PI: Associate Professor Emelie Engström, Department of Computer Science, Lund University
co-PI: Professor Nauman bin Ali, Department of Software Engineering, Blekinge Institute of Technology
HUMOND investigates how human-centered monitoring can ensure that autonomous AI-enabled cyberphysical systems (CPS) remain aligned with stakeholder intent and safety constraints throughout the DevOps lifecycle and during long-term operation. HUMOND treats alignment as a system-level property that emerges over time through interactions across development and operational processes, rather than as a property of individual AI components. The project develops alignment theory in the form of (i) a calibrated catalogue of alignment-drift indicators derived from heterogeneous operational data, and (ii) a catalogue of human-inthe-loop integration patterns for embedding alignment review into CI/CD workflows and feedback loops in autonomous CPS, informed by retrospective and longitudinal industrial case studies. The project builds on established industrial collaboration networks in automotive and connected-manufacturing sectors. The expected contribution is a software engineering method that makes alignment drift in autonomous AI-enabled CPS measurable and actionable in industrial DevOps practice.
Project number: H3