IoT + Recycling: Simplifying Waste Sorting for Households
Developing a prototype using IoT technology to enhance household recycling. The goal was to develop a solution for sorting and disposing of waste, simplifying the process for users.
As companies place more responsibility on consumers to recycle, this system leverages IoT to make recycling more intuitive, empowering households to contribute to environmental protection.
Role
Me
Aashika Parakh
Macy Bosworth
Arduino Uno
Ardunio IDE
Visual Studio Code
ml.p5
Problem Space
The current household recycling process is often inefficient due to contamination and lack of user knowledge. Many consumers are unsure of what can be recycled, leading to high contamination rates and ineffective recycling efforts.
The goal was to create a functional prototype that leverages IoT technology to automatically sort waste and recyclables, reducing user errors and improving recycling efficiency.
Results
We successfully developed a functional MVP using an Arduino circuit, servo motors, and an object detection algorithm with ml5.js. The system utilizes a computer's camera to identify whether an item is recyclable or waste. Once classified, the Ecosorter opens the corresponding door using servo motors to automatically sort the item.
We were then able to test with different users and found a high accuracy rate* when sorting
*The object detection model was trained on a limited database of recyclable objects
Process
Research into Problem Space
Secondary
Research
Activities: comparative analysis of existing technology
Outcomes:
Identified key factors for recycling technology:
efficiency
cost-effectiveness
environmental impact
practicality
Existing Solutions:
Solar powered trash compactors
Recycling sorters
E-waste kiosks
Interviews
Goal: understand typical recycling habits
Outcomes:
Identified pain points
time-consuming
sorting trash and recycling is difficult
cleaning recyclables
unsure of what's recyclable
two bins take up space
Digital Prototyping Technology
Goal:
decide on most practical digital technology to implement within our prototype
understand functionality in higher fidelity
Outcomes:
Object detection ML:
automate the recognition and classification of items
camera captures the item in real-time and compares it to the ML model
Arduino + Serial Communication
using arduino UNO and servo motors to physically prototype the sorting mechanism
serial communication allows real-time classification (ML) + sorting (arduino)
Process
Design + Prototyping
Sketching
1. Analyzed the pros and cons of different design elements
convenience
efficiency
environmental impact
Low-Fi Prototyping
1. Analyzed the pros and cons of different design elements
Object recognition
Pros: convenience & efficiency, automated recognition & sorting
Cons: object recognition tech is expensive, recognition errors
2. Decided on a design that balances:
convenience
efficiency
environmental impact
Mid-Fi Prototyping
Next, I created a 3D model using Sketchup to visualize and iterate the sorting mechanism
One issue identified from concept testing that we focused on was the sliding panels for the sorting mechanism. We used the model above to help us visualize the sliding panels. However, we noticed functionality errors that wouldn’t allow proper sorting and movement of the doors. Therefore, we used another program called Tinkercad to help us iterate this mechanism.
Process
Digital Prototyping
Object detection using ml5.js
ML5.js provides a library of machine learning algorithms and models that have been built off of TensorFlow.js. This database requires no additional dependencies to be downloaded or a virtual environment to be created
Arduino + Serial Communication
After figuring out how to detect trash and recycling with p5 + ml5.js, we wanted to better physically represent the movement of the moving doors with the Arduino. Within our p5.js code, we added serial communication to enable connection between the object detection code and our arduino.
Arduino circuit on tinkercad
Testing with users
Showcase
We were able to have users use the digital prototype during our end of the year project showcase, which allowed us to test the object detection and physical prototype together. We gained valuable feedback and identified potential limitations and areas for improvement.
This project was a comprehensive exploration of designing a smart home recycling assistant. From initial research to final prototype, the team faced and overcame numerous challenges, resulting in a functional MVP.