The reason d-orbitals make a difference is that electrons in d-orbitals do not screen nuclear charge as effectively as those in s and p orbitals. This is because of something called penetration .
The mathematical shapes of d-orbitals prevent them from allowing electrons to penetrate very closely to the nucleus, compared with electrons in s or p-orbitals. In gallium, you have $10$ electrons in the filled $3$d-subshell, and each of these electrons is doing a slightly worse job (relatively speaking) of screening the nuclear charge than the electrons in the s and p orbitals. Therefore, the effective nuclear charge in gallium is slightly higher than that in aluminum, so the increase in the radius is a quite a bit smaller than would be expected based on the difference between boron and aluminum, or gallium and indium.
The trend goes:
$\pu{82 pm}$ ($\ce{B}$)
$\pu{118 pm}$ $\ce{(Al)}$
$\pu{126 pm}$ $\ce{(Ga)}$
$\pu{144 pm \ce{(In)}}$
[covalent radii from www.webelements.com].
This effect is generally known as the d-block contraction. (It can be more or less pronounced depending on how you define the atomic radii.)
A similar thing happens (in principle) when you go from indium to thallium; except in this case you are now dealing with adding a filled f-subshell to the valence shell.
Electrons in f-orbitals are even worse at screening nuclear charge than those in d-orbitals, therefore again, the effective nuclear charge in thallium is a bit larger than it is in indium, so again the jump in radius is fairly small (from $144$ to $148$ pm). This effect (of the filled f-subshell) is generally known as the lanthanide contraction.