[论文] [论文] OSWAM: An Effective World Action Model with One-Step Action Gener...

论文概要

研究领域: CV
作者: Guowen Zhang, Liyi Chen, Xu Hu, Pengfei Wang, Jingfan Chen et al.
发布时间: 2026-10-26
arXiv: 2610.35850

中文摘要

直接策略式世界动作模型（WAM）已证明：把动作生成与显式未来视觉预测解耦有助于推理效率，但它们仍需数十步去噪才能达到有竞争力的动作质量，延迟可观，阻碍实时部署。简单减少去噪步数会把非线性多步去噪轨迹塌缩成粗糙更新，动作质量大幅下降。为在减步数的同时实现实时动作生成，我们提出 OSWAM——一步动作生成的有效世界动作模型。我们认为，正是随机噪声初始化中结构化动作先验的缺失，迫使直接策略 WAM 靠迭代去噪恢复连贯动作序列。因此我们构建「想象 token」初始化扩散过程，提供物理先验与动作条件表征，并引入带未来条件化的特权信息蒸馏增强未来感知动作生成。结果 OSWAM 只需一次扩散步即可高效生成高质量动作。实验中 OSWAM 在自动驾驶与具身控制上达到甚至优于需数十步的直接策略 WAM 与 Joint-WAM；相比直接策略基线约 3 倍加速，在 RTX 5090 上编译推理延迟低至 40ms。

原文摘要

Direct-policy World Action Models (WAM) have demonstrated the benefits of decoupling action generation from explicit future visual prediction for efficient inference in autonomous driving and embodied control. However, they still require tens of denoising steps to achieve competitive action quality, incurring substantial latency and hindering real-time deployment. Simply reducing the number of denoising steps will collapse the nonlinear multi-step denoising trajectory into coarse updates, substantially degrading action quality. To reduce the denoising steps while enabling real-time action generation, we present an effective world action model with one-step action generation, namely OSWAM, for autonomous driving and embodied control. We argue that it is the absence of structured action priors in random noise initialization that forces direct-policy WAMs to recover coherent action sequences through iterative denoising. Hence, we construct imagination tokens to initialize the diffusion process, providing physical priors and action-conditioned representations. We further introduce privileged information distillation with future conditioning to enhance future-aware action generation. As a result, OSWAM can efficiently and effectively generate high quality actions in just one diffusion step. Our experiments demonstrate that OSWAM achieves comparable or even better results in both autonomous driving and embodied control than direct-policy WAM and Joint-WAM that require tens of steps. It obtains approximately \(3\times\) speedup over direct-policy WAM baselines and supports real-time compiled inference with latency as low as 40ms on an NVIDIA RTX 5090 GPU. The project page is available at https://gwenzhang.github.io/OSWAM/

自动采集于 2026-10-28

#论文 #arXiv #CV #小凯