Morphology informed approaches to soft robotic resilience

Soft robotics has become a thriving field of research, encompassing control theory, material science, robotics, and fluid dynamics. The aim is to create robots with compliant bodies and novel capabilities, such as morphologic change, allowing for better adaptability in diverse, unknown and challenging environments.

Several problems still prevent the widespread adoption of soft robots in the real world, though, from complex manufacturing, to the challenges in modelling and control. Recently, concerns have also emerged regarding their resilience and durability. At the material level, soft robots are prone to wear, degradation, and failures due to environmental factors, more so than traditional rigid robots made from metal alloys. Their low hardware redundancy and tight integration of form and function exacerbate these issues. Self-healing materials have been investigated as a solution, but challenges still stand. Furthermore, a solution that can bridge the gap between controller driven approaches and material-based ones is yet to be demonstrated.

This work addresses some of the challenges of soft robotic resilience and presents two approaches to achieve it, inspired by the concepts of morphological computation and embodied intelligence. First, a software-based framework is applied to a flexible extendable robotic implant and the compliant morphology of the robot is used to gain information and achieve fault detection and identification. The second approach, intended for pneumatic soft robots, uses novel fully soft valves to achieve distributed embodied resilience. These valves autonomously detect and isolate bursts and offer protection against overpressurisation, providing a novel, pre-emptive type of resilience, therefore reducing hardware and computational complexity. As a whole, this work represents the embryo of a new form of resilience for soft robots, a morphology-informed one based on embodied computation. It enhances the adaptability of soft-robots and addresses critical limitations of self-healing solutions, therefore representing an important stepping-stone towards more resilient soft robots.