Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
GTR-Loc: Geospatial Text Regularization Assisted Outdoor LiDAR Localization
Authors: Shangshu Yu, Wen Li, Xiaotian Sun, Zhimin Yuan, Xin Wang, Sijie Wang, Rui She, Cheng Wang
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on challenging large-scale outdoor datasets, including QEOxford, Oxford Radar Robot Car, and NCLT, demonstrate the effectiveness of GTR-Loc. Our method significantly outperforms state-of-the-art approaches, notably achieving a 9.64%/8.04% improvement in position/orientation accuracy on QEOxford. |
| Researcher Affiliation | Academia | 1School of Computer Science and Engineering, Northeastern University, Shenyang 110819, China 2Fujian Key Laboratory of Sensing and Computing for Smart Cities, Xiamen University, China 3Key Laboratory of Multimedia Trusted Perception and Efficient Computing, Ministry of Education of China, Xiamen University, China 4School of Artificial Intelligence and Software Engineering, Nanyang Normal University, China 5Nanyang Technological University, Singapore 6Beihang University, China EMAIL EMAIL |
| Pseudocode | No | The paper describes its methodology through figures (Figure 2, Figure 3) and textual descriptions in Section 3 and its subsections, but it does not contain explicit pseudocode blocks or algorithm listings. |
| Open Source Code | Yes | Our code is available at https://github.com/PSYZ1234/GTR-Loc. |
| Open Datasets | Yes | We evaluate GTR-Loc on three commonly used large-scale outdoor datasets: Oxford Radar Robot Car (Oxford) [2], QEOxford [23], and NCLT [26]. |
| Dataset Splits | Yes | Further details on the data splits can be found in Tab. 1 and Tab. 2. Table 1: Details of the Oxford dataset. ... Table 2: Details of the NCLT dataset. ... Both tables list specific sequences and categorize them as 'Train' or 'Eval' split. |
| Hardware Specification | Yes | All experiments are conducted on a single NVIDIA RTX 4090 GPU. |
| Software Dependencies | No | GTR-Loc is implemented in Py Torch [28] and Minkowski Engine [10]. The paper mentions the software used but does not provide specific version numbers for PyTorch or Minkowski Engine. |
| Experiment Setup | Yes | We adopt the Adam W optimizer with a one-cycle learning rate schedule ranging from 5e 4 to 5e 3. The model is trained for 25 epochs on Oxford/QEOxford and 30 epochs on NCLT. Input point clouds are voxel-downsampled with a voxel size of 0.25m on Oxford/QEOxford and 0.3m on NCLT. The number of districts z is set to 100, and the number of directions d is set to 16. α in Eq. 3 is set to 0.1. β1 and β2 in Eq. 6 are set to 1, β3 is set to 0.1. |