通过协同感知-规划蒸馏，提升自动驾驶模型的长期稳定性与感知精度。
EvoDriveVLA: Evolving Driving VLA Models via Collaborative Perception-Planning Distillation
- 采用自锚定视觉蒸馏与未来感知轨迹优化，实现双路径知识迁移。
- 在nuScenes开环测试中达当前最优，在NAVSIM闭环测试中性能显著提升。
- 适合研究自动驾驶多模态模型泛化与长期规划的开发者参考。
视觉-语言-动作模型在自动驾驶中展现巨大潜力，但解冻视觉编码器后感知能力下降，长期规划易产生累积不稳定性。为此，我们提出EvoDriveVLA——一种协同感知-规划蒸馏框架，融合自锚定感知约束与未来感知轨迹优化。具体而言，自锚定视觉蒸馏利用自锚定教师模型提供视觉锚定约束，通过轨迹引导的关键区域感知正则化学生模型表征；同时，未来感知轨迹蒸馏采用具备粗到细轨迹精炼能力的未来感知教师模型，并结合蒙特卡洛丢弃采样生成模拟推理轨迹，使学生模型内化教师的前瞻性决策知识。EvoDriveVLA在nuScenes开环评估中达到当前最优（SOTA）表现，并在NAVSIM闭环评估中显著提升性能。代码已公开：https://github.com/hey-cjj/EvoDriveVLA。
原文摘要 · Abstract (English)
Vision-Language-Action models have shown great promise for autonomous driving, yet they suffer from degraded perception after unfreezing the visual encoder and struggle with accumulated instability in long-term planning. To address these challenges, we propose EvoDriveVLA-a novel collaborative perception-planning distillation framework that integrates self-anchored perceptual constraints and future-informed trajectory optimization. Specifically, self-anchored visual distillation leverages self-anchor teacher to deliver visual anchoring constraints, regularizing student representations via trajectory-guided key-region awareness. In parallel, future-informed trajectory distillation employs a future-aware oracle teacher with coarse-to-fine trajectory refinement and Monte Carlo dropout sampling to synthesize reasoning trajectories that model future evolutions, enabling the student model to internalize the future-aware insights of the teacher. EvoDriveVLA achieves SOTA performance in nuScenes open-loop evaluation and significantly enhances performance in NAVSIM closed-loop evaluation. Our code is available at: https://github.com/hey-cjj/EvoDriveVLA.
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。