Harrison Bradley, Jeff Bertel and Marc Welsh studied on the Advanced Mechatronics course at Georgia Tech. They’ve created an apparatus which takes in a piece of waste and then automatically sorts it depending on what the object is identified as: cans to the left, bottles to the right, landfill to the centre. It is currently a prototype but they have a patent pending.

They decided to go with ‘machine vision’ as their approach. A photograph is taken and then fed back to a Raspberry Pi for processing. They wrote some AI software which could be “taught” what objects were which. The system has an integrated 10-inch touchscreen monitor and a program running on the Raspberry Pi to display statistics and status data.

Once the waste is identified, the bin moves to the correct slot and the waste is deposited in the correct sorting container. This was done using stepper motors controlled by an Arduino Mega. They have implemented various safety features using solenoids to stop, for example, anyone putting their hand through the hatch while the mechanism was moving.

Much cheesey music abounds in this promotional video for the invention.