Matplotlib demo for electrocardiogram

[vsquared]

I'm trying to use py5 for an app which uses a serial port connected to an Arduino UNO. Imported data is then plotted using Processing's line() call. However I get an error message that says "this.g" is null. How can I initialize "this.g" in Python? The same issue occurs with setting the background color. Source code follows:

# Uses imported mode for py5
# Cannot invoke "processing.core.PGraphics.line(float, float, float, float)" because "this.g" is null
import serial

_wndW = 600
_wndH = 400

global x
x = 0
global y
y = 0
global y1
y1 = 0

def setup():
    size(_wndW, _wndH)
       
ser = serial.Serial('/dev/tty.usbmodem14301')  # open serial port   
print(ser.name)         # check which port was really used

  
while 1:    
    val = str(ser.readline())
    val = val.lstrip("b")
    val = val.strip("'")
    val = val.rstrip("\\r\\n")
    print(val)
    if len(val) > 0:
        y = int(float(val))
        
    line(x, _wndH-y-50, x-4, _wndH-y1-50)
    x = x+4
    y1 = y
    if (x > _wndW):
        x = 0
        background(0,0,255)

Arduino code:

byte outputValue = 0;

void setup() {
  Serial.begin(9600);
}

void loop() {
    outputValue += 4;
    Serial.println(outputValue);
}

[villares]

Cheers @vsquared!

If you are not using "static mode", in your case you are using "imported mode", but it would be the same in "module mode", you can't have a drawing method like line() outside draw() or some other py5-main-loop-controlled-functions.

So just replace while 1: by def draw(): and believe it should work.

[vsquared]

It looks like 'while 1' is required to pull in the data but it seems to block anything else, eg, drawing into the window. I added a separate draw() function similar to what I do in Processing but all I get is a rotating beachball; there is no drawing.

[hx2A]

@vsquared , for your first example, I would try creating a draw() function for drawing with py5 methods and then move the serial port stuff to its own function and have py5 call it repeatedly with launch_repeating_thread(). You wouldn't need the while 1 loop to repeatedly process data from the Adruino and the drawing code would be taken care of by py5 in draw().

https://py5coding.org/reference/sketch_launch_repeating_thread.html

[vsquared]

The data points just keep getting scrunched up and the frame rate gets slower and slower (probably some setting I'm not aware of):

[hx2A]

Can we see the code? When you create the deque you should use the maxlen parameter to set the maximum length.

[vsquared]

plt.axis([0, None, 0, 300])
There is no top end (xmax); I want data points to be dropped off the bottom end when xmax exceeds the window width.

import matplotlib.pyplot as plt
from matplotlib.animation import FuncAnimation
from matplotlib import style
import numpy as np
import random
import serial

global x,y

# initialize serial port
ser = serial.Serial()
ser.port = '/dev/cu.usbmodem14301' #Arduino serial port
ser.baudrate = 9600
# ser.timeout = 10 #specify timeout when using readline()
ser.open()
if ser.is_open==True:
    print("\nSerial port is open. Configuration:\n")
    print(ser, "\n") #print serial parameters

# Create figure for plotting
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
xs = []
ys = []
x=0
y=0

# This function is called periodically from FuncAnimation
def animate(i, xs, ys):
    global x,y
    #Aquire and parse data from serial port
    val=str(ser.readline())      #ascii
    val = val.lstrip("b")
    val = val.strip("'")
    val = val.rstrip("\\r\\n")
#     print(val)
    print("["+str(x)+"]"+val)
    if val.isnumeric():
        val = int(val)
        ys.append(val)
        xs.append(x)
    x = x + 1
    print(ys[x-1])
    
    ax.clear()
    ax.plot(xs, ys)

    # Format plot
    plt.xticks(rotation=45, ha='right')
    plt.subplots_adjust(bottom=0.10)
    plt.title('Sawtooth Model')
    plt.ylabel('Value')
#     plt.legend()
# xmin,xmax,ymin,ymax
    plt.axis([0, None, 0, 300]) #Use for arbitrary number of trials

animation = FuncAnimation(fig, animate, fargs=(xs,ys), interval=1)

plt.show()

[vsquared]

I figured out how to show just a certain amount of data points:
plt.axis has to be set to ([None, None, 0, 300]) and you have to use deque's for the data points with the maxlen set, eg

xs = deque([],maxlen=550)
ys = deque([],maxlen=550)

I'll edit the above post later.

