This paper presents a high level strategy planning framework for the UAV, in which the strategy planning problem under uncertain conditions is abstracted into Markov Decision Processes with uncertain parameters, and the mission requirements are specified using Linear Temporal Logic Language. The objective is to compute a robust satisfying policy to maximize the robustness to the uncertainty while satisfying the desired requirements of system performance or the mission specification. The info-gap decision model is used to describe the uncertain parameters of MDP, i.e., the transition probability, and thus we propose a new model as Info-gap based MDPs (IMDPs). The LTL formula of mission specifications is converted to Deterministic Rabin Automaton (DRA). A product IMDP is constructed by combing the IMDP with DRA in the form of Cartesian product. Based on robust dynamic programming, we propose a robust satisfying policy generation algorithm to solve the product IMDP. An example of UAV high level strategy planning is given to verify our algorithm and the resulting policy can maximize the robustness while satisfying the mission specifications.
Info-gap decision; Linear Temporal Logic; Markov Decision Process; Robust satisfying policy