Leveraging V2X for Collaborative HD Maps Construction Using Scene Graph Generation

TL;DR

This paper introduces HDMapLaneNet, a framework that uses V2X communication and scene graph generation to collaboratively build HD maps in real-time, reducing costs and improving map accuracy for autonomous vehicles.

Contribution

It proposes a novel V2X-based collaborative approach utilizing scene graph generation for real-time HD map construction, addressing limitations of traditional mapping methods.

Findings

Superior association prediction on nuScenes dataset

Effective lane centerline extraction from camera images

Enhanced real-time map updating capabilities

Abstract

High-Definition (HD) maps play a crucial role in autonomous vehicle navigation, complementing onboard perception sensors for improved accuracy and safety. Traditional HD map generation relies on dedicated mapping vehicles, which are costly and fail to capture real-time infrastructure changes. This paper presents HDMapLaneNet, a novel framework leveraging V2X communication and Scene Graph Generation to collaboratively construct a localized geometric layer of HD maps. The approach extracts lane centerlines from front-facing camera images, represents them as graphs, and transmits the data for global aggregation to the cloud via V2X. Preliminary results on the nuScenes dataset demonstrate superior association prediction performance compared to a state-of-the-art method.