Grouping factors in more general ones

I have to work on this database on R. I have tried to make a tree model, but all the variables are factors with too many levels. I tried changing the column "operator" in three levels (keeping private, having military for all "military - state" and air company for all the others) using something like getData$operator[grep("Military", getData$operator)] <- "Military" but it changes into NAs since it's not a valid factor.

I could change it to string and make it work, but it doesn't solve the "if not rivate or military, change to air company" part. Is there any other way?