Learning to See the Wood for the Trees: Deep Laser Localization in Urban and Natural Environments on a CPU
Open accessAbstract
Localization in challenging, natural environments, such as forests or woodlands, is an important capability for many applications from guiding a robot navigating along a forest trail to monitoring vegetation growth with handheld sensors. In this letter, we explore laser-based localization in both urban and natural environments, which is suitable for online applications. We propose a deep learning approach capable of learning meaningful descriptors directly from three-dimensional point clouds by comparing triplets (anchor, positive, and negative examples). The approach learns a feature space representation for a set of segmented point clouds that are matched between a current and previous observations. Our learning method is tailored toward loop closure detection resulting in a small model that can be deployed using only a CPU. The proposed learning method would allow the full pipeline to run on robots with limited computational payloads, such as drones, quadrupeds, or Unmanned Ground Vehicles (UGVs).
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 1× · see citing works
“The main drawback of classical approaches that use LIDAR data is that they rely on handcrafted features which do not perform as well as learned features [20].”
BackgroundAdam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 2× · see citing works
“We use an initial learning rate of η = 10−3, decaying to minimum η = 10−6 with ADAM [30] as optimizer.”
Method(2014) · Cited by 33,419 · see citing works
Fischler, M. A., & Bolles, R. C. (1981) · Communications of the ACM · doi:10.1145/358669.358692 · Cited by 25,893 · cited 2× · see citing works
“used the same geometric consistency algorithm for all baselines substituting RANSAC [31] with PROSAC [9] for the proposed method to incorporate the quality of the features.”
MethodVision meets robotics: The KITTI dataset
Geiger, A., Lenz, P., Stiller, C., & Urtasun, R. (2013) · The International Journal of Robotics Research · doi:10.1177/0278364913491297 · Cited by 9,913 · cited 4× · see citing works
“These types of approaches typically assume a certain inherent structure in the scene, image features are dependant on repeatable camera viewpoint [3], and methods are often tested in urban environments which guide the robot along the route in question [4].”
BackgroundMethodInfluentialPointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric SpacePreprint
Qi, C. R., Yi, L., Su, H., & Guibas, L. J. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1706.02413 · Cited by 9,834 · cited 1× · see citing works
“The issue of handling point clouds was addressed in [23], [24] by learning”
BackgroundPointNet: Deep Learning on Point Sets for 3D Classification and Segmentation
Charles, R., Su, H., Mo, K., & Guibas, L. J. (2017) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2017.16 · Cited by 9,767 · cited 1× · see citing works
“The issue of handling point clouds was addressed in [23], [24] by learning”
BackgroundMulti-scale context aggregation by dilated convolutions
Yu, F., & Koltun, V. (2015) · arXiv · doi:10.48550/arxiv.1511.07122 · Cited by 7,738 · cited 1× · see citing works
“In order for the top X -conv operators to see a larger portion of the point cloud a dilation rate (D) is applied [27].”
MethodORB-SLAM: A Versatile and Accurate Monocular SLAM System
Mur-Artal, R., Montiel, J. M. M., & Tardos, J. D. (2015) · IEEE Transactions on Robotics · doi:10.1109/tro.2015.2463671 · Cited by 6,778 · see citing works
Distance Metric Learning for Large Margin Nearest Neighbor Classification
Weinberger, K. Q., & Saul, L. K. (2009) · Journal of Machine Learning Research · doi:10.5555/1577069.1577078 · Cited by 4,651 · cited 1× · see citing works
“ation of the triplet loss [28] is used.”
MethodORB-SLAM: A Versatile and Accurate Monocular SLAM System
Mur-Artal, R., Montiel, J. M. M., & Tardos, J. D. (2015) · Zaguan (Universidad de Zaragoza) · Cited by 3,153 · see citing works
A solution to the simultaneous localization and map building (SLAM) problem
Dissanayake, M. W. M. G., Newman, P., Clark, S., Durrant-Whyte, H. F., & Csorba, M. (2001) · IEEE Transactions on Robotics and Automation · doi:10.1109/70.938381 · Cited by 2,661 · see citing works
A Point Set Generation Network for 3D Object Reconstruction from a Single Image
Fan, H., Su, H., & Guibas, L. J. (2017) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2017.264 · Cited by 2,410 · see citing works
Fischler, M. A., & Bolles, R. C. (1987) · Elsevier eBooks · doi:10.1016/b978-0-08-051581-6.50070-2 · Cited by 2,192 · see citing works
Real-time loop closure in 2D LIDAR SLAM
Hess, W., Kohler, D., Rapp, H., & Andor-Ardó, D. (2016) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2016.7487258 · Cited by 1,945 · cited 1× · see citing works
“keeping the computational cost low [15].”
