Skip to content

Latest commit

 

History

History
99 lines (73 loc) · 3.08 KB

README.md

File metadata and controls

99 lines (73 loc) · 3.08 KB
Raw

Trajectory optimization in flow fields

NOTE: The article "Schnitzler et al., General extremal field method for time-optimal trajectory planning in flow fields, DOI 10.1109/LCSYS.2023.3284339" refers to release v1.0.0 of the source code.

Presentation

This module tackles trajectory optimization problems in strong, non-uniform and unsteady flow fields, in particular wind fields.

The module performs origin-to-destination trajectory optimization (wrt. time) by sampling extremal trajectories. Extremal trajectories are Hamiltonian-minimizing trajectories. The time-optimal trajectory of a problem, when it exists, is a particular extremal trajectory. So the set of all extremal trajectories (parametrized by a real value, the initial angle $\theta_0 \in [0, 2\pi[$) is guaranteed to contain the time-optimal trajectory, and this motivates the extremal integration method.

The module provides the following features for flow fields:

  • Python code for analytic flow fields (see dabry/flowfield.py)
  • Custom numpy zip (.npz) format definition for discrete flow fields (see docs/flow_format.md)
    • Translation from GRIB2 files to the npz format

A demonstration notebook is given at examples/Gyre.ipynb and can be viewed using nbviewer.org.

The module supports 2D planar environment as well as spherical problems.

Cloning the repo

Clone the module using

git clone [repo-url]

Installation as Python module

After cloning the repo using the previous command, install the project as a Python module using

python3 -m pip install -e ./dabry

Base examples

You can now run base examples calling dabry as Python module:

python3 -m dabry case [case_name]

Results will be automatically saved to a new folder with the case's name in the current working directory. The different output files format specification can be found in docs.

Available default problems for case_name are listed in dabry/problems.csv

Real data

To directly run trajectory optimization on real wind fields, you have to install the cdsapi module.

python3 -m pip install cdsapi

Then, configure your CDS Python API key for the cdsapi module to be allowed to extract wind fields from the CDS database.

After that you can run trajectory optimization on real problems using

python3 -m dabry real [lon_start] [lat_start] [lon_target] [lat_target] [date_start] [airspeed] [altitude]

Note that an automatic CLI generator for dabry's real cases is available on the website.

Visualization

Make sure the dependencies from requirements_display.txt are installed.

If the previous computation put the results in the "movor (scaled)" directory (for instance), then the interactive display can be launched using

python -m dabry.display "movor (scaled)"

If easygui is installed, then the command

python -m dabry.display .

launches an interactive prompt to select the example to display from current directory.