Carbin

Encourage student participation in responsible waste management with smart bins that guide proper disposal.

Comment

Problem 🌍

Buildings are significant contributors to greenhouse gas emissions, with three main sources: electricity, natural gas, and waste. While efforts to reduce energy-related emissions are ongoing, waste management has been an overlooked area. In the US, landfills contribute over 15% of methane emissions, nearly 3 times more than coal mining. Alarmingly, 75% of landfill waste could have been recycled, while 25% of recycling loads are contaminated and end up in landfills.

🚨 These statistics reveal a critical missed opportunity for waste reduction and environmental protection.

Inspiration 🌱

Our inspiration stems from the potential of merging artificial intelligence with everyday waste disposal while addressing overall building emissions. We aim to make building-wide emissions management, especially waste handling, more:

  • Intuitive 🧠
  • Accurate ✅
  • Engaging 🎯

By developing AI-powered smart bins and a holistic emissions tracking platform, we strive to:

  • Simplify the recycling process for users 🔄
  • Boost awareness and education around proper waste sorting and energy usage 📚
  • Deliver real-time feedback on recycling habits and overall building emissions 📊
  • Create a data-driven approach to optimize waste management and energy use strategies 💻

Our project bridges the gap between intention and action in recycling, empowering people to make eco-conscious decisions effortlessly. By harnessing technology to make recycling more accessible, we aspire to significantly reduce landfill-bound recyclables and curb methane emissions.

What it does 🚀

Carbin is a two-pronged solution designed to revolutionize responsible waste disposal:

  1. Smart Bins: Equipped with friendly interfaces and AI signage, guiding users on proper waste sorting. 🗑️
  2. Management Dashboard: Empowers building owners and organizers to track their carbon footprint reduction and access detailed emission statistics. 📈

How we built it 🛠️

  1. AI-Powered Sorting: Trained a Roboflow machine learning algorithm to accurately identify and categorize different types of waste.
  2. Real-Time Data Logging: Smart Bins automatically update to our Azure Cosmos database, providing insights on bin activity and usage patterns.
  3. Other Emissions Tracking: Recognizing that waste is only part of building emissions we also built in electricity and natural gas emission tracking. We offer two choices
    • Monthly tracking: Utilizes Azure Document Intelligence to quickly process gas and electric bills.
    • Yearly tracking: Leverages Azure OpenAI for comprehensive analysis.
  4. Web Dashboard: Deployed a user-friendly webapp showcasing simple graphs and logs of building's emissions and waste data.

Challenges we ran into 🧗

While developing Carbin, we encountered obstacles in:

  • Turning pdf files into a form that is readable by Azure OpenAI without destroying complex table structures.
  • Attempting to train/serve Roboflow through Azure.
  • Finding and labeling hundreds of trash and waste related images.
  • Deploying a docker instance on Azure a to host our nextJS web app.

Accomplishments that we're proud of 🏆

  • Getting a vision model to run on device in the browser to drastically reduce deployment costs.
  • Creating a clean and professional user interface.
  • Our pipeline that turns complex electricity/gas billing PDFs into data points without the user doing any tedious data entry.

What we learned 🧠

  • How to use Azure Cosmo DB in place of Firestore.
  • How to send and receive files and not just text data for our own backend APIs.
  • How the canvas browser API can be used to draw simple shapes like bounding boxes without css.
  • The importance of waste management!

What's next for Carbin 🚀

  1. Cost-Effective Scaling: With projected upgrade costs under $50 per bin, Carbin offers a cost-effective solution for widespread adoption. But we need to go from proof of concept in the web browser to cheap iot hardware like a Raspberry Pi or ESP32.
  2. Enhanced AI Accuracy: Continual improvement of our waste identification algorithms for even greater accuracy. And possibly even more detailed carbon estimates by estimating trash size/weight and not just classification.
  3. Gamification: Introducing competitive elements and rewards to further engage users and promote positive recycling habits.
  4. Integration with Smart Cities: Partnering with municipalities to incorporate Carbin into broader urban sustainability initiatives.

By making responsible waste disposal accessible, engaging, and data-driven, Carbin aims to create a significant impact on reducing landfill emissions and promoting a more sustainable future. 🌱🌎

Built With

Share this project:

Updates