Mechanics of Soft-Rigid Interactions: From Wearable Robots to Stingray Skeletons

Abstract

Across robots and biological organisms, both rigid and soft materials are frequently encountered, providing either strength/precision or adaptability/safety, respectively. In nature, materials of different stiffnesses are frequently coupled together, producing highly capable "hybrid-stiffness" structures that leverage the benefits of each material in tandem. In robotics, however, most systems are either fully rigid or fully soft, leaving a vast, largely unexplored regime of hybrid-stiffness mechanisms in the middle. This dissertation seeks to study the rigid/soft interactions that are critical to this regime, looking at two particular model systems: a soft wearable robot to assist the shoulder joint (which is heavily influenced by interactions with the rigid-body kinematics of the human shoulder complex) and batoid fish pectoral fins (which comprise an intricate arrangement of rigid skeletal elements embedded within soft muscular tissue). In the first part of the dissertation, an experimental procedure was developed to characterize the mechanical performance of the shoulder assistance robot in-situ on the human body. Unlike prior characterization techniques that were limited to benchtop settings, this procedure captures the rich rigid-soft interactions between the device and the human body. Significant differences in behavior are noted between these boundary conditions, motivating the importance of studying wearable robots in their actual on-body setting. The results from this investigation then motivated the second portion of the dissertation, investigating and predicting the mechanical hysteresis exhibited by these devices on the body. The proposed modeling framework can reliably predict on-body hysteresis using minimal data collected on each individual wearer, and was demonstrated to outperform existing models from the literature. In the final portion of the dissertation, the rigid-soft interactions of batoid fish pectoral fin skeletons were studied. The skeletal geometry was parameterized, and variations in these parameters across the fins of three model species were studied. Finite element modeling was then employed to map these geometric parameters onto effective material properties. It was found that the skeleton served primarily to stiffen the in-plane membrane response of the fin while softening the out-of-plane bending response. Moreover, the spatial variations in skeletal geometry led to spatial modulation of the bending properties, with increased compliance observed in regions that exhibit exaggerated bending motions. These projects involved a combination of experimental characterization, data-driven reduced-order modeling, and physics-informed numerical simulation. Each of these efforts contributes toward the long-term goal of designing highly capable hybrid-stiffness robots that exceed the capabilities of systems constructed with fully rigid or fully soft materials, alone. By studying the rigid-soft interactions of these two models model systems, this dissertation represents an initial step toward this eventual vision.