(588cn) Streamlining the Integration of Machine Learning and Automation for Materials Development and Scale-up

To achieve a truly closed loop laboratory, there must be seamless integration of algorithmic decision making, data analytics and automated experimentation. Unfortunately, there is no unified approach for AI and automation integration for chemical laboratories due to the low connectivity of conventional lab equipment, and lack of a common framework for establishing communication. In this talk, we describe the development of FLAB, a new, open-source framework for building autonomous or self-driving laboratories. Built with python, FLAB creates an open, object-oriented architecture that allows users to easily develop systems with multi-task, synchronous operations, communicate with multiple devices and utilize the latest AI and machine learning packages available. We present the development of FLAB across published case studies from automated formulation, to nano-cellulose synthesis, to scale-up of pharmaceutical intermediates. We show how 1) simplification 2) agility and 3) user-friendliness drive adoption of automated and autonomous systems. Finally, we showcase the development of a new, code-free implementation of FLAB - AMLearn - which allows users to access the benefits of FLAB without the requirement of coding experience.