.png){kind=icon}
.png){kind=icon}
Return
DMLL: Differential-Map-Aided LiDAR-Based Localization
.png){kind=icon}
PRE
AI
DOI：10.1109/TIM.2023.3324349.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This article presents differential-map-aided LiDAR-based localization (DMLL), a localization framework that combines map-based localization and LiDAR odometry to achieve accurate and reliable robot pose estimation in partially known environments. In DMLL, measurement from map-matching are novelly modeled as differential constraints in the factor graph, rather than absolute constraints. This simplifies the fusion process and avoids initial alignment errors. We design uncertainty quantization for both the LiDAR odometry measurement and the map-matching measurement. Furthermore, we design an adaptive decision mechanism that selectively triggers the map-matching module based on the quantified uncertainty of these measurements. This mechanism achieves a balance between resource consumption and localization accuracy while also being capable of detecting and discarding erroneous data, thus improving the overall localization reliability. Our proposed method is extensively evaluated on both public datasets and our own field-collected data. The result indicates that our method can achieve highly accurate localization in practical scenarios.
Keywords:
Location awareness
Laser radar
Simultaneous localization and mapping
Odometry
Reliability
Robots
Real-time systems
LiDAR odometry
map-based localization
simultaneous localization and mapping (SLAM)
Journal
IF:
5.9
Papers:
1.9W
Citations:
5.8W