Inspiration

According to the U.S. Fire Administration, the traditional use of sirens and lights often proves ineffective, especially in high-traffic areas or when vehicles have soundproofing and closed windows. This vital data coupled with personal stories of delayed response times of emergency services, our inspiration for this project stemmed from the current challenges in effectively alerting drivers about approaching emergency vehicles, such as ambulances and fire trucks. This delay in clearing the road for emergency responders can be a critical factor in life-threatening situations, where every second counts. Recognizing this gap, we decided to develop a more efficient solution aimed at reducing traffic congestion and improving response times during emergencies. By providing real-time alerts and more accurate traffic predictions directly to drivers through an integrated digital system, we aim to enhance the speed and safety of emergency vehicle navigation, potentially saving lives in critical moments.

What it does

Facilitating emergency response travel can be categorized into 2 main factors: the route the ambulance will take to its destination (which ideally will be fed by these emergency services directly), and the drivers who happen to be on that route who are in some level of traffic congestion. This service will determine when the ambulance is a close distance behind the cars on its route and alert them beforehand, ensuring that they can make way for the ambulance to go through. For example, all cars when alerted will be directed to move away from the left most lane, ensuring no traffic congestion delays an ambulance from reaching its patient in time.

How we built it

We started by analyzing the data using Postman, which helped us understand its structure and brainstorm ways to leverage it for our goals. Next, we integrated open-source APIs such as Google Maps and the Distance Matrix API to calculate distances between coordinate points. With this foundation, we drafted the core of our algorithm and initially tested it using dummy data, refining it until it was robust. We then incorporated real-time congestion data by integrating the INRIX API with AWS Bedrock and Llama, developing functions to fetch metrics like car counts and congestion levels. Replacing our test data with these real-time inputs, we cleaned and processed the data further using vector calculations to identify the traffic camera closest to the client’s coordinates. To ensure seamless updates, we automated image and token management using Python’s OS library. Finally, we polished our frontend with Next.js and connected it to the backend, providing users with an interactive interface and real-time updates.

Challenges we ran into

One of the main challenges we faced during the project was dealing with inaccurate readings from traffic images, as many of the cameras provided blurry or unclear visuals. Furthermore, Llama’s model through AWS Bedrock would sometimes produce outputs in formats that we struggled to extract data from. Over time, we managed to gradually improve our approach by refining our algorithms and preprocessing techniques. Additionally, integrating data from Postman involved scraping and structuring it for use in our predictive algorithm, which required significant trial and error to optimize. Additionally, we encountered limitations with the Inrix API, as the data provided only produced live footage every 5 minutes from relatively sparse camera feeds. Despite this, we maximized the efficiency of what was available, and the product’s performance and accuracy could be significantly enhanced and is easily scalable with access to live video feeds or higher-quality live image data.

Accomplishments that we're proud of

Throughout this project, we gained valuable experience in using various AWS tools, including Bedrock, Llama, and EC2, which enhanced our ability to integrate and analyze real-time data. We also deepened our understanding of vector mathematics, applying it to optimize data processing and improve accuracy in identifying the closest traffic cameras. Our work with APIs was extensive, as we mastered the implementation of several open-source APIs, such as Google Maps, Distance Matrix, and INRIX, to fetch and utilize location and congestion data effectively. Additionally, we honed our skills in web development by connecting the backend to the frontend using Next.js, creating a seamless and interactive user interface that dynamically updates with real-time data. This project not only expanded our technical expertise but also improved our ability to integrate diverse tools and technologies into a cohesive solution that has a powerful impact on emergency services saving lives.

What we learned

Throughout this project, we gained valuable experience in using various AWS tools, including Bedrock and Llama, which enhanced our ability to integrate and analyze real-time data. We also deepened our understanding of vector mathematics, applying it to optimize data processing and improve accuracy in identifying the closest traffic cameras. Our work with APIs was extensive, as we mastered the implementation of several open-source APIs, such as Google Maps, Distance Matrix, and INRIX, to fetch and utilize location and congestion data effectively. Additionally, we honed our skills in web development by connecting the backend to the frontend using React, creating a seamless and interactive user interface that dynamically updates with real-time data. This project not only expanded our technical expertise but also improved our ability to integrate diverse tools and technologies into a cohesive solution.

What's next for EmergenSee

In the future, we plan to implement several enhancements to improve the accuracy and functionality of our system. One key improvement is leveraging Amazon SageMaker to train a machine learning model for more precise predictions of traffic density and congestion, using a broader dataset and advanced analytics. We also aim to integrate our platform with major phone operating systems, such as Apple iOS and Android, to provide seamless, real-time updates directly to users' devices. Additionally, we see potential in partnering with navigation platforms like Google Maps and Waze to incorporate their extensive traffic and routing data, further refining our congestion metrics and delivering more accurate, real-time traffic insights. These enhancements would significantly boost the accuracy, reliability, and user experience of our application.