LIO-SAM

1 Introduction

LIO-SAM is a framework for tightly-coupled lidar inertial odometry via smoothing and mapping, that achieves highly accurate, real-time mobile robot trajectory estimation and map-building. LIO-SAM formulates lidar-inertial odometry atop a factor graph, allowing a multitude of relative and absolute measurements, including loop closures, to be incorporated from different sources as factors into the system. The estimated motion from inertial measurement unit (IMU) pre-integration de-skews point clouds and produces an initial guess for lidar odometry optimization. The obtained lidar odometry solution is used to estimate the bias of the IMU. To ensure high performance in real-time, it marginalize old lidar scans for pose optimization, rather than matching lidar scans to a global map. Scan-matching at a local scale instead of a global scale significantly improves the real-time performance of the system, as does the selective introduction of keyframes, and an efficient sliding window approach that registers a new keyframe to a fixed-size set of prior “sub-keyframes.”
Here are some reference links of code and paper: code link. paper link.

Image description — Fig. 1. Workflow diagram of LIO-SAM

2 Run Process

2.1 Requirement

ubuntu 20.04 、ros-noetic、cmake、gcc、g++、eigen、pcl1.10、gtsam、code、bag .

During the installation of third-party libraries, there may be many version mismatch issues. Below are just examples of installation.Please make sure to match the versions, otherwise compilation will have significant problems. If there are any issues during the appeal process, please search online to resolve them.

Install dependence


              sudo apt-get update 

              sudo apt-get install git gcc g++ vim make cmake

Install Eigen3


              download Eigen3 

              cd eigen3

              mkdir build && cd build

              cmake ..

              make

              sudo make install

Install gtsam


              wget -O ~/Downloads/gtsam.zip https://github.com/borglab/gtsam/archive/4.0.3-alpha2.zip

              cd ~/Downloads/ && unzip gtsam.zip -d ~/Downloads/

              cd ~/Downloads/gtsam-4.0.3-alpha2/

              mkdir build && cd build

              cmake ..

              make

              sudo make install

All the necessary components for the algorithm have been installed here, and it is important to ensure the corresponding version relationships, otherwise there will be trouble later.

2.2 Build LIO-SAM

Clone the repository and catkin_make:


            cd ~/catkin_ws/src

            git clone https://github.com/TixiaoShan/LIO-SAM.git 

            cd ../

            catkin_make

            source ~/catkin_ws/devel/setup.bash

2.3 Run LIO-SAM

Run the algorithm and play the dataset, in addition to recording the trajectory topic generated by the algorithm. The process of running the algorithm is shown on the screen in Figure 2.


            roslaunch lio_sam run.launch 

            rosbag play night_liosam_traj.bag 

            rosbag record -o night_liosam_traj.bag /lio_sam/mapping/odometry

3 Evaluation

First of all, we need to convert the recorded trajectory bag into TUM or txt format for later evaluation using evo. You can use the commands provided with evo to convert to TUM format, or use Python script to convert to txt format.



          evo_traj bag night_liosam_traj.bag /lio_sam/mapping/odometry  --save_as_tum               

          python bagtotum.py night_liosam_traj.bag /lio_sam/mapping/odometry liosam_night_tum.txt

And then we use evo_traj evaluates the trajectory and ground truth generated by the algorithm.


          evo_traj tum lio_sam_mapping_odometry.tum --ref=gt.txt -va -p

Afterwards, we use evo_ape to evaluate the absolute error between the trajectory generated by the algorithm and the ground truth.


          evo_ape tum lio_sam_mapping_odometry.tum gt.txt -a  -p                  

          evo_ape tum liosam _night_tum.txt gt.txt -a  -p

The results generated by the evaluation trajectory are shown below as Fig3 Trajectory of LIO-SAM and GT. Fig4 APE of LIO-SAM and GT. Fig5 XYZ of LIO-SAM and GT. Fig6 RPY of LIO-SAM and GT.

