SkinnyPi

Losing Weight Triggers Party Music & Lights
A Laravel, Raspberry Pi, FitBit & MQTT Project

I thought it would be fun to setup a Raspberry Pi so that every time I lost weight, it would flash lights and play music. I referred to this as “Party mode” or “Happy Dance.” It would be positive reinforcement and a fun way to start my morning. Hopefully this is useful for someone or at least a great example of how to get your cloud services to communicate with your Raspberry Pi.

Visit my site to see all the fun projects I create: https://jazmy.com

Github Source Code:
https://github.com/jazmy/laravel-skinnypi
https://github.com/jazmy/raspberrypi-skinnypi

Demo:
Skinny Pi in Action — Video Demo

Requirements

1. A Fitbit Compatible Scale
   Any scale that allows you to sync your weight wirelessly to the Fitbit site. I am using the Fitbit Aria
2. A Raspberry Pi (with wifi)
   I am using the Raspberry Pi 3 Model B
3. Hosted Cloud Server
   I am using Cloudways which is easy for Laravel
4. Hosted MQTT Server
   I am using the free version of CloudMQTT

How it Works (simplified version):

1. You step on the Fitbit Aria Scale and it syncs over wifi to the fitbit server.
2. The Fitbit API triggers that a new weight was added and does a call to your cloud server.
3. Your cloud server stores the new weight and calculates to see if it was higher or lower than the previous weight.
4. If the weight is lower it sends a message to your MQTT service.
5. Your Raspberry Pi is subscribed to the MQTT service so it receives the message and triggers a python script.
6. The Python script plays music and flashes lights.

Part 1 — The Raspberry Pi

Requirements

• Raspberry Pi 3 Model B
• 32 GB Micro SD Card
• Pimoroni Blinkt Leds
• Speakers 8 ohm 0.25W
• Clear Case for Pi 3

1.1 Assembling Your Pi

I’ll assume if you are reading this that you know how to put together a raspberry pi. All the parts together will be about $60 so it is a fairly expensive project to build.

1.2 Installing Software

1.2.1 Installing Raspbian from your PC:

1. You will need to install Raspbian on your Raspberry Pi: https://www.raspberrypi.org/downloads/
2. To get the Raspbian image on your Raspberry Pi you need a tool called win32 disk imager:https://sourceforge.net/projects/win32diskimager/
3. Unzip the raspbian software which is just an image file. It may takes a couple minutes because it’s 2 GB.
4. Open Win32 disk and browse for the raspbian image.
5. Select the “Device” which is the drive your SD card is on. When selecting the device, make sure you select the correct letter because it will delete everything on that drive. It’s a common issue for people to accidentally delete a usb backup drive they may have also had connected.
6. Then click the “Write” button. It takes a few minutes to write depending on the speed of your card.
7. Once done, windows may get confused because it can no longer recognize the drive and it will ask you to format your sd card. Ignore that error. Your SD card is complete, now you plop it in your raspberry pi

1.2.2 Setting up your Pi the First Time

1. Turn on your Raspberry Pi
2. Login with default credentials: Username: pi and Password: raspberry
3. It will ask you to change those
4. It will do updates
5. It will ask you to connect to your wifi
6. Open the terminal window
7. Type “ifconfig” to figure out your IP address (Another option is to type: sudo /etc/rc.local )
8. Click on the Raspberry icon for a drop down menu and select “Preferences” then “Raspberry Pi Configurations”
9. Enable SSH

1.2.3 Install Mosquitto

There are several applications that can be used to send and receive through MQTT, but the simplest on the Raspberry Pi is probably Mosquitto. We will install this on the Raspberry Pi first:

sudo apt-get install -y mosquitto mosquitto-clients

1.2.4 Install Paho MQTT

In order to allow your Pi to subscribe to your MQTT channel and listen for messages, you need to install Paho-mqtt. Tutorial:https://tutorials-raspberrypi.com/raspberry-pi-mqtt-broker-client-wireless-communication/

Install both python 2 and 3 versions. If you use Thonny to test you coe it will look for python3 version. If you run your code in the terminal then it will default to python2. From my experience, life is easier when you install both.

sudo pip install paho-mqtt 
sudo pip3 install paho-mqtt

1.2.5 Install Blinkt Scripts

Blinkt are eight super-bright RGB LED lights that you can add to your raspberry pi and control from python scripts. Each pixel on Blinkt! is individually controllable and dimmable allowing you to create gradients, pulsing effects, or just flash them on and off like crazy.

Blinkt Python Scripts for SkinnyPi:
https://github.com/jazmy/raspberrypi-skinnypi

Blinkt Python Code Examples: https://github.com/pimoroni/blinkt/tree/master/examples

Create a folder “skinnypi” /home/pi/skinnypi/skinnypi.py

Place your audio files in wav format in that folder. It’s best to name your audio files: 1.wav, 2.wav, 3.wav, etc…

Make sure your audio is set to “analog” If you have your raspberry pi connected to a monitor via HDMI then it will default to playing the audio on HDMI. To make that change, right click on the audio icon in the upper right hand corner of the raspberry pi OS.

Execute the file

sudo python skinnypi/skinnypi.py

1.2.6 Setup your Blinkt Scripts to Automatically Run on Boot

https://learn.sparkfun.com/tutorials/how-to-run-a-raspberry-pi-program-on-startup#method-1-rclocal

sudo nano /etc/rc.local

This is tricky but you need to ensure that you wait 10 seconds to give your pi enough time to boot and connect to the network before you run your script. This will create a log file if there are problems.

sudo bash -c '(sleep 10;/usr/bin/python3 /home/pi/skinnypi/skinnypi.py > /home/pi/skinnypi/skinnypi.log 2>&1)' &

Part 2 — The Cloud Server

If you are looking for a managed web host, with easy laravel site creation, then I highly recommend Cloudways. Cloudways will setup a laravel site and mysql database in minutes.

2.1 Laravel App

Requirements

• Laravel 5.7
• MySQL
• Laravel Authentication — php artisan make:auth

Dependencies Included with Package

This project includes Fitbit Provider for OAuth 2.0 Client https://github.com/djchen/oauth2-fitbit

Installation Instructions

This repository is an entire Laravel site and not just a package. It takes a couple steps to install but I will try to make it as simple as possible.

composer require jazmy/laravel-skinnypi

Note: The package will automatically register itself using Laravel’s package discovery feature for versions 5.6 and above. This means you do not need to update your config/app.php file.

Step Three:

We need to add the additional database tables so run the following command

bash php artisan migrate

2.2 Setup a Fitbit App (client id & client secret)

Fitbit API Documentation: https://dev.fitbit.com/
Creating a Laravel fitbit application is really tricky but here are the basic steps for how it works:

• The user authorizes the application
• Fitbit returns the user back to the callback url with the authorization token

Example Callback URL: https://your.domain.com/authorize When you setup the fitbit application you need to give it a callback URL that it will send the token. We MUST store that authorization token so that we will be able to make subsequent requests on behalf of the user later.

• We use the id of the authorization token that was stored to create a subscription for the user on our subscriber endpoint. That means every time a change happens, fitbit will fire your callback page.

Example endpoint subscriber URL: https://your.domain.com/callback

• Once the subscription is complete, the user sees a message on the screen saying that their subscription has been created or that the subscription already exists if it already does.
• When the user’s weight data comes into Fitbit, fitbit will notify our application by calling the subscriber url.https://your.domain.com/callback
• Fitbit only NOTIFIES us that the user weight data has changed. It does not send us the actual weight.
• In the getCallback method in the FitBitController, we have to quickly save the notification data that was sent to us and return a 204 response to fitbit so that fitbit will know that we received the notification.
• There is no way the application can receive the notification data in the callback, fetch the notification details from fitbit, parse the notification, fetch the affected user, save the new weight log, do the weight comparison check, send the requested mqtt message and reply fitbit with the required ‘HTTP 204 No Content’ status all within 3 seconds in the callback. So you need to break it down into a couple steps to prevent fitbit from banning your api access.
• The solution is to setup a Laravel queue listener so that it is always running. When fitbit triggers the callback you add a job to the queue. The queue listener should process the notification within 5 seconds or so of receiving the push from the callback.
• ‘Supervisor’ is a process monitor program that will monitor the queue listener process and automatically restart it in case the process crashes.

The job you create will do the following:

• fetch the notification details from fitbit
• parse the notification
• fetch the affected user
• save the new weight log
• do the weight comparison check
• send the requested mqtt message

Then we will run the http://your.domain.com/notification-detail/notification-details route to go through the notifications we have saved and create a request to fitbit to fetch the details of each of those notifications. It is in the response that we get from fitbit that we will have the weight that has changed.

The weight data will then be written to a file inside the storage/app/weight_logs folder.

Part 3 — MQTT Server

MQTT is a way to push and subscribe to messages and works well on Internet of Things (iOT) devices like a Raspberry pi. You setup the Pi to subscribe to a MQTT channel so whenever your server pushes a message to that MQTT channel then MQTT will automatically push it to any device subscribed.

MQTT Introduction: http://www.steves-internet-guide.com/mqtt/

If the user has lost weight it should send this json to MQTT:

$message = { "color": "1", "style": "1", "seconds": "10", "audio": "1"}

If the user has not lost weight it should send this json to MQTT:

$message = { "color": "2", "style": "2", "seconds": "10", "audio": "2"}

*Disclaimer - Links to recommendations, in this article, include my affiliate code.