FitFinder

Inspiration

Our team was inspired by the iconic movie, Clueless, where the main character, Cher, has cutting-edge closet technology. Every morning, she can input the style she wants to wear into her computer, and it will suggest the best outfit for that day.

When people sleep in, they wind up rushing in the morning, often literally getting dressed in the dark. FitFinder takes the guesswork out of it. It ensures that when you choose to prioritize sleep over style, your outfit doesn't suffer like your punctuality.

What it does

FitFinder is a gpt-5-mini wrapper that has a detailed description of your wardrobe, including unique tags for outfits that describe their aesthetic and color. The browser app pulls user descriptions of desired aesthetics, occasion, and other factors relevant to proper styling, throwing that data into gpt-5-mini, and returning the style tags that best suit that request. The program matches the tags to the existing clothing items in the wardrobe to find the most suitable and stylish outfit for any occasion. This technology elevates the everyday morning to new heights, taking the everyday inconvenience of early-morning decision fatigue and eradicating it entirely.

Once the items are identified, their locations are sent to our Arduino and motor, which spin the clothes rack to give you the identified pieces!

How we built it

Our team divided into a hardware and a software team. Our hardware team (consisting of Mechanical Design Lead and Electrical Hardware Lead), and our software team (consisting of our AI Research Lead, Software Lead) divided the work for the product depending on the main discipline.

The hardware team performed extensive research into stepper motors, motor controls, limit switches, Arduinos, as well as mechanical design and manufacturing processes. The hardware team utilized programs such as SolidWorks to visualize and provide design basis for the manufactured components.

The software team also performed extensive research, both independently and through guided seminars, on how to incorporate AI agents (primarily OpenAI) into our programs. We decided on gpt-5-mini for our application. Then, the software team collaborated to code the application in Python, utilizing VS Code. The software team also worked on integration within our browser app, communication between the Arduino and the program, and fine-tuning the rotation of the motor once it was completed.

The base of the rack was built with plywood. In the final design, the stepper motor spins the central shaft and closet rack. The 3D printed components were all made of polylactic acid (PLA) filament.

Challenges we ran into

The primary challenge that the team ran into was with electrical hardware malfunctioning. We fried 4 motor drivers and one Arduino, discovered a faulty breadboard, and wound up needing to pivot several times.

Accomplishments that we're proud of

The team is incredibly proud of the leaps that we made in terms of utilizing LLMs in our program to interpret user input, the ingenuity that we called on in order to come up with new and adventurous solutions to get the system to rotate, and the way that the team rallied through exhaustion and frustration to make it to the end.

What we learned

The entire team became familiar with Python programming, as debugging the code was an all-team task. We learned a lot about stepper motors, their limitations, and the value of using new and functional parts. Each member of the team was brand new to every electrical and software skill utilized in this project. We rapidly learned and applied our new knowledge every step of the way.

What's next for FitFinder

FitFinder has some exciting plans in the near future! With a little more time, the team would have liked to utilize a more powerful stepper motor and driver module to improve the performance of the system. Additionally, the team has aspirations to expand the number of clothing items that can be included in the system, exploring opportunities to incorporate shoes, belts, hats, etc.