AI and the World's Hard Problems · Chemistry & materials · arXiv · 2026-10-20
Autonomous adaptive spectral analysis with an LLM agent for self-driving laboratories
Self-driving laboratories need reliable automated analysis of measured spectra, but the analysis can fail on some measurements. The authors built an LLM agent that supervises a fixed analysis script and, for failed spectra, re-runs it with bounded parameter adjustments or returns NaN when no reliable value can be obtained. On optical spectra from 203 β-Ga2O3 films, the agent recovered 16 of 20 failed cases and returned NaN for the other 4.