Modeling, Self-Sensing, and Control of Novel Coiled String Actuators with Application to Compliant Manipulation

This award supports research that contributes new knowledge related to a compliant actuation mechanism, thereby promoting both the progress of science and advancing the national health and prosperity. Compliant actuators are constructed from soft or compliant materials and can generate shape changes under external physical or chemical stimuli. Compliant actuators technology is crucial to enable dexterous and compact compliant robots, which have a wide application in areas that demand safe interaction with humans in highly unstructured environments.

Many compliant actuators, including different types of artificial muscles, have been studied. However, there has not yet been a compliant actuator that exhibits properties or performance metrics comparable to those of biological muscles. This project will address this critical gap by providing fundamental knowledge on the modeling, sensing, and control of a promising compliant actuator, namely the coiled string actuator, which overcomes common limitations of existing compliant actuators.

Results from this research will benefit the U.S. economy and society for applications in manufacturing, agriculture, healthcare, and wearable devices. This research involves several disciplines including material science, robotics, control theory, and machine learning. The multi-disciplinary approach will help broaden participation of underrepresented and underserved groups in research and positively impact engineering education.

Besides curriculum enrichment, this project will provide research training opportunities for students and a number of outreach activities for students and the public.

This research aims to make fundamental contributions to enable the coiled string actuator to produce predicable and desirable outputs, and further realize compliant robotic manipulators with superior compactness and dexterity. The coiled string actuator can overcome common limitations existing compliant actuator technologies have, ranging from small strain or stress outputs, high power requirement, low bandwidth, poor power efficiency, and bulkiness.

However, the non-uniform actuation during string coiling, together with sophisticated, coupled electro-mechanical dynamics and time-varying material nonlinearities, presents significant challenges in precisely controlling the coiled string actuators. This project aims to model the coiled string actuators’ non-smooth coiling-induced actuation and its strain self-sensing property, and control them to obtain smooth and desirable actuation.

This research will (1) develop a physics-based model, a kinematic model, and a machine learning-based, control-oriented model to capture the coiling-induced actuation, (2) create a model to realize strain self-sensing by correlating the electro-mechanical property and material time-varying nonlinearity, (3) construct new inverse compensation and self-sensing-based adaptive control strategies for controlling systems with non-uniform coiling-induced actuation, and (4) experimentally validate the projected modeling, self-sensing, and control approaches on a compliant robotic manipulator with superior compactness and dexterity.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.