Building a Self-Driving Lab from Scratch

The development of autonomous laboratories, or self-driving labs, represents a paradigm shift in experimental research by automating the entire experimental process—from planning and execution to analysis and iteration. In this talk, we will explore the process of building a self-driving lab from scratch, outlining three key approaches to laboratory autonomy and how they can be implemented to accelerate discovery. The first approach is the development of an all-in-one automated platform, where the "brain" (AI-driven experimental planning) directs the workflow, and the "eyes" (computational feedback) inform decision-making algorithms in real-time. The second method focuses on modular integration, allowing flexibility through the addition of robotic arms and components. This approach replaces high-fidelity characterization with lower-fidelity, higher-throughput proxies, enhancing speed while maintaining effective feedback. Lastly, open-source hardware is discussed as a means of democratizing science with low-cost modular autonomous setups. With custom-built robots, labs can be tailored to specific research needs at a lower cost, making advanced automation accessible to more scientists. Through these three approaches, we will demonstrate how to build a fully functional self-driving lab from the ground up, revolutionizing how experiments are conducted and accelerating scientific discovery across disciplines.