About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
AI-Enabled Materials Processing: Integrating Accelerated Experimental Workflows and Processing-Aware Machine Learning
|
| Presentation Title |
Closed-Loop Bayesian Optimization of Multi-Material 3D-Printed Tensegrity-Inspired Energy Absorbers |
| Author(s) |
Marcus Earl Madsen, Audrey Kate Christiansen, Jinkwan Han, Jeffrey R. Hill, Sterling G. Baird |
| On-Site Speaker (Planned) |
Marcus Earl Madsen |
| Abstract Scope |
Tensegrity-inspired architectures—rigid struts in a continuous flexible network—offer tunable nonlinear force–displacement responses and high energy absorption per mass, but their design space (strut geometry, connectivity topology, unit-cell tiling) is too large for trial and error. We present a closed-loop, experiment-driven campaign that co-prints rigid PLA struts and flexible TPU tension elements by multi-material fused deposition modeling and uses Bayesian optimization to jointly tune architecture and FDM process parameters—nozzle temperature, print speed, layer height—from physical measurements, not calibrated simulation. Moving toward a self-driving lab, we couple printer and orchestration software via direct Python integration, track data and metadata provenance in the cloud, and keep a human in the loop, raising autonomy. A Gaussian-process surrogate over measured peak force, specific energy absorption, and compaction efficiency proposes batches via noisy multi-objective acquisition, with print failures as a feasibility constraint. We report how Pareto trade-offs evolve, distilling lessons for accelerated, processing-aware machine learning. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Additive Manufacturing, Machine Learning, |