The Araucaria Project: High-precision orbital parallax and masses of eclipsing binaries from infrared interferometry

Context. The precise determinations of stellar mass at $\sim$1% provide important constraints on stellar evolution models. Accurate parallax measurements can also serve as independent benchmarks for the next Gaia data release. Aims. We aim at measuring the masses and distance of binary systems with a precision level better than 1% using a fully geometrical and empirical method. Methods. We obtained the first interferometric observations for the eclipsing systems AI Phe, AL Dor, KW Hya, NN Del, $\psi$ Cen and V4090 Sgr with the VLTI/PIONIER combiner, which we combined with radial velocity measurements to derive their three-dimensional orbit, masses, and distance. Results. We determined very precise stellar masses for all systems, ranging from 0.04% to 3.3 % precision level. We combined these measurements with stellar effective temperature and linear radius to fit stellar isochrones models and determined the age of the systems. We also derived the distance to the systems with a precision level as high as 0.4%. Conclusions. The comparison of theoretical models with stellar parameters shows that stellar models are still deficient in simultaneously fitting the stellar parameters (Teff , R and M) with such level of precision on individual masses. This stresses the importance of precisely measuring the stellar parameters to better calibrate stellar evolution models. The precision of our model-independent orbital parallaxes varies from 24$\mu$as as to 70$\mu$as and they provide a unique opportunity to check on the future Gaia measurements for possible systematic errors.