How to Load Spaces like you can in settings in c#?

EricAlbers

We need a way to do what the Load button in Settings->Spaces does in code.
Our customers will be in spaces which the hololens will have extreme difficulty accurately mapping to a particular space (Its dozens of corridors, all look VERY similar)...but we can tell from a QR code where it is and need to Load a particular space.
Any idea how to do this?
Thanks!
Just to add to this, in long corridors of spaces which all look geometrically very similar, if there is wifi, its sketchy, and all the ssids are exactly the same over extremely long stretches of similar looking corridors.

WE KNOW where the device is at specific locals due to QR code placement, so I need to be able to have the hololens be TOLD what space it is in, the Load button in Settings does this, I just need to be able to do that in code.

If its not possible, is there a way to hook the UI in the hololens to open settings and click a load button on a space by hooking the mouse/click system?

stepan_stulov

Hey, @EricAlbers

There is unfortunately no (known to me) API for anything Space related. Spaces are OS-level. Yor app has to work with the space the OS decided the HoloLens finds itself in. Basically a slave-mode. All your app can get is a processed, derivative, reconstructed spatial mesh of the space at a certain update rate.

There are no APis to force the HoloLens to think it's in a different space or to anyhow aid it.

There are indirect ways of telling which space the HoloLens in I think, such as having a unique world anchor per space and seeing which one can get located. There has been a thread earlier on the forum with all kinds of tricks for how to achieve it.

Long story short, it's impossible.

PS: Arguably it's good that it's impossible. One "smart" app editing/unloading spaces and thus making life worse for other apps is arguably a bad idea. I could imagine it could of course be implemented differently where you'd have some kind of per-app sandboxes. But it's not that way right now. The whole concept of augmented reality is that, augmenting the reality, not telling what reality it is. This whole paragraph is all very arguable.

EricAlbers

Thanks Stepan,
I had suspected as much, we are looking at using a raspberry pi acting as a access point to 'fake' a SSID to indicate the area and try to force the proper space load. Lots of work getting the area from the QR code to the raspberry pi, then having it switch SSID's and hoping the Hololens figures it out...not sure how long it will take the hololens to get synched though.

It seems very weird for the development team to assume that the hololens is the only thing which can figure out where it is, with CNN neural networks being used all the time, you would think they would have given the developers a way to inform the device where it is. The trashing of spaces by one app is simply fixed by allowing a copy of the space to be made if requested, or by allowing spaces to be kept on a per app guid type thing....seems a trivial task.

Unfortunately, we are looking at dropping the hololens until a company like Sony or some such comes out with a proper device with a real development API.

Imagine if your writing for a playstation or wii type console and you can't access the GPU hardware directly, no one would develop for your device, or they might, but they would laugh at the company.

Unfortunately, we have an application for the Hololens with large companies which have big $$ to spend, but because of a simple lack of API developer support the really awesome use is going to be lost until a new company can put together proper access to the device hardware.

Perhaps one day MS will develop something and actually allow developers to write applications for it without putting them in boxes and limiting their creativity....but that, I suppose what large companies always do, limit creativity and out of box thinking, till a startup comes and shows them the proper way to do it.
Sigh.

stepan_stulov

A couple of thoughts.

If you corridors are identical and the difference is really only in some kind of identification number, perhaps you could (just as you suggested) not rely on finding yourself in the right space but on a different clue. You can then build your scenario as a "sub-space" where the space is the same but an additional sticker/qr/vumark helps identifying the exact location.

it's a tricky case, because of its "impurity". If I understood you correctly, the rooms are neither completely the same nor distinct enough to allow HoloLens to definitively identify them. Perhaps, due to this "gray zone" you could drop relying on spaces altogether and identify the room by markers only. If you additionally put a constraint onto the mutual marker layout and where the markers are in the room you can then be fairly precise having looked at each marker as a room entry calibration process. Once having looked at each marker you derive world anchors from them and the HoloLens takes it from there tracking-wise. Best of both worlds: precision of the markers and continuity of world anchors. It is an overhead for the user of course...

Hope this helps.