CoVLM-Bench：协同驾驶问答与规划的真实世界基准
原标题：CoVLM-Bench: A Real-World Benchmark for Cooperative Driving Question Answering and Planning
AI 摘要
研究者提出 CoVLM-Bench，一个基于 V2X-Seq-SPD 真实车路配对图像构建的协同驾驶基准，同时支持协同驾驶问答（CDQA）与协同规划（CP）两项任务，包含 2,196 对帧、35,136 条问答记录和 2,129 个完整规划样本。基于该基准，作者提出统一 VLM 基线 CoVLM-Drive，直接使用配对视角完成问答与轨迹预测。实验显示 CDQA 适配提升问答准确率，问答初始化与理由监督各自降低规划误差，其标准 CP 配置的最终位移误差低于对比的 V2X 规划器。
正文节选
CoVLM-Bench: A Real-World Benchmark for Cooperative Driving Question Answering and Planning Abstract Vision–language models (VLMs) have made substantial progress in autonomous driving, but their success has primarily been studied in ego-centric scenes. Infrastructure-side observations provide views beyond the ego vehicle’s field of view, yet conventional cooperative-driving systems typically transform them into geometric representations for downstream perception and planning. Directly incorporat