Stanford-NavLab/chimera_fgo

Spoofing-Resilient LiDAR-GPS Factor Graph Optimization (FGO) with Chimera authentication.

Paper: https://ieeexplore.ieee.org/abstract/document/10139945

@inproceedings{dai2023spoofing,
  title={Spoofing-Resilient LiDAR-GPS Factor Graph Localization with Chimera Authentication},
  author={Dai, Adam and Mina, Tara and Kanhere, Ashwin and Gao, Grace},
  booktitle={2023 IEEE/ION Position, Location and Navigation Symposium (PLANS)},
  pages={470--480},
  year={2023},
  organization={IEEE}
}

Setup

Clone the GitHub repository:

git clone https://github.com/Stanford-NavLab/chimera_fgo.git

Create and activate conda environment:

conda create -n chimera python=3.9
conda activate chimera

Install dependencies:

pip install -r requirements.txt

Install chimera locally from directory containing setup.py

pip install -e .

Data

We use data from the KITTI dataset for our experiments. The pre-processed data can be found at this Google drive link. Inside chimera_fgo, create a data folder and with the following structure (you may have to unzip the oxts folders):

chimera_fgo
|  data
   |  kitti
      |  0018
         |  icp
         |  svs
         |  oxts      
      |  0027
      ...

Running experiments

The ephermeris_fgo.ipynb notebook steps through an example of Chimera FGO.

The script testing_script.py is used to run Monte Carlo simulations over various different FGO settings (window size, spoofing attack size), and calls the chimera_fgo function in chimera_fgo.py.