Hello Everyone, hope all is well with you guys. I was watching a video on you tube about the new Tesla S model car.

https://www.youtube.com/watch?v=pkO5u4AJkwQ

At 39:18 the narrator talks about how the car is moved to a place called "General Assembly" to fit the "Panoramic glass roof"
A few seconds later he says..."The Installation robot takes a picture of the top of the car. It uses the digital picture to measure the exact orientation and bond the glass."
Can some one tell me how this might work? After taking the picture how would the robot know where to align the glass? Does the robot's software some how pick arbitrary points in the digital picture itself for reference?

Even though the robot has a picture how does the software recognize the hole on top of the car? Let alone align the glass?
I would appreciate any and all info on how these robots align items.

Thank You

 

Lots to read about how it works, here - https://www.google.com/search?q=vis...la:en-US:official&client=firefox-a&channel=sb

When still working, my company used 'vision' to measure wiring terminals as they were made. Presses made 1300 parts per minute and 'vision' cheked 6 or more points on every terminal. To 0.001 inch tolerance!

 

Image recognition.

Even though the robot has a picture how does the software recognize the hole on top of the car?

It learns what the hole looks like and then identifies it with different rotations and variations. Once it figure this out, it can determine its distance since it knows how big the hole it is looking at is.

 

Vision systems feature all sorts of edge detection/shape recognition,etc.. and from there it can determine coordinates/rotation angles,etc...
pick-place machines do the same for picking parts out of the feeders and for placement on the boards..

 

Almost all of the car makers use this type of windshield instal, not just Tesla. The rear window is done the same way.