软体机器人的仿生物理智能

Bio-inspired physical intelligence for soft robotics

生物可以在各种非结构化自然环境中生存,其身体中所蕴含的物理智能至关重要,涉及材料、结构和形态等要素.通过融合仿生物理智能,有望降低软体机器人的控制成本,提高机器人系统的响应速度和极端环境下的鲁棒性,以及使微型机器人更加智能化.本文阐述了自然界生物的材料、结构、形态学物理智能特征及其原理,介绍了软体机器人实现仿生物理智能的目的及相关的关键技术与方法,列举了软体机器人仿生物理智能的典型应用,最后展望了软体机器人仿生物理智能的未来发展及挑战.软体机器人仿生物理智能有望在高速动态作业、极端环境探索及微型机器人智能化等方面发挥独特的优势,相关研究将进一步促进生物、机器人、材料、化学和计算机学科之间的交叉.

Biological organisms can efficiently move, prey, mate, and grow in complex, unstructured natural environments. These remarkable behaviors are endowed not only by the computational intelligence in the biological brains but also by the physical intelligence encoded in their bodies. Physical intelligence can be defined as encoding sensing, actuation, control,memory, logic, computation, adaptative material and structure, self-learning, self-healing, and self-decision-making into physical bodies of the biological or robotic agents. Physical intelligence can also be generated during the interaction between the agent’s body and the environment over time. The previous focus of intelligent robots is primarily focused on the computational intelligence. As a new paradigm, physical intelligence is expected to boost intelligent robots in realworld applications.Soft robots commonly use soft stimuli-responsive materials and intelligent structures and maintain high stretchability and considerable deformation, therefore, have intrinsic environmental conformable physical property. Thus, soft robots are essential platforms for testing the hypothesis of physical intelligence in nature. This review mainly focused on three key physical intelligence elements encoded in the natural organisms’ bodies: Material, structure, and morphology. Through bioinspired design, smart soft materials, smart soft structures, and adaptable morphologies can be integrated into the robot’s body, thus introducing bio-inspired physical intelligence into the robots. By integrating bio-inspired physical intelligence,soft robots could reduce the cost of control, improve the response speed of the systems, enhance the robustness of robots in extreme environments, and embed intelligence into the micro-and small-scale robots. The research on bio-inspired physical intelligence may also promote the multi-disciplinary collaboration of biology, robotics, materials science,chemistry, computer science, etc.In this review, we first describe the characteristics and principles of physical intelligence in natural organisms’ material,structure, and morphology. Then we introduce the purposes and related key technologies and methods of realizing bioinspired physical intelligence of soft robots. Furthermore, we enumerate the typical applications of bio-inspired physical intelligence of soft robots. Finally, we highlight the trends and challenges of bio-inspired physical intelligence of soft robots in the future.The research on bio-inspired physical intelligence for soft robots is still in the primary stage. There are several critical questions and challenges to be addressed. First, researchers should investigate the basic principles of physical intelligence in biomechanics and then apply the basic principles to guide the development of soft robots. Second, intelligent materials need to meet the challenges of fully integrating sensing, actuation, memory, computation, and communication in soft robots. To this end, an interesting approach is to use stimulus-responsive materials as the basic building blocks to rationally design and integrate different functions into a single composite material. Third, smart structures, like mechanical metamaterials can be a promising research direction in the field of intelligent structures for soft robots. Aided by artificial intelligence and 3 D printing, mechanical metamaterials will boost the performance of soft robots in the foreseeable future.Fourth, the potential of adaptive morphology can be realized by endowing soft robots with the ability to self-optimize their morphologies in the environments. Finally, the integration of intelligent materials, structures, and morphologies in the soft robot system will significantly improve physical intelligence. We believe that bio-inspired physical intelligence for soft robotics will be an important research direction in the future.