Published January 3, 2018

Car Alarm Tracker

Create a device that will send you a text when your car's alarm goes off.

IntermediateFull instructions provided3 hours2,875

Things used in this project

Hardware components

Raspberry Pi 3 Model B

For testing purposes only

Raspberry Pi Zero Wireless

USB 2.0 Mini Microphone for Raspberry Pi 3

Micro-USB to USB Cable (Generic)

Hologram Global IoT SIM Card

Hologram Nova

Story

Abstract

Create a device that will recognize when a car alarm is sounding and text the user that his/her car alarm is going off and they should investigate. The device shouldn't go off if another car is just honking for a few seconds or if a loud bang occurs - it should only go off if there is a sustained car alarm.

Introduction

Car alarms are present on almost all new cars. Some cars will let you set the tolerance setting for when the car alarm goes off and various other settings but this is only helpful if you are: 1) near enough to the car to hear/see the alarm go off or 2) can recognize your car's alarm sound from a distance so you can go investigate. To address this issue, and using a Nova Hologram, a Raspberry Pi W, a USB microphone and a portable battery, I've assembled a device that will recognize the car alarm and alert you.

Methods

The requirements are outlined above and in order to address them I had to make a couple of assumptions:

I assume that the device is going to go inside the glove box of the car. The reason being is the glove box is relatively insulated compared to the rest of the car. This allows distant noises (like birds chirping or general traffic) to be diminished, allowing the tolerance level detected to be lower

Initially, I set the tolerance low for testing. If you are building this for your car, you can set the tolerance level higher as needed.

I had to distinguish between a car loudly stopping or a car honking for a couple seconds from a sustained car alarm. The assembled device listens for several seconds and if the tolerance is above that threshold for that period of time, it sends the alarm. If the device only hears one peak and then doesn't hear another peak within a waiting period, it will reset the count.

Honking is a sustained noise for a period of time, Loud Stopping is an infrequent noise, Car Alarm is a frequent noise

.This project was initially developed on a Raspberry Pi B+ and then the SD card was transferred to the Raspberry Pi Zero W after most of the testing was complete. This was done because the Raspberry Pi B+ has four USB ports = two for keyboard and mouse, one for the microphone USB plug and one for the Nova Hologram.

The code was developed in python 2.7 and required the following libraries:

Hologram Cloud library - please follow the excellent guide to initially setting it up at https://projects.hologram.io/hologram/add-cellular-to-a-raspberry-pi-with-hologram-nova-ea5926

Time library - this was used to create a delay in the code that would "wait" a certain amount of time, send the text and then delay again before starting

audioop - this python library is incredibly useful. The rms function was used to take the root mean squared of the 1/2 second interval that is sampled.

pyaudio - this was used in tandem with the audioop library to take in samples.

In order to import the pyaudio library into python, use:

sudo apt-get install python-pyaudio

In case this is your first time using python, to verify that you have imported the libraries correctly, open up a window and type:

import pyaudio

If it doesn't result in an error, you've done imported the library correctly.

After you've imported all of the libraries, the next step is verify that you can text from hologram.

Here is the link to the Python SDK for Hologram: https://hologram.io/docs/reference/cloud/python-sdk/

The most important part of this is where it talks about HologramCloud(credentials, network='',authentication_type='topt'). The default mode for the authentication_type is totp but in order to send texts from the device, the authentication_type must be csrpsk. credentials are set as:

credentials = {'devicekey':'your 8 letter device key found in the hologram overview'

The device utilizes the four usb ports and the power supply was initially from the wall but final installation is using a lipo battery.

The red and blue device is the hologram, above that is the usb microphone

See attached code for the commented program.

The initial setup included playing a car alarm audio off of my computer and setting the threshold down to 100.

Car Alarm showing rms values (it dips down and above 100 so that is where I set the threshold)

After 10 seconds of the noise, I received a text.

Text showing that the car alarm is going off

To make the script run when the device boots, open a command window on your raspberry pi and type:

crontab -e

It will ask you which one you would like to boot into, select option 2.

Enter in at the bottom of the script:

@reboot sudo python /home/pi/test.py

test.py is what I called the script I wrote. You can change based off of your naming.

