Evolutionary models of brown dwarfs are plagued by a lack of observational constraints. The complex molecular chemistry of their atmospheres leaves a relatively wide parameter space for models to span. Placing accurate mass and luminosity data to observationally populate the mass-luminosity relationship provides a major contribution to an understanding of brown dwarf evolutionary models. To date, individual dynamical masses are known for only a handful of brown dwarfs. Radial-velocity measurements provide only a lower limit on the measured masses due to the unknown orbital inclination. Therefore, directly imaging these candidates is needed to break that degeneracy and provide constraints on the dynamical mass of the companion, serving as a benchmark substellar object. I present a systematic approach to hunt for these benchmark objects using the radial-velocity CORALIE survey with over 20 years worth of data containing a volume-limited sample of 1647 low-mass main sequence stars within 50 parsecs. From this survey I have identified targets that I have directly imaged using VLT/SPHERE, which has led to the direct detection and spectral characterisation of several new benchmark brown dwarfs. This includes the coldest companion ever imaged to a main sequence star, and the first T/Y transition object with a measured dynamical mass. These objects are crucial in probing a parameter space in mass, separation, and age where the occurrence rate of these objects is not well understood, and where the mass-luminosity-age relation is unconstrained, providing important clues to the formation and evolution of these ultracool objects.