Poster

Simon Seidel, Dr. Simon Rihm, Dr. Jörg Schemberg, Christian Schipp, Prof. Markus Kraft, Prof. Peter Neubauer, Dr. Mariano Nicolas Cruz Bournazou (TU Berlin)

Automated bioprocess laboratories generate heterogeneous research data across bioreactors, analytical instruments, automation systems, and human-driven workflows. Although individual laboratory systems are increasingly automated, their data often remain fragmented, weakly contextualized, and difficult to integrate or reuse. Within the ABM-KIMa project, supported by the Federal Ministry for Economic Affairs and Energy, TU Berlin and CMPG are developing an Autonomous Biolab Manager (ABM) that integrates research data management with experimental planning and coordination.

The ABM uses the knowledge graph-based The World Avatar ecosystem to represent laboratory equipment and capabilities, experiments, samples, process parameters, measurements, analytical results, and provenance within a shared semantic environment. The architecture follows a hierarchical control concept: process-critical and time-sensitive control tasks remain within dedicated local control systems, such as PID controllers, model predictive controllers, and device-specific automation. The ABM operates at a supervisory level, where it can support tasks such as design of experiments, experimental planning, selection of operating conditions, workflow coordination, and adaptation of subsequent experiments based on previously acquired results.

The initial implementation integrates data from a KLF bioreactor, a Cedex Bio HT biochemical analyzer, and an Agilent LC/Q-TOF system. This combines online process data such as pH, dissolved oxygen, temperature, agitation, gas flow, and dosing information with offline metabolite measurements and chromatographic and mass-spectrometric analytical results.

By capturing experimental context and provenance alongside the generated data, the ABM aims to make bioprocess data findable, interoperable, reusable, and AI-ready. The poster presents the system architecture and initial integration pathway toward transparent, human-supervised autonomous bioprocess experimentation.