We have written a number of papers combining linear temporal logic (LTL) and reinforcement learning. See also our work on reward machines and on symbolic planning with LTL and automata.