BADAS: Context Aware Collision Prediction Using Real-World Dashcam Data
This work addresses the issue of false alerts in collision prediction for autonomous driving systems, though it is incremental as it builds on existing benchmarks and methods.
The paper tackles the problem of collision prediction methods failing to distinguish ego-vehicle threats from random accidents, leading to false alerts, by introducing BADAS, a family of models trained on a real-world dashcam dataset, which achieves state-of-the-art AP/AUC and outperforms a forward-collision ADAS baseline.
Existing collision prediction methods often fail to distinguish between ego-vehicle threats and random accidents not involving the ego vehicle, leading to excessive false alerts in real-world deployment. We present BADAS, a family of collision prediction models trained on Nexar's real-world dashcam collision dataset -- the first benchmark designed explicitly for ego-centric evaluation. We re-annotate major benchmarks to identify ego involvement, add consensus alert-time labels, and synthesize negatives where needed, enabling fair AP/AUC and temporal evaluation. BADAS uses a V-JEPA2 backbone trained end-to-end and comes in two variants: BADAS-Open (trained on our 1.5k public videos) and BADAS1.0 (trained on 40k proprietary videos). Across DAD, DADA-2000, DoTA, and Nexar, BADAS achieves state-of-the-art AP/AUC and outperforms a forward-collision ADAS baseline while producing more realistic time-to-accident estimates. We release our BADAS-Open model weights and code, along with re-annotations of all evaluation datasets to promote ego-centric collision prediction research.