材料內部所具有的微觀結構直接影響了材料的巨觀行為，而生物體內的結構多數也藉此概念演化出優異的機械性質，例如人類骨骼由不同尺度下觀察，小至分子間排列，大至肉眼可見之孔洞，經過排列之後形成了兼具高強度及輕量化的結構。透過觀察不同尺度的生物結構材料，有許多值得效仿的生物結構，如蜂巢結構(Honey comb)、磚泥結構(Brick-and-mortar)、多孔螺旋結構(Gyroid)以及旋轉夾板結構(Bouligand)，這些微結構背後的力學機制也值得深入探討。
本研究的目的在於建立一個快速且準確的設計平台，針對鞋中底結構設計的問題，進行孔洞材料分布的設計，期望讓鞋中底具備高吸能、低反力且輕盈等多樣特性。本研究主要以三重週期性最小面積結構(Triply periodic minimal surface, TPMS)為主，TPMS結構由連續光滑的表面組成，具有大的表面積及內部通道，屬於孔洞材料，可以很好的避免應力集中的狀況發生，其中又挑選了Gyroid、Sch-D、Sch-P 三種型態作為比較，三種型態都可由三角函數組成的方程式表現，其中由於Gyroid的壓縮性能較為突出，因此本研究主要目標為利用Gyroid極高的強度重量比，選擇兩個不同強度的Gyroid單元做為軟、硬材料進行布置，藉由結構本身具有很好的韌性表現來設計出一款同時具有輕量化、高吸能、高強度的高性能鞋墊。
本研究使用光固化(Stereolithography)3D列印機Form 3製造試體進行實驗，將不同條件的壓縮實驗完成後進行後處理可以得到標準曲線(Master Curve)，藉此則可以預測特定條件如相對密度(Relative density)、型態(Type)或是平台應力(Plateau stress)等等試體的結構強度，找到不同密度的孔洞材料作為軟材以及硬材，各自賦予給一個元素，稱為簡化模型(Reduced model, RM)，每種RM的排列會作為輸入狀態，排列完使得模型具有相應的韌性值作為獎勵，藉由強化學習(Reinforcement learning, RL)在指定的設計空間內如對稱條件下的4*方格或是8*8方格，排列出壓縮性值最符合設計目標的組合，至此完成設計平台。最後針對需求功能的鞋中底(Midsole)進行設計進而供以製造。經由對稱條件下8*8方格之設計結果顯示，通過等比例的結構材料設計，以全硬材料60%的材料使用量，保留47%的應變能卻降低43%反力以及86%的中心應力，成功找到吸能且舒適的抗壓縮鞋墊設計，並且實際透過3D列印製造出來進行實驗驗證。
本研究利用人工智慧模型進行仿生材料之性能設計，開發技術與設計準則可以為先進材料產業帶來重要競爭力，複合材料的設計以及製造困難的鴻溝將被跨越；未來也能以模組化的方式，針對各種產業的需求進行設計，例如汽車業的輕量化車體，強化學術以及產業的交流，對於產業升級提供重要協助。

In recent years, the footwear industry in Taiwan has transformed from a manufacturing field into a research and design field. The development of shoes is to strengthen specific functions for various purposes. Bio-inspired microstructural materials are suitable for designing shoe midsoles. Being one of bio-inspired microstructural materials, gyroid has many advantages, such as lightweight, high-energy absorption, and high strength, so gyroid is the primary goal material in this study. According to the characteristics of the staggered soft and hard materials of biological materials, this research uses the distribution of soft and hard materials to design the impact-resistant component. Furthermore, we use reinforcement learning to optimize the solution which can meet our targets. Through the rewards obtained by each action in the environment, the model will learn how to obtain the highest reward in the current environment and find the best arrangement in the end. In this study, reinforcement learning is used to design the bio-inspired microstructural components.
As a result, we successfully designed a component with 47% strain energy remaining, 43% reaction force, and 86% stress reduced. RL is a new method to solve complex situations like multi-parameter problems more precisely.

1. Yu, C.-H., Qin, Z. &Buehler, M. J. Artificial intelligence design algorithm for nanocomposites optimized for shear crack resistance. Nano Futur. (2019) doi:10.1088/2399-1984/ab36f0.
2. Yu, C. H., Qin, Z. &Buehler, M. J. Artificial intelligence design algorithm for nanocomposites optimized for shear crack resistance. Nano Futur. 3, (2019).
3. Nigg, B. M., Cigoja, S. &Nigg, S. R. Effects of running shoe construction on performance in long distance running. Footwear Sci. 12, 133–138 (2020).
4. NIGG &BM. Experimental techniques used in running shoe research. Biomech. Run. Shoes 27–61 (1986).
5. Nigg, B. M. &Liu, W. The effect of muscle stiffness and damping on simulated impact force peaks during running. J. Biomech. 32, 849–856 (1999).
6. Nigg, B. M. Biomechanics, Load Analysis and Sports Injuries in the Lower Extremities. Sports Medicine: An International Journal of Applied Medicine and Science in Sport and Exercise vol. 2 367–379 (1985).
7. Bahlsen: Biomechanics of running shoes - Google Scholar. https://scholar.google.com/scholar?cluster=8114382741131920189&hl=en&oi=scholarr.
8. Dura, J.V., Garcia, A. C. &Solaz, J. Testing shock absorbing materials: the application of viscoelastic linear model. Sport. Eng. 5, 9–14 (2002).
9. Bouix, R., Viot, P. &Lataillade, J. L. Polypropylene foam behaviour under dynamic loadings: Strain rate, density and microstructure effects. Int. J. Impact Eng. 36, 329–342 (2009).
