2022 article

Control of a dynamic load emulator for hardware-in-the-loop testing of fluidic artificial muscle bundles

BIOINSPIRATION, BIOMIMETICS, AND BIOREPLICATION XII, Vol. 12041.

co-author countries: United States of America 🇺🇸
author keywords: Fluidic artificial muscles; hardware-in-the-loop; load emulator; cyber-physical system; bioinspired robotics
Source: Web Of Science
Added: August 29, 2022

Fluidic artificial muscles (FAMs) have emerged as a viable and popular robotic actuation technique due to their low cost, compliant nature, and high force-to-weight-ratio. In recent years, the concept of variable recruitment has emerged as a way to improve the efficiency of conventional hydraulic robotic systems. In variable recruitment, groups of FAMs are bundled together and divided into individual motor units. Each motor unit can be activated independently, which is similar to the sequential activation pattern observed in mammalian muscle. Previous researchers have performed quasistatic characterizations of variable recruitment bundles and some simple dynamic analyses and experiments with a simple 1- DOF robot arm. We have developed a linear hydraulic characterization testing platform that will allow for the testing of different types of variable recruitment bundle configurations under different loading conditions. The platform consists of a hydraulic drive cylinder that acts as a cyber-physical hardware-in-the-loop dynamic loading emulator and interfaces with the variable recruitment bundle. The desired inertial, damping and stiffness properties of the emulator can be prescribed and achieved through an admittance controller. In this paper, we test the ability of this admittance controller to emulate different inertial, stiffness, and damping properties in simulation and demonstrate that it can be used in hardware through a proof-of-concept experiment. The primary goal of this work is to develop a unique testing setup that will allow for the testing of different FAM configurations, controllers, or subsystems and their responses to different dynamic loads before they are implemented on more complex robotic systems.