Robust Monte Carlo localization for mobile robots
Thrun, S., Fox, D., Burgard, W., & Dellaert, F. (2001) · Artificial Intelligence · doi:10.1016/s0004-3702(01)00069-8 · Cited by 1,809 · see citing works
3D-R2N2: A Unified Approach for Single and Multi-view 3D Object Reconstruction
Choy, C., Xu, D., Gwak, J., Chen, K., & Savarese, S. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-46484-8_38 · Cited by 1,741 · cited 2× · see citing works
“In [21] the authors used a 32 × 32 × 32 voxelized grid, resulting in a dimensionality of 32768,”
MethodDistance Metric Learning for Large Margin Nearest Neighbor Classification
Weinberger, K. Q., Blitzer, J. C., & Saul, L. K. (2005) · Cited by 1,693 · see citing works
Monte Carlo localization for mobile robots
Dellaert, F., Fox, D., Burgard, W., & Thrun, S. (2003) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/robot.1999.772544 · Cited by 1,622 · cited 2× · see citing works
“Robot localization has been heavily explored using different sensors such as LIDAR [10], vision [11], GPS [12] or”
MethodMatching with PROSAC — Progressive Sample Consensus
Chum, O., & Matas, J. (2005) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2005.221 · Cited by 1,178 · see citing works
Agarwal, S., Furukawa, Y., Snavely, N., Simon, I., Curless, B., Seitz, S. M., & Szeliski, R. (2011) · Communications of the ACM · doi:10.1145/2001269.2001293 · Cited by 1,077 · cited 2× · see citing works
“These types of approaches typically assume a certain inherent structure in the scene, image features are dependant on repeatable camera viewpoint [3], and methods are often tested in urban environments which guide the robot along the route in question [4].”
BackgroundComparing ICP variants on real-world data sets
Pomerleau, F., Colas, F., Siegwart, R., & Magnenat, S. (2013) · Autonomous Robots · doi:10.1007/s10514-013-9327-2 · Cited by 764 · cited 1× · see citing works
“When the assumption of a local neighbourhood is lost most of these approaches require ICP alignments or similar costly alternatives [16].”
BackgroundPointNetVLAD: Deep Point Cloud Based Retrieval for Large-Scale Place Recognition
Uy, M. A., & Lee, G. H. (2018) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2018.00470 · Cited by 689 · cited 1× · see citing works
“In [25] a query point cloud is downsampled and matched”
BackgroundMap-Based Precision Vehicle Localization in Urban Environments
Levinson, J., Montemerlo, M., & Thrun, S. (2007) · Robotics: Science and Systems III · doi:10.15607/rss.2007.iii.016 · Cited by 477 · cited 3× · see citing works
“Robot localization has been heavily explored using different sensors such as LIDAR [10], vision [11], GPS [12] or”
MethodLearning descriptors for object recognition and 3D pose estimation
Wohlhart, P., & Lepetit, V. (2015) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2015.7298930 · Cited by 452 · cited 2× · see citing works
“To train our descriptor we use a variation of the triplet loss similar to the one defined in [29] - in our case we use a squared L2 norm as shown in Eq.”
MethodSegMatch: Segment based place recognition in 3D point clouds
Dubé, R., Dugas, D., Stumm, E. S., Nieto, J. I., Siegwart, R., & Cadena, C. (2017) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2017.7989618 · Cited by 404 · cited 6× · see citing works
“We provide comparisons against a popular method for segment-based localization, SegMatch,1 as proposed in [5], a data-driven incremental approach SegMap1 [6], and our previous work NSM [7].”
