LEGOLAS: a low-cost kit for learning autonomous science
HERBIE's predecessor, built for teaching the closed loop of plan, experiment, measure, and learn.
The first step into self-driving labs.
LEGOLAS is a low-cost autonomous research kit designed for teaching. Like HERBIE, it closes the loop: it runs an experiment, measures the result, and uses machine learning to decide what to try next.
Built from LEGO parts, 3D-printed parts, a Raspberry Pi, and aluminum extrusions for under $300, LEGOLAS is a modular kit that mixes food-safe household acids and bases at varying ratios and measures the resulting pH. A simpler color-mixing version, using food coloring and a camera, suits younger students.
Students use Gaussian process regression and active learning to map the titration curve and hit a target pH in fewer than 10 experiments, then go further with Bayesian inference and symbolic regression to rediscover the Henderson–Hasselbalch equation.
LEGOLAS has been used in machine learning courses at the University of Maryland, where student teams work through multi-week challenges with the kits. After several years in the classroom, it was published in MRS Bulletin in 2022. Read the paper →
Demonstrated on Capitol Hill and at a White House conference.
Senate Robotics Showcase and Demo Day
Hart Senate Office Building, Washington, DC
AI Aspirations: R&D for Public Missions — A White House Conference
Hosted by the White House Office of Science and Technology Policy at the Johns Hopkins University Bloomberg Center, Washington, DC