🤖 AI Summary
Long-tail “corner cases” (CCs)—unpredictable, potentially hazardous scenarios—pose a critical challenge to the large-scale deployment of autonomous driving systems.
Method: This paper proposes the first machine-learning-oriented, computationally tractable description paradigm for CCs. It integrates knowledge-driven taxonomy construction with data-driven representation modeling to design a machine-parsable descriptive language, systematically formalizing CC definitions, root causes, and multidimensional representations, while conducting data distribution analysis.
Contribution/Results: We introduce the first structured CC description framework that bridges domain knowledge with ML requirements. This framework enables efficient offline CC mining and enhances online system robustness, directly alleviating key bottlenecks in CC acquisition, annotation, simulation, and verification. Empirical evaluation demonstrates substantial improvements in both coverage and scalability of CC handling across the autonomous driving development pipeline.
📝 Abstract
Scaling the distribution of automated vehicles requires handling various unexpected and possibly dangerous situations, termed corner cases (CC). Since many modules of automated driving systems are based on machine learning (ML), CC are an essential part of the data for their development. However, there is only a limited amount of CC data in large-scale data collections, which makes them challenging in the context of ML. With a better understanding of CC, offline applications, e.g., dataset analysis, and online methods, e.g., improved performance of automated driving systems, can be improved. While there are knowledge-based descriptions and taxonomies for CC, there is little research on machine-interpretable descriptions. In this extended abstract, we will give a brief overview of the challenges and goals of such a description.