BackgroundMethodInfluentialVisual teach and repeat for long‐range rover autonomy
Furgale, P., & Barfoot, T. D. (2010) · Journal of Field Robotics · doi:10.1002/rob.20342 · Cited by 263 · cited 1× · see citing works
“Robot localization has been heavily explored using different sensors such as LIDAR [10], vision [11], GPS [12] or”
Method3D Point Cloud Registration for Localization Using a Deep Neural Network Auto-Encoder
Elbaz, G., Avraham, T., & Fischer, A. (2017) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2017.265 · Cited by 241 · cited 1× · see citing works
“[19] used segments produced from a Random Sphere Cover Set overlapping point cloud spheres where each point of the original point cloud could be part of more than one sphere.”
MethodExperience-based navigation for long-term localisation
Churchill, W., & Newman, P. (2013) · The International Journal of Robotics Research · doi:10.1177/0278364913499193 · Cited by 225 · cited 2× · see citing works
“To label the data we combined four consecutive Velodyne scans using the motion estimator from [2] and extracted segments using distance-based segmentation algorithm.”
MethodSegMap: 3D Segment Mapping using Data-Driven Descriptors
Dubé, R., Cramariuc, A., Dugas, D., Nieto, J. I., Siegwart, R., & Cadena, C. (2018) · Robotics: Science and Systems · doi:10.15607/rss.2018.xiv.003 · Cited by 182 · cited 8× · see citing works
“We provide comparisons against a popular method for segment-based localization, SegMatch,1 as proposed in [5], a data-driven incremental approach SegMap1 [6], and our previous work NSM [7].”
BackgroundMethodInfluentialRobust LIDAR localization using multiresolution Gaussian mixture maps for autonomous driving
Wolcott, R. W., & Eustice, R. M. (2017) · The International Journal of Robotics Research · doi:10.1177/0278364917696568 · Cited by 169 · cited 1× · see citing works
“ample, the approach of [14] uses a prior map of reflectivity and exploits road marks to reliably localize a vehicle in an urban environment.”
BackgroundSegMap: 3D Segment Mapping using Data-Driven Descriptors
Dubé, R., Cramariuc, A., Dugas, D., et al. (2018) · Repository for Publications and Research Data (ETH Zurich) · doi:10.3929/ethz-b-000319620 · Cited by 67 · see citing works
Scan segments matching for pairwise 3D alignment
Douillard, B., Quadros, A., Morton, P., et al. (2012) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2012.6224788 · Cited by 56 · cited 1× · see citing works
“The concept of segment localization was presented in [17] but more specifically our work is”
BackgroundA dynamic programming approach for fast and robust object pose recognition from range images
Zach, C., Peñate-Sánchez, A., & Pham, M.-T. (2015) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2015.7298615 · Cited by 48 · cited 1× · see citing works
“Features created from LIDAR data are less informative or unique than visual features, thus approaches often prune the candidate matches to reduce the percentage of outliers before performing a robust geometric validation [18].”
Method
Flagged first, then by citations
3 of the 37 references couldn’t be matched to a known paper.
Research works printed but not linked: 2
- li, 0, PointCNN: Convolution on X-transformed points, Proc Adv Neural Inf Process Syst, 828
- 10.1109/iros.2018.8594042 · DOI
Cited by
Sorted by citations
A survey of deep learning techniques for autonomous driving
Grigorescu, S., Trasnea, B., Cocias, T., & Macesanu, G. (2020) · Journal of Field Robotics · doi:10.1002/rob.21918 · Cited by 1,809 · see citing works
“uses a deep neural network to build a learned representation of the driving scene from LiDAR sweeps and intensity maps. The localization of the vehicle is obtained through convolutional matching. In (Tinchev et al., 2019), laser scans and a deep neural network are used to learn descriptors for localization Figure 6: E…”
MethodA survey of state-of-the-art on visual SLAM
Kazerouni, I. A., Fitzgerald, L., Dooly, G., & Toal, D. (2022) · Expert Systems with Applications · doi:10.1016/j.eswa.2022.117734 · Cited by 396 · see citing works
Chen, X., Läbe, T., Milioto, A., Röhling, T., Behley, J., & Stachniss, C. (2021) · Autonomous Robots · doi:10.1007/s10514-021-09999-0 · Cited by 152 · see citing works
“Tinchev et al. (2019) propose a learning-based method to match segments of trees and localize in both urban and natural environments.”
