KF-GINS is an open-source, Extended Kalman Filter (EKF)-based GNSS/INS integrated navigation system. It implements a loosely-coupled approach using an error-state vector to fuse GNSS positioning and IMU data, targeting researchers and developers in the field of inertial navigation and sensor fusion. The system provides a robust solution for navigation tasks by compensating for IMU biases and scale factor errors.

How It Works

KF-GINS employs an EKF with a 21-dimensional error state vector, encompassing position, velocity, attitude (using the Phi-angle model), IMU biases, and scale factor errors. The navigation solution utilizes a two-sample mechanization algorithm with second-order coning and sculling motion compensation. This approach is advantageous for its comprehensive error modeling and efficient state estimation, providing accurate navigation solutions even with low-cost IMUs.

Quick Start & Requirements

• Installation: Compile using CMake. Recommended environment is Linux with g++.
• Dependencies: Eigen3, abseil-cpp, yaml-cpp (included as third-party libraries).
• Execution: ./bin/KF-GINS ./dataset/kf-gins.yaml (after compilation).
• Documentation: KF-GINS README

Highlighted Details

• Implements a 21-dimensional error state vector for comprehensive error modeling.
• Utilizes the Phi-angle model for attitude error representation.
• Supports initial alignment with pre-defined states in a configuration file.
• Defines specific text file formats for IMU, GNSS, and output data.

Maintenance & Community

• Developed by the Integrated and Intelligent Navigation (i2Nav) Group, GNSS Research Center, Wuhan University.
• Contact points: Liqiang Wang (wlq@whu.edu.cn), Hailiang Tang (thl@whu.edu.cn).
• Chinese users can join QQ group: 481173293.

Licensing & Compatibility

• License: GPLv3.
• Commercial Use: Requires explicit contact with Prof. Xiaoji Niu (xjniu@whu.edu.cn).

Limitations & Caveats

The system currently only supports initial alignment with all initial states provided in the configuration file. The README indicates ongoing code reliability improvements.