Amartya Sen has written extensively about how to do just this, though he wouldn't call it utility either (it's one of the cornerstones of the capability approach). He formalizes it in terms of the real option sets available to an individual rather than "free will" though. The main difficulty is how to quantify and value different option sets. (You can't just look at the size of the sets, because different options are likely to be differentially valuable qua options, and you need to incorporate that somehow.)

Many (even most?) people do have freedom as part of their utility function, but at different weights, so redefinition is unnecessary if you grant a moral imperative to increase the utility of others.

The problem is that most people interpret such an imperative as "increase the utility others would have if they shared my own utility function", which is not at all correct. Simply redefining utility in the general case to include freedom is in this class of mistakes.