HERBIE: an open-source autonomous research robot
A low-cost platform for teaching and doing autonomous science, supported by a growing community of researchers, educators, and students.
Plan, experiment, measure, learn — on its own.
- 1 · Plan
- 2 · Experiment
- 3 · Measure
- 4 · Learn ↺
HERBIE is a compact robot that runs experiments on its own. It moves across a well plate, dispenses liquids, and takes measurements, then uses machine learning to decide which experiment to run next. That closed loop of plan, experiment, measure, and learn is the core of autonomous science.
HERBIE is built from 3D-printed parts and off-the-shelf components: stepper motors, a Raspberry Pi, a syringe pump, mass and pH sensors, and cameras. Its design, software, and documentation are open source, so users can build, modify, and extend it.
HERBIE is also modular. New synthesis and characterization capabilities can be added through additional device modules, so the platform can grow with your teaching and research.
Four ways to bring HERBIE to your group.
Education
Course materials, tutorials, and workshops on the fundamentals of autonomous science and engineering research.
Kits
Complete kits of parts for institutions that want to build their own HERBIE.
Assembled systems
Preassembled and tested systems, ready to use out of the box.
Software support
Ongoing maintenance of HERBIE's open-source software.
Contact us for pricing.
Built for teaching. Ready for research.
Education
HERBIE teaches the fundamentals of autonomous science and engineering research: how robots and AI work together to plan experiments, run them, and learn from the results.
HERBIE has been part of a hands-on course in machine learning for autonomous science at the University of Maryland since 2020. The course is now in its seventh year.
Research
HERBIE also supports research projects in:
- Materials science
- Chemistry
- Artificial intelligence
From LEGOLAS to HERBIE.
- 2020 Hands-on ML for autonomous science course begins at UMD Now in its seventh year
- 2022 LEGOLAS, HERBIE's predecessor, published After several years of use in the classroom Read the paperAbout LEGOLAS
- April 30, 2024 Senate Robotics Showcase and Demo Day HERBIE's predecessor demonstrated at the Hart Senate Office Building, Washington, DC
- June 13, 2024 AI Aspirations: R&D for Public Missions — A White House Conference HERBIE's predecessor demonstrated at this Office of Science and Technology Policy (OSTP) meeting, held at the Johns Hopkins University Bloomberg Center, Washington, DC
- June 2027 · planned HERBIE commercial release Expected release date
HERBIE is supported by principal investigators and their teams of undergraduate and graduate students at organizations including:
- University of Maryland
- University of Tennessee at Chattanooga
- Johns Hopkins University
- Loyola University Maryland
- Pacific Northwest National Laboratory
- National Institute of Standards and Technology
- Kohler Co. (industry partner)
Join the HERBIE community
To learn more, bring HERBIE to your classroom or lab, or join the open-source community, get in touch with us.