Having been on long flights, we became aware of the many inefficiencies of in-flight communication.
What it does
Our platform provides a web-based chat portal that allows passengers to send personalized requests to flight attendants in their native languages through speech or text, streamlining operations for flight attendants and promoting meaningful interactions.
In addition, the platform provides a suit of features including a bathroom queuing system and in-flight information regarding times and weather.
How we built it
We began with a simple front end built with Pug and SCSS, filled with placeholder data. Different information APIs were added to get estimated flight arrival, time remaining, and weather at the destination city. After we had a simple interface working, the next step was a bathroom queue.
The queue brought the first challenge due to the separation of queues for different flights. We used micro-services to facilitate the queuing, and used a simple model to predict time remaining until the bathroom would become available for use.
The final part of the project was a messaging system for passenger-attendant interaction. We began with a simple chat app built with socket.io, then slowly expanded to make it easier to use and innovative. We used Microsoft Cognitive Services to run natural language processing algorithms to predict the type of message the passenger sent (food, temperature, etc.) and a sentiment analysis algorithm to predict the nature of the message. We then added a translation service that would allow a passenger to send a message in their preferred language, translate it into the flight attendant's preferred language, and translate the attendant's response back into the passenger's preferred language. To top it all off, we added a speech to text service.
Challenges we ran into
One of the challenges was allowing multiple flights to run in parallel. This was solved by giving each flight it's own set of variables in a map with flight number as a key. Another challenge we ran into was getting audio input from the user. We ended up using RecordRTC to get the data in a webm blob, sending the blob to the back-end where it was converted to wav format by ffmpeg then sent to Microsoft's Speech to Text service. Finally, the text was sent back to the user.
Accomplishments that we're proud of
We're proud of the multifaceted API integration and the appealing, easy to use interface. We are also very happy with our use of Microsoft's API to solve the problem of cross-lingual communication on international flights.
What we learned
We learning about the cutting edge technologies provided by Microsoft through their Cognitive Services API and American Airlines through their mock API. We sharpened our full stack web-development skills to create an app that could communicate information through RESTful APIs and quick, reliable messages through web sockets. We improved our interpersonal skills and learned a ton about the various companies impacting technology today.
What's next for OnTheFlyCommunications
We plan on utilizing Big Data to improve message classification, sentiment analysis, and flight attendant work flow. We are currently working on incorporating IoT devices to interact with the restroom queue to improve efficiency. A real-time seat map notification interface is in the works to improve flight attendant efficiency.