IndoorGraphLocalization is a multi-user indoor localization system utilizing graph-based optimization for sensor fusion. The repository contains the code, the dataset and the scripts making it possible to reproduce results of our research.
If you found code useful in your academic work, please cite:
@article{SensorsMN,
title={A Multi-user Personal Indoor Localization System Employing Graph-based Optimization},
author={Nowicki, M. R. and Skrzypczynski, P.},
journal={MDPI Sensors},
volume={},
number={},
pages={},
doi = {},
year={2019}
}
The installation can be performed with the provided build.sh script:
cd IndoorGraphLocalization
chmod +x build.sh
./build.sh
The results presented in the article can be reproduced with provided script:
cd IndoorGraphLocalization
python scripts/run.py
After running the script, the directory results should contain four directories:
- iUC_false_mKP_1.0_sLE_false_run_0
- iUC_false_mKP_1.0_sLE_true_run_0
- iUC_true_mKP_1.0_sLE_false_run_0
- iUC_true_mKP_1.0_sLE_true_run_0
The iUC stands for connections between users, while sLE stands for step length estimation. Each directory should contain results and figures for the evaluated approach. After running the script, the console should present list of the obtained results that are later sorted according to the RMSE error:
Config RMSE AvgErr Sigma MaxErr
iUC_false_mKP_1.0_sLE_false_run_0 6.24 2.42 5.76 85.67
iUC_false_mKP_1.0_sLE_true_run_0 5.19 2.21 4.69 57.25
iUC_true_mKP_1.0_sLE_false_run_0 4.77 2.16 4.25 42.41
iUC_true_mKP_1.0_sLE_true_run_0 4.65 2.11 4.15 42.64
--------------
iUC_false_mKP_1.0_sLE_false_run 6.24 2.42 5.76 85.67
iUC_false_mKP_1.0_sLE_true_run 5.19 2.21 4.69 57.25
iUC_true_mKP_1.0_sLE_false_run 4.77 2.16 4.25 42.41
iUC_true_mKP_1.0_sLE_true_run 4.65 2.11 4.15 42.64
The code is available under the GPLv3 license. For the convenience, the directory EXTERNAL contains the code of the g2o library (https://github.com/RainerKuemmerle/g2o) used for the optimization that is available mostly under the BSD license. Some external dependencies of the g2o library are available under GPLv3, so please check EXTERNAL/g2o/README.md for details. If other licensing is needed please contact the author: michal.nowicki (at) put.poznan.pl.