Imagine you’re at home, at a classroom, or in an unfamiliar space and you hear the tornado sirens or the building begins to shake. Research shows that the brain often freezes in spaces where they are terrified, and a natural disaster is no exception. Additionally, after a natural disaster, people may lose connection and be unable to send their location to emergency crisis centers. We wanted to reduce the number of injuries and deaths in a natural disaster by providing clear cut instructions and sending locations to 311 (an emergency line that would be able to track people if they’re unable to find them after the disaster occurs) and build an app that does the following:
What it does
An iOS app that gives users the best course of action during a natural disaster by detecting the environment they are in and surrounding objects using machine learning . A message automatically sends user location to emergency services.
How we built it
Node.js, IBM Visual Recognition (training our own ML algorithm), Google Cloud (specifically Google App Engine), axios, Microsoft Azure Cosmos DB, React Native
Challenges we ran into
Although we had big ambitions of serving a Swift app to iphones and integrating camera functions to take a live scan of the room and give you instructions based on that, we ran into several problems: both technical and the practicality of several functions. Yet, the biggest issue we had was attempting to learn Swift and apply it to the app. None of us had ever used Swift before and ultimately (after several hours of updating XCode and installing different dependency managers) it proved too difficult to integrate the several APIs (which often only connected to JS or C) into the Swift coding language. Thus, around midnight, we moved to Node.js and ReactNative.
Yet, another challenge we faced is making the app as easy to access and use during panic moments in a natural disaster. The natural disasters chosen were purposefully chosen to be ones without much reaction times. Whereas hurricanes may allow some time to evacuate the area, tornadoes and earthquakes are preceded by either a siren or a shaking of the house, and that can create a large amount of panic within a user. We did not want the user to have to think about taking the phone out, go through a list of selections to determine what kind of disaster is occuring, or even take out the phone and scan the room. None of these options were short enough for the user. Thus, we had to think about ways to attain that information about the room beforehand and ensure that the user only had to press a button in a moment of panic to get what they needed.
Accomplishments that we're proud of
We were very proud of our unique idea. As far as we know, nobody has thought of a “panic button” of sorts in a hackathon. We thought it was a clever implementation of a simple concept and a very needed thing. We are also very proud of our trained ML algorithm that is quite successful at determining the type of room and giving ideas based on the room on what to do.
What we learned
We each learned a lot from our parts. All of us spent time learning new software like Swift, Node.red, Google App Engine, and Microsoft Azure. Each brought its own unique challenges. Learning how to connect the databases and APIs proved to be a difficult challenge as well. We also learned a lot about keeping ourselves safe in a natural disaster.
What's next for us?
With more time, we would love to let the user take a video panorama of their surroundings and store them in a database everywhere they go. We could then store the videos globally so that wherever users go, there will be a surrounding map of the room. We can also scrape the web for weather data so that natural disaster information can be found on the fly and users in those locations can be notified beforehand. Furthermore, we can integrate this idea with wearable tech. As an example, a user can press a button whenever a disaster occurs and the phone can speak the instructions aloud. Lastly and most widely-applicable, we can extend our application to any disaster or emergency, such as feeling unsafe at night or during a crime like active shooting/robbery.