Towards Strong Cyclic Planning under Partial Observability

Strong Cyclic Planning aims at generating iterative plans that only allow loops so far as there is a chance to reach the goal. The problem is already significantly complex for fully observable domains; when considering partially observable domains, even providing a formal definition is far from straightforward. In this work, we provide a formal definition of Strong Cyclic Planning under Partial Observability, which makes clear how several degrees of solution are possible and equally interesting, depending on the admissible delay between achieving the goal and detecting that it has been achieved.

• Type: Conference
• Year: 2006
• Where: AIPS
• Authors: Piergiorgio Bertoli, Alessandro Cimatti, Marco Pistore
• Keywords: AIPS 2006, Artificial Intelligence, Formal Definition, Observable Domains, Strong Cyclic Planning