Inspiration - It is an innate human instinct to "Take Care". Be it our family, friends, pets and everything else on Earth. We are all in this together and we need to take care of it. The growing list of challenges we as humanity are facing is predominantly because of exploiting our environment. This project is one step in that direction in reducing our footprint.

What it does - The Can I Recycle This Chatbot, will help citizens answer questions regarding product recyclability. Users can upload a photo of the product or ask directly in the conversation if a product/item is recyclable. Based on their location and other factors the chatbot will answer if the product is recyclable or not. Every County's in the US has different recycling laws and hence we need the user's location to give an accurate answer about a product's recyclability in their county.

How I built it - Facebook's Messenger ChatBots have been a blessing to get this idea off the ground and reach our potential users easily. Components like the "Generic Template" have helped us visually communicate to the user their next action minimalistically. The "Persistent Menu" component from Messenger has helped us build an easy to use navigation menu for our users. Users can also upload an image of a product to check recyclability. In such cases, we parse them to an Image Recognition Engine (Cloud Sight) to understand the product and provide recyclability results accordingly. We used "Node Js" to interact with Facebook Messenger API. Our database is built on "Algolia" & "Firebase" to index recyclability data in real-time. User intents in the messenger conversation are identified using Google's Dialog Flow Engine.

We have built a growing database of products/items and categorized them based on We have a team of environmental engineers who bring environmental domain expertise in analyzing if a product is recyclable or not.

Challenges I ran into - Curating the database of all products from scratch was a critical challenge. To be able to answer questions that pertain to all products, we needed to maintain all kinds of products in our database. To do this we came up with a novel way of organizing data (Sorry, but we cannot mention it here since it is our IP) that relatively made the task less complex. We also ran into challenges with the accuracy of the Image Recognition Engine. To overcome this we built a separate training algorithm that helped us increase the accuracy.

Accomplishments that I'm proud of - While we are taking baby steps to increase recyclability and ensure people do it correctly, we are happy with the response we have been getting. Citizens are quickly adapting this tool to ensure they make correct recycling decisions. When a product isn't recycled correctly, there is a cost associated with it both financially and environmentally.

What I learned - The complexity in pushing citizens to recycle has always been challenging for governments throughout the world. Developing nations find it even harder to execute this since sustainability comes with a price when there is no proper infrastructure. However, the recent penetration of the internet and mobile devices have made it easier for governments to reach their citizens and provide them with accurate data. Until a few years ago such outreach programs were expensive and inefficient. In the process, we learned how we can help governments achieve their sustainability goals by using cutting edge technology, at a lesser cost and more efficiently.

What's next for Can I Recycle This - We are currently providing this service in Athens, GA and the team is actively expanding to other cities. We are also looking to bring this solution to What's App, Instagram, and other voice assistants.

Share this project: