A Hybrid-OT Runtime for Behavior Tree-Based Skill Orchestration in Heterogeneous Manufacturing Systems

Abstract

Modern manufacturing increasingly depends on the ability to extend existing production systems with new functionality while preserving already established machine behavior. In flexible and adaptive automation, this requires engineering methods that support capability- and skill-based system extension across heterogeneous Information Technology (IT) and Operational Technology (OT) resources. This paper presents a Hybrid-OT runtime architecture for the integrated execution of reusable IT and OT skills in such environments. In the proposed approach, a manually curated Behavior Tree (BT) specifies the desired skill orchestration and serves as the basis for identifying, preparing, verifying, and deploying the required executable skills. The architecture combines BT-based orchestration, standardized machine interface descriptions for the generation of executable BT nodes, Asset Administration Shell (AAS)-based skill management, and Ansible-based deployment. Together these mechanisms enable verification, retrieval, dependency-aware configuration, deployment and execution of reusable skills across ROS 2-based IT components and PLC-based OT components within a common runtime framework. The approach is validated in an applesorting scenario in which an existing quality inspection system is extended with a robotic arm for physical sorting without duplicating established inspection functionality. The results show that the architecture facilitates reproducible and modular system extension that aims to reduce manual engineering effort. The paper thus contributes a runtime-oriented engineering method for deploying skill-based extensions in adaptive and heterogeneous manufacturing systems.