RelMap: Enhancing Online Map Construction with Class-Aware Spatial Relation and Semantic Priors

TL;DR

RelMap is an innovative framework that improves online HD map construction for autonomous driving by explicitly modeling spatial relations and semantic priors, leading to enhanced accuracy and generalization.

Contribution

The paper introduces RelMap, which incorporates class-aware spatial relation priors and a mixture-of-experts semantic prior, advancing the state-of-the-art in online HD map construction.

Findings

Achieves state-of-the-art performance on nuScenes and Argoverse 2 datasets.

Effectively models spatial dependencies and semantic relationships among map elements.

Compatible with both single-frame and temporal perception backbones.

Abstract

Online high-definition (HD) map construction is crucial for scaling autonomous driving systems. While Transformer-based methods have become prevalent in online HD map construction, most existing approaches overlook the inherent spatial dependencies and semantic relationships among map elements, which constrains their accuracy and generalization capabilities. To address this, we propose RelMap, an end-to-end framework that explicitly models both spatial relations and semantic priors to enhance online HD map construction. Specifically, we introduce a Class-aware Spatial Relation Prior, which explicitly encodes relative positional dependencies between map elements using a learnable class-aware relation encoder. Additionally, we design a Mixture-of-Experts-based Semantic Prior, which routes features to class-specific experts based on predicted class probabilities, refining instance feature decoding. RelMap is compatible with both single-frame and temporal perception backbones, achieving state-of-the-art performance on both the nuScenes and Argoverse 2 datasets.