SV-WAM: An Efficient Surround-View World-Action Model for End-to-End Autonomous Driving
SV-WAM: An Efficient Surround-View World-Action Model for End-to-End Autonomous Driving
This research paper presents SV-WAM, a surround-view world-action model designed for end-to-end autonomous driving.
The paper introduces SV-WAM, a model that uses future-video prediction during training to supervise action learning while allowing the video branch to be discarded at inference, thereby maintaining six-camera spatial coverage and achieving state-of-the-art planning performance on NAVSIMv2 and nuScenes benchmarks with low latency.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Control
- Reported autonomousOpen source S2
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
arXiv:2609.03602v2 Announce Type: replace-cross Abstract: World models (WMs) have demonstrated strong potential for end-to-end autonomous driving by learning predictive representations of future scene dynamics. However, generating future videos during inference introduces substantial computational overhead, leading many recent driving WMs to adopt a single front camera as input for efficient deployment. This design r
Open source S2
SV-WAM: An Efficient Surround-View World-Action Model for End-to-End Autonomous Driving
Open source S1
ssing drivable boundaries, improving planning safety and boundary awareness. Extensive experiments on the closed-loop NAVSIMv2 benchmark and the open-loop nuScenes benchmark demonstrate that SV-WAM achieves state-of-the-art planning performance with low inference latency and competitive zero-shot transfer capability.
Open source S5
estricts spatial coverage in safety-critical maneuvers such as lane changes, merges, and turns. To address this limitation, we propose SV-WAM, a surround-view world-action model (WAM) that preserves full six-camera observations while maintaining efficient inference. SV-WAM leverages future-video prediction as dense training supervision for action learning within a shared generative model, rather than as an inference-
Open source S3
time output. At the core of this design is an action-centered causal mask that prevents action tokens from attending to future-video tokens during joint action-video denoising. Consequently, the video branch can be discarded at deployment, enabling efficient action-only planning. Furthermore, we introduce a differentiable drivable-area compliance regularizer that penalizes vehicle-footprint corners approaching or cro
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org