Don't Drive Off!

Promotional Art
Live Gameplay Screenshot
Lobby Screenshot - Create or Join a Game
Character Sprite Assets
Power-Up Assets

Inspiration

Well... we kept getting distracted while trying to learn about new AI technologies. And after weeks of procrastinating at our screens, we came together at VTHacks to build something about it.

Don't Drive Off! is our attempt at building a competitive game to help users focus on their computer screens, instead of on any pesky distractions!

What it does

First and foremost, DDO! is a real-time vision detection app. It tracks users' face and eye motions by sending webcam footage to a backend running Google AI Edge's MediaPipe Solutions.

Our backend keeps track of where users are looking, giving them ambient points for focusing on their screen and taking away points for getting distracted. Our frontend rewards focus with Power-Ups to crush the competition. From "Gotta Pee!" that lets you take a break, to "Attention Grabber" that lets you steal your unfocused friends' points - we reward users for staying Locked In!

How we built it

Our backend is a Python Flask app running on Nginx with a Sqlite db to keep track of things. We use React for our frontend, Tailwind to quickly style it up, and some custom assets to give it its artistic flair. We leveraged OpenAI's ChatGPT and Microsoft Co-Pilot in order to quickly iterate on our designs and get a working site up and running in no time.

We chose a lot of the technologies we used in order to ensure we could get something working by the end of the weekend - Flask and React are easy to use and well-documented. We really wanted to focus on AI technologies - we honed our asset-generation skillsets, we tested how well AI can help us code, and we made the AI solve all the tricky vision detection algorithm stuff for us.

Challenges we ran into

Even though these libraries provide a lot of power, getting them all working (and working together!) is never easy in a weekend. Our foremost challenge was getting the app built, and our team's FrontEnd skills were definitely a source of struggle. None of us know enough CSS...

Device-specific calibration for vision detection is tricky, so our app works best on laptops right now. We'd love for the app to be able to handle any device in a way that's smart - without so much human-training on angles/lighting. This is a challenge with any developing AI model and its implementation in real-world situations.

Accomplishments that we're proud of

We got a full-stack app developed in a weekend, which is always a challenge in and of itself. We're especially proud of getting a real-world AI solutions platform running on our backend server, and we tried to make that "professional accomplishment" as fun and useful as possible for a 2-day project.

Each team member focused on a few key skills they wanted to improve on. Kaushik is the best artist of the group and he focused on bringing those skillsets to a real-time web-app - he knows a lot more React because of it. Shorya is the best with AI tools, but he hadn't gotten one running for multiple users on his web server until now. Aarav has the most full-stack experience, but relies on simple designs because of his art skills. I think we're all proud of the project that we made, and we're happy that we pushed each other to build something cool together. None of us could've made this project alone.

What we learned

The pace of the hackathon really helped us prioritize a minimum viable product and get into gear quickly. Lowkey our code was... not always good - also a worthwhile lesson - but we learned the power of some steadfast teamwork and dedication over 24 hours. We also found some viable ways to cut corners, in order to get a product out the door. While we always want to strive for perfection, we're always working on deadlines and it's good to learn ways to simplify our workflows (even if the quality of the work isn't quite the same - sometimes you need something QUICK!)

What's next for Don't Drive Off!

Sleep? Coffee? Probably not much Venture Capital... smh. those guys don't GET IT. this is a bazillion dollar idea!!!!

In the future, we'd like to see this type of functionality tested in real-time scenarios where a lack of focus can cost real lives. Although this is a goofy proof-of-concept video-game, we truly think that the technologies we've used and the skillsets we've improved can help build AI tech that puts users in control of their focus and their technology tooling.