Soft robotics has opened a unique path to flexibility and environmental adaptability, learning from nature and reproducing biological behaviors. Nature implies answers for how to apply robots to real life. To find out how we learn from creatures to design and apply soft robots, in this Review, we propose a classification method to summarize soft robots based on different functions of biological systems: self-growing, self-healing, self-responsive, and self-circulatory. The bio-function based classification logic is presented to explain why we learn from creatures. State-of-art technologies, characteristics, pros, cons, challenges, and potential applications of these categories are analyzed to illustrate what we learned from creatures. By intersecting these categories, the existing and potential bio-inspired applications are overviewed and outlooked to finally find the answer, that is, how we learn from creatures.

翻译：软体机器人开辟了一条通往柔顺性与环境适应性的独特路径，通过向自然学习并复现生物行为。自然界为机器人如何应用于现实生活提供了答案。为探究我们如何从生物身上学习以设计并应用软体机器人，本文提出了一种基于生物系统不同功能的分类方法，将软体机器人归纳为四类：自生长型、自修复型、自响应型与自循环型。文章阐述了基于生物功能的分类逻辑，以解释为何向生物学习；并分析了各类别的前沿技术、特性、优缺点、挑战及潜在应用，以说明我们从生物中学到了什么。通过交叉融合这些类别，本文对现有及潜在的仿生应用进行了综述与展望，最终回答核心问题：我们如何向生物学习。