The Sorting Centers are not the problem, we are. We need to sort our trash better; and what better way to do it by letting someone.. I mean something else do it for you ?

The team decided to try their first hardware hackathon, combining the green hacking challenges of Accenture and Telus, as well as an IoT hack.

Smartbin is a smart trash can solution that uses image recognition to analyze the about-to-be garbage item and redirects it to the correct garbage type.

What it does

The project is a hardware solution for stopping people from throwing their garbage in the wrong category, teaching them about their mistakes, and for making earth a better place to live by minimizing pollution and optimizing recycling.

How we built it

On the software side, three servers are necessary:

  • one for analyzing the uploaded picture from a cellphone;
  • one for parsing the data sent after the picture's analysis
  • and send the correct information to the third server, which is the hardware.

On the hardware side, all materials used for assembling Smartbin are from what the team has found near the hacking activities: soft drink cardboards, recycled papers, and even utensils from meal times. As for the components:

  • 1 Arduino Uno
  • 1 small breadboard
  • 1 ultrasonic sensor
  • 1 IR sensor
  • 1 IR remote
  • 2 servo motors

An electrical diagram is included in the image gallery!

Challenges we ran into

Our hardware project heavily relies on the Qualcomm DragonBoard 410c. Unfortunately for us, the DragonBoards were running an Android OS and didn't allow us to progress any further. It would've been possible to flash the OS back to a Debian OS in order to host a server.

We did not attempt to flash the OS to Debian since the hardware was lent from MLH.

Because we really wanted to work with hardware, we decided to go with an Arduino Uno. However, the downside of using an Arduino instead of a DragonBoard is that there is no internet connection: it won't be possible to send data to an Arduino without a Wifi development tool.

Android and Bluetooth connection is hard. Android, Bluetooth connection with arduino.. even harder. Got a lot of connection refused.

Accomplishments that we're proud of

Having an MVP. Familiarizing ourselves more with Google Cloud Console. Being able to learn to use Arduino and use most of the components.

What we learned

  • Google Cloud Console, how it works, what tools it offers.
  • Docker, deployment hero.
  • Designing hardware is awfully hard. We definitely did not have the necessary training for it (we keep forgetting about gravity ...)

What's next for Smartbin

Fully experience a DragonBoard with the correct OS and unleash the full potential of Smartbin. WITH SOME DATA VIZ HOPEFULLY!


Special thanks to the SnapTrash team who developed a SnapChat lookalike app using machine learning and image recognition for sorting trash and for locating the nearest trash / recycle / compost bin in Montreal.

Share this project: