COLL (an abbreviation for Corners of Last Layer) is a method in the CxLL group that solves the Last Layer corners in one algorithm, while preserving the F2L and the orientation of the LL edges. So, after the COLL step, only edge permutation (EPLL) is left. COLL requires learning 40 algorithms (25 excluding mirrors).
COLL is not to be confused with CLL, which is an abbreviation for exactly the same name. The difference is that CLL preserves the F2L but not the last layer edges orientation, so it leaves the LL edges scrambled and the next step would be full ELL. For some cases the CLL and the COLL are the same algorithm, but for other cases the CLL is much shorter. For CLL and COLL algorithms, see the page on CxLL Algorithms.
One possible extension of COLL is ZBLL. This LL method solves the entire LL if the edges are already oriented, but it has almost 500 cases.