Strategic planning for probabilistic games with incomplete information

Alternating-time Temporal Logic (ATL) [1] is used to reason about strategic abilities of agents. Aiming at strategies that can realistically be implemented in software, many variants of ATL study a setting where strategies may only take available information into account [7]. Another generalization of ATL is Probabilistic ATL [4], where strategies achieve their goal with a certain probability. We introduce a semantics of ATL that takes into account both of these aspects. We prove that our semantics allows simulation relations similar in spirit to usual bisimulations, and has a decidable model checking problem in the case of memoryless strategies (for memory-dependent strategies the problem is undecidable). Categories and Subject Descriptors I.2.4 [Knowledge Representation Formalisms and Methods]: Temporal logic General Terms Theory Keywords Alternating-time temporal logic, incomplete information, probability

• Type: Conference
• Year: 2010
• Where: ATAL
• Authors: Henning Schnoor
• Keywords: Alternating-time Temporal Logic, ATAL 2010, ATL Study, Intelligent Agents, Strategies