A group of scientists are using robotics and machine learning to invent new advanced compounds and chemicals faster and cheaper—without the presence of a human in the lab.
The researchers founded startup company Kebotix in 2017 using this technology to create a “self-driving lab” that could create new compounds with beneficial properties—from absorbing pollution to fighting drug-resistant infections to improving electronics.
Kebotix’s software uses machine learning methods to design new chemical compounds. It then feeds 3D models of the compounds into a neural network that learns a statistical representation of the compounds’ properties. The algorithm then works to develop potential new compounds that could fit the molecular model. Another neural network is used to evaluate and discard designs that don’t fit into the model and would likely be unsuitable.
The company uses its robotics system to test the new materials the way a human researcher would: repeatedly experimenting with them. The results are fed back into the first machine learning system to further refine the molecular model. This process is repeated until the right chemical compound is found.
“The AI predicts and plans what to do next; the robot automation system very rapidly tests our new molecule,” said Christoph Kreisbeck, Kebotix’s chief product officer. “The machine can learn from the database and make a better decision for the next round.”
The company is starting with molecules for electronic applications but plans to expand to new polymers and alloys.
Using automation to develop advanced materials isn’t new. It is particularly common in the pharmaceutical industry. But, the process requires massive amounts of data that can be difficult and time consuming to collect for even the simplest chemical compound. Kebotix believes that using AI and robotics in the lab could significantly cut down the time and cost, potentially reducing research that could take decades down to months.
“Discovery is too slow,” said Jill Becker, Kebotix CEO. “You have an idea for a material. You try to make it, and you test it. Few ideas are tested, with even fewer results.”
Kebotix’s potentially game-changing technology has received plenty of interest. The company recently raised $5 million in seed funding.
“We are building the materials company of the 21st century because how scientists discover new materials has not evolved since the 18th century,” Becker said.
Read more about the use of AI in labs at A Healthy Future for Artificial Intelligence in Healthcare.