TRASH.PY - A SMART SYSTEM ENSURING PROPER WASTE SORTING AND ELIMINATING USER DECISIONS

Abstract

For proper waste processing, the trash cans at the University of Michigan need to separate waste into different categories. These categories include landfill and recycling, and sometimes some others. The current burden for separating this waste is placed on the trash can user, with different labeled receptacles being provided for the user to decipher and manually separate their waste between. Relying on users to properly classify their waste causes problems. For instance, some users do not care enough to throw their waste into the proper receptacles, resulting in the recyclables becoming contaminated, and potentially recyclable items being taken to a landfill. Additionally, one of the receptacles could become overfull, resulting in the users just throwing all their waste into the only available receptacles. The recycling system at the university also differs from that in Ann Arbor and leads to confusion. Since most students and faculty live off campus, they are used to doing their household recycling for the Ann Arbor standard. We aim to address these problems by redesigning the trash cans at the University with an autonomous waste separation system. Removing the decision from the user on whether their item is recycling or landfill and therefore reducing the contamination in the bins. The bin will be able to detect whether an item dropped into the bin by a user is meant for landfill or recycling, and then navigate that item to the proper storage receptacle contained within the bin. This system will be based on a computer vision model that will be trained to understand the proper sorting standards at the university. We aim to have a trash can that yields more valid/clean recyclables than the current system at the University.