Intelligent Control Systems for Soft Artificial Muscles

NSF’s AI Institute for Artificial and Natural Intelligence awarded Davoodi a grant for robotics project

The emerging field of soft robotics represents the foundation of future robotic systems, with a multitude of applications in human-robot interaction, locomotion, and rehabilitation technologies. The NSF AI Institute for Artificial and Natural Intelligence (ARNI), led by Columbia University, has recently awarded Dr. Mohammadreza Davoodi, assistant professor in Electrical and Computer Engineering, a grant for the project "Intelligent Control Systems for Soft Artificial Muscles," marking a significant advancement in robotics. In this project, Davoodi collaborates with Dr. Nafiseh Ebrahimi from Virginia State University and Dr. Xaq Pitkow at Carnegie Mellon University to develop innovative machine learning/AI algorithms for controlling novel artificial muscle technology developed by Ebrahimi. These artificial muscles are formed by bio-inspired, highly scalable, flexible, and biocompatible Electromagnetic Soft Actuators (ESA).

A human skeletal muscle, with bundles of sarcomeres composed of actin and myosin, behaves like a network of soft actuators. Similarly, a network of ESAs can be integrated inside an artificial muscle. By interconnecting multiple artificial muscles, we can mimic the functionalities of natural skeletal muscle systems. However, effective control of such a system remains a challenge, which this project seeks to address. Our long-term goal is to create and control an embodied agent with multiple artificial muscles.

This project addresses a core ARNI theme that falls under the Neural Mechanisms thrust and the theme of embodied control platforms. Specifically, it will advance the next generation of flexible, powerful, and compliant artificial muscles, with applications in wearable assistive devices and portable active braces. This research will pave the way for future investigations into the design and control of dynamic systems involving networks of soft actuators, establishing a foundation for further advances in soft robotics and biological motor control. Additionally, our team will provide research training opportunities on the foundations of control theory and modern machine learning techniques for students at all collaborating sites.

For more information on this project, contact Davoodi at mdavoodi@memphis.edu.