Workshop: XLOOP 2026: The 8th Annual Workshop on Extreme-Scale Experiment-in-the-Loop Computing
Authors: Woong Shin, Craig A. Bridges, Frédéric Suter, Marshall McDonnell, and Rafael Ferreira Da Silva (Oak Ridge National Laboratory)
Abstract: An autonomous laboratory's inner loops are already closed, but the loop spanning synthesis, measurement, and simulation closes through a person, where a result changes the next decision. An agent placed there is blind, because the state it needs lives in someone's head, in tools that disagree, and in files that cannot be queried. We present \texttt{reach}, an intelligent state layer holding knowledge, execution, and operational state as one navigable graph, kept consistent across its many writers. With that state addressable, an agent plans a campaign and decides each next step while the scientist supervises along a trust gradient. We contribute (i)~the state layer architecture; (ii)~the seam contract, implementable by any long-running external system; and (iii)~an integration study at the Autonomous Chemistry Lab, where the deployed layer drove an agent-planned vial-transfer chain and carried a measured diffraction pattern through a vendor application into agentic Rietveld refinement.
Back to XLOOP 2026: The 8th Annual Workshop on Extreme-Scale Experiment-in-the-Loop Computing Archive Listing
Back to Full Workshop Archive Listing