DAMap: Distance-aware MapNet for High Quality HD Map Construction

2025-10-26

AI Summary

Existing HD map construction methods suffer from task misalignment—specifically, label ambiguity arising from one-to-one matching and feature misalignment caused by shared sampling—hindering simultaneous high classification accuracy and precise geometric localization. To address this, we propose DAMap, a novel framework featuring three key innovations: (1) a distance-aware focal loss to mitigate class imbalance between near and far map elements; (2) a hybrid loss strategy enabling fine-grained label assignment under one-to-many matching; and (3) a task-modulated deformable attention mechanism that decouples and specializes features for classification and localization tasks. Evaluated on NuScenes and Argoverse2 benchmarks, DAMap achieves state-of-the-art performance across multiple metrics, demonstrating strong generalization across diverse backbones, data splits, and training configurations.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Computer Vision: Multi-modal Vision
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Predicting High-definition (HD) map elements with high quality (high classification and localization scores) is crucial to the safety of autonomous driving vehicles. However, current methods perform poorly in high quality predictions due to inherent task misalignment. Two main factors are responsible for misalignment: 1) inappropriate task labels due to one-to-many matching queries sharing the same labels, and 2) sub-optimal task features due to task-shared sampling mechanism. In this paper, we reveal two inherent defects in current methods and develop a novel HD map construction method named DAMap to address these problems. Specifically, DAMap consists of three components: Distance-aware Focal Loss (DAFL), Hybrid Loss Scheme (HLS), and Task Modulated Deformable Attention (TMDA). The DAFL is introduced to assign appropriate classification labels for one-to-many matching samples. The TMDA is proposed to obtain discriminative task-specific features. Furthermore, the HLS is proposed to better utilize the advantages of the DAFL. We perform extensive experiments and consistently achieve performance improvement on the NuScenes and Argoverse2 benchmarks under different metrics, baselines, splits, backbones, and schedules. Code will be available at https://github.com/jpdong-xjtu/DAMap.

Problem

• Addresses misalignment in HD map prediction for autonomous driving safety
• Solves inappropriate task labels from one-to-many matching queries
• Resolves sub-optimal features from task-shared sampling mechanisms

Innovation

• Distance-aware Focal Loss for classification labels
• Task Modulated Deformable Attention for features
• Hybrid Loss Scheme to optimize training