A visual SLAM algorithm based on adaptive inertial navigation assistant feature matching

Xiaoxue JIA; Dongqing ZHAO; Letian ZHANG; Guorui XIAO; Qing XU

doi:10.37188/OPE.20233105.0621

Optics and Precision Engineering, Volume. 31, Issue 5, 621(2023)

A visual SLAM algorithm based on adaptive inertial navigation assistant feature matching

Xiaoxue JIA... Dongqing ZHAO*, Letian ZHANG, Guorui XIAO and Qing XU |Show fewer author(s)

Author Affiliations

Geography Space Information Institute， Information Engineering University， Zhengzhou450001， China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(11)

Fig. 1. Flow chart of IMU aided image feature matching

Download full size

View in Article

Fig. 2. Imaging model of stereo camera

Download full size

View in Article

Fig. 3. Flow chart of determining adaptive feature matching search radius

Download full size

View in Article

Fig. 4. Cumulative distribution function of search radius for adaptive feature matching of sequences

Download full size

View in Article

Fig. 5. Average time-consuming of each feature point matching before and after improvement

Download full size

View in Article

Fig. 6. V101 sequence tracking thread location before and after improvement

Download full size

View in Article

Fig. 7. Absolute position and pose errors of V101 sequence before and after improvement

Download full size

View in Article

Fig. 8. Dense map of the V101 sequence

Download full size

View in Article

Table 1. Sequences chosen for the experiment

View table

View in Article

Table 1. Sequences chosen for the experiment

序列	场景	距离/m	陀螺仪噪声 /（rad·s^-1·Hz $^{- \frac{1}{2}}$ ）	加速度计噪声/（m·s^-2·Hz $^{- \frac{1}{2}}$ ）
MH01	厂房	80.6	0.000 17	0.002
V101	房间1	58.6
V201	房间2	36.5
Corridor1	走廊	305	0.000 16	0.028
Magistrale1	大厅	918
Room1	室内	146
Slides1	滑梯	289

Table 2. Root-mean-square of absolute pose error of each sequence tracking thread

View table

View in Article

Table 2. Root-mean-square of absolute pose error of each sequence tracking thread

数据集	序列	Before/m	After/m	精度提升
EuRoC	MH01	0.374 1	0.217 0	41.99%
	V101	0.525 9	0.327 5	37.73%
	V201	0.226 6	0.103 5	54.32%
TUM	Corridor1	0.123 2	0.077 8	36.85%
	Magistrale1	0.907 2	0.473 9	47.76%
	Room1	0.111 6	0.079 4	28.85%
	Slides1	0.403 7	0.326 5	19.12%

Table 3. Root-mean-square of the absolute pose error of the global pose of each sequence

View table

View in Article

Table 3. Root-mean-square of the absolute pose error of the global pose of each sequence

数据集	序列	Before/m	After/m	精度提升
EuRoC	MH01	0.038 3	0.031 0	19.06%
	V101	0.037 9	0.037 8	0.26%
	V201	0.025 8	0.022 7	12.02%
TUM	Corridor1	0.037 9	0.031 6	16.62%
	Magistrale1	0.894 5	0.466 2	47.88%
	Room1	0.012 4	0.011 1	10.48%
	Slides1	0.368 2	0.337 6	8.31%

Tools

Get Citation

Copy Citation Text

Xiaoxue JIA, Dongqing ZHAO, Letian ZHANG, Guorui XIAO, Qing XU. A visual SLAM algorithm based on adaptive inertial navigation assistant feature matching[J]. Optics and Precision Engineering, 2023, 31(5): 621

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Three-dimensional topographic mapping

Received: May. 27, 2022

Accepted: --

Published Online: Apr. 4, 2023

The Author Email: Dongqing ZHAO (dongqingtree@qq.com)

DOI:10.37188/OPE.20233105.0621

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Sequences chosen for the experiment

Table 1. Sequences chosen for the experiment

Table 2. Root-mean-square of absolute pose error of each sequence tracking thread

Table 2. Root-mean-square of absolute pose error of each sequence tracking thread

Table 3. Root-mean-square of the absolute pose error of the global pose of each sequence

Table 3. Root-mean-square of the absolute pose error of the global pose of each sequence

微信扫一扫：分享