IF-LIO

IF-LIO: Intensity-Enhanced Front-End Matching for Robust LiDAR–Inertial Odometry under Geometric Degeneracy

Introduction

IF-LIO is a degeneration-aware intensity-assisted LiDAR-SLAM framework that improves localization robustness in geometrically degenerate environments by selectively leveraging intensity information as complementary constraints.

1 Build

cd <your workspace>
mkdir src
cd src
git clone https://github.com/usersky-cmyk/IF-LIO.git
cd ..
catkin_make

2 Run

2.1 Newer College Dataset

Download Newer College Dataset from https://ori-drs.github.io/newer-college-dataset/

source devel/setup.bash
roslaunch if_lio Newer_college.launch

2.2 ENWIDE Dataset

Download ENWIDE Dataset from https://projects.asl.ethz.ch/datasets/enwide/

source devel/setup.bash
roslaunch if_lio ENWIDE.launch

2.3 Self-collected Dataset

Download Self-collected Dataset from https://pan.baidu.com/s/1O7TMEtp_MsC5haGQPSLjsw (password: 1234)

source devel/setup.bash
roslaunch if_lio avia.launch

3 Result

3.1 Newer College Dataset

Method (Length m)	Quad-Hard (234.81)	Cloister (428.79)	Stairs (57.04)	Park (2396.20)
LIO-SAM	0.299	0.145	×	1.566
FAST-LIO2	0.049	0.078	×	0.310
RI-LIO	0.237	0.285	×	89.289
COIN-LIO	0.046	0.078	0.102	0.287
PG-LIO	0.054	0.097	0.077	0.331
Ours	0.045	0.067	0.092	0.241

quad-hard

3.2 ENWIDE Dataset

Method (Length m)	TunnelS (251.58)	TunnelD (179.71)	IntersectionS (279.47)	IntersectionD (388.47)	RunwayS (333.57)	RunwayD (357.14)	FieldS (232.70)	FieldD (287.91)	KatzenseeS (242.88)	KatzenseeD (177.20)
LIO-SAM	×	×	×	×	×	×	×	×	5.588	×
FAST-LIO2	×	×	12.437	23.800	×	×	0.163	9.209	1.122	1.02
RI-LIO	×	×	×	×	×	×	1.721	24.851	×	×
COIN-LIO	0.743	0.487	0.466	1.912	1.033	2.437	0.232	0.581	0.412	0.592
PG-LIO	0.635	0.425	0.279	0.473	0.491	3.273	0.274	0.206	0.260	0.355
Ours	1.786	0.937	0.231	0.413	0.480	2.093	0.194	0.192	0.246	0.271

runway_d

3.3 GEODE Dataset

Method (Length m)	Shield_tunnel1 (γ)	Shield_tunnel2 (γ)	Shield_tunnel3 (γ)	Shield_tunnel4 (γ)	Shield_tunnel5 (γ)	Shield_tunnel6 (γ)
LIO-SAM	×	×	×	×	×	×
FAST-LIO2	×	×	×	×	×	×
RI-LIO	--	--	--	--	--	--
COIN-LIO	--	--	--	--	--	--
PG-LIO	--	--	--	--	--	--
Ours	1.189	0.778	1.996	1.972	7.257	0.428

3.4 self-collected Dataset

playground

4 Acknowledgements

Thanks for the below great open-source project for providing references to this work.

• FAST-LIO
• btsa
• RMS
• PV-LIO