arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2610.11723cs.AIcs.CV

MultiWorldBench:独立控制的视图是否描述同一个共享世界?

MultiWorldBench: Do Independently Controlled Views Describe One Shared World?

  • Zhiying Guangnian Technology Co., Ltd.(智影光年科技有限公司)

机构由 AI 辅助整理,请以论文原文为准。

Zhangbo Xu, Ruoxi Zhang, Rui Hu, Yisong Wang

AI总结:

该研究推出MultiWorldBench基准,评估Solaris等多人世界模型,发现现有模型在状态持久性等关键能力上表现差,未构成连贯多人世界,与人类偏好评估结果一致。

AI中文摘要:

多人世界模型必须确保独立控制的视图与同一个持久共享世界保持一致。我们推出MultiWorldBench,这是一个诊断性Minecraft基准,包含7个任务套件、10种能力下的495种案例配置,这些能力包括独立控制、跨视图运动、共享状态同步、持久性、结构推理、并发交互和延迟重访。我们评估了Solaris、Gamma-World和MineWorld,以Engine GT作为参考。在生成的系统中,Gamma-World的10种能力平均得分最高,为21.39,其次是Solaris(20.88)和MineWorld(1.89),而Engine GT达到91.69。Gamma-World在多项控制、共享状态和重访能力上表现更好,而Solaris在跨视图运动和竞态条件一致性方面领先。不过,所有生成的系统在状态持久性上的得分最多为8.00,在结构一致性上最多为1.33,且没有一个系统在空间推理或建筑身份保留上取得成功。人类偏好产生了相同的整体排名,且与自动评估高度一致,维度级平均Spearman相关系数为0.96。这些结果表明,目前看似合理的单个视图尚未构成一个连贯的多人世界。

英文摘要:

Multiplayer world models must ensure that independently controlled views remain consistent with one shared and persistent world. We introduce MultiWorldBench, a diagnostic Minecraft benchmark containing 495 case configurations across seven task suites and ten capabilities, including independent control, cross-view motion, shared-state synchronization, persistence, structural reasoning, concurrent interaction, and delayed revisit. We evaluate Solaris, Gamma-World, and MineWorld, using Engine GT as a reference. Gamma-World achieves the highest ten-capability average among the generated systems at 21.39, followed by Solaris at 20.88 and MineWorld at 1.89, while Engine GT reaches 91.69. Gamma-World performs better on several control, shared-state, and revisit capabilities, whereas Solaris leads in cross-view motion and race-condition consistency. Nevertheless, all generated systems score at most 8.00 on state persistence and 1.33 on structural consistency, and none succeeds in spatial reasoning or building-identity preservation. Human preferences produce the same overall ranking and show strong alignment with the automatic evaluation, with a mean dimension-level Spearman correlation of 0.96. These results show that plausible individual views do not yet constitute a coherent multiplayer world.

补充信息

↑