Pith. sign in

REVIEW 4 cited by

LongDWM: Cross-Granularity Distillation for Building a Long-Term Driving World Model

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2506.01546 v1 pith:W5OF76CR submitted 2025-06-02 cs.CV

classification cs.CV
keywords drivingmodelvideoworldlearninglong-termmodelsdistillation
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

Driving world models are used to simulate futures by video generation based on the condition of the current state and actions. However, current models often suffer serious error accumulations when predicting the long-term future, which limits the practical application. Recent studies utilize the Diffusion Transformer (DiT) as the backbone of driving world models to improve learning flexibility. However, these models are always trained on short video clips (high fps and short duration), and multiple roll-out generations struggle to produce consistent and reasonable long videos due to the training-inference gap. To this end, we propose several solutions to build a simple yet effective long-term driving world model. First, we hierarchically decouple world model learning into large motion learning and bidirectional continuous motion learning. Then, considering the continuity of driving scenes, we propose a simple distillation method where fine-grained video flows are self-supervised signals for coarse-grained flows. The distillation is designed to improve the coherence of infinite video generation. The coarse-grained and fine-grained modules are coordinated to generate long-term and temporally coherent videos. In the public benchmark NuScenes, compared with the state-of-the-art front-view model, our model improves FVD by $27\%$ and reduces inference time by $85\%$ for the video task of generating 110+ frames. More videos (including 90s duration) are available at https://Wang-Xiaodong1899.github.io/longdwm/.

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. ScenarioControl: Vision-Language Controllable Vectorized Latent Scenario Generation

    cs.CV 2026-04 unverdicted novelty 7.0 of 10

    ScenarioControl introduces the first vision-language controllable generator for realistic vectorized 3D driving scenarios with temporal consistency across actor views.

  2. OpenLongTail: Generative Scaling of Long-Tail Driving Data

    cs.CV 2026-07 conditional novelty 6.0 of 10

    Pose-informed diffusion with Plücker rays, depth warps, and cross-view memory converts monocular long-tail videos into multi-view assets that improve closed-loop driving robustness nearly to ground-truth multi-view levels.

  3. WorldLens: Full-Spectrum Evaluations of Driving World Models in Real World

    cs.CV 2025-12 conditional novelty 6.0 of 10

    A five-aspect, 24-metric benchmark, a 26K human-annotated dataset, and an AI evaluator show that today's driving world models cannot simultaneously look real, respect geometry, and behave safely.

  4. A Comprehensive Survey on World Models for Embodied AI

    cs.CV 2025-10 conditional novelty 6.0 of 10

    A unified three-axis taxonomy — functionality, temporal modeling, spatial representation — organizes the world-model literature for embodied AI.

Pith tools