Publication
A Modular Behavior Tree Engine for Smart Manufacturing: An AAS-Driven Framework for Autonomous Skill Execution
Mahdi Rezapour; Abdallah Fawzy Mohamed Mohamed Darwish; Achim Wagner; Daniel Görges; Martin Ruskowski
Zenodo, 2026.
Abstract
The shift toward highly adaptive production ecosystems has intensified the need for automation frameworks that can coordinate complex skills across manufacturing assets. Behavior Trees (BTs) have shown significant potential for coordinating robotic and cyber-physical systems. In this field having a formalized notion of behavior that unifies how skills are described, composed, and executed, would be constructive. Clarifying this concept adds essential semantic structure to BTs, enabling their broader and more reliable application in modular and distributed industrial environments. In parallel, the Asset Administration Shell (AAS) is emerging as the fundamental digital-twin layer of Industry 4.0, offering structured and interoperable representations of assets. However, executable logic is rarely integrated into current AAS implementations, leaving a gap between semantic asset descriptions and operational skill execution.
This work investigates the research questions: 1) How can the notion of “behavior” be formally conceptualized and operationalized within Behavior Trees to enable decentralized and autonomous skill execution in Industry 4.0 manufacturing systems? 2) How can execution and interoperability be formulated within the AAS through BT-based representations? To address these questions, we introduce a Modular BT Engine that (i) formalizes behaviors as composable, execution-ready structures, (ii) read and interprets BT models stored within AAS submodels, and (iii) autonomously executes skills through an integrated IT/OT interface. The engine supports reconfiguration, plug-and-play behavior deployment, and consistent execution semantics across distributed assets. The proposed framework is implemented and validated on real SmartFactoryKL demonstrator where it coordinates robotic tasks and modular production operations. The results indicate that the AAS-integrated BT Engine enables flexible orchestration and adaptive skill execution, thereby advancing the realization of modular and reconfigurable smart manufacturing environments.