关键词: 3D voxel-based simulator biohybrid machines evolutionary algorithms machine learning optimization

来  源:   DOI:10.3389/frobt.2024.1337722   PDF(Pubmed)

Abstract:
Biohybrid machines (BHMs) are an amalgam of actuators composed of living cells with synthetic materials. They are engineered in order to improve autonomy, adaptability and energy efficiency beyond what conventional robots can offer. However, designing these machines is no trivial task for humans, provided the field\'s short history and, thus, the limited experience and expertise on designing and controlling similar entities, such as soft robots. To unveil the advantages of BHMs, we propose to overcome the hindrances of their design process by developing a modular modeling and simulation framework for the digital design of BHMs that incorporates Artificial Intelligence powered algorithms. Here, we present the initial workings of the first module in an exemplar framework, namely, an evolutionary morphology generator. As proof-of-principle for this project, we use the scenario of developing a biohybrid catheter as a medical device capable of arriving to hard-to-reach regions of the human body to release drugs. We study the automatically generated morphology of actuators that will enable the functionality of that catheter. The primary results presented here enforced the update of the methodology used, in order to better depict the problem under study, while also provided insights for the future versions of the software module.
摘要:
生物混合机器(BHM)是由活细胞与合成材料组成的致动器的混合物。它们是为了提高自主性而设计的,适应性和能源效率超出了传统机器人的能力。然而,设计这些机器对人类来说不是一件小事,提供了字段的简短历史记录,因此,设计和控制类似实体的经验和专业知识有限,比如软机器人。为了揭示BHM的优势,我们建议通过为BHM的数字设计开发模块化建模和仿真框架来克服其设计过程的障碍,该框架包含人工智能驱动的算法。这里,我们在一个示例框架中介绍了第一个模块的初始工作原理,即,进化形态发生器。作为这个项目的原理证明,我们使用开发生物混合导管的方案作为能够到达人体难以到达的区域释放药物的医疗设备.我们研究了自动生成的执行器形态,以实现该导管的功能。这里介绍的主要结果强制更新了所用方法,为了更好地描述研究中的问题,同时还为软件模块的未来版本提供了见解。
公众号