Cepheid stars play a considerable role as astronomical distances indicators thanks to the empirical relation between their pulsation period and intrinsic luminosity: the PL relation. The uncertainty on this relation is the largest contributor to the error budget of the Hubble constant, that describes the Universe's expansion. The value of the Hubble constant is currently at the center of a major controversy: while it is estimated at 67.4 +/- 0.5 km/s/Mpc by the Planck satellite, the local measurement based on Cepheids is larger by 4 sigma, with a value of 74.0 +/- 1.4 km/s/Mpc. This discrepancy may provide evidence for physics beyond the standard model: it is therefore critical to improve the PL calibration with precise and accurate distance measurements of Cepheids. In 2018, the second data release of the Gaia satellite (Gaia DR2) provided parallaxes for 1.3 billion stars with an unprecedented precision. However, Cepheids are bright stars and are often saturated in detectors. Moreover, the variations in brightness and color that occurs for variable stars like Cepheids are not yet taken into account in the Gaia data reduction. Therefore, Cepheids parallaxes can be affected by systematics due to their photometric variability. In order to avoid these issues, a solution is to find stable and faint companion stars in the close environment of Cepheids. Using 36 indirect, unbiased and accurate distances based on Gaia DR2, I calibrate the PL relation and revise a previous value of the Hubble constant based on HST measurements of Galactic Cepheids.