Towards an Intention Abstraction Layer for Autonomous Industrial Systems

arXiv cs.AI · 6 hours ago

Researchers propose an Intention Abstraction Layer (IAL) that uses a large language model and formal ontology to convert natural language goals into structured intentions for autonomous industrial systems, detect conflicts before execution, and explain them to users. The proof of concept demonstrates two autonomous agents registering conflicting production and energy intentions that the IAL flags and explains before reaching the execution layer. This approach shifts conflict detection from after failures occur to before autonomous subsystems begin executing conflicting goals.