Research on 2D Laser Scan Matching Algorithm Based on Corner Features

Yanqing Wang; Deqiang Zhou; Hao Xu; Weifeng Sheng; Wenjuan Zuo; Qing Xi; Quyan Chen

doi:10.3788/LOP240994

Laser & Optoelectronics Progress, Volume. 62, Issue 4, 0415006(2025)

Research on 2D Laser Scan Matching Algorithm Based on Corner Features

Yanqing Wang^1、*, Deqiang Zhou¹, Hao Xu¹, Weifeng Sheng¹, Wenjuan Zuo¹, Qing Xi², and Quyan Chen²

Author Affiliations

¹School of Mechanical Engineering, Jiangnan University, Wuxi 214122, Jiangsu , China

²Wuxi Hongyi Intelligent Technology Co., Ltd.Wuxi 214000, Jiangsu , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(18)

Fig. 1. Definition of discrete points and discrete point sets in original point cloud data

Download full size

Fig. 2. SWC algorithm for identifying whether detected point is noise

Download full size

Fig. 3. LiDAR measurement model

Download full size

Fig. 4. Filtering of raw point cloud data using SWC algorithm

Download full size

Fig. 5. Mapping of two-dimensional point cloud in three-dimensional space

Download full size

Fig. 6. Remove outliers and fit extracted lines. (a) Fit lines close to vertical features; (b) fit regular line features

Download full size

Fig. 7. Flowchart for determining the type of intersection

Download full size

Fig. 8. Matching feature points. (a) Point clouds of target frame and matching frame; (b) corresponding feature points after matching

Download full size

Fig. 9. Solve pose transformation by non-mirror similarity affine transformation of successfully matched feature points

Download full size

Fig. 10. Collect 36 frames of data with equiangular rotation 360° at pentagon position, and calculate covariance of intersection coordinates

Download full size

Fig. 11. Calculation time per frame and number of intersections in a welding laboratory environment. (a) Real laboratory scenes; (b) 2D mapping of laboratory

Download full size

Fig. 12. Computation time and number of intersection points for each frame of data in a robotics laboratory environment. (a) Robotics laboratory; (b) corresponding 2D mapping

Download full size

Fig. 13. Test scenario and trajectory diagram. (a) Experimental scenarios; (b) calculation trajectories of different matching algorithms

Download full size

Table 1. Euclidean distance in ascending order between a specific intersection point and remaining intersection points in target frame along with attributes of remaining points of Fig. 6(a)

View table

Table 1. Euclidean distance in ascending order between a specific intersection point and remaining intersection points in target frame along with attributes of remaining points of Fig. 6(a)

Intersection 1		Intersection 2		Intersection 6		Intersection 7
Distance /m	Label	Distance /m	Label	Distance /m	Label	Distancem /m	Label
0	-1	0	-1	0	-1	0	1
0.747325	-1	0.747325593	-1	0.866452	1	0.866452	-1
0.775649	-1	0.773745324	0	1.611255	-1	1.598030	1
1.087829	0	1.072722933	-1	1.823952	1	1.812103	-1
5.062566	0	5.086140606	-1	3.132579	0	3.087903	-1
5.157471	-1	5.115085168	0	3.213890	-1	3.231250	0
6.587579	-1	6.576012817	1	7.423185	-1	7.400298	0
6.635139	1	6.623814912	-1	7.457795	0	7.454086	-1
8.198835	-1	8.174043690	1	8.198835	-1	8.174043	-1
8.224975	1	8.226792213	-1	8.226792	-1	8.224975	-1

Table 2. Feature point descriptions in matching frame of Fig. 6(b)

View table

Table 2. Feature point descriptions in matching frame of Fig. 6(b)

Intersection 1		Intersection 2		Intersection 6		Intersection 7
Distance /m	Label	Distance /m	Label	Distance /m	Label	Distance /m	Label
0	-1	0	-1	0	1	0	1
0.856734	1	0.891182	1	0.856734	-1	0.891182	-1
1.306920	1	1.654421	1	1.248031	-1	1.615621	1
1.528990	-1	1.894363	1	1.532711	1	1.812283	-1
1.654421	-1	2.961341	1	1.615621	1	2.863652	-1
1.812283	1	3.053057	-1	1.894363	-1	3.020219	1
5.796647	-1	7.451069	-1	5.779019	0	7.394640	0
5.860474	0	7.504524	0	5.824379	-1	7.429701	-1
6.549020	-1	8.203441	-1	6.529755	-1	8.145376	-1
6.604060	-1	8.250449	-1	6.569513	-1	8.177174	-1

Table 3. Comparison of accuracy and time consumption between SWC-ICE algorithm and other matching algorithms under different rotation angles

