HARDWARE SETUP
Why is hardware important?
Though the purpose of this project is to ultimately use Digital Signal Processing tools to determine whether an arm is flexing or not, the sensor and hardware systems that we are using are integral to the purpose of this project: To make a prosthetic arm that is both affordable and works well!
Myoware EMG Sensor [1]
EMG Sensor
The standard sensor that is typically used is something called an EMG sensor. The MyoWare Sensor above is currently used by the Michigan Neuroprosthetics team - which our member Maxton is a member of. The EMG sensor measures the electrical activity in response to a nerve's stimulation of the muscle. An EMG sensor is good at pinpointing specific muscles. However, one problem with this sensor is that EMG data is hard to read accurately. It does not always give a clean signal when a user bends their arm for example. The Michigan Neuroprosthetics team has had problems with accurately reading EMG data. Therefore, we will be exploring the cumulative sum algorithm as a method for detecting the changes in the EMG data in order to make the sensor more reliable in reading in flexes.
Reason for Another Type of Sensor
Though EMG sensors are often used in Prosthetic arms, they are not cheap - typically costing around 40$. Along with that, they aren’t always reliable. Sweat and other muscular noise can have a detrimental effect on what the sensor reads and can make it harder to determine when the user flexes. Therefore, we decided to explore another method of muscle detection.
In the Academic paper Trans-Radial Prosthesis with Three Opposed Fingers by Masahiro Yoshikawa[2], it hinted to a way of using an Infrared Line sensor to measure distance. We incorporated the idea into our design to create a cheaper sensor to measure flexes.
QRE 1113GR IR Line Sensor [3]
The IR Sensor Design
We used the sensor above, the QRE 1113GR IR Line Sensor - which measures distance based on the reflection of Infrared Light to make a sensor that could detect muscle flexes.
​
In our design, the IR sensor is placed on top of a foam spacer. The spacer suspends the sensor away from the surface of the arm while also protecting the sensor from any outside radiation and the whole setup is held in place by a band wrapped around the user’s arm. When the user flexes, their arm expands, squeezing the setup and compressing the foam. This brings the sensor closer to the skin, which alters the signal coming from the IR sensor. Using DSP techniques, we can then analyze the signal and determine when the user is flexing and code our microcontroller to have the arm respond accordingly.
​
The benefits of this sensor is that it’s cheap: rangin around $3 per sensor, compared to the $40 EMG sensor, it is much cheaper. Another benefit is it is not directly touching the skin, therefore has less chance of wearing, and getting sweat upon. Also, as you will see in the Data section, the sensor provides pretty clear muscle flexing data.
​
Ultimately, we settled on placing the IR sensor on the outside of the upper arm, near the tricep. We discovered that when the sensor was placed anywhere else, a bend at the elbow was detected as a flex. Selecting this location for the sensor effectively resolved this issue.
IR Sensor Design
