Inspiration

Flying to and from India every summer to visit relatives, we realized that we often get sick during our time in a plane making our first few days of vacation miserable. With the airplane cabins being tightly cramped, the spreading of bacteria is often inevitable.

What it does

Flow starts by recording the sound waves of each passenger via their phone. We identify whether the sounds made are produced from symptoms of disease: cough and sneeze. After uploading this data to a server, we update a visual heat-map indicating the degree to which individuals in the plane are sick as well as change the flow of air in the airplane by closing or opening certain fans.

How we built it

The three components of Flow are the iOS App, the Hardware, and the Web App. Using the American Airlines API, the user is able to check into his flight. In Firebase, each flight along with the passengers and their sickness level is stored. The sickness level is determined through the use of a Sound Recognition Machine Learning model. This data is not only retrieved in the Web App to display changes in the heat map but in the arduino. For each seat, the sickness index of a passenger is measured along with his/her neighbors in order to subsequently modify the stepper motor and adjust the airflow.

Accomplishments that we're proud of

We were able to successfully port sklearn machine learning model for sound recognition to a CoreML model for our iOS app—which was then used successfully to analyze the sounds emitted from the passengers. We were also able to successfully develop communications between the hardware and the iOS app to depict that we could easily hack the airlines' airflow system in a manner to mitigate the transmission of disease within the cabin.

What we learned

We learned how to successfully display real time analysis from a backend which was supporting most of our platform. More specifically, we appreciated learning about Firebase and JavaScript along with the complexities of CoreML in Swift. Finally, we also learned a lot about specific Arduino syntax when developing Serial communications revolving around the hardware components.

What's next for Flow

Next steps for Flow include developing higher quality hardware in terms of stepper motors and the design of the cabin fans. We ideally want to be able to control the direction of the airflow as well which is achievable using rotatable motors and fans. Furthermore, we would like to port a more accurate sound recognition model to CoreML and also be able to utilize more data from the Airlines' API.

Share this project:

Updates