A LiDAR/Visual SLAM Backend with Loop Closure Detection and Graph Optimization

Shoubin Chen,Baoding Zhou,Qingquan Li,Weixing Xue,Changhui Jiang

doi:10.3390/rs13142720

Shoubin Chen, Baoding Zhou + Show 3 more

Open Access

https://doi.org/10.3390/rs13142720

Copy DOI

Abstract

LiDAR (light detection and ranging), as an active sensor, is investigated in the simultaneous localization and mapping (SLAM) system. Typically, a LiDAR SLAM system consists of front-end odometry and back-end optimization modules. Loop closure detection and pose graph optimization are the key factors determining the performance of the LiDAR SLAM system. However, the LiDAR works at a single wavelength (905 nm), and few textures or visual features are extracted, which restricts the performance of point clouds matching based loop closure detection and graph optimization. With the aim of improving LiDAR SLAM performance, in this paper, we proposed a LiDAR and visual SLAM backend, which utilizes LiDAR geometry features and visual features to accomplish loop closure detection. Firstly, the bag of word (BoW) model, describing the visual similarities, was constructed to assist in the loop closure detection and, secondly, point clouds re-matching was conducted to verify the loop closure detection and accomplish graph optimization. Experiments with different datasets were carried out for assessing the proposed method, and the results demonstrated that the inclusion of the visual features effectively helped with the loop closure detection and improved LiDAR SLAM performance. In addition, the source code, which is open source, is available for download once you contact the corresponding author.

Highlights

The concept of simultaneous localization and mapping (SLAM) was first proposed in 1986 by Cheeseman [1,2]
After more than 30 years of development, SLAM technology is no longer limited to theoretical research in the field of robotics and automation; it is promoted in many applications, i.e., intelligent robots, autonomous driving, mobile surveying, and mapping [3]
We hoped that true positive (TP) and true negative (TN) would appear as much as possible, whereas we hoped that FP and false negative (FN) would appear as little as possible or not at all

Summary

Introduction

The concept of simultaneous localization and mapping (SLAM) was first proposed in 1986 by Cheeseman [1,2]. VLOAM has achieved relatively high accuracy in the state estimation of the front-end odometer, but the lack of back-end loop closure detection and global graph optimization will inevitably affect the positioning accuracy and the consistency of map construction, and it will continue to degrade over time. A LiDAR/visual SLAM based on loop closure detection and global graph optimization (GGO) is constructed to improves the accuracy of the positioning trajectory and the consistency of the point clouds map. KTTI datasets and WHU Kylin backpack datasets were utilized to evaluate the performance of the visual BoW similarity-based loop closure detection, and the position accuracy and point clouds map are presented for analyzing the performance of the proposed method.

System Architecture

System

Visual BOW Similarity

Loop Closure Detection Verifying and Its Accuracy

Detection Results Reference

Global Pose Construction

Globe Pose Graph Optimization

Experiments and Results

Results Analysis

Results and Analysis

Comparisons with Google Cartographer

Conclusions