johannesbrand / ecl_ekf_analysis

a python package for PX4 ecl ekf analysis

ecl_ekf_analysis

This repository contains a python package for ekf related analysis on PX4 ULog files (pyulog). The command line scripts of this package analyse the ekf innovations and the internal ekf test metrics for irregularities.

The provided command line scripts are:

• process_logdata_ekf: analyse a single PX4 ULog file.
• batch_process_logdata_ekf: run process_logdata_ekf on multiple PX4 ULog files in a directory.

Installation

Installation with package manager via ssh:

pip install git+ssh://[email protected]/Auterion/ecl_ekf_analysis.git#egg=ecl_ekf_analysis

or using https:

pip install git+https://github.com/Auterion/ecl_ekf_analysis.git#egg=ecl_ekf_analysis

Check the official pip documentation for installing particular branches or commits using git hashes.

Installation from source:

python setup.py build install

Development

For development it might be useful to install the package as a link to the repo so that the requirements are always updated when the repo changes:

pip install -e .

Linting & Testing

This repository includes a Makefile and dockerfiles for conveniently running the linter (pylint) and the unit tests (pytest). Recent versions of docker and docker-compose are required to run the following make targets.

run the linter:

make lint

run the unit tests:

make test

the unit tests run the analysis over a set of golden log files and compare the current analysis results to stored ground truth results. The golden log files are described in the README file.

Usage

The analysis can also be run using the provided docker file. Recent versions of docker and docker-compose are required to run the following make targets.

analyse a single ulog file

run make analyse-file file=PATH/TO/THE/ULOG-FILE. This should create a new .json file ulog_filename.json in the folder of the .ulg file containing the analysis results.

analyse a directory of ulog files

run make analyse-dir dir=PATH/TO/THE/LOG-FOLDER/ to create a new .json analytics result file for every .ulg file in the log folder.