-
-
(body) a structural breakdown
Inspiration
After an event, perhaps after this event too, the various trash and recycling bins around CULC and Klaus are inundated with both trash and recycling. It can't be helped with so many people. Unfortunately, that leads to trash in the recycling bins and recycling in the trash bins, defeating the purpose of having four bins in a row and rendering them all trash bins. Moreover, some items are hard to classify which bin they should go in and often times, items are thrown in all together under the name of recycling, leaving no opportunity to sort them out.
Fear not, we've prototyped a solution: a single bin that sorts itself: the eBin. No longer do people have to think about whether a napkin goes into the mixed paper section or the trash section; all they have to do is throw it in the bin, and the bin does the rest.
What it does
We reinvented the classic trash can, compartmentalizing with wood, four sections for different types of waste: plastic bottles, paper, aluminum cans, and general waste. Above the sections is a rotating filter that turns to where each type of waste should go depending on what the ESP32 Cam sees. The camera communicates through wifi sockets to a computer that analyzes the stream of information in real time and sends back to the smart bin system which section it should turn to.
How we built it
We took a trash can and added electronics and wood and cardboard to it. More specifically, we lasercutted plywood, MDF wood, and cardboard to make the main internal structure. For electronics, we used breadboards, wires, resistors, and jumper cables combined with an ESP32 Cam, Arduino Nano, stepper motor, stepper motor driver, RBG LED to make the embedded system.
Challenges we ran into
As we developed our smart bin, we ran into issues relating to the physical limitations of our breadboard and ESP32. Our arrangement of devices on the breadboard was restricted by its size, and the number of peripherals we could use was restricted by the number of GPIO pins the ESP32 microcontroller has.
Come time to manufacture the bin's architecture, it was a daunting, steep-learning curve for us ECE majors to build from scratch. With just raw material and cumulatively negligible experience, we took to the Invention Studio to lay out and model the compartments, filter assembly, and lid. AJ, the really awesome PI, really pulled through in teaching us the ropes of Solidworks and guiding us through water-jetting and laser-cutting procedures.
On the less physical side, Arduinos are very finicky when it comes to flashing programs onto chained microcontrollers (our ESP32). It was tough figuring out the precise sequence of actions needed to properly load programs in.
On the software side, we initially had challenges setting up OpenCV as many of us had never worked with it. Luckily, documentation and other internet resources helped us push past the initial set up as we learned a smidge about Deep Learning Neural Networks and how to use pre-trained models as classifiers for our recycling bin. In addition, setting up websocket connections was a struggle, likely due to firewalls preventing connection.
Accomplishments that we're proud of
We are proud of creating such a clean yet functional product in the span of a weekend. Not only is the design physically appealing (which definitely surpassed our initial intentions of cardboard-mania), but our CV waste detection performed pretty well. Plus, watching it rotate to the proper section is very satisfying.
What we learned
It was our first time using SolidWorks and EAGLE to make designs. Moreover, it was our first time using lasercutters and waterjets to make the lid and compartments of our bin. Three-fourths of us had had little to no experience with OpenCV. And half of us thought ESP32 was a type of superglue.
What's next for RoboGech
Refining our product
- More refined object detection and specialized sorting.
- Fine-tuning the YOLO model for commonly thrown out recyclable items.
- Developing future versions with different material compartments (like plastic or metal).
- Upgrading the stepper motor to turn faster and more precisely.
- Scaling the product size for commercial buildings or outdoor use.
Future features to consider
- Audio indicators that the bin has configured itself appropriately to the waste.
- Tracking bin fullness levels with ultrasonic sensors.
Log in or sign up for Devpost to join the conversation.