InnoTank

Inspiration

The problem we wanted to tackle is the lack of recycling. Recycling resources is one key aspect of sustainability. Water bottles are not recyclable when there are labels on it or contents inside. So, we made a fun, smart garbage bin that promotes sustainability through the fact that people simply love chucking things into the garbage bin, especially water bottles. Simple, fun, and convenient.

What it does

We made a fun, smart garbage bin that promotes sustainability through the fact that people simply love chucking things into the recycling bin, especially water bottles. Throw a water bottle at it, and it will suck it in. Then, it will separate the content, label, and the bottle itself and stores them.

How we built it

link Our design utilizes machine learning in order to recognize the patterns of different label positions on water bottles in order to rotate the desired amount to remove the label. link This is a top-level design of the inside of the trash can. It separates trash from the bottle if the necessary conditions are fulfilled from the machine.

Challenges we ran into

Creating a pre-trained model from scratch takes a substantial amount of time to implement, and wasn't realistic for the time frame we had to finish this project. In addition to not finding a dataset that matched our needs and not having a fourth group member, we had to divert many tasks appropriately in order to code our solution.

Accomplishments that we're proud of

We managed to use Google Collab to implement an already found pre-trained model to detect water bottle. PyTorch, an open source machine learning framework, was also integrated into our project to help our Innotank detect accurately.

What we learned

We've learned how robust machine learning can be! There's a lot of time that goes into coding machines to perform logic for us instead of coding it ourselves. Additionally, we learned how to use numerous other software environments and open sources, such as PyTorch and Google Collab, in order to bolster our machine learning goals.

What's next for InnoTank

We want to open to door for InnoTank to be able to recognize other types of recyclable waste, such as cardboard with paper (and also false positives like soda cans) in order to be able to make our product more mainstream. Right now, we have the main coding base finished for our product, and all that's left is to expand its capabilities.