"The use of Autonomous Research System (ARES) for carbon nanotube growth optimization.”

Physics Seminar

Friday, January 26, 2024

112 Oelman Hall, 2:30 to 4:00

"The use of Autonomous Research System (ARES) for carbon nanotube growth optimization.”

Dr. Robert Waelder 

Research Scientist in Materials and Manufacturing Directorate with AFRL

Abstract: The nature of scientific advancement is to become more complex over time. As our experiments become more complex, particularly in dimensionality, we begin to lose our ability to see patterns in data, and our standard techniques for planning experiments become insufficient. Materials synthesis is one of these systems: a seemingly simple approach may in fact include upwards of ten input variables, and they are not all guaranteed to be independent. I will discuss how we are approaching this problem using the Autonomous Research System (ARES) to grow carbon nanotubes. In particular, how we are able to perform experiments autonomously, as well as how we automatically summarize experimental outputs to provide an actionable result to our machine learning planning algorithms.

Bio: Dr. Robert Waelder is an Adjunct Professor at Wright State University and Research Scientist with UES, Inc, working in the Materials and Manufacturing Directorate of Air Force Research Lab in Dr. Benji Maruyama’s lab. He received both his PhD in Materials Design and Innovation and BS in Physics at the University at Buffalo.

He is interested in the human/machine/experiment interface in autonomous research, particularly in the context of novel material synthesis, where the input and output spaces for any given system are rich. Benji Maruyama’s group has built the Autonomous Research System (ARES) platform, with which several autonomous laboratories have been built, including two focused on different types of carbon nanotube (CNT) synthesis. Dr. Waelder operates the original system, now dubbed LaserCNT ARES, which is able to perform CNT synthesis experiments ~60x faster than typical systems, and is capable of designing its own experiments in pursuit of a pre-defined campaign goal.