tags would be suggested where, and how much correlation there was between different runs at different times with different people.

1 Like
3

That’s a very good point, thanks for the suggestion, I will keep myself busy with that in the meantime then. I’ll leave the thread open meanwhile for more helpful inputs regarding the other points mentioned.

4

Tough topic. Do you have any prior literature on how accurate sensor data is and what other people we’re able to derive from real measurement series? Is there experience on this topic where you are working in your thesis?
My recommendation would be to do some literature research on that to assess feasability of your idea.

5

Just an ordinary openstreetmapper here: I would be curious how well your map matching performs and which one you use?

6

If you are talking about users that are not in wheelchairs, you will get a lot of “bounce” from just their normal stride, even more if they are holding their phone in their hand since most people swing their arms when they walk or run.

1 Like