10. Yu, C. H., Qin, Z., Martin-Martinez, F. J. &Buehler, M. J. A Self-Consistent Sonification Method to Translate Amino Acid Sequences into Musical Compositions and Application in Protein Design Using Artificial Intelligence. ACS Nano (2019) doi:10.1021/acsnano.9b02180.
11. Maiti, S. K., Gibson, L. J. &Ashby, M. F. Deformation and energy absorption diagrams for cellular solids. Acta Metall. 32, 1963–1975 (1984).
12. Bhattacharya, A., Calmidi, V.V. &Mahajan, R. L. Thermophysical properties of high porosity metal foams. Int. J. Heat Mass Transf. 45, 1017–1031 (2002).
13. Gómez, S., Vlad, M. D., López, J. &Fernández, E. Design and properties of 3D scaffolds for bone tissue engineering. Acta Biomater. 42, 341–350 (2016).
14. Kapfer, S. C., Hyde, S. T., Mecke, K., Arns, C. H. &Schröder-Turk, G. E. Minimal surface scaffold designs for tissue engineering. Biomaterials 32, 6875–6882 (2011).
15. Zhou, H., Zhao, M., Ma, Z., Zhang, D. Z. &Fu, G. Sheet and network based functionally graded lattice structures manufactured by selective laser melting: Design, mechanical properties, and simulation. Int. J. Mech. Sci. 175, (2020).
16. Qin, Z., Jung, G. S., Kang, M. J. &Buehler, M. J. The mechanics and design of a lightweight three-dimensional graphene assembly. Sci. Adv. 3, 1–9 (2017).
17. Pais, A., Alves, J. L. &Belinha, J. Design of functionally graded gyroid foams using optimization algorithms and the finite element method: A simple approach on prototyping from simulation to manufacturing. Int. J. Adv. Manuf. Technol. 725–739 (2021) doi:10.1007/s00170-020-06542-w.
18. Peng, C. &Tran, P. Bioinspired functionally graded gyroid sandwich panel subjected to impulsive loadings. Compos. Part B Eng. 188, 107773 (2020).
19. Deng, F., Liu, L., Li, Z. &Liu, J. 3D printed Ti6Al4V bone scaffolds with different pore structure effects on bone ingrowth. J. Biol. Eng. 15, 1–13 (2021).
20. Jian, Z. et al. 3D bioprinting of a biomimetic meniscal scaffold for application in tissue engineering. Bioact. Mater. 6, 1711–1726 (2021).
21. Huo, C. et al. Hierarchically porous alumina catalyst carrier with biomimetic vein structure prepared by direct ink writing. J. Eur. Ceram. Soc. 41, 4231–4241 (2021).
22. Kuntjoro, W., Roslan, I. S., Muta, A. B. A., Anwar, N. I. A. K. &Nasir, R. E. M. a Study of 3D Printed Box Structure. 53, 1–4 (2020).
23. Tiwari, N., Gagare, S. W. &Shaikh, A. A. Shape recovery analysis of the additive manufactured 3D smart surfaces through reverse engineering. Prog. Addit. Manuf. 6, 281–295 (2021).
24. Lu, J., Dong, P., Zhao, Y., Zhao, Y. &Zeng, Y. 3D printing of TPMS structural ZnO ceramics with good mechanical properties. Ceram. Int. 47, 12897–12905 (2021).
25. Watkins, C. J. C. H. &Dayan, P. Q-Learning. vol. 8 (1992).
26. Mnih, V. et al. Playing Atari with Deep Reinforcement Learning. (2013).
27. Im Breisgau, F. et al. Robotics: Science and Systems ssss OIL: Observational Imitation Learning. https://sites.google.com/kaust.edu.sa/oil/ (2019) doi:10.0/Linux-x86_64.
28. Shin, S. Y., Kang, Y. W. &Kim, Y. G. Obstacle avoidance drone by deep reinforcement learning and its racing with human pilot. Appl. Sci. 9, 5571 (2019).
29. Brooks, R. A. A Robust Layered Control System for a Mobile Robot. (1985).
30. Balch, T. &Arkin, R. C. Behavior-based formation control for multirobot teams. IEEE Trans. Robot. Autom. 14, 926–939 (1998).
31. Bonasso, R. P., Kortenkamp, D., Miller, D. P. &Slack, M. Experiences with an architecture for intelligent, reactive agents. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) vol. 1037 187–202 (Springer Verlag, 1996).
32. Fuchs, A., Heider, Y., Wang, K., Sun, W. C. &Kaliske, M. DNN2: A hyper-parameter reinforcement learning game for self-design of neural network based elasto-plastic constitutive descriptions. Comput. Struct. 249, 106505 (2021).
33. Whitelam, S. &Tamblyn, I. Learning to grow: Control of material self-assembly using evolutionary reinforcement learning. Phys. Rev. E 101, (2020).
34. Luo, C., Ning, S., Liu, Z. &Zhuang, Z. Interactive inverse design of layered phononic crystals based on reinforcement learning. Extrem. Mech. Lett. 36, (2020).
35. Timercan, A., Sheremetyev, V. &Brailovski, V. Mechanical properties and fluid permeability of gyroid and diamond lattice structures for intervertebral devices: Functional requirements and comparative analysis. Sci. Technol. Adv. Mater. (2021) doi:10.1080/14686996.2021.1907222.
36. Maskery, I. &Ashcroft, I. A. The deformation and elastic anisotropy of a new gyroid-based honeycomb made by laser sintering. Addit. Manuf. 36, (2020).
37. Gonzalez, O. Exact energy and momentum conserving algorithms for general models in nonlinear elasticity. Comput. Methods Appl. Mech. Eng. 190, 1763–1783 (2000).