A-BDD: Leveraging Data Augmentations for Safe Autonomous Driving in Adverse Weather and Lighting

TL;DR

This paper introduces A-BDD, a large synthetic dataset with diverse weather and lighting augmentations, to improve autonomous vehicle perception in challenging conditions, demonstrating the effectiveness of data augmentation strategies.

Contribution

The paper presents A-BDD, a new large-scale synthetic dataset with diverse weather and lighting augmentations, and novel quality metrics for data selection to enhance autonomous driving perception.

Findings

Data augmentation improves perception accuracy in adverse conditions.

Feature-based metrics like FID and CMMD effectively select useful data.

Augmented data helps close performance gaps in challenging environments.

Abstract

High-autonomy vehicle functions rely on machine learning (ML) algorithms to understand the environment. Despite displaying remarkable performance in fair weather scenarios, perception algorithms are heavily affected by adverse weather and lighting conditions. To overcome these difficulties, ML engineers mainly rely on comprehensive real-world datasets. However, the difficulties in real-world data collection for critical areas of the operational design domain (ODD) often means synthetic data is required for perception training and safety validation. Thus, we present A-BDD, a large set of over 60,000 synthetically augmented images based on BDD100K that are equipped with semantic segmentation and bounding box annotations (inherited from the BDD100K dataset). The dataset contains augmented data for rain, fog, overcast and sunglare/shadow with varying intensity levels. We further introduce novel strategies utilizing feature-based image quality metrics like FID and CMMD, which help identify useful augmented and real-world data for ML training and testing. By conducting experiments on A-BDD, we provide evidence that data augmentations can play a pivotal role in closing performance gaps in adverse weather and lighting conditions.