Awesom high definition map methods collection with detailed interpretations of some key papers.

1. Online HD Map

1.1 Roadmap

1.2 Paper list

2024

1. StreamMapNet: Streaming Mapping Network for Vectorized Online HD Map Construction [code] WACV2024
2. SuperFusion: Multilevel LiDAR-Camera Fusion for Long-Range HD Map Generation [code] ICRA2024
3. MapNeXt: Revisiting Training and Scaling Practices for Online Vectorized HD Map Construction Arxiv
4. Stream Query Denoising for Vectorized HD Map Construction Arxiv
5. EAN-MapNet: Efficient Vectorized HD Map Construction with Anchor Neighborhoods Arxiv
6. HIMap: HybrId Representation Learning for End-to-end Vectorized HD Map Construction[code, not release] CVPR2024
7. MapUncertaintyPrediction: Producing and Leveraging Online Map Uncertainty in Trajectory Prediction [code] CVPR2024
8. MGMap: Mask-Guided Learning for Online Vectorized HD Map Construction [code] CVPR2024
9. MapVision: CVPR 2024 Autonomous Grand Challenge Mapless Driving Tech Report CVPR2024
10. LGmap: Local-to-Global Mapping Network for Online Long-Range Vectorized HD Map Construction CVPR2024
11. Leveraging SD Map to Assist the OpenLane Topology CVPRChallenge2024
12. UniHDMap: Unified Lane Elements Detection for Topology HD Map Construction CVPRChallenge2024
13. HybriMap: Hybrid Clues Utilization for Effective Vectorized HD Map Construction Arxiv
14. MapTracker: Tracking with Strided Memory Fusion for Consistent Vector HD Mapping [code] ECCV2024 oral
15. ADMap: Anti-disturbance framework for reconstructing online vectorized HD map [code] ECCV2024
16. MapQR: Leveraging Enhanced Queries of Point Sets for Vectorized Map Construction [code] ECCV2024
17. HRMapNet: Enhancing Vectorized Map Perception with Historical Rasterized Maps [code] ECCV2024
18. MapBEVPrediction: Accelerating Online Mapping and Behavior Prediction via Direct BEV Feature Attention [code] ECCV2024
19. Mask2Map: Vectorized HD Map Construction Using Bird’s Eye View Segmentation Masks [code ,not release] ECCV2024
20. DTCLMapper: Dual Temporal Consistent Learning for Vectorized HD Map Construction [code] IEEE T-ITS2024
21. PrevPredMap: Exploring Temporal Modeling with Previous Predictions for Online Vectorized HD Map Construction [code] WACV2025
22. PriorMapNet: Enhancing Online Vectorized HD Map Construction with Priors Arxiv
23. GNMap: Neural HD Map Generation from Multiple Vectorized Tiles Locally Produced by Autonomous Vehicles Arxiv
24. Online Temporal Fusion for Vectorized Map Construction in Mapless Autonomous Driving Arxiv
25. GenMapping: Unleashing the Potential of Inverse Perspective Mapping for Robust Online HD Map Construction Arxiv
26. GlobalMapNet: An Online Framework for Vectorized Global HD Map Construction Arxiv
27. ExelMap: Explainable Element-based HD-Map Change Detection and Update Arxiv(ECCV2024)

2023

1. VectorMapNet: End-to-end Vectorized HD Map Learning [code] ICML2023

2. MapTR: Structured Modeling and Learning for Online Vectorized HD Map Construction [code] ICLR2023

3. MapTRv2: An End-to-End Framework for Online Vectorized HD Map Construction [code] Arxiv

4. MapVR: Online Map Vectorization for Autonomous Driving: A Rasterization Perspective[code1, code2] NIPS2023

5. BeMapNet: End-to-End Vectorized HD-Map Construction With Piecewise Bezier Curve [code] CVPR2023

6. PivotNet: Vectorized Pivot Learning for End-to-end HD Map Construction ICCV2023 [code] ICCV2023

Before 2023

1. HDMapNet: An Online HD Map Construction and Evaluation Framework [code] ICRA2022,CVPR2021

1.3 Auxiliary Knowledge or methods (Distillation)

1. P-MapNet: Far-seeing Map Constructer Enhanced by both SDMap and HDMap Priors [code] ICLR2024 (reject)

2. SatForHDMap: Complementing Onboard Sensors with Satellite Map: A New Perspective for HD Map Construction [code] ICRA2024

3. Leveraging SD Map to Assist the OpenLane Topology CVPR2024

4. MapDistill: Boosting Efficient Camera-based HD Map Construction via Camera-LiDAR Fusion Model Distillation [code ,not release] ECCV2024

5. Enhancing Online Road Network Perception and Reasoning with Standard Definition Maps Arxiv

6. Driving with Prior Maps: Unified Vector Prior Encoding for Autonomous Vehicle Mapping Arxiv

2. 衍生工作

2.1 鲁棒性

1. Is Your HD Map Constructor Reliable under Sensor Corruptions? [code] Arxiv 这篇蛮有意思，后续工作适合做场景迁移、半监督学习

2.2 矢量化场景表示

1. VAD: Vectorized Scene Representation for Efficient Autonomous Driving [code] ICCV2023 (MapTR同作)

2. Is Ego Status All You Need for Open-Loop End-to-End Autonomous Driving? [code] CVPR2023 VAD -> BEV-Planner

2.3 车道线角度角度

1. OpenLane-V2: A Topology Reasoning Benchmark for Unified 3D HD Mapping [code] NeurIPS2023
2. Augmenting Lane Perception and Topology Understanding with Standard Definition Navigation Maps [code] Arxiv

2.4 AIGC相关

1. Panacea: Panoramic and Controllable Video Generation for Autonomous Driving [code] CVPR2024

2. Panacea+: Panoramic and Controllable Video Generation for Autonomous Driving [code] Arxiv

2.5 众包建图优化

1. CrowdSourcing Live High Definition Map via Collaborative Computation in Automotive Edge Computing T-VT2024

2. VI-Map: Infrastructure-Assisted Real-Time HD Mapping for Autonomous Driving MobiCom2023

3. Edgemap: Crowdsourcing high definition map in automotive edge computing ICC2022

3. 资源开销统计

算法	训练资源	Modal
BEVFormer	8 GPUs 后续较多工作都是基于BEVFormer，基本都是8卡起步	c
BEVDistill	8 NVIDIA A100 GPUs	c+l
HDMapNet	这篇是属于分割方法论的，所以训练开销明显少一些，bs默认4	c+l
MapTR v1/v2	8 NVIDIA GeForce RTX 3090 GPUs	c+l
MapDistill	8 NVIDIA RTX A6000 GPUs	c+l -> c
HRMapNet	8 NVIDIA A100 GPUs bs 8 × 4	c
P-MapNet	4 NVIDIA GeForce RTX 3090 GPUs，基于HDMapNet，所以开销也不高	c+l
SuperFusion	1 NVIDIA A100-40G GPU bs 1x4，也是基于HDMapNet，只支持单卡训练	c+l
StreamMapNet	8 GTX3090 GPUs bs 8x4	c
Mask2Map	4 GTX3090 GPUs bs 4x4，效果提升有点夸张	c+l
DTCLMapper	4 NVIDIA RTX A6000 bs 16	c
PrevPredMap	4 NVIDIA A100 GPUs bs 4x4	c

4. Reference Repo

https://github.com/LoveFaFa2333/Awesome-Online-HDMap

https://github.com/Honminden/awesome-hd-map-construction

http://t.csdnimg.cn/WosD6