SUV: Future Scene Understanding as Video Generation for End-to-End Driving

• URL: http://arxiv.org/abs/2608.03084v1
• Date: Tue, 04 Aug 2026 03:51:53 GMT
• Title: SUV: Future Scene Understanding as Video Generation for End-to-End Driving
• Abstract summary: We introduce SUV, a unified end-to-end driving framework that casts future Scene Understanding as Video generation.<n>SUV models future appearance, semantics, relative depth, and instance-level dynamics as video streams with a shared video expert.<n>Experiments show that SUV directly predicts all four future streams, while controlled ablations show that structured future supervision and direct future-stream access yield higher trajectory planning scores.
• Score: 71.39332228514006
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: End-to-end driving requires a coherent understanding of future scenes, yet existing methods model these scenes using task-specific heads and output formats, with limited scalability. Can video generation instead provide a shared predictor? We introduce SUV, a unified end-to-end driving framework that casts future Scene Understanding as Video generation using a pretrained video foundation model. SUV models future appearance, semantics, relative depth, and instance-level dynamics as video streams with a shared video expert, without stream-specific visual prediction heads. Through joint video-action attention, the action expert attends to the latent representations of all future streams and generates the ego trajectory. Experiments show that SUV directly predicts all four future streams, while controlled ablations show that structured future supervision and direct future-stream access yield higher trajectory planning scores. With only a single front camera and no candidate-trajectory selection, SUV outperforms a broad set of recent state-of-the-art methods on both NAVSIM-v2 splits, achieving 91.0 EPDMS on navtest and 36.9 on navhard. On the long-tail WOD-E2E benchmark, SUV achieves a competitive RFS of 7.94.

Related papers

• SimWAM: A Simple World Action Model for End-to-End Autonomous Driving [66.93548079427477]
  We present SimWAM, a simple yet effective WAM that leverages future-video prediction as a training-time supervision signal.<n>An isolated attention mask keeps action prediction independent of future frames, allowing trajectory prediction without explicit future-frame generation at inference.<n>Our SimWAM achieves 91.5 PDMS on NAVSIM, surpasses state-of-the-art WAM-based planners with substantially lower latency, and transfers zero-shot to nuScenes.
  arXiv  Detail & Related papers  (2026-08-07T17:59:09Z)
• UNIVERSE: Unified Video Action Models for Autonomous Driving with Flexible Mask-Modulated Modality Generation [23.675212404889805]
  World Action Models (WAMs) have shown strong potential for improving action generalization in autonomous driving.<n>Existing cascaded or dual-DiT designs separate video imagination from action prediction.<n>We propose UNIVERSE, a unified video-action model built upon a single mask-modulated Transformer.
  arXiv  Detail & Related papers  (2026-07-06T14:18:18Z)
• DriveWAM: Video Generative Priors Enable Scalable World-Action Modeling for Autonomous Driving [19.231942908215174]
  We present DriveWAM, a driving world-action model that adapts a pretrained video diffusion transformer into an autoregressive video-action policy.<n>DriveWAM organizes video and action streams into a unified temporal token sequence and trains them under a joint flow-matching objective.<n>Experiments on NAVSIM and the PhysicalAI-Autonomous-Vehicles benchmark show that DriveWAM achieves strong planning performance.
  arXiv  Detail & Related papers  (2026-05-27T14:36:26Z)
• LMGenDrive: Bridging Multimodal Understanding and Generative World Modeling for End-to-End Driving [60.31765454895336]
  We present LMGenDrive, the first framework that combines multimodal understanding with generative world models for end-to-end closed-loop driving.<n>We propose a progressive three-stage training strategy, from vision pretraining to multi-step long-horizon driving, to improve stability and performance.
  arXiv  Detail & Related papers  (2026-04-09T19:13:14Z)
• Future-Aware End-to-End Driving: Bidirectional Modeling of Trajectory Planning and Scene Evolution [96.25314747309811]
  We introduce SeerDrive, a novel end-to-end framework that jointly models future scene evolution and trajectory planning.<n>Our method first predicts future bird's-eye view (BEV) representations to anticipate the dynamics of the surrounding scene.<n>Two key components enable this: (1) future-aware planning, which injects predicted BEV features into the trajectory planner, and (2) iterative scene modeling and vehicle planning.
  arXiv  Detail & Related papers  (2025-10-13T07:41:47Z)
• ProphetDWM: A Driving World Model for Rolling Out Future Actions and Videos [13.630119246378518]
  We argue that driving world models should have two additional abilities: action control and action prediction.<n>We propose ProphetDWM, a novel end-to-end driving world model that jointly predicts future videos and actions.
  arXiv  Detail & Related papers  (2025-05-24T11:35:09Z)
• GenAD: Generalized Predictive Model for Autonomous Driving [75.39517472462089]
  We introduce the first large-scale video prediction model in the autonomous driving discipline. Our model, dubbed GenAD, handles the challenging dynamics in driving scenes with novel temporal reasoning blocks. It can be adapted into an action-conditioned prediction model or a motion planner, holding great potential for real-world driving applications.
  arXiv  Detail & Related papers  (2024-03-14T17:58:33Z)
• Policy Pre-training for End-to-end Autonomous Driving via Self-supervised Geometric Modeling [96.31941517446859]
  We propose PPGeo (Policy Pre-training via Geometric modeling), an intuitive and straightforward fully self-supervised framework curated for the policy pretraining in visuomotor driving. We aim at learning policy representations as a powerful abstraction by modeling 3D geometric scenes on large-scale unlabeled and uncalibrated YouTube driving videos. In the first stage, the geometric modeling framework generates pose and depth predictions simultaneously, with two consecutive frames as input. In the second stage, the visual encoder learns driving policy representation by predicting the future ego-motion and optimizing with the photometric error based on current visual observation only.
  arXiv  Detail & Related papers  (2023-01-03T08:52:49Z)
• Vision-Guided Forecasting -- Visual Context for Multi-Horizon Time Series Forecasting [0.6947442090579469]
  We tackle multi-horizon forecasting of vehicle states by fusing the two modalities. We design and experiment with 3D convolutions for visual features extraction and 1D convolutions for features extraction from speed and steering angle traces. We show that we are able to forecast a vehicle's state to various horizons, while outperforming the current state-of-the-art results on the related task of driving state estimation.
  arXiv  Detail & Related papers  (2021-07-27T08:52:40Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.