Novel sensory-motor coupling in soft robotic systems

在机器人系统中使用柔软、灵活的技术是从传统的机器人设计基础上的根本转变。顺应性，可变形的材料允许许多生物学原理被模仿，这种新兴的软方法可以创建机器人，这些机器人可以鲁棒地处理不确定性，更安全地与人类互动，并顺应地在非结构化环境中移动。可以利用软机器人系统中的固有顺应性来将计算从中央高级处理器转移。这种方法从自然界中获得灵感，在自然界中，章鱼等生物通过将智能体现在触角中来减少对大脑的计算需求。生物物种通过精心调整的形态参数和肌肉和感觉元素的高度整合来实现这种先进的行为。这个博士将研究软机器人系统的人工肌肉如何与局部感觉受体整合，以通过本体感受和环境相互作用实现低级计算。受自然界的启发，感觉神经元通常在脊髓而不是大脑中形成突触，将开发新技术以实现动态低水平反应。通过考虑如何利用不同的感觉模式来创建致动的突触通路，这种方法将得到扩展。在软机器人中，将计算分散到传感器-电机元件将导致多自由度系统的动态响应性的进步，并随之减少中央处理器的计算负荷。分散软机器人中人工肌肉的控制也将提高它们对不确定性的鲁棒性，因为局部感觉-运动相互作用可能会产生紧急行为。据设想，该博士将为下一代软机器人系统做出贡献，该系统可以安全有效地产生动态，多功能行为。

The use of soft, flexible technologies in robotic systems is a fundamental shift from the traditional basis of how a robot should be designed. Compliant, deformable materials allow many biological principles to be emulated and this emerging soft approach can create robots that robustly deal with uncertainty, interact more safely with humans and compliantly move through unstructured environments. The inherent compliance in soft robotic systems can be exploited to devolve computation away from the central high-level processor. This approach takes inspiration from nature, where organisms such as octopuses reduce computational demands on their brain by embodying intelligence into their tentacles. Biological species achieve such advanced behaviour through carefully tuned morphological parameters and a high integration of muscle and sensory elements.This PhD will investigate how a soft robotic system's artificial muscles can be integrated with local sensory receptors to enable low-level computation through proprioceptive and environmental interaction. Inspired by nature, where sensory neurons often synapse in the spinal cord rather than the brain, novel technologies will be developed to enable dynamic low-level responses. This approach will be expanded by considering how different sensory modes can be exploited to create synaptic pathways to actuation. Devolving computation to distributed sensory-motor elements in soft robots will lead to advances in the dynamic responsiveness of multiple degree-of-freedom systems and concomitant reductions in the computational loading of the central processor. Decentralising the control of the artificial muscles in soft robots will also improve their robustness to uncertainty as emergent behaviours can arise from localised sensory-motor interactions. It is envisaged that this PhD will make contributions towards the next generation of soft robotic system that can generate dynamic, multi-functional behaviours in a safe and efficient manner.