Train ML models for different use cases

We have various features to add to the app, such as flight phase detection, flap position, and in-flight speech recognition, each requiring separate ML model training, data preparation, and data labeling.

Solution

We studied each feature case carefully, prepared the data set, picked the ML models, and trained the model with a supervised ML algorithm.

‍

Lack of relevant data set

To reach the client’s global goal – make the flight more safe and efficient, it was important to detect the jet’s maneuvers in real time. Despite the vast number of logs and data that the jet gathers during the flight, there was no single defined characteristic that could clearly indicate the maneuvers.

Solution

We started to think about how we could solve this issue. Having data such as the altitude, speed angle of inclination, etc., our idea was to create an algorithm that could identify and tell us when a maneuver is happening.

‍

Data visualization

Pilots' attention is crucial when they are on board, so we had to design the app's UI to be informative without being distracting.

Solution

Our design for AboardAI is seamless in interaction while remaining informative and useful for pilots to ensure a safe and efficient flight.