Workflow automation in robotic cultivation facilities

Recent advances in robotic experimentation have set high expectations for accelerating scientific discovery in natural sciences and engineering. Particularly, in bioprocess engineering, robotic cultivation systems combined with mathematical models enable more efficient strain selection and optimization of operating conditions under process-relevant settings. However, progress is still hindered by the lack of seamless integration of lab robotics, digital infrastructures, and workflow management systems that automate, manage, and allow to share experimental workflows to enhance reproducibility. To address these challenges, this work presents: (1) The integration of a small-scale cultivation system into a liquid handling station for automated cultivations. (2) The integration of a mobile robot, demonstrating flexible device integration in an open laboratory layout. (3) The development and implementation of an open-source workflow management system to orchestrate experimental and computational tasks, facilitate metadata tracking, and contribute to reproducible experimentation. The workflow management system follows the Workflow-as-Code paradigm, allowing users to define, schedule, and execute workflows as Python code, represented as Directed Acyclic Graphs. The robotic system is tested through multiple biotechnological use-cases, demonstrating its ability to support diverse experimental setups, including model-based strain characterization, online redesign of feeding rates in parallel fed-batch cultivations, and strain responses towards glucose oscillations compared to continuous glucose supply. By enabling seamless integration of various devices, heterogeneous computational tools, and automated workflow execution, this approach enhances reproducibility, interoperability, and data quality in bioprocess development.