Bio-inspired robot control systems for space application

用于空间应用的机器人系统需要在最少的人类交互作用下增强自主性。虽然机器人技术在认知层面上得到了很多强调，但最近更多的强调是在情境机器人技术上取得了不同程度的成功，但在扩展到更复杂的行为方面已经达到了极限。仿生学方法提供了另一个层次的研究-在生物动物中，形态结构与其控制结构共同进化。事实上，生物系统中的结构和机械配置是由控制系统利用的。此外，生物与其环境的相互作用是生物进化的驱动力（因为这定义了生存直到繁殖）。主体-环境相互作用的这一方面一直被忽视，但它是生物学的基础。造成这种忽视的部分原因是生物和工程中使用的材料不同——工程材料的设计是从刚度到变形，而生物材料的设计是从韧性到断裂。生物系统利用结构和机械顺应性作为与环境物理相互作用的动态缓冲。由于反馈控制回路的固有时间延迟会产生不稳定性，因此这种动态相互作用力将是混沌的。前馈机制在生物学中也被用来作为预测因素，但也有局限性。此外，生物系统利用多个控制回路，包括严密的控制回路，其中传感器和执行器的概念在探索活动中是模糊的。该研究项目旨在通过利用生物学的经验教训，进一步了解机器人与环境相互作用的这些方面，从而产生更强大的自主机器人空间系统。由于需要远程但强大和自适应的自主性，这将应用于操纵器、漫游器、样本采集设备和可部署的空间科学仪器。本研究计划将探讨几种可供利用的方法。

Robotic systems for space application require enhanced autonomy with minimal human interaction. Although much emphasis in robotics has been at the cognitive level, more recent emphasis has been in situated robotics with mixed success but which has reached a limit in scaling to more complex behaviours. A biomimetic approach provides another level of investigation - in biological animals, morphological structures have co-evolved with their control structures. Indeed, structural and mechanical configuration in biological systems is exploited by the control system. Furthermore, the interaction of the organism with its environment is the driver of biological evolution (as this defines survival until procreation). This aspect of agent-environment interaction has been much neglected yet it is fundamental in biology. Part of this neglect results from different materials used in biology and engineering - engineering materials are designed for stiffness to deformation while biological materials evolved with toughness to fracture. Structural and mechanical compliance is exploited by biological systems as a dynamic buffer to physical interaction with the environment. Such dynamic interaction forces would otherwise be chaotic as inherent time delays of feedback control loops generate instabilities. Feedforward mechanisms are also exploited in biology to act as predictors but these also have limitations. Furthermore, biological systems make use of multiple control loops including tight control loops in which the concept of sensor and actuator are blurred during exploratory activity. This program of research seeks to further understand these aspects of robot-environment interaction by exploiting lessons from biology, thereby generating more robust autonomous robotic space systems. This will be applied to manipulators, rovers, sample acquisition devices and deployable scientific instruments for space application due to the need for remote but robust and adaptive autonomy. There are several approaches to be exploited which shall be be explored in this research programme.