-
-
SafeSphere Logo
Inspiration
Just a year ago, Kevin, I, and a few other friends were being driven home by my father from a friend's birthday celebration. Nothing seemed serious, but the weather was bad. Our visibility decreased as the rain worsened on the way home. We continued because Google Maps indicated a clear path. Then we came upon this completely flooded, low-lying area. We were frightened in the backseat as the car became stuck and began to drift. We were saved by emergency services, but the emotion stayed with us. We knew we wanted to create something that would help others avoid the near-catastrophe we went through when we were choosing a Technica concept.
What it does
SafeSphere is a responsive mobile web app that enables users in New Jersey to stay safe during floods by synthesizing flood-aware navigation, real-time alerts, and early risk prediction. It enhances Google Maps with Smart Routing, showing flood-risk markers, road closures, and alternative paths, while its integrated AI Assistant is capable of responding to messages like "I see water ahead" and immediately routing users around the flood zone for safety. A gradient boosting machine learning model predicts flood risk 2-6 hours in advance based on data from USGS stream gauges, National Weather Service rainfall forecasts, seasonal patterns, and known flood-prone areas. SafeSphere updates live every 30 minutes, with a Community Safety Board for users to report flooding, blockages, or weather warnings; it also has a secure login system that saves user settings and enables participation in community updates.
How we built it
We got real-time data from USGS stream gauges and the NWS API, and we developed SafeSphere using Flask and Python for the backend. On the responsive mobile experience side, we used JavaScript, Bootstrap, and Google Maps API.
Challenges we ran into
One of our biggest technical challenges was combining several data sources like Google Maps, NWS alerts, and USGS stream gauges into one system. When testing, we had to adjust our data fetching strategy due to API rate limits. It took multiple iterations for user authentication and session management to function consistently.
Accomplishments that we're proud of
The accomplishments that we're proud of are developing a flood prediction system that really does have the potential to save lives. Mainly, we used our own experience of becoming stranded in floodwaters to create a tool that can shield others from the same danger.
What we learned
While building SafeSphere, we had to figure out how to use environmental data APIs and make a prediction model even though we didn’t have a ton of training data to work with. We also learned that an emergency tool has to be incredibly clear and easy to understand, because in a stressful moment, users don’t have time to think twice. Making the app work smoothly on different screens taught us how important accessibility really is. We also saw how much of a difference language support makes when you’re trying to help whole communities, not just a single group.
What's next for SafeSphere
In the future, we want SafeSphere to work outside of New Jersey. A lot of other places deal with flooding, and we want the app to help them too. We also want to bring in more kinds of data, like traffic cameras or quick reports from people on the road, so our updates are as accurate as possible. Another thing on our list is building an actual mobile app with offline mode and push alerts, since people don’t always have service during storms.
Log in or sign up for Devpost to join the conversation.