Reshaping Diverse Planning

Abstract

The need for multiple plans has been established by various planning applications. In some, solution quality has the predominant role, while in others diversity is the key factor. Most recent work takes both plan quality and solution diversity into account under the generic umbrella of diverse planning. There is no common agreement, however, on a collection of computational problems that fall under that generic umbrella. This in particular might lead to a comparison between planners that have different solution guarantees or optimization criteria in mind. In this work we revisit diverse planning literature in search of such a collection of computational problems, classifying the existing planners to these problems. We formally define a taxonomy of computational problems with respect to both plan quality and solution diversity, extending the existing work. We propose a novel approach to diverse planning, exploiting existing classical planners via planning task reformulation and choosing a subset of plans of required size in post-processing. Based on that, we present planners for two computational problems, that most existing planners solve. Our experiments show that the proposed approach significantly improves over the best performing existing planners in terms of coverage, the overall solution quality, and the overall diversity according to various diversity metrics.

Top-Quality Planning: Finding Practically Useful Sets of Best Plans

“…To be able to do that, we need to come up with a reformulation of a planning task that forbids all plans with the exact number of appearances for each action. Similar reformulation was recently suggested by Katz and Sohrabi (2020) for diverse planning. The reformulation, which we refer to as FPMA for Forbidding a Plan as a Multi-set of Actions, can forbid a single multi-set, and thus for a set of plans, the union of their multi-sets was forbidden in each consecutive iteration.…”

“…In this case, we do not need to maintain the mapping between the reformulated and original actions and keep the reformulated task size smaller. In the rest of this section we extend the definition of Katz and Sohrabi (2020) to a set of plans (as multisets), present an algorithm that exploits the adapted definition to derive top-quality solutions, and prove its soundness and completeness. We start by presenting a simplified variant of the original definition.…”

“…Diverse planning focuses on difference between plans, evaluating a set of plans by aggregating over the pairwise differences between plans in the set. Recent work in diverse planning introduced additional restrictions on solution quality, requiring each plan in the set to also be of bounded cost (Vadlamudi and Kambhampati 2016;Katz and Sohrabi 2020). Top-k planning is a generalization of cost-optimal planning.…”

“…A top-k planner would have to generate all these plans before it can get to a plan of a higher cost. Diverse planners (Bryce 2014;Nguyen et al 2012;Coman and Muñoz-Avila 2011;Roberts, Howe, and Ray 2014;Vadlamudi and Kambhampati 2016) tackle the issue by defining diversity criteria over a set of plans, but only a handful of works take the plan quality into consideration (Roberts, Howe, and Ray 2014;Vadlamudi and Kambhampati 2016;Katz and Sohrabi 2020). While some computational problems in diverse planning do require to provide some guarantees on solution quality (e.g., Katz and Sohrabi 2020), existing diverse planners still do not provide such guarantees.…”

“…Furthermore, we propose a first planner for unordered topquality planning that iteratively finds a single plan of top quality and forbids at once all plans found so far, including all their valid reorderings. For that, we adapt a recently proposed diverse planning reformulation that forbids a single multiset of actions (Katz and Sohrabi 2020) to forbid exactly a collection of multisets. Our adaptation of the existing reformulation allows us to forbid multiple sets of plans at each iteration while preserving soundness and completeness of our approach.…”