TU Faculty Develop Robotic Actuators that Mimic Muscles
Most robotic actuators are discrete units, think motors or hydraulic actuators, that move a single part on a robotic system. If these actuators fail, then that part of the robot ceases to operate and the consequences can range from reduced performance to complete failure. Dr. Joshua Schultz is working to build fundamentally new robotic actuators that will prevent these catastrophic failures and improve the reliability of robotic systems. His approach is to develop intelligent actuation systems that are made from modular components, including force generating units, computational units, and sensing units. Unlike the monolithic actuators we are familiar with, failure of any given module does not lead to failure of the entire actuator, similarly to how failure of a single muscle fiber does mean the muscle fails. These actuators include embedded computational modules that allows the system to operate with a decentralized control architecture, where the embedded computation makes the actuator appear to a high-level controller as a "black box" with simple system dynamics, even though its internal behavior is complicated, much like what happens with biological systems.