Engineering AI planning systems

Abstract

Planning and acting adaptively in complex environments is fundamental to human intelligence. We navigate such complexity by simulating, evaluating, and selecting among possible courses of action using internal models of the world. The field of AI planning seeks to computationally reproduce this form of intelligent behavior. This habilitation thesis presents research on engineering AI planning systems that can effectively address complex, real-world problems, with the broader objective of enabling users to design, develop, and operate such systems. The thesis offers and advocates for a holistic approach to AI planning, one that integrates algorithmic, knowledge, and systems engineering to support the entire lifecycle of planning technology, from conceptualisation to deployment. By bridging foundational research with real-world application, it strives to widen the academic dialogue around AI planning and to establish a pathway toward making AI planning technology a practical tool for innovation. The habilitation thesis is cumulative and accompanied by an executive summary synthesising insights from sixteen publications. The executive summary begins by outlining the research context, key challenges, and research approach employed. It then turns to the published contributions, with particular focus on systems engineering and knowledge engineering, including methodological and artefactual advancements. These contributions are supported by applied research in the domain of smart energy systems. The summary concludes by reflecting on the implications of the work and offering directions for future work.