Research Performed at the University of New Brunswick Institute of Biomedical Engineering under Dr. Peter Kyberd
Force Sensor-Controlled Prosthetics Abstract:
There are multiple problems associated with myoelectric control, currently the most popular form of prosthetic control. Myoelectrodes are expensive, require extensive processing to remove noise, must sometimes be implanted to receive the best signal, and often receive a noisy signal when used externally. One out of every twenty times, myoelectrodes inaccurately predict muscle bulge. Force sensors, a new control method being tested in this project, measure muscle bulge directly, rather than the electricity produced by the muscle. Force sensors are inexpensive, require little or no signal processing, and are used externally. To test this control method, an operational prosthetic hand prototype was built. MATLAB programming language was employed to write a program that could take readings through the computer, from both myoelectrodes and force sensors, and compare their accuracy. The program used Linear Discriminant Analysis to analyze the input voltages and convert them into a signal that would be capable of commanding movement for a given degree of freedom in a prosthetic device. Results show that force sensors can accurately differentiate between different forearm muscles with little training, indicating that in the future they could provide a low-cost, low-maintenance control method for amputees. Research was supported by a grant from Mu Alpha Theta.
This project was independently orchestrated and researched, with advisory support from Professor Peter Kyberd of the University of New Brunswick. You can learn more about the progression of this project in some of my blog posts.
Research Details and Results:
|One of the first steps in my prosthetics research was to develop an operable prototype to demonstrate that a simple prosthetic hand could be reliably controlled using input from a force sensor, instead of myoelectrodes. Utilizing a BasicStamp 2pe Microcontroller, Servo Motor, Plastic Hand Model, Metal chassis, Force Sensor Cast (see below), Piezo Vibration Sensor/Amplifier, and Digital Readout, I constructed a prototype. The video below shows the progression through sever versions of my prototype, with each adding more functionality, including automatic slip-arrest, proportional force control, and more.|
|With help from the University of New Brunswick’s World-Class Biomedical Research Facility, I modeled a cast to my arm and outfitted it with force-sensitive resistors (FSRs) on key muscle-bulge sites. These sites are outlined in the paper and video below. The FSRs feed into an amplifier board that I designed and built, which filters and amplifies the low-voltage signals picked up by sensors. This amplifier board then feeds into a National Instruments Data Acquisition Device which interfaces with MATLAB software developed by me for performing signal analysis.|
|Shown here is an example of an acquired, amplified signal from a force sensor in my cast located above the Extensor Carpi Ulnaris muscle. As one would expect, when I flexed my fingers, the sensor above the responsible muscle showed the greatest response, with a clear distinction from the other force sensor signals. The research paper linked below shows all the signals and how they compare to more traditional myoelectric control. Using linear algebra matrix filtering techniques (also outlined in the paper) signals were analyzed to extract signals that can be used for prosthetic control. Results from this data showed that force sensor control is less prone to cross-talk interference than myoelectric control.|
Videos of my Prosthetic Prototypes, and Me Explaining my Research:
|Intel Science Talent Search (STS) Inspiration Video
||Prototype Model Progression (Played at my Intel STS Booth)
Below materials are provided via a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.
Publications & Papers:
- Blum, J. E., (2007) “Using Force Sensors to Effectively Control a Below-Elbow Intelligent Prosthetic Device”, Intel Science Talent Search Comprehensive Research Paper.
- Anders Fougner, Marthe Sæther, Øyvind Stavdahl, Peter J. Kyberd, and Jeremy Blum, (2008) “Cancellation of Force Induced Artifacts in Surface EMG Using FSR Measurements“, Proceedings of the 2008 MyoElectric Controls/Powered Prosthetics Symposium.
Posters and Presentation Files
- Download the 20×30 Summary Poster (PDF)
- Download the 48×48 Presentation Poster (PDF)
- Download the 30 slide PowerPoint (PDF)
Blog Posts on my Prosthetics Research & the Intel Science Talent Search Competition:
Visit the Prosthetics Research Category in my Blog.
News Coverage of my Prosthetics Research:
- 02/01/2007 – “$2000 Grant gives Teen a Hand” in The journal News (PDF)
- 03/14/2008 – “Best Inventions of 2008 from Intel Science Talent search” on InventorSpot
- 04/28/2008 – “Young Einsteins” on the Journal News Hall Monitor Blog
- 05/30/2008 – Parallax Winners Circle (Scroll Down)