Sliding Window MSCKF
Authors: Doom
!!!!NOTICE!!!! This implementation is still under constrution. Please DO NOT fork yet!
MSCKF (Multi-State Constraint Kalman Filter) is a well known Visual Inertial Navigation System (VINS) method, proposed by Anastasios I. Mourikis et al. I have great interest in using this method in my own project, but there are no available code yet. So I decide to create a complete implementation of this method using C++. However, the most tough part of MSCKF is the feature/keyframe culling, in order to make this code working efficently and easily, I use a sliding window instead. The propagation and update part of our method using the derivation of MARS's quaternion report.
Related Papers
- A Multi-State Constraint Kalman Filter for Vision-aided Inertial Navigation, Anastasios I. Mourikis and Stergios I. Roumeliotis, ICRA '07
- Indirect Kalman Filter for 3D Attitude Estimation A Tutorial for Quaternion Algebra, Nikolas Trawny and Stergios I. Roumeliotis, Department of Computer Science & Engineering University of Minnesota
Sensor
ZED + Xsens
Change Log
v0.1 (2016/12/20)
- Create this repo.
- Create the basic framework.
v0.2 (2016/12/21)
- Create frame and feature classes.
- Create imu and some utils classes.
- Update CMakeLists.txt
v0.3 (2016/12/22)
- Fix some bugs of frame class.
- Create sliding window algorithm classs.
- Achieve the propagation part of filter as the derivation of quaternion report @ UMN.