Think very seriously about whether you want to do theory, observation, data analysis or instrumentation.
You may end up doing things you never imagined out of necessity (like theorists go take observations, cause if they don’t no one else will; or observers running simulations, or building the instrument they need to do the observations etc etc).
…
Finally: READ!!! Pro-actively.
Check arXiv regularly and thoroughly. Read the papers relevant to you and anything else that looks interesting.
Read the references! They are there for a reason. Read the citations - if a paper is interesting, papers which cite it are also likely to be interesting. Use the ADS “C” option liberally and look through it quickly. If in doubt ask you advisor, or just read it anyway.

Next, the slightly tricky issue of what we actually “do”, research wise type of thingy. Might take a while…