Deriving the expansion history of the Universe is a major goal of modern
cosmology. To date, the most accurate measurements have been obtained with Type
Ia Supernovae and Baryon Acoustic Oscillations, providing evidence for the
existence of a transition epoch at which the expansion rate changes from
decelerated to accelerated. However, these results have been obtained within
the framework of specific cosmological models that must be implicitly or
explicitly assumed in the measurement. It is therefore crucial to obtain
measurements of the accelerated expansion of the Universe independently of
assumptions on cosmological models. Here we exploit the unprecedented
statistics provided by the Baryon Oscillation Spectroscopic Survey (BOSS) Data
Release 9 to provide new constraints on the Hubble parameter $H(z)$ using the
em cosmic chronometers approach. We extract a sample of more than 130000 of the
most massive and passively evolving galaxies, obtaining five new
cosmology-independent $H(z)$ measurements in the redshift range $0.3<z<0.5$,
with an accuracy of $\sim$11-16\% incorporating both statistical and systematic
errors. Once combined, these measurements yield a 6\% accuracy constraint of
$H(z=0.4293)=91.8\pm5.3$ km/s/Mpc. The new data are crucial to provide the
first cosmology-independent determination of the transition redshift at high
statistical significance, measuring $z_{t}=0.4\pm0.1$, and to significantly
disfavor the null hypothesis of no transition between decelerated and
accelerated expansion at 99.9\% confidence level. This analysis highlights the
wide potential of the cosmic chronometers approach: it permits to derive
constraints on the expansion history of the Universe with results competitive
with standard probes, and most importantly, being the estimates independent of
the cosmological model, it can constrain cosmologies beyond -and including- the
$\Lambda$CDM model.