During the project, Dunia progressed the core technical development of Dunia’s autonomous catalyst discovery platform, IRIS, with the goal of delivering an integrated, reliable system that can design, run, and learn from experiments with minimal human intervention. The main focus was on building the full end-to-end workflow: laboratory automation, experiment orchestration, data capture and storage, and machine-learning models operating together as one operational platform.
A major achievement was the development and validation of IRIS Platform V2. We integrated robotics and automated lab workflows with software orchestration, data management, and ML components into a single system and ran extended trials to test stability and robustness. In total, 2,533 automated platform tasks were executed during the first half of the project only. System reliability improved steadily, and by the end of the first half of the project the overall failure rate decreased to 0.19%, demonstrating strong progress toward stable and repeatable autonomous operation.
In parallel, we built a unified data infrastructure that allows structured ingestion, tracking, and versioning of experimental results, performance metrics, degradation/stability data, and analytical outputs. This enables consistent descriptor generation, model training, and closed-loop optimisation, and establishes the foundation for reproducible, large-scale CO2 reduction research. Building on these capabilities, we initiated AI-driven catalyst discovery campaigns in closed-loop mode, where algorithms select candidates, the platform runs automated experiments, and models are updated based on results. This demonstrates meaningful progress toward identifying high-performance and stability-aware CO2 reduction catalysts and toward maturing the overall system for industrially relevant operation.