MethodRange Image-based LiDAR Localization for Autonomous Vehicles
Chen, X., Vizzo, I., Läbe, T., Behley, J., & Stachniss, C. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561335 · Cited by 142 · see citing works
“[27] propose a learningbased method to match segments of trees and localize in both urban and natural environments.”
MethodA Survey on Global LiDAR Localization: Challenges, Advances and Open Problems
Yin, H., Xu, X., Lu, S., Chen, X., Xiong, R., Shen, S., Stachniss, C., & Wang, Y. (2024) · International Journal of Computer Vision · doi:10.1007/s11263-024-02019-5 · Cited by 143 · see citing works
“In these feature-based global matching methods (Dubé et al., 2017, 2018; Cramariuc et al., 2021; Tinchev et al., 2018, 2019), point cloud segments are aligned with low-dimensional and distinctive descriptors for global matching.”
InfluentialZhou, Z., Zhao, C., Adolfsson, D., Su, S., Gao, Y., Duckett, T., & Sun, L. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9560932 · Cited by 119 · see citing works
“SegMatch-like methods [4], [5], [6], [7]”
BackgroundMethodInfluentialLoGG3D-Net: Locally Guided Global Descriptor Learning for 3D Place Recognition
Vidanapathirana, K., Ramezani, M., Moghadam, P., Sridharan, S., & Fookes, C. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9811753 · Cited by 111 · see citing works
“Recently features extracted from larger local regions such as point segments have shown superior performance [12], [13].”
BackgroundOnline LiDAR-SLAM for Legged Robots with Robust Registration and Deep-Learned Loop Closure
Ramezani, M., Tinchev, G., Egor, I., & Fallon, M. F. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196769 · Cited by 57 · see citing works
“[26], which is based on matching individual segments in pairs of point clouds using a deeplylearned feature descriptor.”
MethodSelf-citationLiDAR-Based Place Recognition For Autonomous Driving: A Survey
Zhang, Y.-J., Shi, P., & Li, J.-Y. (2023) · ACM Computing Surveys · doi:10.1145/3707446 · Cited by 86 · see citing works
“Tinchev et al. [171] utilize convolution to obtain an embedding space suitable for urban and natural scenarios.”
Learning an Overlap-based Observation Model for 3D LiDAR Localization
Chen, X., Labe, T., Nardi, L., Behley, J., & Stachniss, C. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9340769 · Cited by 44 · see citing works
“[29] propose a learning-based method to match segments of trees and localize in both urban and natural environments.”
LIRO: Tightly Coupled Lidar-Inertia-Ranging Odometry
Nguyen, T., Cao, M., Yuan, S., Lyu, Y., Nguyen, T. H., & Xie, L. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9560954 · Cited by 45 · see citing works
“Another approach could be to train a neural network on recognizing segmented objects from the pointcloud for place recognition and loop closure [27].”
Background3D point cloud-based place recognition: a survey
Luo, K., Yu, H., Chen, X., Yang, Z., Wang, J., Cheng, P., & Mian, A. S. (2024) · Artificial Intelligence Review · doi:10.1007/s10462-024-10713-6 · Cited by 41 · see citing works
“To adapt to online applications, Tinchev et al. (2019) then explored laser-based localization in both urban and natural environments and proposed a deep learning approach capable of learning meaningful descriptors directly from 3D point clouds as well as a feature space representation for the set…”
3D Semantic Map Construction Using Improved ORB-SLAM2 for Mobile Robot in Edge Computing Environment
Cui, X., Lu, C., & Wang, J. (2020) · IEEE Access · doi:10.1109/access.2020.2983488 · Cited by 41 · see citing works
“The feature extraction network Darknet53 of the YOLOv3 real-time target detection model is composed of 53 convolutional layers and 24 residual layers [27], [28].”
MethodOneShot Global Localization: Instant LiDAR-Visual Pose Estimation
Ratz, S., Dymczyk, M., Siegwart, R., & Dubé, R. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9197458 · Cited by 41 · see citing works
“One of the main differences to previous segment based approaches [3]–[5, 23] is that our system works with a single LiDAR scan at a time.”
