Deterministic planning in the fifth international planning competition: PDDL3 and experimental evaluation of the planners

Abstract

The international planning competition (IPC) is an important driver for planning research. The general goals of the IPC include pushing the state of the art in planning technology by posing new scientific challenges, encouraging direct comparison of planning systems and techniques, developing and improving a common planning domain definition language, and designing new planning domains and problems for the research community. This paper focuses on the deterministic part of the fifth international planning competition (IPC5), presenting the language and benchmark domains that we developed for the competition, as well as a detailed experimental evaluation of the deterministic planners that entered IPC5, which helps to understand the state of the art in the field.We introduce an extension of PDDL, called PDDL3, allowing the user to express strong and soft constraints about the structure of the desired plans, as well as strong and soft problem goals. We discuss the expressive power of the new language focusing on the restricted version that was used in IPC5, for which we give some basic results about its compilability into PDDL2. Moreover, we study the relative performance of the IPC5 planners in terms of solved problems, CPU time, and plan quality; we analyse their behaviour with respect to the winners of the previous competition; and we evaluate them in terms of their capability of dealing with soft goals and constraints, and of finding good quality plans in general. Overall, the results indicate significant progress in the field, but they also reveal that some important issues remain open and require further research, such as dealing with strong constraints and computing high quality plans in metric-time domains and domains involving soft goals or constraints.

Over-Subscription Planning with Boolean Optimization: An Assessment of State-of-the-Art Solutions

“…Thus, our compilation allows to find plans with optimal metrics at fixed makespan. Further, note that while literals related to goal preferences can be implicitly considered to hold only "at the end" modality [22], i.e., at the final makespan, this is not the case for the ones related to preconditions that can, in general, hold at any makespan, unless we know that, instead, STRIPS actions can be only executed once (e.g., this is the case for well known real-world planning domain like blocks-world and logistics). The changes in SATPLAN were mainly related to the creation of formulas in the formats accepted by the various solvers employed instead of the DIMACS format for SAT formulas in CNF.…”

Solving satisfiability problems with preferences

“…In particular, (1)(2)(3)(4)(5) are Formal Verification instances ((1-2) from the Bejing'96 competition, (3)(4)(5) by Ofer Shtrichman); (6)(7)(8)(9)(10)(11)(12)(13)(14) are planning problems from SATPLAN; (15)(16)(17)(18)(19)(20) are Data Encryption Standard instances; (21)(22)(23)(24)(25)(26) are quasi group instances. The domains contain, in general, more benchmarks that the one we show: The instances showed representative of the 9 A literal l is pure in a set of clauses ϕ if l does not belong to any clause in ϕ.…”

“…Indeed, we allow for preferences to be partially ordered and in many applications the number of preferences is relatively low, as, e.g., in planning with soft goals [16,15]. Summing up, given a set of clauses ϕ, the main contributions of the paper are:…”

“…Indeed, in many applications, including planning with soft goals [16], we can expect problems with a few (in the order of tens) preferences, and with such a low number of preferences there is hardly any difference between the performances of nOPTSAT and those of MINISAT when run on the original SAT instance (see [15]). …”

Preference-Based Planning via MaxSAT

“…The 2006 International Planning Competition (IPC-2006) created a track on this topic which resulted in the extension of the standardized Planning Domain Description Language (PDDL) to support the specification of preferences [6]. In PDDL3, desirable properties of a plan are expressed as preference formulae.…”