Compliant, Large-Strain, and Self-Sensing Twisted String Actuators with Applications to Soft Robots
Loading...
Authors
Bombara, David Paul
Issue Date
2022
Type
Thesis
Language
Keywords
Alternative Title
Abstract
The twisted string actuator (TSA) is a rotary-to-linear transmission system that has been implemented in robots for high force output and efficiency. The basic components of a TSA are a motor, strings, and a load (to keep the strings in tension). The twisting of the strings shortens their length to generate linear contraction. Due to their high force output, energy efficiency, and compact form factor, TSAs hold the potential to improve the performance of soft robots. Currently, it is challenging to realize high-performance soft robots because many existing soft or compliant actuators exhibit limitations such as fabrication complexity, high power consumption, slow actuation, or low force generation. The applications of TSAs in soft robots have hitherto been limited, mainly for two reasons. Firstly, the conventional strings of TSAs are stiff and strong, but not compliant. Secondly, precise control of TSAs predominantly relies on external position or force sensors. For these reasons, TSA-driven robots are often rigid or bulky.To make TSAs more suitable for actuating soft robots, compliant, large-strain, and self-sensing TSAs are developed and applied to various soft robots in this work. The design was realized by replacing conventional inelastic strings with compliant, thermally-activated, and conductive supercoiled polymer (SCP) strings. Self-sensing was realized by correlating the electrical resistance of the strings with their length. Large strains are realized by heating the strings in addition to twisting them. The quasi-static actuation and self-sensing properties are accurately captured by Preisach hysteresis operators. Next, a data-driven mathematical model was proposed and experimentally validated to capture the transient decay, creep, and hysteretic effects in the electrical resistance. This model was then used to predict the length of the TSA, given its resistance. Furthermore, three TSA-driven soft robots were designed and fabricated: a three-fingered gripper, a soft manipulator, and an anthropomorphic gripper. For the three-fingered gripper, its fingers were compliant and designed to exploit the Fin Ray Effect for improved grasping. The soft manipulator was driven by three TSAs that allowed it to bend with arbitrary magnitude and direction. A physics-based modeling strategy was developed to predict this multi-degree-of-freedom motion. The proposed modeling approaches were experimentally verified to be effective. For example, the proposed model predicted bending angle and bending velocity with mean errors of 1.58 degrees (2.63%) and 0.405 degrees/sec (4.31%), respectively. The anthropomorphic gripper contained 11 TSAs; two TSAs were embedded in each of the four fingers and three TSAs were embedded in the thumb. Furthermore, the anthropomorphic gripper achieved tunable stiffness and a wide range of grasps.