Inspiration
We all do our best to recycle. When presented with the classic quartet of "bottles/cans", "paper", "compost", and "trash", we're happy to help out. But what about at home? What happens to the waste we create, when we don't have the space or time to commit to sorting it? About 76% of recyclables in the US households are lost to trash, meaning only a small fraction of our recycling efforts are used again towards new products. We wanted to create a solution that made recycling our waste easy and practical. Our solution is Sortify, a waste sorting solution that can be integrated into our everyday lives.
What it does
Sortify sits in the same footprint of a normal home trashcan. A computer vision system automatically classifies items placed on the lid into one of four internal bins: trash, plastic, paper, and soda cans. A differential gearbox tilts and rotates the main plate to redirect items to their respective zone.
How we built it
The differential gearbox was designed using OnShape and most of its components were 3D printed. Two stepper motors control the pitch and yaw of the plate. The relationship of the two gears controls the motion of the top gear, resulting in a compact system. We chose to use an ESP32 as the controller because it allows for precise stepper motor control and also has WIFI which allowed for us to easily communicate with it from our other components. We have a 12V input into our breadboard and later buck it down to 5v for all our control logic, we did this in order to easily share the grounds between all our electronic components to make sure communication signals work well. The vision structure was made using 80/20 extrusion and also acts as a handle.
Challenges we ran into
After we built the differential, we ran a stress test for over 10 minutes to make sure it would be able to handle the larger weight of the plate. After the test, we noticed that both of the motors had rounded out their bores on the bevel gear. To fix this we revised the design of the bevel gears to use a metal insert to spread the load of the motor across a larger area, in order to prevent the motor from rounding out the gear.
Accomplishments that we're proud of
Overall, we're proud to have created a high-quality product that matched our expectations! The differential worked smoothly, and the vision integration with Gemini was both fast and reasonably accurate for our first prototype.
Log in or sign up for Devpost to join the conversation.