BioMeld Concludes: Establishing the Foundations for Biohybrid Machines

Machines that are partly alive — this is the bold idea behind BioMeld. Over the past three and a half years, a team of seven European research institutions set out to build the scientific and engineering foundations for a new class of devices called biohybrid machines (BHMs): miniature robots that integrate living muscle cells with engineered materials to move, sense, and respond to their environment.

Rather than developing a single device in isolation, BioMeld’s ambition was broader: to create a reusable design and manufacturing framework that could be applied to a wide range of future biohybrid devices. As proof of concept, the team built a biohybrid catheter: a tiny, flexible tube actuated by real skeletal muscle tissue, capable of bending on demand when electrically stimulated. 

What was achieved: 

By the end of the project, the consortium had produced a working biohybrid catheter, integrating bioengineered muscle rings that generate bending angles when stimulated. The living tissue maintains cell viability thanks to a biochamber and can be cryopreserved and revived after months in storage. These are critical steps toward practical shelf life. A custom-designed flexible electronic platform provides both electrical stimulation and real-time sensing. On the computational side, the project delivered an AI-assisted design pipeline connecting user requirements through to physical morphology generation, machine-learning surrogates that replicate complex simulations, and a validated digital twin that can predict muscle force responses under electrical stimulation with high fidelity.

An important component of this pipeline is a Retrieval-Augmented Generation (RAG) system that allows designers to describe a
new biohybrid device in plain language. The system searches a library of validated design templates, identifies the closest matching
precedents, and, with the help of a large language model, automatically translates the natural language description into a formal, machine-readable set of design constraints. The system is confidence-aware: high-confidence matches proceed automatically,
while ambiguous or novel queries prompt the human expert for clarification before any automated steps continue. This makes complex BHM design accessible without sacrificing rigor. All of this was formalized in a Human-in-the-Loop Design Framework that captures the lessons learned across design, fabrication, and validation — creating a reusable knowledge base to guide future biohybrid machine development beyond the project’s lifetime.