An interactive enhanced driving dataset for autonomous driving

Engineering notes

Key topics: autonomous driving, bev, nuplan, waymo, level 5, perception. See the paper for implementation details and experimental results.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Abstract Driving interaction data are important for training and evaluating autonomous driving Vision-Language-Action (VLA) models, but existing datasets contain limited dense interaction samples and weak alignment between trajectories, visual inputs, and language annotations. This work presents the Interactive Enhanced Driving Dataset (IEDD), a large-scale interaction-oriented dataset constructed from five naturalistic trajectory datasets: Lyft Level 5, Waymo, nuPlan, INTERACTION, and SIND. IEDD contains 7.31 million ego-centric interaction segments, including 6.66 million multi agents cases, covering head-on, car-following, merging, and crossing interactions. Each segment is associated with trajectory-derived interaction metrics describing interaction intensity and efficiency. Based on these annotations, IEDD-VQA further provides trajectory-reconstructed BEV videos, structured interaction semantics, and multi-turn question-answer pairs. The dataset can support interaction mining, long-tail scenario analysis, VLA instruction tuning, and hierarchical evaluation of perception, behavior description, physical quantification, and counterfactual reasoning.

Links and sources

Comments