Why does it need to know anything in the Z dimension higher than 6"?

For visual SLAM I presume it's useful to see typically unobstructed clean edges between the walls and the ceiling.

But it should be done on the device, without sending images anywhere.

So it can know where it is in that 6"

https://en.wikipedia.org/wiki/Simultaneous_localization_and_...

It's cheaper to have a camera read the celling photo data and combine that with it bumping into walls to find its way around than to use a sensible solution like a lidar.

It's just a cost cutting measure.

I see, so its more of a positional feature than it is used in the function of mapping, "so it knows where it is in the map" so to speak?

Right. It doesn't care per say if your wall is red or green, but knowing that can help it find the room it is in.

Because things came look very same-y at that height. Looking higher helps to disambiguate.

Why do you need to know what things are? You are a vacuum cleaner. You just need to see that you aren’t on the edge of a step (achieved with foor facing sensors for a xiaomi vacuum) and that you don’t have an obstruction ahead of you (detected by the onboard lidar and confirmed with a pressure plate on the xiaomi).

Thats not how this device works. You can pick it up, walk around, set it down in the den, and push the button, and it will say "vacuuming den". It draws a 2d map of your floor, using 3d images

It's how it should work, that's the point.

Thats flawed because the room can change. My vacuum uses lidar and just makes a new scan, you don’t need to cache any data to plot a route on the floor.