Architectural Governance for Autonomous Robotic Systems under the EU AI Act

Autonomous robotic systems increasingly integrate artificial intelligence capabilities operating in dynamic real-world environments. As these systems scale across logistics, manufacturing, and service domains, regulatory frameworks such as the EU Artificial Intelligence Act introduce requirements for governance, monitoring, and lifecycle control of high-risk AI systems. This paper proposes an architectural perspective on how governance mechanisms embedded within robotic platforms can support regulatory compliance. The proposed framework maps regulatory requirements such as risk management, human oversight, monitoring, and traceability to concrete architectural components including capability lifecycle management, authorization control, and runtime safety enforcement. By introducing governance layers that regulate the activation and execution of robotic capabilities, the architecture enables bounded and verifiable operation of autonomous robotic systems and distributed robot fleets. The work contributes to emerging research on AI governance architectures for cyber-physical systems and explores how technical system architectures can operationalize regulatory AI governance requirements.