Bridging Instruments, Automation, and Machine Learning in Complex Scientific Workflows

Speaker: Dr. Matteo Michiardi (UBCV Physics and Astronomy; Quantum Matter Institute) This talk is part of the Characterization @ UBC seminar series, held biweekly in person at UBC Vancouver. Website - https://characterization.ubc.ca Subscribe here - https://ubc.ca1.qualtrics.com/jfe/for... ------------------------------------- Modern laboratories often bring together many instruments, each with its own proprietary control system and data format. While vendors provide domain‑specific solutions for their own devices, these tools do not scale to laboratories where data must move freely between heterogeneous instruments in a distributed environment. At the same time, the rise of AI - both in pattern recognition and domain‑specific reasoning - and the growing investment in self‑driving materials discovery highlight the need for equally capable, automated characterization workflows. Achieving this in complex heterogeneous cutting-edge research labs, however, presents a distinct set of challenges. In this talk, I will discuss several efforts in laboratory integration and automation informed by our experience developing software infrastructure in the ARPES laboratory and through custom solutions built at TapyrLabs. I will outline approaches for orchestrating multiple instruments in a decentralized manner, considerations around data ethics and reproducible workflows, and our recent work in machine learning for autonomous decision-making.