Expressive Planning Through Constraints

The real-world applicability of automated planners depends on the expressiveness of the problem modeling language. Contemporary planners can deal with causal features of the problem, but only limited forms of temporal, resource and relational constraints. These constraints should be fully supported for dealing with real-world applications. We propose a highly-expressive, action-based planning language which includes causal, relational, temporal and resource constraints. This paper also contributes an approach for solving such rich planning problems by decomposition and constraint reasoning. The approach is general with respect to the types of constraints used in the problem definition language, in that additional solvers need only satisfy certain formal properties. The approach is evaluated on a domain which utilizes many features offered by the introduced language.