发表机构
Alaya Lab; Peking University; Institute of Science Tokyo（Alaya实验室; 北京大学; 东京科学大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究提出MASS模型，通过将世界动态与视图渲染解耦，解决现有视频世界模型在多人环境中的缺陷，在多人Snake基准测试中实现更优状态精度，为多智能体世界模拟提供实用基础。
AI 中文摘要
当前视频世界模型在多人环境中表现不佳，因为它们将世界状态与依赖视图的视觉潜变量纠缠在一起，导致计算冗余、视图不一致性和可扩展性差。我们提出MASS（具有权威共享状态的多人世界模型）来解决这一局限。受多人游戏架构启发，MASS将世界动态与视图渲染解耦。一个学习得到的逻辑引擎从联合动作推进全局的、权威的类型化状态，无需任何手动编写的转移函数，充当唯一的循环记忆和同步参考。从该共享状态出发，一个学习得到的渲染引擎可按需为任意请求的相机生成独立且一致的视图。这种显式解耦使MASS在匹配的多人Snake基准测试中，相较于最先进的多视图基线，实现了更优的状态精度和更低的跨视图不一致性。它推进了包含1024个并发玩家的预测世界，共10000个循环步骤。我们的结果表明，显式的权威状态建模为可扩展且一致的多智能体世界模拟提供了实用基础。
英文摘要
Current video world models struggle in multiplayer environments because they entangle world state with view-dependent visual latents, leading to redundant compute, view inconsistencies, and poor scalability. We propose MASS (Multiplayer world models with Authoritative Shared State) to resolve this limitation. Inspired by multiplayer game architectures, MASS disentangles world dynamics and view rendering. A learned Logic Engine advances a global, authoritative typed state from joint actions without any hand-written transition function, acting as the sole recurrent memory and synchronization reference. From this shared state, a learned Rendering Engine generates independent and consistent views for any requested camera on demand. This explicit disentangling allows MASS to achieve superior state accuracy and lower cross-view inconsistency compared to state-of-the-art multi-view baselines on a matched multiplayer Snake benchmark. It advances predicted worlds with 1,024 concurrent players for 10,000 recurrent steps. Our results show that explicit, authoritative state modeling provides a practical foundation for scalable and consistent multi-agent world simulation.
CommentsProject Page: https://alaya-lab.github.io/MASS/
