China News Service, Beijing, January 24 (Reporter Sun Zifa) A new chemistry paper was published in Nature Communications, an academic journal of Springer Nature. Researchers reported on two independent laboratories located in Singapore and Cambridge, UK. This approach could help make certain types of research more efficient by increasing the flow of data and materials between laboratories in different parts of the world, which has so far been difficult.
According to the paper, automating laboratory equipment is a complex task that requires a team of experts and expensive equipment to work on a large scale. A lack of standardization also means that no two autonomous laboratories are the same, each based on the specific needs of the researcher. and restrictions tailored.
However, this flexibility, while useful on a small scale, makes sharing data and facilities difficult.
Research is therefore often conducted within a single organization by large research teams, with little global collaboration.
In this study, an overview of the method called "The World Avatar" that can connect autonomous laboratories around the world (drawn by the author of the paper).
Springer Nature/Photo provided
Based on this, in this study, the corresponding author of the paper, Markus Kraft of the University of Cambridge, UK, and his colleagues and collaborators used a dynamic knowledge graph to associate abstract knowledge and language with specific hardware execution, thereby linking two independent autonomous systems. laboratory and optimized a simple chemical reaction experiment.
Each laboratory selected promising experimental conditions following a reinforcement learning scheme and added the results to a common knowledge base as the basis for decisions on new experiments.
Although the two laboratories work on different platforms and communicate entirely through the Internet, the laboratory cooperation has accelerated data generation, improved resilience to software and hardware failures, and demonstrated effective communication and cooperation.
The authors conclude that their proof-of-concept research could lead to improvements in sharing and collaborating on autonomous laboratory equipment, similar to how cloud computing shares available hardware resources.
(over)