Characterisation, Modelling, and Development of the Reactive Peano Muscle

Abstract

Wearable robots for effctive healthcare and morphing aerofoils for efficient transport can benet enormously from actuators with muscle-like properties. These properties can make actuators inherently adaptive, productive, unobtrusive, and controllable. Motivated by the usefulness of muscle-like properties, this thesis aims to contribute knowledge to develop actuators that have a soft structure, biological motion and force capability, sheetlike form, and reactivity to their environment. Mettam's Peano muscle was found as a promising solution, although it has been barely investigated since its invention in the 1950s. This thesis lays the foundation to mature the Peano muscle as an enabler of new robotics technologies. Specically, it tackles the following problems with the Peano muscle: the lack of understanding of how its behaviour is inuenced by its geometry and material properties; the absence of validated tools to model its static, hysteretic, and dynamic behaviour; and its inability to react to its environment. First, an experimental study conrmed the Peano muscle is biologically capable and provided insight into how it can be optimised to meet the requirements of various existing and imagined applications. Second, two tools were developed to make it easier to characterise and control. The MECHanical Approximation Lumped Parameter (MECHALP) static model includes signicant eects not found in previous models, improving its usefulness and accuracy for Peano muscle design and control. The Multivariable Arbitrary Piecewise MOdel REgression (MAPMORE) algorithm generates accurate models of systems with nonlinear static, hysteretic, and damping behaviour such as the Peano muscle and dielectric elastomer sensors. Third, a reactive Peano muscle was built with embedded strain, pressure, and force sensors. These sensors allow it to monitor and control its interaction with uncertain environments. Implemented in a wrist orthosis, its sheetlike form and reactivity enabled the orthosis to be uniquely concealable and have the potential to assess its wearer's impairment; both features invaluable to rehabilitation devices. Each of these contributions is a step towards enabling a new generation of robotics made possible by mimicking the properties of muscles. Further steps should focus on improving the robustness and relevancy of the models and muscle fabrication methods for real-world applications.