FireWatch AI

Inspiration:

This project was inspired by the California wildfires. We have noticed that the fires caused a lot of damage and destruction around California and effected many different individuals personally. Seeing this impact affect so many individuals, we wanted to develop something that could possibly be utilized by citizens in danger of future wildfires, in order to mitigate possible harm done by these wildfires and overall, provide people with information to be better prepared.

What it does:

We developed this interface including both a map and an AI chat box to support the users of this website and give them intel about fires near their vicinity. On the map there are markers to pinpoint locations of the fires based on the data provided by the API powered by FIRMS, this data being the latitude and the longitude coordinates of the fires. The AI chat box is for users to ask a specific question about the fires around them and potential shelters to which they can refugee at. It informs the user about their location based on provided longitude and latitude coordinates.

How we built it:

To build this website we've used various API's. The API's we have used in our project are Google Maps, NASA, and Gemini AI. We used the API of Google maps to locate where our user is based off of their longitude/latitude coordinates, and to allow us to display where the fires are on the map. Then we used the API from NASA to get the longitude and latitude data of the wildfires, and based off of this data, we would be able to accurately display it on the map. Then the API from Gemini AI was used to create the chat box feature we have in our project. We fed the AI information about the wildfires fetched from FIRMS API, and the location of the user, and gave Gemini system instructions so that it answers the prompt given by the user while staying on the topic of wildfires.

Challenges we ran into:

Some problems we ran into was finding an API that provides data about the wildfire locations. We were surfing on the web to find an API for the data we needed for our project.

Accomplishments that we're proud of

We are most proud of training the AI for our website and situation, and feeding the relevant data and context for it to respond appropriately and to our needs.

What we learned

We learned how to construct the AI to respond off the data we fetched from the API, and for it to respond more like a wildfire assistant.

What's next for FireWatch AI

We planned to add a predictive that better assists the user, and know what to expect in a future time frame. This will be supported by weather, and machine learning to reinforce its predictions.