Collaborative Research: Collaborative Sequential Experimentation for Bio-Additive Manufacturing Systems

This award supports research to make bio-additive manufacturing more efficient, reliable, and collaborative. Bio-additive manufacturing uses three-dimensional printing and related processing steps to create biologically relevant products, including resorbable tissue scaffolds that may help guide the repair of damaged nerves. Designing these products is difficult because material formulation, part geometry, printing conditions, and post-processing all influence whether a scaffold achieves the needed shape, strength, and degradation behavior. Each trial can require substantial time, specialized equipment, and costly testing, so a single printer, manufacturer, or laboratory may struggle to identify high-quality designs quickly. This research addresses that challenge by developing methods that allow multiple printers, laboratories, and organizations to learn from one another during experimentation while protecting private data and intellectual property. The work serves the national interest by advancing the science of biomanufacturing, supporting the efficient development of regenerative medicine products, and helping United States manufacturers improve quality with fewer resources. The research also supports education through publicly available software and data, demonstrations for industry and academic audiences and new course modules on collaborative experimentation and bio-additive manufacturing. The technical goal is to establish and validate a framework for collaborative sequential experimentation in bio-additive manufacturing systems. The framework addresses four connected tasks. First, consensus-based optimal design enables each participating printer or laboratory to learn a local surrogate model and identify a useful candidate experiment, then reach a consensus that steers every participant toward experiments closer to the optimum by borrowing strength from others, without sharing raw data or response measurements. Second, conditioned surrogate models extend collaboration to asynchronous settings, letting a participant choose experiments by drawing on historical data, expert knowledge, or other participants while filtering information that conflicts with local evidence, thereby reducing negative transfer under process heterogeneity. Third, geometry and toolpath information is incorporated through point-cloud representations, boundary features, and physics-informed learning to predict performance for unseen scaffold geometries and connect those predictions with sequential decision-making. Fourth, a cyber-enabled testbed links multiple extrusion-based printers through a cloud-based orchestrator, fabricating nerve guidance scaffolds from printable thermoplastic materials and comparing collaborative and non-collaborative experimentation using dimensional accuracy, yield stress, Young’s modulus, time, and cost. The resulting methods, testbed, data, and software are expected to provide a general approach for collaborative optimal design in manufacturing settings where experiments are costly and processes are heterogeneous, with privacy protected whenever needed. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.