FireGrid.ai

Inspiration

Wild fires across California have caused major physical, financial, and emotional distress. Wild fires can rapidly spread, especially in certain environments, including water deprived areas. Insurance companies are decreasing coverage or discontinuing altogether due to the high disaster risk. Our goal for this project was to develop a product that is able to take measurements of temperature, soil moisture, and humidity in order to see trends and possibly predict and prepare for wild fires.

What it does and how we built it

Our projects consists of two parts - a hard ware portion consisting of an Arduino board with multiple sensors connected to a Raspberry Pi, and a web application that tracks real time metrics from the sensors, with a robust LLM that uses our metrics to answer questions. The workflow is as follows: the Arduino board tracks metrics such as temperature, humidity, and soil moisture. This data is pushed to the Raspberry Pi, which uploads the data to a non-relational database. Our full-stack web application, which is built on Next.js and various UI frameworks, watches the database and updates the graphs and LLMs accordingly. Additionally, if the temperature goes above a certain limit, a push notification is sent directly to a designated phone. This product aims to make wildfire monitoring more effective and efficient.

Challenges we ran into

Finding the most efficient method of data collection, transfer and analysis was a significant challenge we faced. We also initially planned on using 2 Cloudflare Workers to manage database management and LLM integration, but discovered that it didn't have great support for the web framework we were using, Next.js. We pivoted to using MongoDB for our database - which we chose for ease of use, since our data was in JSON format - and using the OpenAI API for our LLM.

Accomplishments that we're proud of

We are proud that we developed a prototype that is able to address a prevalent issue in our society affecting thousands of people in a significant manner. Apart from physical damage and financial loss, people have lost theire homes, businesses, and much more. Although our product is far from where it needs to be to become wide spread, we felt happy to provide a possible solution to the problem that thousands of people face. Moreover, we felt proud of exploring new ideas and implementing the hardware in a way that effectively solve the problem at hand.

What we learned

Some team member brought very technical skills while others heavily influenced the creative and logical side of the project. We were able to merge concepts and develop ideas off one another in order to develop the product we are presenting. We also did significant research into databases, artificial intelligence, among other things, and brushed up on the basics of coding and hardware. We learned hoe to prompt AI in order to give us the most relevant and helpful results. We also incorporated AI into the web app that we developed to have a chat bot, which gives real time output to the user.

What's next for FireGrid.ai

The next step for FireGrid,ai is making our product resistant to the elements. Given that there were some limitations to the hardware being used, there is the possibility of optimizing the product, both in strength and performance. The product is designed to be outdoors, meaning it must withstand heat, water, and other natural elements. Optimizing production, material cost, and manufacturing is the next step in growing this product.

Built With

• arduino
• mongodb
• nextjs
• openai
• python
• raspberry-pi
• react