I have some idea about make something and make some small production. Idea is to have something that measure distance between two objects. First one called A size doesn't matter, can be connected to pc or something and measure distance to B. B have to be insignificant size. This will operate with maximum distance 5m (10m even better) in xyz with precision of mm. Is there some technology to measure this, and what technology? And A to B is not necessary to be at line of view, and speed of motions is speed of human motions. B can be attached on for example human who is moving some room or open space.
I've seen three basic options for distance measurement: triangulation, measuring power drop from a source of known power, and measuring travel time for a signal of known velocity.
Signal power drop is an issue over 10 meters because of the inverse square law: if a signal can be detected 10m away, it's going to be 10000x stronger at a distance of 10 cm. For magnetic fields, for example, that's a difference between 60 uT (about the Earth's magnetic field) and 6 T (a very very strong magnet, which would rip the metal off anything nearby). You might could use RF power drop, but I expect that there would be similar safety issues, not to mention the FCC coming down on you like a ton of bricks.
You've got two basic options for timing signal propagation: sound, and EM fields. EM fields can't really give you the precision you want using travel time as your distance measurement, because 1 mm is about 3 picoseconds at the speed of light. I could be wrong, but that's orders of magnitude beyond what's a reasonable granularity for timing measurements with any hardware I've ever seen.
Sound is easier to work with, but can't really propagate at the distances you're talking about at frequencies over about 250 kHz. (http://www.katho.be/apps.aspx?smid=2688) The wavelength of 250 kHz is about 1.4 mm, so no one microphone is going to give you an arrival time down to the precision you want. Possibly some sort of interferometer? Not to mention that those high frequencies aren't going to work so well without line of sight.
Your best bet is probably triangulation. That requires three locations, not two, and might give you acceptable results if a third station is something you can work with. That changes the question to: how do I determine the direction to something, when it might change location by 1mm 10 meters away, giving a needed precision of five thousandths of a degree?
I'm afraid I don't know the answer to that one. :)
You should investigate the technology that Polaroid first used in its cameras to perform autofocus. It uses pulses of sound at about 50 kHz. Similar devices are still available, e.g. the SensComp 7000 is a complete system comprising a transmitter, receiver and transducer. It claims a range up to about 10 meters and a 17 degree beamwidth and is relatively low cost (about $55). Accuracy and range resolution are not specified but should be in the order of 10 mm, maybe better since this technology was used to focus a camera lens. You may have problems in your application if the target is small and/or moves fast. Some sort of scanning may be necessary if the bearing of the target B relative to your A can vary a lot and rapidly.
If you have a accurate high resolution clock, you can measure the time difference a data frame uses to travel from host to device and back. This principle is used in the Network Time Protocol (NTP) that in turn is used to make PC's keep accurate time. Once NTP is in sync, you know the delay (the distance between sender and receiver in seconds) and the jitter (accuracy). Now the fun part about this is that NTP is an open protocol and therefore extensively explained on Internet, the downside is that it is quite complicated to grasp first time. This might be a good start if you are interested in how NTP works. On the upside, you are only in it for measuring delay, not for exact time keeping. This may simplify the programming efforts.