Overview the Elemyo-library for Arduino.

This example is intended for Arduino firmware for visualization and signal processing in Python and MatLab. It is enough to compile this example and flash Arduino, then run the sketch in Python (download the example) or MatLab (download the example). The example is compatible with the entire range of sensors. In Fig. 1 shows the MYO to Arduino Uno connection diagram used in the example. Other sensors you can connect in the same way.

This example implements the simplest function: reading sensor readings on analog input A0 and writing the signal to Serial. If a sensor with adjustable gain is used, the gain can be set using the MyoSensor.gain (value) command, where one of the following gain values is used as the argument "value": "x1", "x2", "x3", "x4", "X5", "x8", "x10", "x16", "x32". The connection diagram is similar to Fig. 1.

This example demonstrates working with an array of sensors. In the example, the simultaneous connection of 6 sensors is analyzed (more or fewer sensors are connected by analogy).
Sensors are connected according to Fig. 2.

Sensor gain control is done through SPI. If you plan to control the gain of each sensor individually, then the CS pins of each sensor must be connected to individual Arduino pins (it can be either PWM or digital). If you plan to set the same gain for all sensors, then the CS pins of each sensor can be connected to one common Arduino pin (it can be either PWM or digital). SCLK pins are connected to pin 13, SI pins to pin 11.
After flashing Arduino with this example, you can start the plotter via a serial connection to draw signals from each sensor.

This example demonstrates one of the signal smoothing methods. As an anti-aliasing method, the method of exponential moving average of arbitrary order is taken. Information on the moving average method and the exponential moving average method of arbitrary order is available here.

Let us have some function f (t) (signal) that changes its value over time (for example, temperature, pressure, EMG, etc.). We are faced with the task of highlighting in real time the main trends in the behavior of a function against the background of short-term fluctuations. To do this, we need to average the values of the function. One of the ways of such averaging is the method of exponential moving average of arbitrary order. The idea of the method is that each new averaged value is obtained based on the current and previous values of the function. The method is constructed as follows:

EMA_tⁿ = a* EMA_t^n-1 + (1 - a)* EMA_t-1ⁿ,

where EMA_tⁿ - this is the value of the exponential moving average of the n-th order at the current time t. EMA_t-1ⁿ this is the previous value of the exponential moving average of the n-th order. EMA_t^n-1 – this is the current value of the exponential moving average of the (n-1) th order. a - smoothing coefficient.

As an example, we consider how signal smoothing occurs by the 3rd order exponential moving average method:

First, we calculate the value of the exponential moving average of the first order.

EMA_t = a*f(t) + (1 - a)*EMA_(t-1), where EMA_t – first order exponential moving average. f(t) – function value at the current time, EMA_(t-1) – first-order exponential moving average calculated over the previous time period.

Then, the second order:

DMA_t = a*EMA_t + (1 - a)*DMA_(t-1), где DMA_t – second order exponential moving average.

Then, the third order:

TMA_t = a*DMA _t + (1 - a)*TMA_(t-1) – third order exponential moving average.

Initial values of EMA₀, DMA₀, TMA₀ can be set equal to f(0).

If we take the EMG signal directly as a function, the smoothing result will be direct, since the EMG signal is symmetrical with respect to the reference voltage. If the EMG signal is rectified with respect to the reference voltage (Fig. 4), then the smoothing curve will pass between the peak values and the value of the reference voltage, while the amplitude of the smoothed signal will be ¼ of the signal amplitude (Fig. 5-7).

The program works as follows: reads the readings from the analog input A0 Arduino, if the value is less than the value of the reference voltage, then the sensor reading is symmetrically "transferred" relative to the value of the reference voltage (Fig. 4). Next, the calculation of EMA_tⁿ. The program is implemented EMA_t³, but if necessary, it can be replaced by any other order.

An example of the algorithm is shown in Fig. 5-7.

You can adjust the "depth" of smoothing using the coefficient a. The closer a is to 1, the mean curve is less sensitive to signal changes.
If the recorded EMG signal turns out to be small in amplitude, then set the increasing gain using the function MyoSensor.gain (value), where one of the following gain values is used as the argument "value": "x1", "x2", "x4", "X5", "x8", "x10", "x16", "x32".

This example demonstrates how to use the EMG signal to set two control commands. The program works as follows: if the muscle is tensioned for 1 second or less, the program will display the value "Type 1", if more than a second, then the value "Type 2". This program can be used as the easiest way to control a robotic arm, motor or just an LED. For example: if we stretch the arm for less than 1 s, then the robotic arm will make a fist-type grip, if for more than 1 s, then the robotic arm will make a pinch grip. In the case of the LED: when the hand voltage is short, the LED will blink 1 time, in case of continuous hand voltage, the LED will blink twice.

The program constantly monitors the magnitude of the amplitude of the recorded signal. As soon as the amplitude value exceeds a certain threshold thresholdStart, the program understands that the muscle is tense and starts counting the time (EMG_TimeStart). As soon as the amplitude of the EMG signal becomes less than the threshold value thresholdEnd, the program detects that the muscle is no longer strained and calculates the time interval EMG_Time, during which the EMG was recorded. Depending on the value of EMG_Time, the value "Type 1" or "Type 1" is displayed.

For the program to work correctly, it is important to set the thresholdStart and thresholdEnd parameters that are individual for each user. To do this, first run the "Terminal" example above and display the EMG signal. thresholdStart set as half the value of the maximum amplitude, thresholdEnd as the doubled value of the interference value in the absence of an EMG signal. If the recorded EMG signal turns out to be small in amplitude, it sets the increasing gain using the MyoSensor.gain (value) function, where one of the following gain values is used as the argument "value": "x1", "x2", "x4", "X5", "x8", "x10", "x16", "x32".

Thanks for attention!
We hope you find this review helpful. If you have any questions - feel free to contact us at info@elemyo.com. We will answer you!