Graduate Research Assistant Georgia Institute of Technology
Aiding those who have difficulty completing everyday tasks can dramatically improve their quality of life. While an exoskeleton which employs electromyography (EMG) and deep learning motion classification in conjunction with lightweight wearable pneumatic actuators can address the issue, such a development poses clear technical challenges in processing and integration. Here, we introduce a framework which wirelessly connects multiple bio-patch devices placed on major muscle sites, predicts the user’s intended motion in real-time, and controls the actuators accordingly. The small flexible circuitry of the bio-patch devices samples skin-surface nanomembrane electrodes at 1 kHz each and transmits the EMG signals through Bluetooth for high-fidelity monitoring of muscular activity. For processing, a custom software application implemented on the exoskeleton synchronizes these datastreams and filters them of noise. Then, a convolutional neural network is trained on the datasets and embedded into the framework to provide a correlation between muscle activity and the user’s intended motion for six distinct tasks with an accuracy of 95%. Simultaneously, a pressure sequence designed to support these tasks is sent through a serial connection to a micro-controller with enough pulse width modulators to control the pneumatics and achieve complex motions. Finally, an inertial measurement unit in each bio-patch tracks the motion of the user to ensure the intended motion is achieved through a closed feedback loop and dynamic actuation. Thus, by integrating these diverse components into one cohesive exoskeleton and implementing complex real-time processing, reliable aid can be given to those who need assistance completing everyday tasks.
Bryan Starbuck1, Jinwoo Lee1, and Woon-Hong Yeo1,2,* 1 George W. Woodruff School of Mechanical Engineering, Center for Human-Centric Interfaces and Nanoengineering at Institute for Electronics and Nanotechnology, Georgia Institute of Technology, Atlanta, GA, 30332 USA 2 Wallace H. Coulter Department of Biomedical Engineering, Parker H. Petit Institute for Bioengineering and Biosciences, Neural Engineering Center, Institute for Materials, Institute for Robotics and Intelligent Machines, Georgia Institute of Technology, Atlanta, GA, 30332 USA *Email: whyeo@gatech.edu
.jpg "Bryan Starbuck, MS photo")