AutoGraph: Predicting Lane Graphs from Traffic Observations

TL;DR

AutoGraph predicts lane graphs from traffic observations using motion patterns, eliminating the need for hand-annotated data, and achieves comparable performance to supervised models on urban datasets.

Contribution

It introduces a novel method that leverages traffic participant motion for lane graph prediction without human annotations.

Findings

AutoGraph performs on par with models trained on hand-annotated data.

The approach effectively uses traffic motion patterns for lane graph estimation.

Model and dataset will be publicly available.

Abstract

Lane graph estimation is a long-standing problem in the context of autonomous driving. Previous works aimed at solving this problem by relying on large-scale, hand-annotated lane graphs, introducing a data bottleneck for training models to solve this task. To overcome this limitation, we propose to use the motion patterns of traffic participants as lane graph annotations. In our AutoGraph approach, we employ a pre-trained object tracker to collect the tracklets of traffic participants such as vehicles and trucks. Based on the location of these tracklets, we predict the successor lane graph from an initial position using overhead RGB images only, not requiring any human supervision. In a subsequent stage, we show how the individual successor predictions can be aggregated into a consistent lane graph. We demonstrate the efficacy of our approach on the UrbanLaneGraph dataset and perform extensive quantitative and qualitative evaluations, indicating that AutoGraph is on par with models trained on hand-annotated graph data. Model and dataset will be made available at redacted-for-review.