In a world where every second counts, especially during emergencies, we were inspired to create a solution that leverages the power of AI to save lives and provide immediate assistance. Our inspiration came from stories of people who faced critical situations and couldn't get help in time. We envisioned an AI that could streamline the 911 calling process and lift the burden off of 911 operators.
OpenSOS is an Automated Emergency Response Agent that provides instant, accurate, and potentially life-saving guidance during various emergencies. The project interacts with the caller, extracts important information and understand the nature of the crisis, provides a user-friendly interface for the human operator and contact emergency services, and even relay critical information to first responders. The system has the potential to be used by the government for fast and accurate responses, improving the lives of the residents and contributing to the development of a smart city.
We used voice recognition and text to audio python libraries to process caller and linked to GPT3.5 API for analysis. We used prompt engineering to guide the LLM to obtain any missing information in a conversational style. We also implemented a simple example using amazon web services such as amazon connect and amazon lambda to demonstrate scalability of the project.
A major challenge we encountered was syncing the various components of the project including front-end, back-end, and cloud services. Through rigorous testing and with help from mentors, we were able to eventually combine the different parts into a coherent system.
As our first time doing AI-integrated systems, we learned to work with the OpenAI library and API. It was a huge accomplishment to see the system interact in real time and output GPT-processed responses.
On top of technical aspects, we were able to work in a team and think as one. We also researched and learned about the nuances of various emergency protocols and the responsibilities of 911 operators.
The future of OpenSOS is promising. We're looking to integrate real-time location-based services to provide even more precise assistance. We're also exploring partnerships with smart home and urban infrastructure systems to enhance the reach and effectiveness of OpenSOS. We look forward to apply this system to real databases and work with exterior parties such as hospitals to provide accurate recommendations.