Mobile hdEMG Exoskeleton
Electromyography (EMG) is a way of categorizing muscle activity through electrical signals from the nervous system. This is useful for allowing those with impairments like amputation and paralysis to use exoskeletons and prosthetics to regain functionality. I developed a high-density EMG controller for a variety of exoskeletons and EMG devices used in rehabilitation research at Shirley Ryan AbilityLab. In addition, I created a latency analyzer device with accompanying software packages to obtain the time duration of each software and hardware interaction within the exoskeletal system. The system is lightweight and compact enough to be used for dynamic walking tasks.
The system is compatible with the Muovi+Pro, SyncStation, and Quattrocento EMG devices. It is compatible with the Fourier X2, M1, and Technaid H3 exoskeletons. Up to four 64 channel EMG grids can be currently connected but can easily be modified to accommodate more. The system supports individual EMG channel removal for detached channels on the grid, and a probe visualization selector to plot a specific muscle group in real-time. It is compatible with root-mean-squared (RMS) and cumulative spike train (CST) muscle activity estimation.
Collaborators: Jackson Levine, Ashwat Dhamotharan, Dr. Matthew Elwin, Dr. Jose Pons
GitHub for using M1, X2, and H3 exoskeletons (separate packages not included): https://github.com/daviddorf2023/hdemg_exo_control
GitHub with H3 exoskeleton packages included: https://github.com/daviddorf2023/mobile_hdemg_exo
System Architecture
The power system for the exoskeleton had to be overhauled to allow for patients to wear the entire system on their upper thigh. Previously, patients had to be lying down for purely isometric movements. The power for most components was connected to wall power outlets, and the exoskeleton came with a bulky backpack battery. Now, the system is completely mobile and has significantly less components.
The old system employed far more components overall, and used the Quattrocento as its EMG device: Quattrocento (otbioelettronica.it). The new system uses the more modern Muovi+Pro, which are small wireless EMG probes that can be recharged at a SyncStation+:Â Muovi+Pro (otbioelettronica.it).
In order to have the system be completely wireless, it had to be lightweight,19V regulated, and have enough capacity to run experiments for an hour before recharging.
Old system hardware layout
New system hardware layout
CST vs RMS vs Torque Patient Datastream
Data Collection and App - Made with Clipchamp.mp4Latency Analyzer
Latency was decreased from the old system by approximately 21ms. This was the result of changes to the number of system components, the simplification of communication protocols, and modifications to how EMG data was recorded.
Processing, visualizing, and outputting raw EMG data was done in parallel with separate nodes, greatly increasing the overall speed that the CST and RMS methods could output data.
The latency of the EMG device hardware was also tested with a PWM input of 50Hz. Both the Muovi+Pro and Quattrocento did not return a square wave as expected. However, the pipeline and hardware setup for testing latency with a PWM signal has been established, opening the potential for future investigation into characterizing this aspect of the system latency.
