Interferometric measurements of optical path length differences of stars over large baselines can deliver extremely accurate astrometric data. The interferometer GRAVITY will simultaneously measure two objects in the field of view of the Very Large Telescope Interferometer (VLTI) of the European Southern Observatory (ESO) and determine their angular separation to a precision of 10 micro arcseconds in only 5 minutes. To perform the astrometric measurement with such a high accuracy, the differential path length through the VLTI and the instrument has to be measured (and tracked since Earth's rotation will permanently change it) by a laser metrology to an even higher level of accuracy (corresponding to 1 nm in 3 minutes). Usually, heterodyne differential path techniques are used for nanometer precision measurements, but with these methods it is difficult to track the full beam size and to follow the light path up to the primary mirror of the telescope. Here, we present the preliminary design of a differential path metrology system, developed within the GRAVITY project. It measures the instrumental differential path over the full pupil size and up to the entrance pupil location. The differential phase is measured by detecting the laser fringe pattern both on the telescopes' secondary mirrors as well as after reflection at the primary mirror. Based on our proposed design we evaluate the phase measurement accuracy based on a full budget of possible statistical and systematic errors. We show that this metrology design fulfills the high precision requirement of GRAVITY.

GRAVITY is an adaptive optics assisted Beam Combiner for the second generation VLTI instrumentation. The instrument will provide high-precision narrow-angle astrometry and phase-referenced interferometric imaging in the astronomical K-band for faint objects. We describe the wide range of science that will be tackled with this instrument, highlighting the unique capabilities of the VLTI in combination with GRAVITY. The most prominent goal is to observe highly relativistic motions of matter close to the event horizon of Sgr A*, the massive black hole at center of the Milky Way. We present the preliminary design that fulfils the requirements that follow from the key science drivers: It includes an integrated optics, 4-telescope, dual feed beam combiner operated in a cryogenic vessel; near-infrared wavefrontsensing adaptive optics; fringe-tracking on secondary sources within the field of view of the VLTI and a novel metrology concept. Simulations show that 10 {\mu}as astrometry within few minutes is feasible for a source with a magnitude of mK = 15 like Sgr A*, given the availability of suitable phase reference sources (mK = 10). Using the same setup, imaging of mK = 18 stellar sources in the interferometric field of view is possible, assuming a full night of observations and the corresponding UV coverage of the VLTI.