ROAR: Robust Accident Recognition and Anticipation for Autonomous Driving

2025-11-09

AI Summary

Existing autonomous driving accident prediction methods exhibit insufficient robustness under realistic conditions—such as sensor failures, environmental disturbances, and incomplete data—and overlook inter-vehicle differences in driving behavior and accident propensity. To address these limitations, we propose a multimodal robust prediction framework: (1) discrete wavelet transform (DWT) is employed to enhance noise suppression for incomplete time-series data; (2) an adaptive target-aware module models spatiotemporal dependencies to improve high-risk vehicle identification; and (3) a dynamic focal loss function mitigates severe class imbalance. This work is the first to jointly tackle noise robustness, target bias correction, and class imbalance. Evaluated on DAD, CCD, and A3D benchmarks, our method achieves significant improvements in average precision (AP) and mean time-to-accident (mTTA) over state-of-the-art approaches, demonstrating superior accuracy and generalization in real-world scenarios.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Natural Language Processing: Safety and Robustness

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Accurate accident anticipation is essential for enhancing the safety of autonomous vehicles (AVs). However, existing methods often assume ideal conditions, overlooking challenges such as sensor failures, environmental disturbances, and data imperfections, which can significantly degrade prediction accuracy. Additionally, previous models have not adequately addressed the considerable variability in driver behavior and accident rates across different vehicle types. To overcome these limitations, this study introduces ROAR, a novel approach for accident detection and prediction. ROAR combines Discrete Wavelet Transform (DWT), a self adaptive object aware module, and dynamic focal loss to tackle these challenges. The DWT effectively extracts features from noisy and incomplete data, while the object aware module improves accident prediction by focusing on high-risk vehicles and modeling the spatial temporal relationships among traffic agents. Moreover, dynamic focal loss mitigates the impact of class imbalance between positive and negative samples. Evaluated on three widely used datasets, Dashcam Accident Dataset (DAD), Car Crash Dataset (CCD), and AnAn Accident Detection (A3D), our model consistently outperforms existing baselines in key metrics such as Average Precision (AP) and mean Time to Accident (mTTA). These results demonstrate the model's robustness in real-world conditions, particularly in handling sensor degradation, environmental noise, and imbalanced data distributions. This work offers a promising solution for reliable and accurate accident anticipation in complex traffic environments.

Problem

• Addresses sensor failures and data imperfections in accident prediction
• Handles variability in driver behavior across vehicle types
• Solves class imbalance and environmental noise in autonomous driving

Innovation

• Uses Discrete Wavelet Transform for noisy data
• Employs object aware module for high risk vehicles
• Applies dynamic focal loss for class imbalance