Materials Discovery

Machine learning and robotics are used to propose and test new chemical compounds and materials.

Business impact

• Discovery speed — The article says machine learning and robotics could make the process “much faster.”
• Human effort in experimentation — The article says humans are “barely required” in the setup.

Data requirements

• 3-D molecular models of compounds with known properties (text) — These models are used to train the system and generate new candidates.
• Experimental results from robotic tests (tabular) — The measured outcomes are fed back into the machine-learning pipeline.

AI methods and techniques

• Machine learning — Used to learn from known molecular properties and generate candidate materials.
• Neural networks — One network learns a statistical representation of desired properties and another filters out unlikely designs.
• Robotic automation — A robotic system tests the remaining chemical structures and feeds results back into the loop.
• Feedback loop — Experimental results are returned to the model to refine the next round of candidates in the self-driving lab.

Industries

Real-world evidence

1 documented case study on record.

Companies using this: Kebotix.