Résumé | In LTL synthesis, the task is to construct a reactive system producing an output stream ensuring a given formula of linear temporal logic is satisfied for any input stream. Recent results on translating LTL to automata open avenues to learning-based approaches and heuristics for such problems. In particular, the automata-theoretic approach to LTL synthesis can utilize not only topological information through standard graph algorithms, but can now also profit from semantic information through learning algorithms. As a result, in many cases an optimal solution can be obtained without any computation; in more general cases, the approach might yield more explainable controllers and scale better.
Recording: https://u-bordeaux-fr.zoom.us/rec/share/8DKuNeylXNZZTxxwmxzIO5I-Uvf9wYq_sgQG3GmZXp9AqSUZ1KgBN1x9tGkhn3nU.Tlcq0tUXut3h98_2 |