2019
DOI: 10.1177/0278364919863090
|View full text |Cite
|
Sign up to set email alerts
|

SegMap: Segment-based mapping and localization using data-driven descriptors

Abstract: Precisely estimating a robot’s pose in a prior, global map is a fundamental capability for mobile robotics, e.g., autonomous driving or exploration in disaster zones. This task, however, remains challenging in unstructured, dynamic environments, where local features are not discriminative enough and global scene descriptors only provide coarse information. We therefore present SegMap: a map representation solution for localization and mapping based on the extraction of segments in 3D point clouds. Working at t… Show more

Help me understand this report
View preprint versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

3
90
0

Year Published

2020
2020
2024
2024

Publication Types

Select...
6
3

Relationship

0
9

Authors

Journals

citations
Cited by 173 publications
(100 citation statements)
references
References 52 publications
3
90
0
Order By: Relevance
“…SegMap aims to extract meaningful features for global retrieval while the semantic class types were limited to vehicles, buildings, and others. The performance of SegMap is further improved in [ 110 ].…”
Section: Role Of Deep Learning In Loop Closure Detectionmentioning
confidence: 99%
“…SegMap aims to extract meaningful features for global retrieval while the semantic class types were limited to vehicles, buildings, and others. The performance of SegMap is further improved in [ 110 ].…”
Section: Role Of Deep Learning In Loop Closure Detectionmentioning
confidence: 99%
“…We consider that the reserved points will meet the needs of location positioning from two aspects. Firstly, the local feature based maps are widely used for precise registration or localization [3], [10], [11]. Edges, walls or semantic objects are mainly used for robust matching in these papers.…”
Section: Map Compression Comparisonmentioning
confidence: 99%
“…In this paper, the idea is extended to the LiDAR maps by proposing an optimization strategy to approximate the solution of original programming problem within tractable time. In many other papers, geometric features or semantics can also be extracted for robust localization [3], [10], [11], so we think the geometric properties might also be correlated with the localization performance. Combing with these existing works, the second hypothesis is raised that the observation count is correlated to the geometric property of point cloud map, which makes the method possible to adapt to the new environments.…”
mentioning
confidence: 99%
“…It incrementally integrates semantic information into a dense 3-D map. Dube et al proposed SegMap [ 33 ] which is segment-based mapping using data-driven descriptors. The proposed descriptor was extracted by a neural network and the descriptor was used for mapping.…”
Section: Related Workmentioning
confidence: 99%