HorizonDrive: 用于长时域驾驶仿真的自纠正自回归世界模型
HorizonDrive: Self-Corrective Autoregressive World Model for Long-horizon Driving Simulation
AI总结：
针对自回归驾驶仿真中教师模型因自身预测漂移导致监督退化的问题，提出HorizonDrive框架，通过计划性滚转恢复（SRR）训练抗漂移教师模型，并利用教师滚转分布匹配蒸馏（TRD）实现长时域监督下的高效学生模型部署。
摘要：
闭环驾驶仿真需要超越短时离线片段的实时交互，推动当前驾驶世界模型向自回归（AR）滚转发展。现有的AR蒸馏方法通常依赖于帧沉或学生端退化训练。前者由于快速的自我运动和场景变化，难以迁移到驾驶场景；后者受限于教师单次输出长度，仅提供有限的监督时域。一个自然的问题是：能否通过AR滚转扩展教师本身，以有限的内存成本提供无限时域的监督？关键困难在于标准教师会在自身预测下漂移，污染其提供的监督。我们的关键见解是使教师具备滚转能力，确保从其自身的AR滚转中获得可靠监督。这实例化为HorizonDrive，一个用于AR驾驶仿真的抗漂移训练与蒸馏框架。首先，计划性滚转恢复（SRR）训练基础模型从预测损坏的历史中重建真实未来片段，得到一个在长AR滚转中保持稳定的教师。其次，通过AR滚转扩展具备滚转能力的教师，在有限内存下提供长时域分布匹配监督，同时短窗口学生通过教师滚转DMD（TRD）与之对齐，以实现高效的实时部署。HorizonDrive原生支持在有限内存下的分钟级AR滚转；在nuScenes上，与最强的长时域流式基线相比，HorizonDrive将FID降低52%，FVD降低37%，并将ARE和DTW分别降低21%和9%，同时与单次驾驶视频生成器保持竞争力。
英文摘要：
Closed-loop driving simulation requires real-time interaction beyond short offline clips, pushing current driving world models toward autoregressive (AR) rollout. Existing AR distillation approaches typically rely on frame sinks or student-side degradation training. The former transfers poorly to driving due to fast ego-motion and rapid scene changes, while the latter remains bounded by the teacher's single-pass output length and thus provides only a limited supervision horizon. A natural question is: can the teacher itself be extended via AR rollout to provide unbounded-horizon supervision at bounded memory cost? The key difficulty is that a standard teacher drifts under its own predictions, contaminating the supervision it provides. Our key insight is to make the teacher rollout-capable, ensuring reliable supervision from its own AR rollouts. This is instantiated as HorizonDrive, an anti-drifting training-and-distillation framework for AR driving simulation. First, scheduled rollout recovery (SRR) trains the base model to reconstruct ground-truth future clips from prediction-corrupted histories, yielding a teacher that remains stable across long AR rollouts. Second, the rollout-capable teacher is extended via AR rollout, providing long-horizon distribution-matching supervision under bounded memory, while a short-window student aligns to it with teacher rollout DMD (TRD) for efficient real-time deployment. HorizonDrive natively supports minute-scale AR rollout under bounded memory; on nuScenes, HorizonDrive reduces FID by 52% and FVD by 37%, and lowers ARE and DTW by 21% and 9% relative to the strongest long-horizon streaming baselines, while remaining competitive with single-pass driving video generators.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
