Predicting flight delay with spatio-temporal trajectory convolutional network and airport situational awareness map

To model and forecast flight delays accurately, it is crucial to harness various vehicle trajectory contextual sensor data on airport tarmac areas. These heterogeneous data, if modelled correctly, can be used generate a situational awareness map. Existing techniques apply traditional supervised learning methods onto historical features route information among different airports predict delay are inaccurate only arrival but not departure delay, which essential airlines. In this paper, we propose vision-based solution achieve high forecasting accuracy, applicable the airport. Our leverages snapshot of map, contains trajectories aircraft such as weather airline schedules. We an end-to-end deep architecture, TrajCNN, captures both spatial temporal from Additionally, reveal that map has vital impact estimating delay. proposed framework obtained good result (around 18 minutes error) for predicting at Los Angeles International Airport.