search

cggos / imu_x_fusion

IMU + X(GNSS, 6DoF Odom) Loosely-Coupled Fusion Localization based on ESKF, IEKF, UKF(UKF/SPKF, JUKF, SVD-UKF) and MAP
https://msf.cgabc.xyz/
GNU General Public License v3.0
915 stars 163 forks source link

IEKF != IESKF? #61

Open gyubeomim opened 9 months ago

gyubeomim commented 9 months ago

Hello Gavin Gao,

I just watched your wonderful blog post about IEKF derivation from the Linear Gaussian Model. Then I'm about to start using your imu_x_fusion repo.

post: https://cgabc.xyz/posts/784a80cb/#iekf

Screenshot 2024-02-01 at 8 39 18 PM

I just wanted to know if your posting's update X equation is different from the Fast-LIO, ROVIO's IEKF update X equation. Fast-LIO(arXiv): eq (18)

Screenshot 2024-02-01 at 8 40 42 PM

ROVIO(IJRR 2017) : eq (51)

Screenshot 2024-02-01 at 8 38 14 PM

Is it because your derivation is about only for the IEKF not for IESKF? or just the missing jacobians are hidden by the equations?

Thanks in advance!

ysingfun commented 8 months ago

I have the same doubts about this question, may I ask whether the problem you raised has been solved？

gyubeomim commented 8 months ago

@ysingfun Not yet..

gyubeomim commented 5 months ago

@ysingfun I figured it out myself. I upload the arXiv preprint about this topic (IEKF, IESKF). Please check it if you have still no answer.

Notes on Kalman Filter (KF, EKF, ESKF, IEKF, IESKF) https://arxiv.org/abs/2406.06427

ysingfun commented 5 months ago

Thanks!