AINeutralarXiv – CS AI · 14h ago6/10
🧠
Bridging the Sim-to-Real Gap in Reinforcement Learning-Based Industrial Dispatching through Execution Semantics
Researchers propose a policy-neutral execution layer that bridges the gap between reinforcement learning scheduling policies and real-world industrial deployment by standardizing decision snapshots, defining explicit action admissibility, and attributing execution failures to specific causes rather than treating them as undifferentiated errors.