Press CTRL+X to exit the script and make sure you saved. I followed the following instructions for this: https://www.dexterindustries.com/howto/auto-run-python-programs-on-the-raspberry-pi/

Here is the image of the text I received after setting up the script to start on boot:

Verified to text

Results

The final step was to take the SD card out of the Raspberry Pi 3 and plug it into the Raspberry Pi Zero W and verify it works!

Wiring with wall power

So it successfully boots up and I can run the program without any issues. I plugged it into a portable battery and the same result occurred!

Wiring with a portable charger

Summary and Conclusions

Using a Raspberry Pi W Zero, a Hologram Nova Global 3G/2G Cellular Modem, and a USB microphone, I've created a device that will send alerts to me when my car alarm goes off. Future implementations could be to charge the battery off of the car when it's on so that you wouldn't have to swap the portable charger. I had to adjust the threshold value to be perfect for my car (not as easy of a task as I'd like because I had to remote into it over wifi and my wifi range doesn't extend that far).

Hologram Texting

from Hologram.HologramCloud import HologramCloud
import audioop
import pyaudio
import time

credentials = {'devicekey':'12345678'}
hologram = HologramCloud(credentials, network='cellular',authentication_type='csrpsk')

# this is the threshold that determines whether or not sound is detected. The higher the number, the louder it is
# according to 
THRESHOLD = 10000

p = pyaudio.PyAudio()

chunk = 22050

# 'rate' specifies the desired sampling rate in Hz (samples/seconds)
# 'channels' specify the desired number of channels
# 'input' specifies whether this is an input stream - defaults to False
# 'output' specifies whether this is an output stream - defaults to False
# 'frames_per_buffer' specifies the number of frames per buffer (samples)
# the 'frames_per_buffer' divided by the 'rate' will determine how long the period we're sampling will be
# For example: 22050/44100 = 0.5 seconds so each data chunk will be for 0.5 seconds
# stream will open 
stream = p.open(format=pyaudio.paInt16,
                channels=1,
                rate=44100,
                input=True,
                frames_per_buffer=chunk)

#open your audio stream    

# countAboveThreshold specifies the number of chunks that are above the threshold value
countAboveThreshold = 0

# countSinceLastThreshold specifies the number of chunks that have occurred since the last threshold
# if the countSinceLastThreshold is above a certain number, we reset the countAboveThreshold
countSinceLastThreshold = 0

# wait until the sound data breaks some level threshold
while True:
    # data is the chunk of information from the current stream
    data = stream.read(chunk)
    # take the rms value of that and set it to be rms
    rms = audioop.rms(data,2)
    # if rms is greater than the threshold level
    #print rms
    if rms > THRESHOLD:
        # increase the countAboveThreshold
        countAboveThreshold += 1
        # reset the countSinceLastThreshold
        countSinceLastThreshold = 0
        # if the countAboveThreshold is above 5 (that means that there have been 2.5 seconds of a loud noise)
        if countAboveThreshold > 5:
            # reset coundAboveThreshold
            countAboveThreshold = 0
            # reset the countSinceLastThreshold
            countSinceLastThreshold = 0
            # Send a text message alerting the user that their car is going off and they should look into it
            delay(5)
            hologram.sendSMS("+112345678","Your car's alarm is currently going off")
            delay(5)
            #break

    # if the countAboveThreshold is above 0 that means that we have heard at least one loud noise
    if countAboveThreshold > 0:
        # increase the countSinceLastThreshold by 1
        countSinceLastThreshold += 1
        # if the countSinceLastThreshold is above 9 that means that there has been at least 5 seconds since the last loud noise
        if countSinceLastThreshold > 9:
            # reset the countAboveThreshold. This indicates that the loud noise wasn't long enough to be important
            countAboveThreshold = 0
            # reset the countSinceLastThreshold
            countSinceLastThreshold = 0

Credits

Alex Merchen

24 projects • 39 followers

I'm an EE with a Masters in ECE. I like building things.

Car Alarm Tracker

Things used in this project

Hardware components

Story

Abstract

Introduction

Methods

Results

Summary and Conclusions

Code

Hologram Texting

Credits

Alex Merchen

Comments

Embed the widget on your own site

Car Alarm Tracker

Car Alarm Tracker

Things used in this project

Hardware components

Story

Abstract

Introduction

Methods

Results

Summary and Conclusions

Code

Hologram Texting

Credits

Alex Merchen

Comments

Related channels and tags