LIGO: Tightly Coupled LiDAR-Inertial-GNSS Odometry based on a Hierarchy Fusion Framework for Global Localization with Real-time Mapping

Code, paper, video are coming soon...... instruction for using will be detailed soon

Our paper is published on TRO

Our datasets are uploaded on Google Drive

Developers:

The codes of this repo are contributed by: Dongjiao He (贺东娇)

Properties

LIGO is a multi-sensor fusion framework that maximizes the complementary properties of both LiDAR and GNSS systems. This package achieves the following properties:

1. Competitive accuracy in trajectory estimation across large-scale scenarios.
2. Robustness to malfunctions of either GNSS or LiDAR sensors, enabling seamless handling of added or lost sensor signals during operation.
3. High-output-frequency odometry.
4. Capability of providing globally referenced pose estimations in both indoor and outdoor environments, suitable for ground vehicles and uncrewed aerial vehicles (UAVs).
5. No requirement for GNSS observations to be obtained exactly at the beginning or end time of LiDAR scans.
6. Robustness to large outliers and high noise levels in GNSS observations.

Hardware setups for self-collected datasets

Setup

Platform: DJI Matrice 300
Onboard computer: DJI Manifold 2-c 256G, CPU: Intel i7-8550U
LiDAR: Livox Mid360 and Livox Avia
IMU: Built-in IMU of Livox LiDAR
GNSS receiver: u-blox C099-F9P-2
GNSS antenna: B4QA4GGGB

Recording rates

LiDAR: 10Hz
IMU: 200Hz
GNSS: 10Hz
RTK: 10Hz

Recording software:

Operating system: Ubuntu 20.04
IMU and LiDAR driver: Livox driver
GNSS driver: ublox driver

ROS topics recorded

IMU: /livox/imu
LiDAR: /livox/lidar
RAW GNSS: /ublox_driver/range_meas
GNSS EPHEM: /ublox_driver/ephem and /ublox_driver/glo_ephem
IONO PARAMETER: /ublox_driver/iono_params
Onboard pos solution of ublox: /ublox_driver/receiver_pvt and /ublox_driver/receiver_lla
PPS time info: /ublox_driver/time_pulse_info

Time synchronization

PPS: Livox LiDARs can receive pps and gprmc given by the GNSS receiver
The time difference between LiDAR and IMU is zero, and between LiDAR and GNSS message is 18.0 s

RTK solution

Please follow the Differential GNSS section shown in ublox driver to get the differential GNSS solution online or offline. The self-collected datasets get the online RTK solution which are saved in the topic '/ublox_driver/receiver_pvt', the value of the 'carr_soln' as 1 and 'diff_soln' as 2 indicates the fix RTK solution.

Build

Prerequisites

We test LIGO on ubuntu 20.04 with ROS noetic, and C++17 compiler & Eigen 3 & GTSAM 4 & opencv 4.2.0 & pcl 1.10

Install Boost using command sudo apt-get install libboost-all-dev

Install Livox Driver

Install gnss_comm with its instuction

Make

clone the code to catkin_ws workspace

cd ~/catkin_ws/src/
git clone https://github.com/Joanna-HE/LIGO..git

compile the package

cd ~/catkin_ws/
source /PATH/TO/LIVOX_DRIVER/DEVEL/.setup.bash
source /PATH/TO/GNSS_COMM/DEVEL/.setup.bash
catkin_make
source ~/catkin_ws/devel/setup.bash

Demo

Performance on a sequence with severe LiDAR degeneracy