AI Bot AutoLabs Streamlines Robot-Assisted Scientific Experiments
Researchers have developed an AI agent called AutoLabs that translates scientists' experimental goals into instructions for laboratory robots, significantly accelerating the experimental design process.
Scientists at the Pacific Northwest National Laboratory (PNNL) have created a generative agentic AI system named AutoLabs to speed up the process of setting up experiments on autonomous laboratory robots. Traditionally, translating experimental objectives into robot instructions can take weeks due to the need for collaboration between scientists and engineers. AutoLabs aims to bridge this gap by quickly converting a researcher's experimental goals into executable commands for robotic systems.
The AutoLabs system is currently designed to work with Big Kahuna, an automated robot developed by Unchained Labs, which is used for studying battery materials. The AI can manage multi-step experimental workflows, including essential laboratory tasks such as mixing, heating, stirring, and filtering samples with minimal human oversight. This automation allows researchers to conduct significantly more experiments than manual methods would permit, potentially increasing output by five to tenfold.
Developed by a team including data scientist Gihan Panapitiya, AutoLabs is built upon an OpenAI model and functions using specialized 'sub-agents,' each with distinct expertise, coordinated by a 'supervisor' agent. This architecture allows the AI to interpret a user's experimental request and translate it into precise instructions tailored for the specific capabilities of a robot like Big Kahuna.
Researchers tested AutoLabs by tasking it with translating five increasingly complex experiments into commands for the Big Kahuna robot. These experiments involved mixing and manipulating various chemicals, with more intricate tasks incorporating multiple chemical compounds, specific temperature and stirring parameters, and sample transfers between vials. In all tested scenarios, AutoLabs successfully generated accurate instructions, performing comparably to a highly trained human operator.
AutoLabs represents a significant advancement in human-robot collaboration for scientific research. By abstracting complex robotic operations, it democratizes access to sophisticated automated equipment, accelerating discovery cycles. This development aligns with the broader trend of integrating AI into scientific workflows, enabling faster iteration and potentially in-situ experimentation in fields requiring high-throughput analysis, such as materials science for energy storage.
Edited by the news editor with AI from the original report — please refer to the original source.