“…Specifically, we assume given a suitable lookahead L ∈ N such that the goal in each step is to find a minimum-cost controller strategy to avoid the error states for at least L steps, starting from the current state s. Note that, in this case, we can simply add up the costs of a play with no need for a discount factor. This problem can be formalized using the techniques developed in [8]. For the original game G S = (N, I, M, c) and the error states E, define the set of error nodes N e := {(s, i) | s ∈ E, i ∈ {0, 1}}, and consider the modified game G S,E,L,s = (N , I , M , c ) starting in s, where…”