roost: Robust Optimization Of Structured Trajectories

What is roost?

The Python Library roost is a software package developed by UC3M. It is a tool for robust flight planning within the currently structured airspace. The main features of roost are: 1) integrates horizontal and vertical decision-making, 2) fast performance thanks to GPU-based parallelization, 3) considers climb, cruise, and descent phases 4) incorporate uncertainty in meteorological variables, as well as initial flight time and initial flight mass.

License: roost is released under GNU Lesser General Public License v3.0 (LGPLv3).

Support: Support of all general technical questions on roost, i.e., installation, application, and development, will be provided by Daniel González Arribas (dangonza@ing.uc3m.es) and Abolfazl Simorgh (abolfazl.simorgh@uc3m.es).

Core developer team: Daniel González Arribas, Abolfazl Simorgh, Eduardo Andrés Enderiz, Manuel Soler.

Copyright (C) 2022, Universidad Carlos III de Madrid

How to run the library

The installation is the first step to working with roost. In the following, the steps required to install the library are provided.

1. It is highly recommended to create a virtual environment (e.g., roost): python conda create -n env_roost conda activate env_roost

2. Clone or download the repository. The roost source code is available on a public GitHub repository: https://github.com/Abolfazl-Simorgh/roost. The easiest way to obtain it is to clone the repository using git: git clone https://github.com/Abolfazl-Simorgh/roost.git.

3. Locate yourself in the roost (library folder) path, and run the following line, using terminal (MacOS and Linux) or cmd (Windows), which will install all dependencies: python python setup.py install it will install all the required dependencies.

How to use it

There is a script in the roost (library folder) path, main_run.py, which provides a sample to get started with the library. This sample file contains comments explaining the required inputs, problem configurations, objective function selection (which includes flight planning objectives), optimization configurations, running, and output files. Notice that we use BADA4.0 to represent the aerodynamic and propulsive performance of the aircraft. Due to restrictions imposed by the BADA license, the current version in GitHub is incomplete, as three python scripts related to the used aircraft performance model have been excluded (i.e., bada4.py, apm.py, and badalib.cu). We are currently assessing the existing open-source aircraft performance models in order to make the complete library available to the public.