Date of this Version
Traditional planning assumes reachability goals and/or full observability. In this paper, we propose a novel solution for safety and reachability planning with partial observability. Given a planning domain, a safety property, and a reachability goal, we automatically learn a safe and permissive plan to guide the planning domain so that the safety property is not violated and which can force the planning domain to eventually reach states satisfying the reachability goal no matter how the planning domain behaves. Our technique is based on active learning of regular languages and symbolic model checking. We describe an implementation of the proposed technique and demonstrate that the tool can efficiently construct safe and permissive plans for four examples.
Date Posted: 15 October 2007