A multimodal robotic platform for multi-element electrocatalyst discovery

"Pioneering advances have been achieved in computational predictions and the automation of materials synthesis17. Yet, most materials experimentation remains constrained to using unimodal active learning (AL) approaches, relying on a single data stream. The potential of AI to interpret experimental complexity remains largely untapped8,9. Here we present Copilot for Real-world Experimental Scientists (CRESt), a platform that integrates large multimodal models (LMMs, incorporating chemical compositions, text embeddings, and microstructural images) with Knowledge-Assisted Bayesian Optimization (KABO) and robotic automation."

"One of the goals of AI for Science' is to discover customized materials through real-world experiments. Pioneering advances have been achieved in computational predictions and the automation of materials synthesis17. Yet, most materials experimentation remains constrained to using unimodal active learning (AL) approaches, relying on a single data stream. The potential of AI to interpret experimental complexity remains largely untapped8,9."

AI-driven materials discovery seeks customized materials via real-world experiments, but most experimentation uses unimodal active learning and single data streams, limiting interpretation of experimental complexity. CRESt integrates large multimodal models that combine chemical compositions, text embeddings, and microstructural images with Knowledge-Assisted Bayesian Optimization and robotic automation. CRESt uses knowledge-embedding-based search-space reduction and adaptive exploration–exploitation strategies to accelerate materials design, high-throughput synthesis, and characterization. The platform targets electrocatalysis and fuel-cell applications and enables coupling of multimodal data, automated experimentation, and optimized decision-making to speed practical materials optimization.

Read at www.nature.com