This code tightly couples the visual information coming from a stereo camera and imu measurements via Multi-State Constraint Kalman Filter (MSCKF).
To work with this code:
- Open S_MSCKF.m file and change the directories based upon where the code is stored.
- Many KITTI datasets have been tested for this code. Uncomment one of the datasets you would like to run the software for.
If wanted to use the other KITTI datasets, you should download the data from KITTI datasets http://www.cvlibs.net/datasets/kitti/raw_data.php and use contents of kitti_extraction to track features and have them stored in a specific .mat file.
Note: This code was originally developed by Lee E Clement for mono-msckf (Clement, Lee E., et al. "The battle for filter supremacy: A comparative study of the multi-state constraint kalman filter and the sliding window filter." 2015 12th Conference on Computer and Robot Vision. IEEE, 2015.).
Please cite properly if this code used for any academic and non-academic purposes.
title={Vehicle positioning in GNSS-deprived urban areas by stereo visual-inertial odometry},
author={Ramezani, Milad and Khoshelham, Kourosh},
journal={IEEE Transactions on Intelligent Vehicles},
volume={3},
number={2},
pages={208--217},
year={2018},
publisher={IEEE}
}