Citation: Liu, Y.; Zhang, W.; Li, F.;
Zuo, Z.; Huang, Q. Real-Time Lidar
Odometry and Mapping with Loop
Closure. Sensors 2022, 22, 4373.
https://doi.org/10.3390/s22124373
Academic Editors: Luis Payá, Oscar
Reinoso García and Helder Jesus
Araújo
Received: 15 May 2022
Accepted: 7 June 2022
Published: 9 June 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Real-Time Lidar Odometry and Mapping with Loop Closure
Yonghui Liu
1
, Weimin Zhang
1,2,3,
* , Fangxing Li
1,2,3
, Zhengqing Zuo
1
and Qiang Huang
1,2,3
1
School of Mechatronical Engineering, Beijing Institute of Technology, Beijing 100081, China;
3120200157@bit.edu.cn (Y.L.); wonk2000@bit.edu.cn (F.L.); 3120215092@bit.edu.cn (Z.Z.);
qhuang@bit.edu.cn (Q.H.)
2
Key Laboratory of Biomimetic Robots and Systems, Ministry of Education, Beijing Institute of Technology,
Beijing 100081, China
3
Beijing Advanced Innovation Center for Intelligent Robots and Systems, Beijing 100081, China
* Correspondence: zhwm@bit.edu.cn
Abstract:
Real-time performance and global consistency are extremely important in Simultaneous
Localization and Mapping (SLAM) problems. Classic lidar-based SLAM systems often consist of
front-end odometry and back-end pose optimization. However, due to expensive computation, it is
often difficult to achieve loop-closure detection without compromising the real-time performance
of the odometry. We propose a SLAM system where scan-to-submap-based local lidar odometry
and global pose optimization based on submap construction as well as loop-closure detection are
designed as separated from each other. In our work, extracted edge and surface feature points are
inserted into two consecutive feature submaps and added to the pose graph prepared for loop-closure
detection and global pose optimization. In addition, a submap is added to the pose graph for global
data association when it is marked as in a finished state. In particular, a method to filter out false
loops is proposed to accelerate the construction of constraints in the pose graph. The proposed
method is evaluated on public datasets and achieves competitive performance with pose estimation
frequency over 15 Hz in local lidar odometry and low drift in global consistency.
Keywords:
real-time lidar odometry; submap-based loop-closure detection; pose graph optimization;
simultaneous localization and mapping (SLAM)
1. Introduction
Simultaneous Localization and Mapping is a significant issue for mobile robots and
autonomous driving vehicles. Vision-based and lidar-based SLAM has been widely studied,
proposing a series of noted methods to achieve real-time and high-performance pose
estimation. Achieving real-time pose estimation on devices with limited computational
resources remains a challenge for both vision and lidar SLAM. For vision-based SLAM
using monocular, stereo, or RGB-D cameras, loop-closure detection and relocalization is
not a particularly difficult task because a bag-of-words library can be trained in advance,
which is a creative approach for data association on a global scale.
Compared with vision-based SLAM, there is a lack of research on loop-closure detec-
tion in lidar-based SLAM, although lidar-based methods are more tolerant of illumination
and initialization. In our work, we focus on lidar-based real-time pose estimation and a
mapping method with loop closure.
A great deal of attention has been paid to lidar-based pose estimation and map-
ping methods for the last few years. For instance, a feature-based 3D lidar-based SLAM
framework called lidar odometry and mapping in real-time (LOAM) [
1
] achieved both
low-drift and low computational complexity. Until now, LOAM and many variants of
LOAM have been widely studied because of their state-of-art performance on the public
dataset KITTI [
2
]. Furthermore, usually a Euclidean distance-based loop-closure detec-
tion approach is used to minimize the accumulated error as with LeGO-LOAM [
3
] and
LIO-SAM [4].
Sensors 2022, 22, 4373. https://doi.org/10.3390/s22124373 https://www.mdpi.com/journal/sensors
