The Energy Impact of Domain Model Design in Classical Planning

TL;DR

This work addresses the energy footprint of classical planning by treating domain model design as a primary lever. It introduces the Domain Model Configuration Framework to systematically generate and evaluate domain-model variants across SSC, MRC, and TDC categories, and then empirically assesses five planners over five IPC domains using 32 variants each. The results show energy consumption is largely stable under small syntactic changes but can explode with redundancy (notably increased action arity) and certain dead-end designs, with strong dependence on planner architecture and domain topology. The study demonstrates that energy efficiency in planning is a structural property emerging from interactions among domain design, planner type, and problem structure, and offers practical guidance for energy-aware modelling and preprocessing, as well as a foundation for future predictive energy-cost models.

Abstract

AI research has traditionally prioritised algorithmic performance, such as optimising accuracy in machine learning or runtime in automated planning. The emerging paradigm of Green AI challenges this by recognising energy consumption as a critical performance dimension. Despite the high computational demands of automated planning, its energy efficiency has received little attention. This gap is particularly salient given the modular planning structure, in which domain models are specified independently of algorithms. On the other hand, this separation also enables systematic analysis of energy usage through domain model design. We empirically investigate how domain model characteristics affect the energy consumption of classical planners. We introduce a domain model configuration framework that enables controlled variation of features, such as element ordering, action arity, and dead-end states. Using five benchmark domains and five state-of-the-art planners, we analyse energy and runtime impacts across 32 domain variants per benchmark. Results demonstrate that domain-level modifications produce measurable energy differences across planners, with energy consumption not always correlating with runtime.