Dealing with uncertainty in the context of planning has been an active research subject in AI. Addressing the case when uncertainty evolves over time can be difficult. In this work, we provide a solution to this problem by proposing a temporal logic to reason about quantities and probability. For this logic, we provide a PSPACE SAT algorithm together with a complete calculus. The algorithm enables us to perform planning under uncertainty via SAT, extending a technique used for classic planning. We can show that any obtained plan will have certain properties (desired or undesired). The calculus can also be used to derive the impossibility of a plan, given a set of specifications.
展开▼