Broadening Participation with The Autonomous Formulation Laboratory—An Open Hardware, User-Facility-Based Self-Driving Lab for Formulation Optimization

The Autonomous Formulation Laboratory, developed at NIST, is a flexible, open hardware and software platform for AI-accelerated design, discovery, and optimization of liquid formulations using multimodal characterization and x-ray or neutron scattering. Liquid formulations are ubiquitous, ranging from pharmaceuticals to paints, deicing fluids to dandruff shampoo. These products undergo a continuous need for (re)design driven by new active ingredients, changing regulatory landscapes, consumer demands, ingredient availability in the dynamic supply chain, etc. The platform has demonstrated typical speedups of 3-5 x vs typical human-designed grids in tackling real formulation problems in drug delivery, coatings, personal care products, and other areas, with exceptional cases yielding 25 x speedups.<br/><br/>The project launched in 2020 with a single robot housed at NIST and traveling to other facilities, and has since grown to a fleet of 4 NIST-owned platforms that routinely travel for measurements around the world and 3-5 platforms in existence or in the process of being built by partners. Originally built by nSoft, an industry-government consortium focused on developing neutron-based measurements for US industry, the platform has expanded to other techniques, other user facilities, and academic labs and users. This talk will focus on our experiences and lessons learned in scaling the project, broadening our contributor and user base, and outline our vision for self-driving labs that scale directly and meaningfully from sub-$10k, benchtop-scale hardware to globally unique national user facilities. We believe that the realization of this vision will result in substantially increased access to and uptake of autonomous experimentation in labs across materials science.