This repository contains an aircraft trajectory learning algorithm used for the terminal airspace modeling project.
-i : input files
-o : output files
- model_train.py : learns the deviations of trajectories from procedures, distance vs. transit time, inter arrival-departure times
python3 src/model_train.py -i data/radar_data.csv data/train_input.json
-o output/model.json- model_generate.py : generates synthetic trajectories using trained deviations and test inputs
python3 src/model_generate.py -i output/model.json data/test_input.json
-o output/synthetic_trajs.csv- radar_animate.py : animates actual/synthetic trajectories
python3 src/radar_animate.py -i output/synthetic_trajs.csv data/test_input.json output/animation.html- radar_plot.py : draws different kinds of plots of actual/synthetic trajectories (log-'hist'ogram, 'all' trajectories, 'each' trajectory)
python3 src/radar_plot.py -i output/synthetic_trajs.csv data/test_input.json hist-
CSV files (radar_data.csv, synthetic_trajs.csv)
- Input and output trajectory data are in CSV format.
- Each row is a ENU position of an aircraft from the airport
: [time(seconds), track_id, x(meters), y(meters), z(meters)]
-
JSON files (train_input.json, test_intput.json)
- Airport runway and procedural information are given as input in JSON format.
- Each row of runway/path coordinates is a point in runway/path
: [latitude(degree), longitude(degree), altitude(feet)] - Path weights are the fraction of each path taken by actual trajectories.