[hx2A]

Yes, that's a good use of the deque data structure!

[vsquared]

The following Python matplotlib source code provides an alternative to the initial post by creating a scrolling data plot when connected to an Arduino UNO through a serial port (Arduino code shown in original post). By using two deque arrays for the x and y data points a limit of 550 data points are plotted. When the window is resized only 550 data points are shown at each size. Initially the x-axis is static but starts scrolling after 550 data points are plotted. By setting maxlen in each of the deque arrays, data points are removed from the left side of the plot as new data points are added to the right side. This creates a 'scrolling' graph that appears to be slower than the incoming data. However, there are no obvious missing data points as seen with the Processing model.

import matplotlib.pyplot as plt
from matplotlib.animation import FuncAnimation
from matplotlib import style
from collections import deque

import numpy as np
import random
import serial

global x,y

# initialize serial port
ser = serial.Serial()
ser.port = '/dev/cu.usbmodem14301' #Arduino serial port
ser.baudrate = 9600
ser.open()

if ser.is_open==True:
    print("\nSerial port is open. Configuration:\n")
    print(ser, "\n") #print serial parameters

# Create figure for plotting
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)

xs = deque([],maxlen=550)
ys = deque([],maxlen=550)
x=0
y=0

# This function is called periodically from FuncAnimation
def animate(i, xs, ys):
    global x,y
    #Aquire and parse data from serial port
    val=str(ser.readline())      #ascii
    val = val.lstrip("b")
    val = val.strip("'")
    val = val.rstrip("\\r\\n")
    print("["+str(x)+"]"+val)
    if val.isnumeric():
        val = int(val)
        ys.append(val)
        xs.append(x)
    x = x + 1
       
    ax.clear()
    ax.plot(xs, ys)

    # Format plot
    plt.xticks(rotation=45, ha='right')
    plt.subplots_adjust(bottom=0.10)
    plt.title('Sawtooth Model')
    plt.ylabel('Value')

# xmin,xmax,ymin,ymax
    plt.axis([None, None, 0, 300]) #Use for arbitrary number of trials

animation = FuncAnimation(fig, animate, fargs=(xs,ys), interval=1)

plt.show()

Data limit is honored at all window sizes:

Addendum:
...And it can do something besides sawtooth patterns:

[hx2A]

py5 has the same limitations as Processing when it comes to drawing graphics (has to be done inside draw()).

Well, there are ways to use py5 without a draw() or setup() function, but they don't open an animated window. For example:

https://py5coding.org/reference/py5functions_render.html

I don't think that's the right fit for this situation but it probably would be possible to do something else that was, particularly if it was limited to the default JAVA2D renderer.

[villares]

matplotlib demo that can be used for plotting an electrocardiogram.

This could be a good name!

It doesn't bother me at all using draw() as the main animation loop. Also, sometimes I have a separate thread at a different speed updating a data structure that I can "sample" at draw time.

[vsquared]

The matplotlib demo doesn't use py5; is that a problem?
I have a separate thread at a different speed updating a data structure
That's why Processing has void serialEvent(Serial p) {...} but I don't think we can use that in py5.

[villares]

No problem. If you add py5 to your example afterwards you can update it.

have a look at the easy threading support by py5: https://py5coding.org/reference/#threading maybe the launch_repeating_thread(), as Jim mentioned... or launch_thread() with your while loop inside it...

[vsquared]

have a look at the easy threading support

I already tried that and the output left a little to be desired. There are images in this thread. Losing data points is not good for my purposes.