.png){kind=icon}
Return
Ground texture-based multi-sequence mapping and prior map localization method
.png){kind=icon}
PRE
AI
DOI：10.1088/1361-6501/ae4416.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
With the continuous advancement of mobile robotics technology, an increasing number of enterprises are adopting indoor mobile robots to enhance operational efficiency and service quality. However, environmental changes can significantly impact system accuracy and robustness during robot operation. Given that indoor environments are known and feature stable, flat flooring with minimal alterations, they provide excellent planar prior information constraints. This paper proposes a solution for achieving globally consistent mapping and high-precision localization by collecting ground texture data. The solution establishes a comprehensive system encompassing global map construction, ground database creation, and high-accuracy localization, utilizing downward-facing cameras to capture ground texture information for improved positioning accuracy and map reusability. The system employs multi-sequence data collection and Z-shaped coverage path planning to systematically acquire ground texture data across target areas. During map construction, bundle adjustment optimization is applied to globally refine keyframe poses, thereby enhancing geometric accuracy and consistency. Additionally, the proposed rapid localization framework achieves a mis-matching rate below 3% under typical operating conditions, specifically when the robot velocity is below 0.15 m s−1. Extensive experiments across diverse floor environments demonstrate the solution’s superior accuracy and robustness.
Keywords:
ground texture
multi-sequence mapping
prior map localization
bundle adjustment
indoor mobile robots
Journal
IF:
3.4
Papers:
2.7K
Citations:
2.3W
Organization
Cited Papers
ShareSave
SPROSAC: Streamlined progressive sample consensus for coarse-fine point cloud registration.png)
APPLIED INTELLIGENCE
APPLIED INTELLIGENCEIF3.5
PREAI
ShareSaveShareSaveShareSave
View more