One of the most crucial aspects of sustainable development is dealing with waste. After visiting a waste treatment plant in Richmond, we found that the most challenging issue in waste management lies in sorting mixed waste.

In other words, the key problem is how to identify and sort waste that has already been mixed together before it can be processed and recycled. For example, if wood is kept with wood, and nails are kept with nails, different machines can be used to handle them separately, enabling recycling. But what happens when they are mixed together? If nails are nailed into wood, how do we deal with that? This is a major challenge. In fact, the recycling rate of waste treatment plants in Canada is less than 60%, with much of the work still relying on manual sorting. Using AI in this field has been researched, but never implemented.

Our solution? An AI powered trash bin. Our prototype includes a camera, a processing chip, and a few LEDs. By holding the item to be sorted up to the camera, the AI can identify up to 47 types of garbage. This information is then simplified into an arbitrary number of categories (paper, recycling, landfill, organic, etc.) The LEDs, each corresponding to a category, then light up according to this information. Currently, our prototype uses a Jetson Orin Nano as the processing chip, but future prototypes will use a Luckfox Pico Pro for mass deployments. Our AI model is based on the YOLOv11 architecture designed for classification.

As of now, we are still in the testing phase. We hope this will be the beginning of our exploration of AI waste sorting. Our plan is to implement our product into schools where they will educate teachers and students, developing a noticeable impact on our community and producing environmentally conscious young leaders.