Learning to Annotate Delayed and False AEB Events: A Practical System for Extreme Class Imbalance and Asymmetric Label Noise
Mengxiang Hao, Xin Jiang, Xinghao Huang, Wenliang Su, Zhiteng Wang, Junjie Rao, Xiaotian Yang, Wei Liao, Chengyu Han, Gen Liang, Yulun Song, Zhitao Xu, Xianpeng Lang
AI summary
Problem
Accurately annotating rare delayed and false Autonomous Emergency Braking (AEB) triggers is essential for system optimization but prohibitively expensive due to extreme class imbalance and asymmetric label noise in massive real-world datasets.
Approach
The authors develop a full-stack annotation framework that synthesizes realistic minority-class samples through physics-guided data augmentation and filters mislabeled majority samples using a hardness-driven, probe-guided noise suppression mechanism.
Key results
- First automated AEB annotation model tailored to spatiotemporal trigger data
- Dual-strategy framework combining targeted augmentation and adaptive noise suppression
- 80% improvement in delayed/false trigger recall in production deployment
- 50% reduction in manual annotation workload with a self-improving pipeline
Why it matters
Provides a scalable, cost-effective solution for identifying safety-critical AEB failures, enabling continuous on-vehicle system optimization and reducing engineering bottlenecks.
Abstract
Autonomous Emergency Braking (AEB) optimiza- tion relies on accurately annotated real-world trigger events, particularly rare but critical delayed and false AEB triggers that expose system deficiencies. However, these minority sam- ples comprise less than 5% of thousands of daily triggers, making manual annotation prohibitively expensive at scale. We present the first automated AEB annotation framework to address this problem. During development, we identified two fundamental challenges that severely impair delayed/false trigger annotation accuracy: (1) Extreme class imbalance where delayed/false triggers are overwhelmed by true triggers; (2) Asymmetric label noise where mislabeled majority samples (true triggers) suppress minority samples (delayed/false trig- gers) learning. To overcome these challenges, we propose two key innovations: (1) Specific data augmentation that synthe- sizes realistic samples by manipulating focal target attributes, transplanting ego-vehicle dynamics, and masking non-focal agents; (2) noise suppression using stable hardness estimation and probe-guided adaptive threshold to clean mislabeled true trigger samples. Crucially, we deploy our model as a practical annotation system with full-stack architecture, efficiently iden- tifying critical delayed/false triggers from thousands of daily AEB events. Production results demonstrate 80% improve- ment in recall of delayed/false triggers and 50% reduction in manual workload. Beyond immediate gains, the system enables continuous self-improvement through accumulated high-quality annotations, establishing a necessary data foundation for on- vehicle AEB system optimization.