Fig3. Trajectory of LIO-SAM and GT

Fig4. APE of LIO-SAM and GT

Fig5. XYZ of LIO-SAM and GT

Fig6. RPY of LIO-SAM and GT

4 Result and Discussion

We use APE to evaluate the accuracy of the algorithm, calculate the Translation mean, rmse, and Rotation mean, rmse using the evo tool, and divide them by the total distance of gt, using percentage as the final evaluation indicator.The units are m/% and °/%, respectively. We will validate and evaluate all sequences from all platforms using this algorithm and summarize them in the following table:

4.1 M3DSS-driving

Platform	Sequence	GT distance	Translation			Rotation
Platform	Sequence	GT distance	mean	rmse	MPE	mean	rmse	MRE
Driving	Urban day	0	0	0	0	0	0	0
	Urban night	0	0	0	0	0	0	0
	Urban Tunnel day	0	0	0	0	0	0	0
	Urban Tunnel night	0	0	0	0	0	0	0
	Urban Harsh day	0	0	0	0	0	0	0
	Urban Harsh night	0	0	0	0	0	0	0
	Urban Highway day	0	0	0	0	0	0	0
	Urban Highway night	0	0	0	0	0	0	0

4.2 M3DSS-UGV

Platform	Sequence	GT distance	Translation			Rotation
Platform	Sequence	GT distance	mean	rmse	MPE	mean	rmse	MRE
UGV	Street day	0	0	0	0	0	0	0
	Street night	0	0	0	0	0	0	0
	Parkway day	0	0	0	0	0	0	0
	Parkway night	0	0	0	0	0	0	0
	Elevator	0	0	0	0	0	0	0
	MCR 01	0	0	0	0	0	0	0
	MCR 02	0	0	0	0	0	0	0
	MCR 03	0	0	0	0	0	0	0

4.3 M3DSS-drone

Platform	Sequence	GT distance	Translation			Rotation
Platform	Sequence	GT distance	mean	rmse	MPE	mean	rmse	MRE
drone	Garden day	0	0	0	0	0	0	0
	Garden night	0	0	0	0	0	0	0
	Landfill day	0	0	0	0	0	0	0
	Landfill night	0	0	0	0	0	0	0
	Field translation	0	0	0	0	0	0	0
	Field rotation	0	0	0	0	0	0	0
	MCR 01	0	0	0	0	0	0	0
	MCR 02	0	0	0	0	0	0	0
	MCR 03	0	0	0	0	0	0	0
	MCR 04	0	0	0	0	0	0	0
	MCR 05	0	0	0	0	0	0	0

4.4 M3DSS-Handheld

Platform	Sequence	GT distance	Translation			Rotation
Platform	Sequence	GT distance	mean	rmse	MPE	mean	rmse	MRE
Handheld	Garden day	0	0	0	0	0	0	0
	Garden night	0	0	0	0	0	0	0
	Building day	0	0	0	0	0	0	0
	Building night	0	0	0	0	0	0	0
	Elevator	0	0	0	0	0	0	0
	Escalator	0	0	0	0	0	0	0
	MCR 01	0	0	0	0	0	0	0
	MCR 02	0	0	0	0	0	0	0
	MCR 03	0	0	0	0	0	0	0
	MCR translation 01	0	0	0	0	0	0	0
	MCR translation 02	0	0	0	0	0	0	0
	MCR rotation 01	0	0	0	0	0	0	0
	MCR rotation 02	0	0	0	0	0	0	0
	MCR flip series	0	0	0	0	0	0	0

4.5 M3DSS-Quadruped

Platform	Sequence	GT distance	Translation			Rotation
Platform	Sequence	GT distance	mean	rmse	MPE	mean	rmse	MRE
Quadruped	hku canteen day	0	0	0	0	0	0	0
	hku canteen night	0	0	0	0	0	0	0
	hku garden day	0	0	0	0	0	0	0
	hku garden night	0	0	0	0	0	0	0
	hku forest day	0	0	0	0	0	0	0
	hku mountain day	0	0	0	0	0	0	0
	hku mainbuilding day	0	0	0	0	0	0	0
	hku mainbuilding night	0	0	0	0	0	0	0
	hku MCR 01	0	0	0	0	0	0	0
	hku MCR 02	0	0	0	0	0	0	0