ResQ-agents : AI Agents that save Lives

ResQ-Agents: an AI-powered natural disaster response system that monitors, analyzes, verifies, and alerts in real time with potential to save millions of lives.

Comment

Inspiration

Every hour, somewhere on Earth, nature strikes, a flood swallowing a village, an earthquake tearing through a city, a wildfire turning skies to ash, leaving countless people in danger. Manual search efforts and slow, fragmented data collection put even more lives at risk with every passing minute. We were deeply saddened when we read the statistics of natural disaster impact globally, 343 disasters per year, 16000+ deaths, 16 million affected and little kids being orphaned. We felt honored to support this cause and work towards helping people out.

What it does

The project aims to deliver an automated AI based natural disaster response system, leveraging three agents that work simultaneously to achieve different functions. The first agent is responsible to search real time disaster data form valid and verified sources and verify using real satellite imaging, the second agent is used to analyze the location and divide it into zones based on severity of damage and within minutes calculate the resources needed to be sent to each region including how many helicopters, rescue units, K-9s if needed and type of equipment, and landscape data. The third agent is responsible to send the information and alert to the local authority(which is automatically determined by location api and government api). All of this information is hosted on python based Streamlit dashboard with a modern interface and neatly organized details populating it. Together the three agents and the dashboard make a comprehensive disaster response system which if implemented can potentially save millions of lives.

How we built it

This project was built using Streamlit library, Python, AWS Bedrock LLM, AWS Bedrock AgentCore, NASA EONET API, USGS API, and NOAA api, NASA satellite imaging,

Amazing use of AWS tools Bedrock LLM - smartly used to reason with real-time data and determine quickly the resources needed to be sent to the affected people, and whether a natural disaster is impending or not based on weather data and real-time disaster data, and zone splitting based on severity.

Bedrock AgentCore - Used AgentCore to create the infrastructure for the three agents which then integrate the APIs and the LLM reasoning data and work autonomously.

Streamlit dashboard - Python library allowing the creation of out beautiful and modern dashboard with the capability of easily integrating real-time data.

Python for the backend code and for Streamlit.

Challenges we ran into

It is the first time we worked with agents although we've studied the theoretical concepts and the math behind them, using them in a program was novel and confusing at first but we persisted and learned from the amazing documentation provided by Amazon web services and then finally implemented it.

Accomplishments that we're proud of

Firstly working for a great cause and putting efforts to built a project that we think actually helps people and make an impact, and secondly using the new AWS tools made use better understand the AWS ecosystem and made us explore so much more and we also got to learn about agents in depth which was very helpful and will help in the future. Overall the idea of working for a good cause like this is extremely satisfying.

What we learned

How to implement agents using AWS in code, more AWS tools explored and a passion to make projects that help community more.

What's next for ResQ-agents : AI Agents that save Lives

We would love to try scaling this project and try talking it into the public sphere, we are super excited about it and would love to pursue this idea further, More features that can be included would be more agents handling these tasks instead of two making it even more faster, and maintaining a close contact with the government for better use of this tool, and more research in natural disasters and existing response systems(ours is still novel and unique after a thorough search). We are totally into helping people, if you are too you resQ-agents should already be your favorite by now!!

Built With

Share this project:

Updates