Post by Carlos Ivan Cardona G

Decoding the geometric patterns in the cross-section of bamboo fibre has long been an engineering paradigm of Nature. Our new approach helps to reveal the design principles behind its optimized topology. New paper out in Composites Part A: "Integrating deep learning and geometry-based microstructural analysis of technical bamboo fibres for bio-composite engineering." Quick summary of our work: Digitalizing Fibre’s hierarchies - We benchmarked five deep learning segmentation models on optical micrographs to turn a complex natural cross-section into phase-resolved geometric data. Expansion and contraction gradients exhibit the cellular network to operate as clusters for intermittent load sharing, thanks to the quality of segmentations achieved. Reading the design principles - With this digital representation, a pattern identification workflow exposed bamboo's functionally graded architecture: solid material along the low-stress neutral axis, larger lumens out in the tension zone. Nature placing material where the bending load demands it, now quantified rather than described. Why it matters: capturing these principles in measurable, phase-resolved form is the groundwork for genuinely biomimetic composite design and provides the geometric inputs for modelling and manufacturing. This investigation was carried out in the frame of the SUSPOCO project (Sustainable Polymer Composites) at the Luxembourg Institute of Science and Technology. Check our Paper 📄 https://lnkd.in/dz7JJVuS #Biomimetics #CompositeMaterials #MachineVision #DeepLearning #MaterialsScience #SustainableMaterials