Neuro-Symbolic Closed-Loop Control of Laser Powder Bed Fusion with an In-Loop Ontology

The paper (arXiv:2608.05773) introduces a geometry-conditioned, neuro-symbolic control loop where a standards-aligned ontology sits inside the loop, coupling symbolic reasoning with statistical learning to set targets for a constraint-aware predictive controller. The ontology links process objectives and constraints to controller-observable signals, and a description-logic reasoner converts these into actionable control targets. This is a specialized application for additive manufacturing, potentially improving robustness and explainability in laser powder bed fusion. The work highlights a growing trend of integrating symbolic AI with machine learning for industrial process control. Future work may involve experimental validation on physical systems.