EMMA：面向自动驾驶的端到端多模态模型

以 Gemini 为基础，从摄像头输入与文本描述预测驾驶轨迹、感知对象和道路结构。

EMMA: End-to-End Multimodal Model for Autonomous Driving

为什么值得读

直接观察通用多模态知识如何迁移到驾驶任务，而不是把语言模型与自动驾驶看成两条无关路线。

引用

@misc{hwang2024emma,
  title = {EMMA: End-to-End Multimodal Model for Autonomous Driving},
  author = {Jyh-Jing Hwang and Runsheng Xu and Hubert Lin and Wei-Chih Hung and Jingwei Ji and Kristy Choi and Di Huang and Tong He and Paul Covington and Benjamin Sapp and Yin Zhou and James Guo and Dragomir Anguelov and Mingxing Tan},
  year = {2024},
  eprint = {2410.23262},
  archivePrefix = {arXiv},
  primaryClass = {cs.CV},
  url = {https://arxiv.org/abs/2410.23262}
}

论文与代码版权归原作者；引用及许可请以原始来源为准。

相关研究