ARETE: Attention-based Rasterized Encoding for Topology Estimation using HSV-transformed Crowdsourced Vehicle Fleet Data
Daniel Fritz, Dimitrios Lagamtzis, Michael Mink, Markus Enzweiler, Steffen Schober
TLDR
ARETE uses a DETR-based approach with rasterized crowdsourced vehicle data to generate accurate HD map centerlines and lane dividers.
Key contributions
- Proposes ARETE, a DETR-based method for HD map generation from crowdsourced vehicle trajectories.
- Transforms vehicle trajectories into a rasterized representation encoding presence and direction.
- Predicts vectorized lane representations, including centerlines with direction and constrained lane dividers.
- Evaluated on internal, nuScenes, and nuPlan datasets for robust performance.
Why it matters
Accurate and up-to-date HD maps are crucial for autonomous driving safety and efficiency. This paper offers a novel, data-driven approach to automatically generate and maintain these maps using readily available crowdsourced vehicle data. This reduces reliance on expensive manual mapping.
Original Abstract
The continuous advancement of autonomous driving (AD) introduces challenges across multiple disciplines to ensure safe and efficient driving. One such challenge is the generation of High-Definition (HD) maps, which must remain up to date and highly accurate for downstream automotive tasks. One promising approach is the use of crowdsourced data from a vehicle fleet, representing road topology and lane-level features. This work focuses on the generation of centerlines and lane dividers from crowdsourced vehicle trajectories. We adopt a Detection Transformer (DETR)-based approach, where a rasterized representation of vehicle trajectories is used as input to predict vectorized lane representations. Each lane consists of a centerline with an associated direction and corresponding lane dividers that are geometrically constrained by the centerline. Our method includes the extraction of local tiles, from which crowdsourced vehicle trajectories are aggregated. Each tile undergoes a transformation into a rasterized representation encoding both the presence and direction of each trajectory, enabling the prediction of vectorized directed lanes. Experiments are conducted on an internal dataset as well as on the public datasets nuScenes and nuPlan.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.