What is missing in autonomous discovery: open challenges for the community

Abstract

Self-driving labs (SDLs) leverage combinations of artificial intelligence, automation, and advanced computing to accelerate scientific discovery.

Accurate, Uncertainty-Aware Classification of Molecular Chemical Motifs from Multimodal X-ray Absorption Spectroscopy

“…Gaussian Processes (GPs) are popular choices due to their statistical rigor (GP uncertainty is derived directly from a covariance kernel, and GPs themselves are properly generalizations of the multivariate normal distribution) and the ability to imbue them with prior physical knowledge . GPs have been widely adopted in autonomous experimentation ,,,− as they are fast (assuming small data sets and limited numbers of features), interpretable, and easy to use. Other popular methods are ensembling (whether it be an ensemble of decision trees or random forests), Monte Carlo dropout, and Bayesian neural networks .…”

Autonomous millimeter scale high throughput battery research system

“…Lastly, the MultiAnalyticHub provides a wide spectrum of analytics, from statistical assessments to computer vision methods. 46 The integration of all the Hubs is the prerogative of acceleration in material development, 15,47 and therefore, by following this design philosophy, our Auto-MISCHBARES platform is able to perform fully autonomous electrochemical measurements tailored to study the formation of CEI and monitor oxidation state changes in active materials at different stages of the experiment.…”

“…Mitigation strategies are a crucial aspect in the realization of a truly autonomous laboratory. 47 Thus, our Multi-AnalyticHub is equipped to enhance the platform by implementing mechanisms for providing critical feedback through automated quality control (QC) and real-time analysis throughout the experimental stages (Fig. 1b, Section Quality Control & Analysis).…”

Event-driven data management with cloud computing for extensible materials acceleration platforms

“…Materials Acceleration Platforms (MAPs) comprise the integration of automation and computation in experimental workflows to accelerate the discovery of materials as well as the underlying scientific knowledge. 1,2 Critical analysis within the MAP community has led to the identification of a portfolio of remaining challenges, 3–5 which can be broadly explained as furthering the extensibility and interoperability of MAPs as well as establishing universal data management protocols. Meanwhile, the successes of individual autonomous and self-driving laboratories has inspired and clarified the vision of global, interconnected MAPs.…”