Repository for generating hobby drone trajectories and writing them to output CSV files. Code originally written by E. R. Mueller and interface was written by S. M. Katz. Hobby drone trajectories are generated by sampling a Bayesian network trained on telemetry log files of multi-rotor aircraft trajectories. The flight data was from DroneShare.com, which was a website in which hobbyists could upload their telemetry log files. Data from over 75,000 flights was publicly available for download. For more information about this model, see the following citation:
E. R. Mueller and M. J. Kochenderfer, “Simulation comparison of collision avoidance algorithms for small multi-rotor aircraft,” in AIAA Modeling and Simulation Technologies Conference, 2016, p. 3674
This code was developed on Julia1.1.
Ensure that the following packages are installed in your current version of Julia: BayesNets, LightGraphs, DataFrames, CSV, and JLD2. To install any of these packages that have not yet been installed, simply run:
include("setup.jl")The following lines will generate a trajectory file titled "test.csv" with data for a hobby drone trajectory with a time step of 1 second.
include("HobbyDroneInterface.jl")
generate_trajectory_file(1.0, "test.csv")The default trajectory time length is 120 seconds. Use the tMax variable to generate trajectories of a different length. The follow line will generate a trajectory that is 100 seconds in length.
generate_trajectory_file(1.0, "test.csv", tMax=100.0)If you want to generate more than 1 trajectory per file, you can add this as the last required argument. The following line will generate a file called "test.csv" with 3 trajectories.
generate_trajectory_file(1.0, "test.csv", 3)The first column of the resulting CSV file will be the time in seconds and the following three columns will be the x-, y-, and z-position respectively in feet. To test on your system, include the RUN_HD.jl script. It should generate a file in the output folder that matches the "hd_traj_ex.csv" file in the examples folder.
generate_HD(initBN::BayesNet, tranBN::BayesNet) - function that interfaces with the original hobby drone code to generate a trajectory. The keyword arguments allow for setting things like initial evidence, time step, maximum time, and initial time.
convert_to_position(stHistory::Array{IntruderState}) - converts output of hobby drone model to just get the trajectory positions
setup.jl - file that installs required packages that you do not currently have
HobbyDroneInterface.jl - Wrapper code for hobby drone model. Contains functions for generating trajectories and writing to files.
CreateHobbyDroneTrajectories.jl - file containing code for sampling the Bayesian network to generate a trajectory.
iBN_data_v2.jld2 - contains the initial network for the dynamic Bayesian network model
tBN_data_v2.jld2 - contains the transition network for the dynamic Bayesian network model
RUN_HD.jl - sample script for generating hobby drone trajectory files
output/hd_traj_ex.csv - sample model output for one hobby drone trajectory of length 120 seconds (created by the RUN_HD.jl file)