研究人员正在开发新颖的世界动作模型（WAM），以提高AI代理在视频游戏和机器人中的性能。例如，GameWAM统一了视觉预测和动作生成，以实现原生的游戏控制，并以更少的动作展示了具有竞争力的结果。其他进展包括LAWA，它使用潜在动作在机器人中进行高效的未来想象；RISE，一个优化规划想象展开的自适应框架；4DGS-WAM利用4D高斯溅射进行以对象为中心的建模；以及GlanceWAM，它将想象与控制分离以实现实时推理。 AI
影响 世界动作模型的这些进步可能导致在视频游戏和机器人等复杂环境中出现更强大的AI代理，从而提高效率和性能。
排序理由 多篇研究论文介绍了世界动作模型的新模型和框架。
AI 生成摘要 · Google Gemini · 来自 10 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2608.26200v1 Announce Type: new Abstract: Modern video games combine first-person perception, rapid visual changes, persistent world state, and heterogeneous native controls. Existing game agents map visual and task context directly to actions but lack explicit world dynami…
Game engines provide executable verification and long-horizon trajectories for reinforcement learning post-training of spatial world models, motivating a human-engine verification paradigm.
LAWA improves robot control by using compact latent actions to retain efficient future imagination without generating observations, achieving strong performance with lower latency.
GameWAM is a unified world-action model for native video-game control that jointly predicts future visuals and executable keyboard-mouse actions using block-causal flow matching, mode-specific distributions, and block-cycle replanning.
RISE adaptively decides when to continue or stop imagination rollouts for planning by weighing expected benefit against cost, supported by a counterfactual driving dataset with expert annotations.
arXiv:2608.25956v1 Announce Type: new Abstract: Current world action models (WAMs) typically operate on 2D visual data. These models can achieve exceptional visual quality, but they lack explicit spatial structure for individual objects and repeatedly process redundant background…
arXiv:2608.23927v1 Announce Type: new Abstract: Video generative models provide rich physical priors for robot learning, yet existing world-action models (WAMs) face a fundamental trade-off: synchronous video generation at control rate is latency-prohibitive, while abandoning tes…
arXiv:2608.20430v1 Announce Type: new Abstract: World Action Models (WAMs) improve planning by incorporating future world evolution into action generation, yet existing methods allocate a fixed imagination budget to every scene. We propose RISE (\textbf{R}efining \textbf{I}magina…
📄 Paper 'Agentic Game Development as a Verifiable Trajectory Data Engine for Scaling World Models' hit 132 upvotes on Hugging Face. Using game dev as a data engine for world models is a fresh take on scaling. https:// huggingface.co/papers/2608.255 18 # AI # MachineLearning # Res…
Agentic Game Development as a Verifiable Trajectory Data Engine for Scaling World Models hit 113 upvotes on Hugging Face – using game dev to generate training data for world models opens a new path to scalable, verifiable AI. https:// huggingface.co/papers/2608.255 18 # AI # Mach…