Bio-inspired Smart Attachment and Adaptive Neuromechanical Control for Highly Efficient Locomotion and Adaptation to an Autonomous Climbing Robot

攀爬机器人能够在平坦或粗糙的地面、墙壁或天花板上移动，在公共和国家安全领域、狭小空间探测和城市景观服务中具有广泛的应用前景。腿式攀爬机器人对环境具有很强的适应性，能够穿越复杂的地形表面，具有良好的运动灵活性。目前，它们的控制和动作生成主要依赖于工程方法，其中在线学习和适应尚未完全整合。这限制了机器人应对不断变化的环境的整体性能。相比之下，壁虎、昆虫和蜘蛛可以在不同的底物坡度和粗糙度上高效地行走和攀爬。它们可以稳定地附着和爬行在自然环境中不稳定的垂直(如墙壁)或倒置(如天花板)表面上，主要依靠机械爪子互锁和脚垫粘合，以及两者的协同效应。此外，它们甚至可以迅速调整自己的行动，以适应环境的变化或未知的情况。多才多艺和适应能力是嵌入其感应器运动回路中的几个成分整合的结果。生物学研究表明，关键成分包括生物力学(如粘合垫和爪子)、感觉反馈和自适应神经控制。综合考虑这些因素，并深入研究和了解攀爬动物在稳态和非稳态下的附着机制，将会产生一种适应性强、通用性强、高效的仿生爬行机器人。基于此，本课题的目的是：1)研究和分析基于爪子机械联锁和脚垫粘连协同作用的爬行动物的粘连机理，以及爬行动物多层剥离策略的智能附着控制；2)进行生物实验，利用前人研制的动物全空间运动行为和反作用力测试系统，研究伸展姿势动物的运动行为及其反作用力，建立四足动物的动力学模型；3)开发具有感觉反馈和在线适应的自适应神经机械控制，以实现稳定、高效的爬行策略；4)开发具有粘合衬垫和主动爪子的仿生脚。开发的仿生脚和自适应神经机械控制将在自主攀爬机器人上实现，以实现高效的移动和适应变化的倾斜角度和光滑和不同粗糙度的衬底的墙壁。

Climbing robots are able to move on smooth or rough ground, walls or ceilings, and have wide application prospects in public and national security areas, narrow space detection, and cityscape services. Legged climbing robots, having a strong adaptability to the environment, can cross the complex terrain surface and have good movement flexibility. At present, their control and movement generation mainly rely on engineering approaches where online learning and adaptation have not been fully integrated. This limits the overall performance of the robot to deal with a changing environment. In contrast, geckos, insects, and spiders can efficiently walk and climb on different substrate slope and roughness. They can be stably attached and crawl on unsteady vertical (such as walls) or inverted (such as ceilings) surfaces in the natural environment, relying primarily on mechanical claw interlocking and foot pad adhesion as well as on the synergistic effect of both. Furthermore, they can even quickly adapt their locomotion to a change of their environment or an unknown situation. The versatile and adaptive abilities are the result of an integration of several ingredients embedded in their sensorimotor loop. Biological studies reveal that the key ingredients include biomechanics (e.g., adhesive pads and claws), sensory feedback, and adaptive neural control. Taking these ingredients into account for robot development together with in-depth studying and understanding attachment mechanisms of climbing animals in steady and unsteady states will lead to an adaptive, versatile, and efficient bio-inspired climbing robot. According to this, the purpose of the project is to 1) study and analyze the adhesive mechanisms of climbing animals based on the synergic action of claw mechanical interlocking and pad adhesion, and the multiple-peeling strategy of climbing animals for smart attachment control, 2) perform biological experiments, investigate locomotion behavior and its reaction forces of sprawled posture animals by using the former developed animal full-space motion behavior and reaction force testing system, and establish dynamic model of quadruped, 3) develop adaptive neuromechanical control with sensory feedback and online adaptation for stable and efficient climbing strategies, and 4) develop a bionic foot with adhesive pads and active claws. The developed bionic foot and adaptive neuromechanical control will be implemented on our autonomous climbing robot to achieve highly efficient locomotion and adaptation on changing incline angles and walls with smooth and different roughness substrates.