execution-semantics

This paper proposes a policy-neutral execution and measurement layer to bridge the sim-to-real gap in reinforcement learning-based industrial dispatching, enabling structured attribution of execution errors and improving reliability and interpretability.