Haloe

Haloe protects users by finding the safest walking route using AI and crime data. Additionally, Loe, the AI companion, can chat with users to ease anxiety, making solo walks less stressful.

Comment

Inspiration

No one wants to walk home alone, especially at night. That’s why, for this hackathon, our team sought out solutions to the uneasiness and fear one might feel when navigating the streets of a dangerous city. We wanted to take the knowledge and insight that a local of a city might have and give it to any user of our app to help them make informed decisions on how to get home safely.

What it does

Haloe is an app that protects its users by finding the safest walkable path from their location to their desired destination. It does this by gathering tens of thousands of crime data points for each city and employing an AI model to determine the most secure route home. Along the way, our AI companion, ‘Loe,’ can talk to the user at their request to help them feel less alone and to ease their worries. Overall, Haloe aims to reduce the stress of those walking alone and helps the time pass faster.

How we built it

We used React as our front-end, allowing us to get our UI up and running fast so we could focus on the functionality of the app. In order to find the safest route home, we needed crime data. We pulled ~40,000 lines of data from the city of Chicago in order to figure out which streets were safe and which ones were not so safe. Then, using these scores, we could map any route in the city using a maps API, our data, and AI to find the best route. To build ‘Loe,’ we used ChatGPT to feed in the prompts that the user selected. Then, we fed the stringified response into a TTS API to get the final audio.

Challenges we ran into

There were a lot of challenges in building this app, the most prevalent of which was resolving merge conflicts. Working with four people all at once on a project is difficult, and when our editing of files overlapped, it often created many issues that took away precious time. Additionally, using the data to update the route was difficult, and we ended up having to implement our own server with Node JS to fix that issue.

Accomplishments that we're proud of

We finished a logically complete app in a span of 48 hours Our pathing algorithm successfully avoids dangerous areas of Chicago, as verified by locals we know Our app has a simple, nice looking, and intuitive UI that has minimal button presses after initial registration

What we learned

Mostly, we learned how to work together as individuals with different skill sets and personalities. We all had never worked together on a hackathon before, and half of us had never even completed a full project for a hackathon before. Also, we are composed of one senior, one junior, and two sophomores, all of various programming levels, so for the more inexperienced of us, we learned how to hit the ground running in a real project and keep our heads above water. And for the more experienced, we learned how to take more of a mentorship role and be patient when mistakes were made.

What's next for Haloe

We did not have access to premium GPT or TTS models in order to make ‘Loe’ feel more real. So naturally, we want to implement the newest technologies in both of these areas (and maybe some nice animations) in order to hammer home the idea of ‘Loe’ being a real companion. With a better model, the stories and jokes will be better. We also want to make individual directions for the user, as opposed to just a path on a map. This will make navigation even easier. Finally, we want to expand this to more cities than just Chicago. Doing this would require massive computing and tens of millions of points of data, but doing this would make the app usable for everyone, and ease the burdens of people all around the world.

Built With

Share this project:

Updates