LiDAR has gained popularity in autonomous driving due to advantages like long measurement distance, rich three-dimensional information, and stability in harsh environments. Place Recognition (PR) enables vehicles to identify previously visited locations despite variations in appearance, weather, and viewpoints, even determining their global location within prior maps. This capability is crucial for accurate localization in autonomous driving. Consequently, LiDAR-based Place Recognition (LPR) has emerged as a research hotspot in robotics. However, existing reviews predominantly concentrate on Visual Place Recognition, leaving a gap in systematic reviews on LPR. This article bridges this gap by providing a comprehensive review of LPR methods, thus facilitating and encouraging further research. We commence by exploring the relationship between PR and autonomous driving components. Then, we delve into the problem formulation of LPR, challenges, and relations to previous surveys. Subsequently, we conduct an in-depth review of related research, which offers detailed classifications, strengths and weaknesses, and architectures. Finally, we summarize existing datasets and evaluation metrics and envision promising future directions. This article can serve as a valuable tutorial for newcomers entering the field of place recognition.

• 2024.11: Our has been accepted for ACM Computing Surveys
• 2024.12: The published paper is available: Published Paper
• 2025.02: We upload the figures and pdf

If this work is useful for your research, please cite our paper:

@article{zhang2024lidar,
  title={Lidar-based place recognition for autonomous driving: A survey},
  author={Zhang, Yongjun and Shi, Pengcheng and Li, Jiayuan},
  journal={ACM Computing Surveys},
  volume={57},
  number={4},
  pages={1--36},
  year={2024},
  publisher={ACM New York, NY}
}

• Question 1: It addresses the problem of "where have I ever been, " also known as Loop Closure Detection (LCD). In this context, PR and localization are interdependent, with PR enhancing localization accuracy through loop closure detection
• Question 2: It tackles the issue of "where am I," also known as global localization. In this context, PR is a specialized localization method that directly provides the vehicle’s global pose.

We categorize methods into handcrafted and learning-based types, further subdividing them, and present detailed introductions to pioneering works.

• SGLC: Semantic Graph-Guided Coarse-Fine-Refine Full Loop Closing for LiDAR SLAM | arXiv

• CrossLoc3D: Aerial-Ground Cross-Source 3D Place Recognition | ICCV

• BTC: A Binary and Triangle Combined Descriptor for 3-D Place Recognition | TRO

• A New Horizon: Employing Map Clustering Similarity for LiDAR-based Place Recognition | TIV

• Fast and Accurate Deep Loop Closing and Relocalization for Reliable LiDAR SLAM | TRO

• P-GAT: Pose-Graph Attentional Network for Lidar Place Recognition | RAL

• Effectively Detecting Loop Closures using Point Cloud Density Maps | ICRA

• OverlapMamba: Novel Shift State Space Model for LiDAR-based Place Recognition | arXiv

• OSK: A Novel LiDAR Occupancy Set Key-Based Place Recognition Method in Urban Environment | TIM

• SelFLoc: Selective feature fusion for large-scale point cloud-based place recognition | KBS

• CVTNet: A Cross-View Transformer Network for LiDAR-Based Place Recognition in Autonomous Driving Environments | TII

• Uncertainty-Aware Lidar Place Recognition in Novel Environments | IROS

• CCL: Continual Contrastive Learning for LiDAR Place Recognition | RAL

• BioSLAM: A Bioinspired Lifelong Memory System for General Place Recognition | TRO

• CASSPR: Cross Attention Single Scan Place Recognition | ICCV

• Contour Context: Abstract Structural Distribution for 3D LiDAR Loop Detection and Metric Pose Estimation | ICRA

• BEVPlace: Learning LiDAR-based Place Recognition using Bird's Eye View Images | ICCV

• Uncertainty-Aware Lidar Place Recognition in Novel Environments | IROS

• TReR: A Lightweight Transformer Re-Ranking Approach for 3D LiDAR Place Recognition | ITSC

• STD: Stable Triangle Descriptor for 3D place recognition | ICRA

• Place Recognition of Large-Scale Unstructured Orchards With Attention Score Maps | RAL

• VNI-Net: Vector Neurons-based Rotation-Invariant Descriptor for LiDAR Place Recognition | arXiv

• A fast LiDAR place recognition and localization method by fusing local and global search | ISPRSJ

• RING++: Roto-Translation Invariant Gram for Global Localization on a Sparse Scan Map | TRO

• Place Recognition of Large-Scale Unstructured Orchards With Attention Score Maps | RAL

• GeoAdapt: Self-Supervised Test-Time Adaptation in LiDAR Place Recognition Using Geometric Priors | RAL

• ORCHNet: A Robust Global Feature Aggregation approach for 3D LiDAR-based Place recognition in Orchards | arXiv

• OverlapNetVLAD: A Coarse-to-Fine Framework for LiDAR-based Place Recognition | arXiv

• Binary Image Fingerprint: Stable Structure Identifier for 3D LiDAR Place Recognition RAL

• Spherical Transformer for LiDAR-Based 3D Recognition | CVPR

• SphereVLAD++: Attention-Based and Signal-Enhanced Viewpoint Invariant Descriptor | RAL

• SVT-Net: Super Light-Weight Sparse Voxel Transformer for Large Scale Place Recognition | AAAI

• BoxGraph: Semantic Place Recognition and Pose Estimation from 3D LiDAR | IROS

• RINet: Efficient 3D Lidar-Based Place Recognition Using Rotation Invariant Neural Network | RAL

• Efficient 3D Point Cloud Feature Learning for Large-Scale Place Recognition | TIP

• AttDLNet: Attention-Based Deep Network for 3D LiDAR Place Recognition | ROBOT

• OverlapTransformer: An Efficient and Yaw-Angle-Invariant Transformer Network for LiDAR-Based Place Recognition | RAL

• SeqOT: A Spatial–Temporal Transformer Network for Place Recognition Using Sequential LiDAR Data | TIE

• Simultaneous viewpoint-and condition-invariant loop closure detection based on LiDAR descriptor for outdoor large-scale environments | TIE

• HiTPR: Hierarchical Transformer for Place Recognition in Point Cloud | ICRA

• LoGG3D-Net: Locally Guided Global Descriptor Learning for 3D Place Recognition | ICRA

• MinkLoc3D-SI: 3D LiDAR Place Recognition With Sparse Convolutions, Spherical Coordinates, and Intensity | RAL

• Improving Point Cloud Based Place Recognition with Ranking-based Loss and Large Batch Training | ICPR

• LCDNet: Deep Loop Closure Detection and Point Cloud Registration for LiDAR SLAM | TRO

• One RING to Rule Them All: Radon Sinogram for Place Recognition, Orientation and Translation Estimation: | IROS

• Fresco: Frequency-domain scan context for lidar-based place recognition with translation and rotation invariance: | ICARCV

• High Accuracy and Low Complexity Lidar Place Recognition Using Unitary Invariant Frobenius Norm: | IEEE Sensors

• AdaFusion: Visual-LiDAR Fusion With Adaptive Weights for Place Recognition | RAL

• InCloud: Incremental Learning for Point Cloud Place Recognition | IROS

• A heterogeneous 3D map-based place recognition solution using virtual LiDAR and a polar grid height coding image descriptor | ISPRSJ

• Dh3d: Deep hierarchical 3d descriptors for robust large-scale 6dof relocal- ization | ECCV

• DSC: Deep Scan Context Descriptor for Large-Scale Place Recognition | MFI

• Object Scan Context: Object-centric Spatial Descriptor for Place Recognition within 3D Point Cloud Map | arXiv

• SC-LPR: Spatiotemporal context based LiDAR place recognition | PRL

• Retriever: Point Cloud Retrieval in Compressed 3D Maps | ICRA

• Place recognition and navigation of outdoor mobile robots based on random Forest learning with a 3D LiDAR Journal of Intelligent & Robotic Systems

• NDT-Transformer: Large-Scale 3D Point Cloud Localisation using the Normal Distribution Transform Representation |ICRA

• MinkLoc3D: Point Cloud Based Large-Scale Place Recognition |WACV

• MinkLoc++: Lidar and Monocular Image Fusion for Place Recognition | IJCNN

• EgoNN: Egocentric Neural Network for Point Cloud Based 6DoF Relocalization at the City Scale | RAL

• TransLoc3D : Point Cloud based Large-scale Place Recognition using Adaptive Receptive Fields | arXiv

• Pyramid Point Cloud Transformer for Large-Scale Place Recognition | ICCV

• SSC: Semantic Scan Context for Large-Scale Place Recognition | IROS

• A registration-aided domain adaptation network for 3d point cloud based place recognition | IROS

• SOE-Net: A Self-Attention and Orientation Encoding Network for Point Cloud Based Place Recognition | CVPR

• Scan Context++: Structural Place Recognition Robust to Rotation and Lateral Variations in Urban Environments: | TRO

• Weighted scan context: Global descriptor with sparse height feature for loop closure detection: | ICCCR

• Season-Invariant and Viewpoint-Tolerant LiDAR Place Recognition in GPS-Denied Environments | TIE

• Spherical Multi-Modal Place Recognition for Heterogeneous Sensor Systems | ICRA

• DiSCO: Differentiable Scan Context With Orientation | RAL

• CORAL: Colored structural representation for bi-modal place recognition | IROS

• OverlapNet: a siamese network for computing LiDAR scan similarity with applications to loop closing and localization | Autonomous Robots

• On the descriptive power of LiDAR intensity images for segment-based loop closing in 3-D SLAM | IROS

• PSE-Match: A Viewpoint-Free Place Recognition Method With Parallel Semantic Embedding | TITS

• Voxel-Based Representation Learning for Place Recognition Based on 3D Point Clouds | IROS

• SpoxelNet: Spherical Voxel-based Deep Place Recognition for 3D Point Clouds of Crowded Indoor Spaces | IROS

• SRNet: A 3D Scene Recognition Network using Static Graph and Dense Semantic Fusion | Computer Graphics Forum

• LiDAR Iris for Loop-Closure Detection: | IROS

• Intensity Scan Context: Coding Intensity and Geometry Relations for Loop Closure Detection: | ICRA

• PIC-Net: Point Cloud and Image Collaboration Network for Large-Scale Place Recognition | arXiv

• A Fast and Robust Place Recognition Approach for Stereo Visual Odometry Using LiDAR Descriptors | IROS

• DAGC: Employing Dual Attention and Graph Convolution for Point Cloud based Place Recognition | ICMR

• SegMap: Segment-based mapping and localization using data-driven descriptors | IJRR

• Seed: A Segmentation-Based Egocentric 3D Point Cloud Descriptor for Loop Closure Detection | IROS

• Learning an Overlap-based Observation Model for 3D LiDAR Localization |IROS

• LiPMatch: LiDAR Point Cloud Plane Based Loop-Closure |RAL

• OREOS: Oriented Recognition of 3D Point Clouds in Outdoor Scenarios | IROS

• PCAN: 3D Attention Map Learning Using Contextual Information for Point Cloud Based Retrieval | CVPR

• LPD-Net: 3D Point Cloud Learning for Large-Scale Place Recognition and Environment Analysis | ICCV

• Semantically Assisted Loop Closure in SLAM Using NDT Histograms | IROS

• c-m2dp: A fast point cloud descriptor with color information to perform loop closure detection | CASE

• 1-Day Learning, 1-Year Localization: Long-Term LiDAR Localization Using Scan Context Image | RAL

• SeqLPD: Sequence Matching Enhanced Loop-Closure Detection Based on Large-Scale Point Cloud Description for Self-Driving Vehicles | IROS

• Loop Closure Detection in Closed Environments | ECMR

• Scan Context: Egocentric Spatial Descriptor for Place Recognition Within 3D Point Cloud Map | IROS

• PointNetVLAD: Deep Point Cloud Based Retrieval for Large-Scale Place Recognition | CVPR

• Delight: An Efficient Descriptor for Global Localisation Using LiDAR Intensities | ICRA

• Seeing the Wood for the Trees: Reliable Localization in Urban and Natural Environments | IROS

• Stabilize an Unsupervised Feature Learning for LiDAR-based Place Recognition | IROS

• SegMatch: Segment based place recognition in 3D point clouds | ICRA

• Efficient 3D LIDAR based loop closing using deep neural network | ROBIO

• M2DP: A novel 3D point cloud descriptor and its application in loop closure detection | IROS

• A fast histogram-based similarity measure for detecting loop closures in 3-D LIDAR data | IROS

• Loop closure detection using small-sized signatures from 3D LIDAR data | SSRR

• Learning to close the loop from 3D point clouds | IROS

• Appearance-based loop detection from 3d laser data using the normal distributions transform | ICRA

• Automatic appearance-based loop detection from three-dimensional laser data using the normal distributions transform | JFR

For any inquiries, feel free to contact me:

• Pengcheng Shi {[email protected]}