Ryan D. Lewis is a robotics and software engineer at Argonne National Laboratory’s Rapid Prototyping Lab, where he develops solutions for lab automation, autonomous discovery, and self-driving labs. He is the lead developer of the MADSci (Modular Autonomous Discovery for Science) framework. He has a M.S. in Robotics from the University of Michigan, and an M.S. in Computer Science and B.S. in Computer Science and Computational Mathematics from Northern Illinois University.
Presentation Title:
Building a Modular, Open Source Toolkit for Autonomous Discovery and Self-Driving Labs
Presentation Abstract:
Self-driving laboratories represent the future of scientific discovery, yet most existing automation frameworks are proprietary, domain-specific, or lack sufficient scope for complex autonomous operations. We present MADSci (Modular Autonomous Discovery for Science), an open-source Python framework designed to enable flexible, scalable laboratory autonomy across diverse scientific domains. MADSci employs a microservice-based architecture with clear separation of concerns: Nodes provide standardized device control interfaces, Managers handle lab-wide functionality (experiments, resources, workflows, data), and Applications orchestrate autonomous experimental campaigns. This modular approach allows separation of expertise between device integrators, lab operators, and domain scientists. Built on open standards (REST APIs, Docker, standard databases), MADSci enables definition of entire autonomous experimental campaigns in less than 100 lines of code. We demonstrate the framework’s capabilities through real-world implementations and discuss how MADSci addresses common pain points in open-source laboratory automation infrastructure, providing the foundation for reproducible, collaborative autonomous science.