MethodAutomatic three‐dimensional mapping for tree diameter measurements in inventory operations
Tremblay, J., Béland, M., Gagnon, R., Pomerleau, F., & Giguère, P. (2020) · Journal of Field Robotics · doi:10.1002/rob.21980 · Cited by 37 · see citing works
Online pole segmentation on range images for long-term LiDAR localization in urban environments
Dong, H., Chen, X., Särkkä, S., & Stachniss, C. (2022) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2022.104283 · Cited by 34 · see citing works
Online Range Image-based Pole Extractor for Long-term LiDAR Localization in Urban Environments
Dong, H., Chen, X., & Stachniss, C. (2021) · European Conference on Mobile Robots · doi:10.1109/ecmr50962.2021.9568850 · Cited by 31 · see citing works
“…combine semantic information such as lanes and traffic signs in a Bayesian filtering framework to achieve accurate and robust localization within sparse HD maps, whereas Tinchev et al. [23] propose a learning-based method to match segments of trees and localize in both urban and natural environments.”
SemSegMap – 3D Segment-based Semantic Localization
Cramariuc, A., Tschopp, F., Alatur, N., Benz, S., Falck, T., Brühlmeier, M., Hahn, B. P., Nieto, J. I., & Siegwart, R. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636156 · Cited by 27 · see citing works
“Extensions to the pipeline include different training methods for the descriptor, as well as fine tuning to different environments [33], [34].”
MethodEugênio, F. C., Badin, T. L., Fernandes, P., et al. (2021) · International Journal of Remote Sensing · doi:10.1080/01431161.2021.1975845 · Cited by 21 · see citing works
VMLoc: Variational Fusion For Learning-Based Multimodal Camera Localization
Zhou, K., Chen, C., Wang, B., Saputra, M. R. U., Trigoni, N., & Markham, A. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i7.16767 · Cited by 21 · see citing works
“On the other hand, even though the lidar signal has a wide viewpoint and is invariant to illumination, it is less informative (Tinchev, Nobili, and Fallon 2018; Tinchev, Penate-Sanchez, and Fallon 2019).”
BackgroundVehicle-Related Distance Estimation Using Customized YOLOv7
Liu, X., & Yan, W. Q. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-25825-1_7 · Cited by 17 · see citing works
MINT: Deep Network Compression via Mutual Information-based Neuron Trimming
Ganesh, M. R., Corso, J. J., & Sekeh, S. Y. (2021) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr48806.2021.9412590 · Cited by 15 · see citing works
“Although deep learning yields remarkable performance in real-world problems like medical diagnosis [1], [2], autonomous vehicles [3], [4], and others, they consume a large amount of memory and computational resources that limit their large-scale deployment.”
BackgroundTightly Coupled LIDAR/IMU/UWB Fusion via Resilient Factor Graph for Quadruped Robot Positioning
Kuang, Y., Hu, T., Ouyang, M., et al. (2024) · Remote Sensing · doi:10.3390/rs16224171 · Cited by 12 · see citing works
LiDAR-Based Initial Global Localization Using Two-Dimensional (2D) Submap Projection Image (SPI)
Li, Y., & Li, H. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9560740 · Cited by 12 · see citing works
“For LiDAR-based initial global localization, one main category of methods rely on point cloud segmentation and segment descriptors [7]–[9].”
Method3D ToF LiDAR for Mobile Robotics in Harsh Environments: A Review
Yang, T., Hu, J., Li, Y., Zhao, C., Sun, L., Krajník, T., & Yan, Z. (2024) · Unmanned Systems · doi:10.1142/s230138502530001x · Cited by 10 · see citing works
Evaluation and Deployment of LiDAR-based Place Recognition in Dense Forests
Oh, H., Chebrolu, N., Mattamala, M., Freißmuth, L., & Fallon, M. F. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10801297 · Cited by 10 · see citing works
“Other similar works such as NSM [19] and ESM [18] explicitly aimed for localization in natural environments.”
Self-citationA physics perspective on lidar data assimilation for mobile robots
Berquin, Y., & Zell, A. (2021) · Robotica · doi:10.1017/s0263574721000850 · Cited by 10 · see citing works
“Recent advances in machine learning algorithms have further accelerated this trend [4, 5].”
BackgroundForestLPR: LiDAR Place Recognition in Forests Attentioning Multiple BEV Density Images
Shen, Y., Tuna, T., Hutter, M., Cadena, C., & Zheng, N. (2025) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52734.2025.00624 · Cited by 8 · see citing works
InstaLoc: One-shot Global Lidar Localisation in Indoor Environments through Instance Learning
Zhang, L., Digumarti, S. T., Tinchev, G., & Fallon, M. F. (2023) · Robotics: Science and Systems · doi:10.15607/rss.2023.xix.070 · Cited by 8 · see citing works
Self-citationOn the descriptive power of LiDAR intensity images for segment-based loop closing in 3-D SLAM
Wietrzykowski, J., & Skrzypczyński, P. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636698 · Cited by 8 · see citing works
“[4], [11], [17]), or there is no open source code available, as in [10].”
BackgroundMethodGeneralizable Stable Points Segmentation for 3D LiDAR Scan-to-Map Long-Term Localization
Hroob, I., Mersch, B., Stachniss, C., et al. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3368236 · Cited by 6 · see citing works
“For example, Tinchev et al. [35] pro-pose a learning-based method for segment matching of trees and localizationindiverseenvironments.”
MethodPose-Graph Attentional Graph Neural Network for Lidar Place Recognition
Ramezani, M., Wang, L., Knights, J., Li, Z., Pounds, P., & Moghadam, P. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3341766 · Cited by 6 · see citing works
“CNN-based methods such as Segmap [22] or Efficient Segment Matching (ESM) [23] encode local patches of a point cloud into local embeddings.”
MethodLoop-Closure and Map-Combiner Detection Strategy based on LIDAR Reflectance and Elevation Maps
Aldibaja, M. A., Suganuma, N., Yanase, R., et al. (2020) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc45102.2020.9294634 · Cited by 6 · see citing works
“A more sophisticated approach has been suggested in [6] to use deep learning capabilities to create a feature space Loop-Closure and Map-Combiner Detection Strategy based on”
MethodHuang, Y., Gu, Y., Xu, C., & Kong, H. (2024) · IEEE transactions on field robotics. · doi:10.1109/tfr.2024.3421392 · Cited by 5 · see citing works
Deep Robust Multi-Robot Re-Localisation in Natural Environments
Ramezani, M., Griffiths, E., Haghighat, M., et al. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10341798 · Cited by 5 · see citing works
“Features are either directly used for place recognition, such as works in [17], [18] or aggregated using either a first-order pooling technique, e.g. , GeM [19], NetVLAD [20] or second order pooling employed in [7], [21], to generate a global descriptor of the point cloud [5]–[8], [22].”
MethodLIRO: Tightly Coupled Lidar-Inertia-Ranging OdometryPreprint
Nguyen, T., Cao, M., Yuan, S., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.13072 · Cited by 5 · see citing works
“Another approach could be to train a neural network on recognizing segmented objects from the pointcloud for place recognition and loop closure [27].”
BackgroundZhou, Z., Zhao, C., Adolfsson, D., et al. (2021) · White Rose Research Online (University of Leeds, The University of Sheffield, University of York) · doi:10.48550/arxiv.2103.12292 · Cited by 4 · see citing works
“SegMatch-like methods [4], [5], [6], [7]”
BackgroundMethodInfluentialMulti-L: A Novel Multi-Robot Cooperative Localization Method in Indoor Environment
Zhu, Z., Zhu, K., Zheng, Z., et al. (2022) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc55140.2022.9922050 · Cited by 4 · see citing works
“…LiDAR-based or vision-based, they often also rely on strategies that bind localization information to environmental descriptions, such as point cloud segmentation, segment descriptors [7]– [9] and the bag-of-words(BoW) [10] model often used in visual SLAM to form global localization descriptors.”
MethodEmbedded Attention Network for Semantic Segmentation
Lv, Q., Feng, M., Sun, X., Dong, J., Chen, C., & Zhang, Y. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3126892 · Cited by 3 · see citing works
Certainty Aware Global Localisation Using 3D Point Correspondences
Steiner, R., Cox, M., Borges, P., Bernreiter, L., & Nieto, J. I. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3114956 · Cited by 2 · see citing works
Persistent Mapping of Sensor Data for Medium-Term Autonomy
Nickels, K. M., Gassaway, J., Bries, M., et al. (2022) · Sensors · doi:10.3390/s22145427 · Cited by 1 · see citing works
“[20] use labelled point clouds to train a deep neural network that improves point LiDAR segmentation performance.”
BackgroundTreeLoc++: Robust Six-DoF LiDAR Localization in Forests With a Compact Digital Forest Inventory
Jung, M., Lee, D., Chebrolu, N., et al. (2026) · IEEE transactions on field robotics. · doi:10.1109/tfr.2026.3710448 · Cited by 1 · see citing works
“Other methods, such as NSM [27] and ESM [28], extract descriptors from individual tree instances but still rely on point clouds and overlook intertree relationships, making them prone to perceptual aliasing in environments where tree geometries are indistinguishable.”
Self-citationXiao, X., Yuan, H., Yang, Y., et al. (2026) · Computers and Electronics in Agriculture · doi:10.1016/j.compag.2026.112078 · Cited by 1 · see citing works
Urban Navigation System With Multiple Sub-Maps and Multiple Sub-Navigators
Sattaratnamai, S., Niparnan, N., & Sudsang, A. (2020) · IEEE Access · doi:10.1109/access.2020.2994652 · Cited by 1 · see citing works
“The localization system for mobile platforms uses a 3D laser scan to obtain point cloud of forests or woodlands [4].”
Method(2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2910670
Environmental Technologies for Nature Conservation Incorporating Artificial Intelligence
(2022) · Nature Environmental Protection · doi:10.38007/nep.2022.030403
LiDAR-Based Localization System for Kidnapped Robots
Lasguignes, T., Gobin, G., & Stasse, O. (2023) · IEEE International Conference on Robotic Computing (IRC) · doi:10.1109/irc59093.2023.00013
“[15] presented the concept of segments or how to use the segmentation of a point cloud to reduce its computational cost.”
Devanna, R. P., Torres‐Torriti, M., Saçılık, K., et al. (2025) · Algorithms · doi:10.3390/a18080522
“In [3], a deep learning approach is developed for laser-based localisation in urban and natural environments, creating meaningful descriptors from 3D point clouds for loop closure detection.”
de Lima, L. C., Griffiths, E., Haghighat, M., et al. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11247490
“Deep-learned approaches, on the other hand, use neural networks to extract features, which are either used directly [22], [23] or aggregated into global descriptors [24]– [26] for LiDAR PR. Methods such as EgoNN [27] and LCDNet [28] estimate the relative pose between two point clouds upon success…”
Progressive Relocalization Based on the Point Cloud Map
Huang, Y., Chen, B., Jin, Y., et al. (2025) · doi:10.1109/iotir66925.2025.00024
“Tinchev[26] Vidanapathirana[28] presented Locus, encoding spatial and temporal topological information robust against occlusions and varying viewpoints.”
- Authors
- Tinchev, G., Penate-Sanchez, A., & Fallon, M.
- Affiliations
- University of Oxford (GB)
- Venue
- IEEE Robotics and Automation Letters
- Published
- 2019-04
- Volume
- 4
- Issue
- 2
- Pages
- 1327-1334
- ISSN
- 2377-3766 (ISSN-L)
- 2377-3774
- OpenAlex
- W2951349360
- Semantic Scholar
- 84b62e6d1c277921dc66de3ec804039492baab61
- Cited by
- 51
- References
- 37
- Citations by year
- 2026: 2 · 2025: 4 · 2024: 8 · 2023: 6 · 2022: 6 · 2021: 15 · 2020: 8 · 2019: 2
- FWCI
- 5.70
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Robotics and Automation Letters (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Acceda (Universidad de Las Palmas de Gran Canaria) (submittedVersion)
- arXiv (Cornell University)
- Funders
- EPSRC RAIN and ORCA Robotics Hubs (EP/R026084/1, EP/R026173/1)
- Royal Society University Research Fellowship
- Topics
- Physical Sciences › Engineering › Aerospace Engineering › Robotics and Sensor-Based Localization
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Physical Sciences › Earth and Planetary Sciences › Geology › 3D Surveying and Cultural Heritage
- Keywords
- triplet learning, point cloud matching, loop closure detection, forest navigation, urban environment, CPU deployment, lightweight models, robot localization, unmanned ground vehicles
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