View table

Table 3. Comparison of accuracy and time consumption between SWC-ICE algorithm and other matching algorithms under different rotation angles

Classical matching algorithm
PL-ICP		GICP			NDT			SWC-ICE
Angle /（°）	Time /ms		Angle /（°）	Time /ms		Angle /（°）	Time /ms		Angle /（°）	Time /ms
1.93 $(2.2 \times 10^{- 8}$ ）	2.5	1.82 $(7.3 \times 10^{- 4})$		347.8	1.84 $(8.4 \times 10^{- 2}$ ）		1286.9	1.02 $(1.7 \times 10^{- 1}$ ）		52.1
4.93 $(3.2 \times 10^{- 3})$	3.3	4.92 $(5.2 \times 10^{- 2}$ ）		870.6	4.72 $(8.4 \times 10^{- 2}$ ）		1875.4	4.82 $(8.4 \times 10^{- 2}$ ）		55.8
9.83 $(1.2 \times 10^{- 2})$	3.9	9.85 $(4.2 \times 10^{- 3})$		799.2	11.35 $(1.146$ ）		2515.2	9.35 $(9.1 \times 10^{- 2}$ ）		53.3
4.97 $(4.773)$	5.8	18.42 （ $9.5 \times 10^{- 2}$ ）		1449.0	19.65 $(1.9 \times 10^{- 1})$		4080.1	18.95 $(6.5 \times 10^{- 3})$		59.2
11.60 $(6.7 \times 10^{- 1})$	9.3	11.17 $(2.0 \times 10^{- 1})$		1095.1	30.02 $(2.2 \times 10^{- 1})$		6024.2	30.15 $(9.5 \times 10^{- 3}$ ）		56.4

Table 4. Comparison of accuracy and time consumption between SWC-ICE algorithm and other matching algorithms under different translation distances

View table

Table 4. Comparison of accuracy and time consumption between SWC-ICE algorithm and other matching algorithms under different translation distances

SWC-ICE and other classical matching algorithms
PL-ICP		GICP			NDT			SWC-ICE
Distance /cm	Time /ms		Distance /cm	Time /ms		Distance /cm	Time /ms		Distance /cm	Time /ms
10.0 $(8.2 \times 10^{- 6})$	2.6	9.8 $(1.4 \times 10^{- 3}$ ）		386.1	10.5 $(5.5 \times 10^{- 2})$		5258.5	9.9 $(3.3 \times 10^{- 2}$ ）		56.5
20.0 $(1.4 \times 10^{- 4}$ ）	2.3	19.7 $(2.4 \times 10^{- 3}$ ）		280.6	21.7 $(3.3 \times 10^{- 2})$		2847.3	19.0 $(2.9 \times 10^{- 2}$ ）		43.0
30.0 $(2.6 \times 10^{- 4}$ ）	3.1	26.1 $(9.7 \times 10^{- 2}$ ）		260.9	28.4 $(9.2 \times 10^{- 2})$		2639.8	27.1 $(5.2 \times 10^{- 2})$		38.9
40.0 $(1.9 \times 10^{- 4}$ ）	2.2	26.6 $(3.2 \times 10^{- 3})$		185.0	23.5 $(1.4 \times 10^{- 1})$		3767.8	37.3 $(5.3 \times 10^{- 2}$ ）		39.7
50.0 $(3.2 \times 10^{- 4}$ ）	1.6	18.3 $(3.2 \times 10^{- 3}$ ）		135.2	36.6 $(1.5 \times 10^{- 1}$ ）		4313.1	47.3 $(2.9 \times 10^{- 2}$ ）		41.0

Table 5. Path estimation bias and length of different algorithms at 0.4 m displacement, 20° rotation, and average matching time

View table

Table 5. Path estimation bias and length of different algorithms at 0.4 m displacement, 20° rotation, and average matching time

Estimation method	x position error /m	y position error/m	Angular error /（°）	Path length /m	Average time consumption /ms
NDT	10.00	14.20	89.7	36.23	326.8
PL-ICP	29.70	0.74	79.8	31.64	3.8
GICP	20.44	0.14	70.7	21.87	248.8
SWC-ICE+PL-ICP	0.15	0.17	8.6	35.55	58.9

Tools

Get Citation

Copy Citation Text

Yanqing Wang, Deqiang Zhou, Hao Xu, Weifeng Sheng, Wenjuan Zuo, Qing Xi, Quyan Chen. Research on 2D Laser Scan Matching Algorithm Based on Corner Features[J]. Laser & Optoelectronics Progress, 2025, 62(4): 0415006

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Machine Vision

Received: Mar. 29, 2024

Accepted: Jul. 29, 2024

Published Online: Feb. 14, 2025

The Author Email:

DOI:10.3788/LOP240994

CSTR:32186.14.LOP240994

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology