An extendable and generic SLAM dataset analysis and characterization framework suitable for characterizing visual-inertial SLAM datasets based on an extended number of characterization metrics.
A characterization metric is a measurement of a certain quantity that can be used as a partial or a full descriptor of the SLAM dataset. In our system, we currently support general non-sensory metrics, inertial metrics, and visual metrics. The total number of supported metrics is currently at 73 different metrics.
It is advised to run the docker version of the system in order to have a smooth operation. To do so, you need to first build the docker image:
./build_docker.sh
Then, you have to run the docker image, but change the path to mount the dataset you want to characterize.
./run_docker.sh
Make sure to update the path to your dataset inside:
in_data/you_ds_config_file.txt
This project is developed by Islam A. Ali, as part of my PhD thesis in Computing Science at the University of Alberta, Edmonton, Canada. For researchers/developers wishing to leverage or utilize the code, please use the citation of the paper below:
@inproceedings{ali2022we,
title={Are We Ready for Robust and Resilient SLAM? A Framework For Quantitative Characterization of SLAM Datasets},
author={Ali, Islam and Zhang, Hong},
booktitle={2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
year={2022},
pages={2810--2816},
organization={IEEE}
}
The codebase is licensed under the GNU General Public License v3 (GPL-3)
