Those instruments already exist, using light:
http://www.zygo.com/
http://www.nanometrics.com/products/clp_9010.html

Do a websearch:
http://www.dogpile.com/info.dogpl/search/web/surface%2Bprofilometry  

TTFN

Thank you! I got frustraded poking around the web and finding nothing. Again thanks for the help.

This case is a bit more harder, curved object I'm trying to measure is glass. Thus no light can be used. I've should have said so earlier.

Why is that a problem?  

Most of the systems available are used to measure objects with glass surfaces.

Alternately, you can use an optical interferometer, which is specifically used to measure optics

TTFN

There has previous attempts to measure the surface with laser and they have failed. Objects that are being measured are 2 square meters in size, and the measuring distance would be about 0,5 meters. I'm not sure if the white light measuring has ever been tested, I'll start finding out if that could be used. Still I have a feeling that radio waves would be the best option.

What exactly are you try to do?  What horizontal spatial resolution are you looking for?

RF spatial resolution is very poor.  Additionally, glass is a dielectric and will not reflect RF well.

Glass optics that large are usually done with a master negative, e.g., an inverse shape that can be placed on top of the unit under test and look for fringes.

If you keep dribbling out information, it's going to take longer to figure a solution.

TTFN

The machine should be able to measure depth of a car windshield that is facing down and moving. It is a bit hard to descripe the process. Windshield is bended to all three dimensions and the depth of (z-axis) is important for the optics.
 Measurement should be taken from bottom to the top of the glass and it should happen pretty fast, roughly 3s/m.
 The optics of the glass doesn't matter at this point only the shape of it (so that it would fit into a car).
Thank you for being patient!

What lateral resolution is required?  if there is a 1" square bump, do you want to detect it?

TTFN

Based on your description, you actually only want to measure the mounting surface?

TTFN

I would only need to measure the depth of the glass, when measured the glass would be facing down (outside down), inside up. And I would only need to know the depth of the glass from bottom edge to upper edge on a very narrow "shred".
 The lateral resolution doesn't matter, but all the windshields are different in size and shape so the measurement should be taken from all over the glass, and leave the computer software to pick up the necessary points (certain points for a certain product). Basically it should work like an echograph.

I don;t understand why you don't think you have a lateral resolution requirement.  If there is a 1" section of the window that's out by 1/8", won't the window seal develop a leak there?

TTFN

That is not possible due to the mold that the glass is bended with, depth is measured about 15 cm from the side edges and from the middle.
 Straight line from upper edge to lower edge and that difference from glass surface. For example think the shape as a bow, string horizontally (straight line) arc down. I would need to measure the distance from the string to the arc.

I'm still looking for an answer on the original question, can distance from glass be measured with RF? Or is it just a waste of time?

In my opinion, the only way to do such a measurement (if it can be done at all), would be at a very high frequency where the wavelength is small.  I am thinking something above 100 GHz.  I suggest you contact someone at the University of Massachusetts--Amherst, where they are specialists in millimeter wave radiometry, and ask them if a radiometer can be used for a distance or interference type measurement.

Also, what type of glass matters a lot, especially if some are metal coated.

http://www.ecs.umass.edu/ece/labs/mirsl/Radiometry/

If you take the RF frequency high enough, you will find yourself in the optical domain!.

I think the solution will be found with a optical measurement solution. The glass will not be transparent to all optical frequencies - choose the optical frequency accordingly.

There are laser distance measurement devices made for use in process applications. Some have very high resolution at a distance up to several meters.

Comcokid: Can you link me to any page that would tell more about those laser distance measurement devices. I am giving up on RF, but it seems rational that with white light these measurements should work. One of the problems is that windshields are made from different colours, but since at least some of the white light frequences should bounce back.