"AI and Automation-Driven" KAIST and POSCO Holdings Establish Autonomous Exploration Laboratory

An autonomous exploration laboratory that develops cathode materials for secondary batteries using artificial intelligence (AI) and automation technology without researcher intervention has been established.

KAIST announced on August 3 that the research team led by Professor Seo Donghwa from the Department of Materials Science and Engineering, in collaboration with the POSCO Holdings research team, has established the autonomous exploration laboratory through industry-academia cooperation.

The autonomous exploration laboratory is a platform that autonomously designs, conducts, and analyzes experiments to discover optimal materials.

Interior view of the autonomous exploration laboratory for cathode materials. Provided by KAIST

The development of cathode materials for secondary batteries inevitably involves processes such as quantitative measurement and transfer of samples, mixing, pelletizing, sintering, and analysis. Additionally, because it is necessary to explore a wide range of cathode compositions (the ratio of component elements in the compound) and experimental variables, it requires the labor of skilled researchers and a long development period.

In contrast, the autonomous exploration laboratory can perform quantitative measurement, mixing, pelletizing, sintering, and analysis of samples without researcher intervention through an automated system based on an AI model, and it can interpret and learn from the analyzed data to select the best candidate materials.

The joint research team increased experimental efficiency by reducing the reliance on robotic arms. They achieved this by constructing the processes of quantitative measurement, mixing, pelletizing, sintering, and analysis as individual device modules, which are then handled by a central robotic arm.

When the intelligent experimental automation system is operated 24 hours a day, it is possible to secure more than 12 times the amount of experimental data, and the time required for material discovery can be reduced by approximately 93% compared to previous methods.

For example, if material discovery requires 500 experiments, the conventional method where researchers conduct each experiment would take 84 days, whereas the automated system can complete the process in just 6 days. This significantly improves research time and manpower efficiency, according to the joint research team.

Overview of Autonomous Exploration Laboratory for Cathode Materials. Provided by KAIST

In the development process of the autonomous exploration laboratory, POSCO Holdings was responsible for the overall planning and management of the project, jointly reviewing the overall platform design, designing partial modules, and developing AI-based experimental design models.

Professor Seo Donghwa's team was in charge of practical system implementation and operation, including overall platform design, partial module design and fabrication, algorithm development, experimental verification based on the automation system, and error correction.

Professor Seo Donghwa stated, "The automation system will help address the issue of declining research personnel due to low birth rates in the future," adding, "Our research team will secure high-quality material data and accelerate the development of secondary battery materials to strengthen global competitiveness."

This research was led by KAIST Department of Materials Science and Engineering Ph.D. candidate Lee Hyunki and master's students Bae Sungjae and Kim Dongwoo, with joint participation from Senior Researcher Park Jungwoo and Senior Researcher Park Incheol at the Energy Materials Research Institute and LIB Materials Research Center of POSCO Holdings Future Technology Research Institute.