Comment by utopiah
3 days ago
How I'd do it :
- setup PeerTube via Docker https://docs.joinpeertube.org/install/docker
- record a video of a path on my phone
- upload video on PeerTube server (via Web or its new mobile app)
- print a QR code of that resulting URL
- slap QR code(s) at location start
optionally
- write a PeerTube plugin to play/pause https://docs.joinpeertube.org/contribute/plugins#write-a-plu... and https://developer.mozilla.org/en-US/docs/Web/API/Acceleromet...
- properly tag videos with geolocation data https://developer.mozilla.org/en-US/docs/Web/API/Geolocation...
- add monetization, e.g. https://www.npmjs.com/package/peertube-plugin-web-monetizati...
my overall point though is that IMHO the core of the project is a video server. There is already at least a quality one, well maintained with a rich community making plenty of plugins.
All of this sounds great for outdoor navigation, but then again, if you're outdoors with good GPS, you don't really need video navigation, a map will do the trick. Video is best for indoor navigation, which is where positioning is the hardest to do, so your approach doesn't work there.
Would be what you would assume - but here is the reality. Pilots with 1st person POV nav have been highly successful for consumer indoor navigation. For asset tracking, position matters. For consumers trying to get where they are going, they could care less about positioning.
Blue dot navigation has failed because it doesn't translate to I door spaces. Video with motion aware playback gives the illusion of an AR experience - but avoids the drift issues that plagued AR driven experiences.
I can cite for you plenty of 1st person POV pilots where the consumer feedback has been exceptional
Are you saying this doesn't do any kind of positioning? Just plays when the accelerometer shows movement and stops when it doesn't? I need to try this at some point because it really feels like that wouldn't work. People walk at different speeds so the video would desync, take a wrong turn or incorrectly recognise landmarks so the video would be completely wrong... What's the error recovery procedure here?
Might just need to try it if you say it works well. Hopefully I find some time to nuke the blockchain out of this and set up a test instance for our university.
1 reply →
The other side to the equation is you have a ton of CXOs that got burned on cap ex trying to use bluetooth beacons to implement effective indoor way finding solutions
You tell a CXO they can have indoor wayfinding to improve patient experience scores and there is no cap ex because there is no hardware - there is literally zero reason to turn it down.
They have nothing to lose and everything to gain
Well yes, that's kind of my point. I understood "just use a peertube instance and the web geolocation api" to mean using GPS for location. Outdoor navigation is more or less solved, but indoor doesn't benefit from GPS at all, so you need some other positioning system. I've worked on a few deployments of different systems and yeah, expensive, annoying, and usually still unreliable.
Like I said in the other comment, if you can make it good without any indoor positioning, that's a killer feature. I just...have my doubts...
1 reply →
This is great feedback - thank you. Video server yes - but what is really critical to me is the creator studio feature that enables nav overlays (these overlays become the throughput for accessibility for vision impaired/language barriers)
Will def look into Peertube