Planning with a Language for Extended Goals

Planning for extended goals in non-deterministic domains is one of the most significant and challenging planning problems. In spite of the recent results in this field, no work has proposed a language designed specifically for planning. As a consequence, it is still impossible to specify and plan for several classes of goals that are typical of real world applications, like for instance “try to achieve a goal whenever possible”, or “if you fail to achieve a goal, recover by trying to achieve something else”. We propose a new goal language that allows for capturing the intended meaning of these goals. We give a semantics to this language that is radically different from the usual semantics for extended goals, e. g. , the semantics for LTL or CTL. Finally, we implement an algorithm for planning for extended goals expressed in this language, and experiment with it on a parametric domain.

Authors

• Ugo Dal Lago
• and Paolo Traverso ITC-IRST

Keywords

No keywords are indexed for this paper.