FaunaSafe

DEMO VIDEO - HUGE THANKS TO ANTHROPIC AND CLAUDE - we use Anthropic to power the response suggestions after the Monte Carlo simulation and we think this is a really powerful use case. https://youtu.be/1BlVI6K2Whk

Why we chose our project

When communities are evacuated during wildfires, wildlife is often left without protection. Emergency resources are frequently deployed too late, and these delays significantly increase ecological damage. Our approach focuses on reducing response time to minimize harm to ecosystems.

Two charities that inspired us to pursue this project are:

World Wide Fund for Nature (WWF)
Wildlife Aid Foundation

What our project does

FaunaSafe provides a map view of the world, where a focal point of a forest fire can be picked to begin a simulation. Upon selecting a focal point, local wildlife is added to the map and the fire begins. As the fire spreads, impact on the local wildlife is evaluated, and is reported for each simulation step. As the simulation runs, the user has the option to place fire breaks. These limit the spread of fire, allowing forward planning of methods used to prevent the spread of fires.

How we built it

At the start of our project, we decided on a web stack consisting of React JS with Vite on the front end and fastAPI on the back end. The fire simulation is a grid-based Markov chain with feature based transition probabilities, such that

$$ \mathbb{P}(\text{ignition}) = \sigma (\beta _0 + \Sigma \beta _i f_i) $$

Where $ \beta $ are parameters to the model, and $f$ are the features of a grid cell.

The animals are modeled as groups with species specific characteristics. Each species has a speed, fire detection distance, habitat modifier and stamina modifier. Different species are also given a trait from flee, caution, territorial and herd where their levels of alertness differ.

Using data from the OSM map API, the front end ties all of this functionality in a a neat and user friendly interface with some impressive data visualization.

Challenges we ran into

Our team had a very limited knowledge of front end development coming into this project. We assumed it would be a trivial task of just "vibe coding" all the UI. Little did we expect the complexities behind a system like React JS is more than AI was able to consistently manage. This put a lot of pressure on the team to adapt quickly and come up with an acceptable solution despite our limited knowledge.

Accomplishments that we're proud of

The scope of this project was difficult to manage. There were times we thought that we had been too ambitious. It took a lot of stress and a total sacrifice of sleep for our project to be presentable. I'm proud of the entire team, for persevering despite the difficulties and doubt experienced on the way.