HorizonDrive

Self-Corrective Autoregressive World Model for Long-horizon Driving Simulation

Authors & Affiliations

Conglang Zhang^1,* Yifan Zhan^2,* Qingjie Wang³ Zhanpeng Ouyang³ Yu Li⁴
Zihao Yang⁵ Xiaoyang Guo⁶ Weiqiang Ren³ Qian Zhang³ Zhen Dong¹
Yinqiang Zheng² Wei Yin^3,‡ Zhengqing Chen^3,†

¹ Wuhan University ² The University of Tokyo ³ Horizon Robotics ⁴ Tsinghua University
⁵ University of Science and Technology of China ⁶ The Chinese University of Hong Kong

^* Equal contribution ^‡ Project lead ^† Corresponding author

arXiv 2605.11596

GitHub HorizonDrive

Film mode Preview

HorizonDrive as a Closed-Loop Driving Simulator

We scale up HorizonDrive to seven views and use it as a simulator for multi-round interaction with a vision-language-action (VLA) model.

Long-horizon rollouts

Driving beyond short clips

Nine nuScenes scenes compare HorizonDrive with long-horizon streaming baselines.

Abstract

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.

Key Features

Controllable Driving scene generation

Long-Horizon Stable generation quality

Interactive AR rollout

Scalable Diverse driving scenes

Closed-Loop Simulation ready

No Explicit 3D representations

Method Overview

A rollout-capable teacher stabilizes long-horizon generation and distills it into an efficient student.

Overview of HorizonDrive framework. We first train a conditional driving world model, then improve its autoregressive stability through scheduled rollout recovery, and finally distill long-horizon teacher rollouts into a few-step, short-chunk student via teacher-rollout DMD.

20-Second AR Results on Nuscenes

Nine nuScenes scenes compare HorizonDrive with long-horizon streaming baselines.

30-Second AR Results on Self-Collected Dataset

Diverse self-collected scenes test scaling and generalization beyond nuScenes.

Minute-Level AR Video Generation

Minute-scale autoregressive rollouts sustain visual and motion consistency over long horizons.

Quantitative Results

Metrics across both datasets summarize fidelity, motion quality, and long-horizon stability.

nuScenes val

Method	FID ↓	FVD ↓	Qual. ↑	Mot. ↑	Img. ↑	ARE ↓	DTW ↓
Long-horizon interactive world model frameworks
Matrix-Game3	35.69	338.22	78.99	93.78	60.44	N/A	N/A
Helios	30.53	218.23	79.02	95.03	58.82	N/A	N/A
Causal-Forcing	49.07	373.29	74.35	92.42	59.00	N/A	N/A
HY-WorldPlay	33.51	580.72	76.58	99.48	58.60	N/A	N/A
LingBot-World	37.67	325.55	77.08	92.87	55.55	N/A	N/A
Long-horizon streaming methods (re-trained on our base model and data)
Self-Forcing	41.53	161.00	79.27	94.17	59.65	3.47	6.22
Self-Forcing++	28.84	147.57	79.47	93.92	60.25	3.78	3.61
LongLive	29.05	161.41	79.35	93.46	60.80	3.28	3.65
HorizonDrive (Ours)	13.82	92.99	79.53	93.85	62.50	2.60	3.27

Self-collected dataset

Method	FID ↓	FVD ↓	Qual. ↑	Mot. ↑	Img. ↑	ARE ↓	DTW ↓
Long-horizon streaming methods (re-trained on our base model and e2e data)
Self-Forcing	58.23	561.11	76.68	94.48	63.18	5.43	14.13
Self-Forcing++	66.93	534.36	74.54	92.70	59.12	7.32	18.40
LongLive	28.39	374.94	78.18	94.57	62.53	4.05	8.11
HorizonDrive (Ours)	12.01	117.27	80.12	95.22	67.65	3.67	5.29

Citation

If you find our work useful, please cite it as:

@misc{zhang2026horizondriveselfcorrectiveautoregressiveworld,
  title={HorizonDrive: Self-Corrective Autoregressive World Model for Long-horizon Driving Simulation},
  author={Zhang, Conglang and Zhan, Yifan and Wang, Qingjie and Ouyang, Zhanpeng and Li, Yu and Yang, Zihao and Guo, Xiaoyang and Ren, Weiqiang and Zhang, Qian and Dong, Zhen and Zheng, Yinqiang and Yin, Wei and Chen, Zhengqing},
  year={2026},
  eprint={2605.11596},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2605.11596},
}