Vision-language generative AI has demonstrated re-markable promise for empowering cross-modal scene understanding of autonomous driving and high-definition (HD) map systems. However, current benchmark ...