Impromptu VLA: Open Weights and Open Data for Driving Vision-Language-Action Models
This work addresses a critical bottleneck in autonomous driving by providing open data and benchmarks for unstructured scenarios, though it is incremental as it builds upon existing datasets and methods.
The paper tackles the problem of Vision-Language-Action models faltering in unstructured corner case scenarios for autonomous driving by introducing the Impromptu VLA Dataset, which includes over 80,000 curated video clips and leads to substantial performance gains, such as improved closed-loop NeuroNCAP scores and near state-of-the-art L2 accuracy in open-loop nuScenes trajectory prediction.
Vision-Language-Action (VLA) models for autonomous driving show promise but falter in unstructured corner case scenarios, largely due to a scarcity of targeted benchmarks. To address this, we introduce Impromptu VLA. Our core contribution is the Impromptu VLA Dataset: over 80,000 meticulously curated video clips, distilled from over 2M source clips sourced from 8 open-source large-scale datasets. This dataset is built upon our novel taxonomy of four challenging unstructured categories and features rich, planning-oriented question-answering annotations and action trajectories. Crucially, experiments demonstrate that VLAs trained with our dataset achieve substantial performance gains on established benchmarks--improving closed-loop NeuroNCAP scores and collision rates, and reaching near state-of-the-art L2 accuracy in open-loop nuScenes trajectory prediction. Furthermore, our Q&A suite serves as an effective diagnostic, revealing clear VLM improvements in perception, prediction, and planning. Our code, data and models are available at https://github.com/ahydchh/Impromptu-VLA.