Home

MADSci¶

Modular Autonomous Discovery for Science — an open-source Python framework for building modular, autonomous scientific laboratories.

What is MADSci?¶

MADSci makes it easier to build self-driving labs (SDLs) — laboratories that close the loop between experiment design, execution, and analysis so that AI and optimization algorithms can drive discovery with minimal human intervention.

Standing up an SDL usually means integrating a zoo of heterogeneous instruments, robots, and data systems — a large and repetitive systems-engineering burden. MADSci lowers that barrier by providing a common standard for wrapping any instrument or robot as a Node, plus a suite of independent manager services you can use on their own or compose into a full autonomous lab:

• Instrument automation & integration — wrap any device as a MADSci Node behind a common (REST) interface, in any language.

• Workflow management — define and run flexible workflows across one or more nodes.

• Experiment management — run closed-loop autonomous experiments combining workflows, compute, decision-making, and analysis.

• Resource management — track labware, consumables, equipment, samples, and assets.

• Event & data management — distributed logging and data capture across the whole lab.

• Location management — coordinate physical locations and their node/resource representations.

• Observability — built-in OpenTelemetry tracing, metrics, and log correlation.

How to cite¶

If you use MADSci in your work, please cite our paper in the Journal of Open Source Software:

Lewis, R. D., Ginsburg, T. S., Ozgulbas, D., Stone, C., Stroka, A., Cleary, A., Foster, I. T., & Paulson, N. (2026). MADSci: A modular Python-based framework to enable autonomous science. Journal of Open Source Software, 11(119), 9416. Lewis et al. (2026)

@article{Lewis2026MADSci,
  title     = {{MADSci}: A modular {Python}-based framework to enable autonomous science},
  author    = {Lewis, Ryan D. and Ginsburg, Tobias S. and Ozgulbas, Doga and Stone, Casey and Stroka, Abraham and Cleary, Aileen and Foster, Ian T. and Paulson, Noah},
  journal   = {Journal of Open Source Software},
  publisher = {Open Journals},
  year      = {2026},
  volume    = {11},
  number    = {119},
  pages     = {9416},
  doi       = {10.21105/joss.09416},
  url       = {https://doi.org/10.21105/joss.09416}
}

Instruments & robots¶

MADSci and its predecessor tooling have been used to integrate 40+ instruments and robots as reusable modules across the Rapid Prototyping Lab’s autonomous laboratories and collaborating sites. A representative sample:

• Robotic arms & motion — PF400, UR5e, PAL, Hudson PlateCrane & Sciclops, Vention linear rail

• Mobile robots — MiR250

• Liquid handling — Opentrons OT-2, Tecan, Hudson SOLO, Big Kahuna, Barty

• Sample handling & prep — Azenta heat sealer, Brooks XPeel, BioStack, Liconic & Tekmatic incubators, Hig centrifuge, Biometra thermal cycler

• Detection & analysis — BMG & Epoch2 plate readers, Hidex, PerkinElmer ICP, Metrohm, Phenom SEM

• Sensors, control & vision — cameras, LabJack, Arduino, Zigbee, Eurotherm nanodac, object detection, AprilTag / RPL tags

• Emerging & embodied — OpenArm, DexMate Vega, and SO-ARM101 humanoid/teleoperation platforms

• Human-in-the-loop — a person node for capturing human input during a run

Browse the full, up-to-date list of modules on the AD-SDL GitHub organization, and see the Equipment Integrator Guide to integrate your own.

Contributors¶

MADSci is developed by the Rapid Prototyping Lab at Argonne National Laboratory and a growing community. The paper authors and core contributors are Ryan D. Lewis, Tobias S. Ginsburg, Doga Ozgulbas, Casey Stone, Abraham Stroka, Aileen Cleary, Ian T. Foster, and Noah Paulson.

See the full list of everyone who has contributed on the contributors graph. Contributions are welcome — see the Contributing Guide to get involved.

Developed by the Rapid Prototyping Lab at Argonne National Laboratory.
Argonne is a U.S. Department of Energy laboratory managed by UChicago Argonne, LLC.