# Geometry-Guided Representations for Coherent Lane and Traffic Topology Reasoning in Driving Scenes

arXiv: https://arxiv.org/abs/2506.13553 · HTML: https://mlsanity.com/paper/2506.13553/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in VLMs for autonomous driving, one of 9 directions in End-to-end autonomous driving.
- Builds on 33 of its 51 references that are in the map.
- First posted on arXiv 16 Jun 2025.

## Placed in

- End-to-end autonomous driving / [VLMs for autonomous driving](https://mlsanity.com/driving/vlms-for-autonomous-driving/)

## Builds on (most cited first, 30 of 33)

- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [End-to-End Object Detection with Transformers](https://mlsanity.com/paper/2005.12872/) (2020)
- [nuScenes: A multimodal dataset for autonomous driving](https://mlsanity.com/paper/1903.11027/) (2019)
- [Unsupervised Feature Learning via Non-Parametric Instance-level Discrimination](https://mlsanity.com/paper/1805.01978/) (2018)
- [Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression](https://mlsanity.com/paper/1902.09630/) (2019)
- [BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers](https://mlsanity.com/paper/2203.17270/) (2022)
- [Self-supervised Learning: Generative or Contrastive](https://mlsanity.com/paper/2006.08218/) (2020)
- [StreamMapNet: Streaming Mapping Network for Vectorized Online HD Map Construction](https://mlsanity.com/paper/2308.12570/) (2023)
- [PersFormer: 3D Lane Detection via Perspective Transformer and the OpenLane Benchmark](https://mlsanity.com/paper/2203.11089/) (2022)
- [3D-LaneNet: End-to-End 3D Multiple Lane Detection](https://mlsanity.com/paper/1811.10203/) (2018)
- [Structured Bird's-Eye-View Traffic Scene Understanding from Onboard Images](https://mlsanity.com/paper/2110.01997/) (2021)
- [OpenLane-V2: A Topology Reasoning Benchmark for Unified 3D HD Mapping](https://mlsanity.com/paper/2304.10440/) (2023)
- [End-to-End Vectorized HD-map Construction with Piecewise Bezier Curve](https://mlsanity.com/paper/2306.09700/) (2023)
- [Making Large Language Models Better Reasoners with Alignment](https://mlsanity.com/paper/2309.02144/) (2023)
- [Graph-based Topology Reasoning for Driving Scenes](https://mlsanity.com/paper/2304.05277/) (2023)
- [Online Map Vectorization for Autonomous Driving: A Rasterization Perspective](https://mlsanity.com/paper/2306.10502/) (2023)
- [Anchor3DLane: Learning to Regress 3D Anchors for Monocular 3D Lane Detection](https://mlsanity.com/paper/2301.02371/) (2023)
- [HIMap: HybrId Representation Learning for End-to-end Vectorized HD Map Construction](https://mlsanity.com/paper/2403.08639/) (2024)
- [LATR: 3D Lane Detection from Monocular Images with Transformer](https://mlsanity.com/paper/2308.04583/) (2023)
- [Online Vectorized HD Map Construction using Geometry](https://mlsanity.com/paper/2312.03341/) (2023)
- [CurveFormer: 3D Lane Detection by Curve Propagation with Curve Queries and Attention](https://mlsanity.com/paper/2209.07989/) (2022)
- [InstaGraM: Instance-level Graph Modeling for Vectorized HD Map Learning](https://mlsanity.com/paper/2301.04470/) (2023)
- [T2SG: Traffic Topology Scene Graph for Topology Reasoning in Autonomous Driving](https://mlsanity.com/paper/2411.18894/) (2024)
- [Topology Preserving Local Road Network Estimation from Single Onboard Camera Image](https://mlsanity.com/paper/2112.10155/) (2021)
- [3D-LaneNet+: Anchor Free Lane Detection using a Semi-Local Representation](https://mlsanity.com/paper/2011.01535/) (2020)
- [Learning to Predict 3D Lane Shape and Camera Pose from a Single Image via Geometry Constraints](https://mlsanity.com/paper/2112.15351/) (2021)
- [TopoLogic: An Interpretable Pipeline for Lane Topology Reasoning on Driving Scenes](https://mlsanity.com/paper/2405.14747/) (2024)
- [Relation DETR: Exploring Explicit Position Relation Prior for Object Detection](https://mlsanity.com/paper/2407.11699/) (2024)
- [Enhancing 3D Lane Detection and Topology Reasoning with 2D Lane Priors](https://mlsanity.com/paper/2406.03105/) (2024)
- [RoadPainter: Points Are Ideal Navigators for Topology transformER](https://mlsanity.com/paper/2407.15349/) (2024)

## Built on by (most cited first, 3 of 3)

- [RelMap: Enhancing Online Map Construction with Class-Aware Spatial Relation and Semantic Priors](https://mlsanity.com/paper/2507.21567/) (2025)
- [HGeo-TopoMap: Boosting Topological Mapping with Hierarchical Geometric Priors](https://mlsanity.com/paper/2607.21281/) (2026)
- [TopoCurve: Geometry-Aware Topology Reasoning via B'ezier Curves in Autonomous Driving](https://mlsanity.com/paper/2610.09118/) (2026)

## Abstract (arXiv)

Road topology reasoning is fundamental for autonomous driving, requiring both accurate perception of road elements and understanding of their complex connectivity, including lane connectivity (Lane-to-Lane, L2L) and traffic regulation (Lane-to-Traffic signs, L2T). However, existing methods typically treat perception and topology reasoning as fragmented tasks, ignoring their potential for mutual enhancement. Crucially, while topology is inherently relational, prior works often overlook geometric relationships during feature extraction, relying instead on brittle post-processing or coordinate-based heuristics applied only at inference time. To bridge this gap, we propose CoPo (Coherent Perception and toPology), a unified framework that integrates geometry-guided relational modeling across three levels: 1) Perception-level: We introduce a relation-aware lane detector that utilizes geometry-biased self-attention and curve-guided cross-attention to enrich lane representations with structural priors; 2) Reasoning-level: We design relation-enhanced topology heads, including a geometry-enhanced L2L head and a cross-view L2T head, which effectively align features to infer connectivity; and 3) Supervision-level: We implement a contrastive InfoNCE strategy to regularize relational embeddings, pulling connected pairs closer in the latent space. This coherent multi-level design enables end-to-end joint optimization of perception and reasoning. Extensive experiments on OpenLane-V2 demonstrate that CoPo significantly outperforms existing methods, achieving gains of +3.1 in DETl, +5.3 in TOPll, +4.9 in TOPlt, and +4.4 overall in OLS, setting a new state-of-the-art.