Accelerating battery research with an AI interface between FINALES and Kadi4Mat
Active learning agent discovers Pareto-optimal sodium-ion coin cell protocols faster.
The formation process for sodium-ion coin cells is notoriously time-consuming and directly impacts battery longevity. A team from Karlsruhe Institute of Technology, Helmholtz Institute Ulm, and Technical University of Munich has created an AI-driven interface connecting the FINALES experiment planning framework with the Kadi4Mat research data management ecosystem. This combination uses an active-learning agent that employs multi-objective batched Bayesian optimization to efficiently explore the trade-off between formation time and end-of-life (EOL) performance.
By minimizing the number of physical experiments while still identifying candidate solutions that approximate the Pareto front, the framework accelerates battery research and reduces resource consumption. The interoperability between FINALES and Kadi4Mat allows coordinated, distributed collaboration across automated systems and human-operated workflows, bridging multiple research centers. The authors emphasize this approach is transferable to a wide range of materials science and engineering optimization problems, making it a template for data-driven discovery beyond battery research.
- Optimizes sodium-ion coin cell formation protocols by balancing formation time and EOL performance.
- Uses multi-objective batched Bayesian optimization within an active-learning agent to reduce experiments.
- Interoperable framework bridges FINALES and Kadi4Mat for automated, distributed research collaboration.
Why It Matters
Makes battery development faster and cheaper by dramatically cutting experiments while improving cell longevity.