Using ESKF to fuse IMU and GPS data to achieve global localization.

The code is implemented base on the book "Quaterniond kinematics for the error-state Kalman filter"

For formula derivation, please refer to：https://zhuanlan.zhihu.com/p/152662055 and https://blog.csdn.net/u011341856/article/details/114262451.

1.Dataset

https://epan-utbm.github.io/utbm_robocar_dataset/

2.Demo

3.Prerequisites

3.1 Ubuntu and ROS

Ubuntu >= 18.04 (Ubuntu 16.04 is not supported)

ROS >= Melodic. ROS Installation

3.2 Nmea_navsat_driver

sudo apt-get install ros-melodic-nmea-navsat-driver libgps-dev

3.3 Mapviz

sudo apt-get install ros-$ROS_DISTRO-mapviz ros-$ROS_DISTRO-mapviz-plugins ros-$ROS_DISTRO-tile-map ros-$ROS_DISTRO-multires-image

4.Build

cd ~/catkin_ws/src
git clone https://github.com/KalmanSLAMer/eskf-localization.git
cd ..
catkin_make
source devel/setup.bash

5.Direct run

roslaunch imu_gps_localization imu_gps_localization.launch
rosbag play YOUR_DOWNLOADED.bag

6.How to use mapviz

how to How to set mapviz, please refer to csdn blog.

roslaunch mapviz mapviz.launch

7.TODO

• add localization by fusing lidar and imu

• add normal deduction

• add EKF

• clean code

• Visualization via mapviz

8.Acknowledgments

Thanks for the book "Quaterniond kinematics for the error-state Kalman filter", eskf_qk, eskf, mapviz, and how to use mapviz.