Flight Insight | Devpost

Inspiration

In the last year we have seen lots of cancellations, delays, and poor customer satisfaction in the airline industry. We decided that we could do something to help improve those issues that were facing the industry by letting people know how the weather could impact their travel and breaking down the barriers in communication that exist between flight crews and their passangers.

What it does

Flight Insight is a mobile focused web app for both customers and employees of American Airlines. The platform provides a comprehensive source of information for the purpose of providing the ultimate comfortable experience for those who choose to fly American. Interaction with our application starts before a customer even boards their flight. Using our pre-flight check list, customers learn what to expect from their flight. In addition to existing airline apps, which focus on protocols and procedure, our app focuses on the environment of the aircraft. We deliver information about the physical conditions inside the aircraft and provide users with advice and resources to optimize the comfort of their flight. No more guessing whether you'll be freezing or sweating the whole flight; no more wondering if you're dressed for your destination; our application will inform you ahead of time so that you can board confidently. Once your board the aircraft, many disturbances, from the baby on the otherside of the aisle to unexpected turbulence at 12,000 ft, may plague your trip. We value transparency and believe that, by providing users with timely information about the conditions of their flight, users will be properly prepared for any disturbances and will be less annoyed by them. Our application uses sensors and LAN inside the cabin to alert users about disturbances, such as turbulence, and providing users with resources do deal with potential sensitivities, such as motion sickness.

How we built it

When we started the project, we wanted to predict flight conditions using ML and provide users with advice on rebooking flights. However, halfway through the hackthon, we realized that working through government data to construct training data for the model, while eventually possible, was not feasible in this time scale. At midnight, we pivoted directions and began constructing a real-time system to provide live results to users. We compliment this sytem with systems that facilitate instant communication with flight crew. We decided to use Node.js through Svelte to build the front- and back-end of the web app. This allows it to run on embedded systems inside the plane. We also interfaced with an isolated database to provide reliable data persistence to our users.

Challenges we ran into

One of our goals when we started this project was to develop a machine learning model to help users predict how weather patterns could impact their travel plans. About halfway through our project we realized that we would not have enough time to finish cleaning the large data set and still have time to train the model up to the standards we had set for the rest of the project. This forced us to evolve our project and think of ways that we could help our users that we hadn't considered before. We are extremely proud that we were able to overcome this challenge and it motivated us to build a better project and expand some of the ideas we hadn't previously considered.

Accomplishments that we're proud of

One of the accomplishments that we are very proud of was the large range of technologies and skills that we used in order to implement our project. We made use of computer engineering, systems design, high and low level software engineering, and a wide variety of other project management and planning skills. While we have gotten to build plenty of web apps in the past, this one was particularly motivating as we were able to push beyond that and build a full system from the ground up and push our technical skills.

What we learned

Since our system integrated over multiple different technologies and systems, we learned lots of things about computer and data interaction. We improved our knowledge of Micro-controllers, Servers, Computer Communications, and various programming languages and frameworks such as SvelteKit, Python, and C/C++. One of the most valuable lessons we learned was how to effectively engineer a system under heavy constraints of time and limited tech knowledge. We pivoted our original idea due to a lot of technical difficulties and managed to finish the project just under 12 hours by leveraging the power of flexibility and rigorous planning.

What's next for Flight Insight

We had many ideas for FlightInsight that did not get realized due to the short amount of time for our project and trouble finding data. However, in the future, we would like to build out our ideas of leveraging machine learning for predictions, applying academic research knowledge to best serve the comfort of airline customers, and improving the look and feel of our interfaces. Our biggest hope for FlightInsight™️ is its adoption and further development by the American Airlines software engineers