Elevated high-precision mapping and localization technology for periodic inspections
Abstract Due to the absence of Global Navigation Satellite System (GNSS) information within tunnels, mobile mapping systems encounter issues with cumulative errors, making it difficult to obtain precise absolute pose information. Consequently, reproducing point cloud maps becomes challenging, thereb...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-02-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-86133-5 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Due to the absence of Global Navigation Satellite System (GNSS) information within tunnels, mobile mapping systems encounter issues with cumulative errors, making it difficult to obtain precise absolute pose information. Consequently, reproducing point cloud maps becomes challenging, thereby limiting the capability of tunnel defect detection methods to conduct repeated inspections, posing a major obstacle in assessing defect progression trends. To address this challenge, this paper proposes a high-dimensional, multi-constraint framework, which integrates not only front-end odometry techniques based on the Error-State Kalman Filter but also back-end optimization techniques utilizing factor graphs. Additionally, we introduce hundred-meter marker detection and recognition methods to enhance loop closure detection, effectively eliminating accumulated errors and achieving precise initialization of train positioning even in the absence of GNSS signals. Multiple experiments conducted in subway tunnel scenarios demonstrate the robust trajectory estimation capability of the proposed method in long tunnel scenarios, showcasing clear advantages over classical multi-sensor fusion methods prone to failure due to sensor degradation. Specifically, the method exhibits accumulated errors in map trajectory consistency below 0.02% in tunnel scenes, with an average positioning error of only 0.05 m, demonstrating its high accuracy and reliability. Currently, the method has been successfully validated in subway scenarios, showing promising application prospects. |
|---|---|
| ISSN: | 2045-2322 |