让语言只在关键帧使用，提升自动驾驶模型效率与性能。
BLUE: Toward Better Language Use in Efficient Vision-Language-Action Models for Autonomous Driving
- 用轻量门控机制判断每帧是否需要语言，不改主干网络。
- 在Bench2Drive上达76.2%成功率，推理速度提升2.54倍。
- 适合追求高效语言增强的自动驾驶系统开发者。
我们提出BLUE，一种面向自动驾驶视觉-语言-动作（VLA）模型的高效语言使用方法。通过分析发现，语言仅在少数路线上显著影响性能，多数帧使用语言浪费计算资源。我们发现预训练VLA隐状态已能预测语言是否有益，即使场景复杂度和运动特征无法准确判断。基于此，BLUE在冻结的VLA隐状态上训练一个仅0.11M参数的轻量门控模块，决定每帧是否生成语言或直接预测动作，无需修改主干或人工标注。在两个基准上均达到新最优：Bench2Drive成功率达76.2%，Longest6 v2驾驶得分为36，同时实现2.54倍推理加速和8.9%成功率提升。代码、数据、日志及模型权重均已开源。
原文摘要 · Abstract (English)
We present BLUE, a minimal method for better language use in vision-language-action (VLA) models for autonomous driving (AD). Through extensive analysis, we reveal that language matters on only a small fraction of routes, but on those routes it can greatly improve or degrade performance. Generating language at every frame is therefore inefficient, since most computation is spent on frames that do not benefit from language. We further show that pretrained VLA hidden states potentially already encode whether language will benefit a given frame, even though scene complexity and kinematic features alone struggle to predict this. Based on this finding, BLUE trains a lightweight gate on frozen VLA hidden states to decide per frame whether to activate language generation or predict actions directly, without modifying the backbone or requiring additional human annotation. With just a 0.11M-parameter gate, BLUE sets a new state of the art on both benchmarks, achieving 76.2% success rate on Bench2Drive and 36 driving score on Longest6 v2, while delivering 2.54x inference speedup and 8.9% success rate improvement over the backbone. BLUE provides a practical path toward efficient language-augmented AD, showing that VLA models can retain the benefits of language at a fraction of the cost. Our code, data, logs and checkpoints are fully available on https://github.com/George-Ling3/BLUE.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。