日本フィジカルAI新聞

世界のフィジカルAIを、日本語で。

週刊ニュースレター購読
世界モデルarXiv:2610.18447

Thinker-WM: 身体性知能のための共有マルチモーダル表現を持つ統合世界モデル

Thinker-WM: A Unified World Model with Shared Multimodal Representations for Embodied Intelligence

シェア:XThreadsFacebookLINEはてブBluesky

RGB・深度・ロボット動作を共有の動的表現で統合し、動画生成や政策推論など多機能を単一モデルで実現した世界モデル。

詳しい要約

1. どんなもの?

- Thinker-WMは、embodied AI向けの統合multimodal world model。 - RGB、metric depth、robot actionを共通のintermediate表現で jointly にモデル化・生成。 - 共有のUnified Dynamics Transformerとmodality-specific encoders/decodersで構成。 - 単一パラメータでjoint video-action generation、world prediction、policy inference、action-conditioned generation、video-to-action generation、inverse dynamicsをサポート。

2. 先行研究と比べてどこがすごい?

- 既存のworld modelはmodalityや機能がfragmentedで、visual dynamics・geometry・actionを別interfaceやdownstream moduleで結合。 - Thinker-WMはrepresentation levelでmultimodal unificationを実現し、appearance・geometry・actionを共通dynamic representationに変換。 - 共有パラメータでjointly evolveしつつ、各出力構造を保持。 - 単一parameterizationで多機能を実現する点が先行研究と異なる。

3. 技術・手法の肝は?

- 共有intermediate Unified Dynamics Transformerが核。 - modality-specific encodersでRGB・metric depth・actionを共通dynamic representationへ変換。 - 共有パラメータでjointly evolveし、modality-specific decodersで各出力を生成。 - 軽量motion probeを導入し、frozen visual representationにaction関連motion情報が含まれるか検証。

4. どうやって有効だと検証した?

- 約20K時間のegocentric video、10K時間のreal-robot data、1K時間のsimulation dataで学習。 - うち約4K時間のreal-robot dataにはaligned depth supervision。 - EWMBench、WorldOlympiad、WorldArenaで1位を獲得。 - controlled studiesでaction conditioning、geometric interaction、multi-function trainingの利点を確認。

5. 議論はある?

- 共有representationがaction関連情報を捉えるかをmotion probeで検証。 - frozen visual representationにrecoverable motion informationが含まれることを示す。 - action conditioning、geometric interaction、multi-function trainingの利点をcontrolled studiesで議論。 - その他の限界や議論は要旨からは不明。

6. 次に読むべき論文は?

- 要旨で参照/比較されている研究は明示されていない。 - 関連手法としてworld model、video-action generation、policy inference、inverse dynamics、EWMBench、WorldOlympiad、WorldArenaが挙げられる。 - 同分野の定番としてDreamer、World Models、VideoGPT、Diffuserなどが考えられるが、要旨に記載はない。

※ AIが要旨から生成した要約です。正確性は原文をご確認ください。

著者: Thinker Team, Peng Gao, Jichao Jiao, Chen Liao, Daqin Luo, Yiming Ma, Baiyu Pan, Hailin Shi, Jiyuan Wang, Xiangming Xu, Xiongwei Zhao

分類: cs.AI, cs.CV, cs.RO

原文アブストラクト

World models have emerged as a promising paradigm for embodied AI, yet existing systems often remain fragmented in the modalities they model and the functions they support. In particular, visual dynamics, geometric structure, and executable action are commonly coupled through separate interfaces or downstream modules rather than being jointly represented and evolved within a shared dynamics model. In this work, we present Thinker-WM, a unified multimodal world model that jointly models and generates RGB, metric depth, and robot action through a shared intermediate Unified Dynamics Transformer with modality-specific encoders and decoders. The model performs multimodal unification at the representation level: appearance, geometry, and action are transformed into a common dynamic representation and jointly evolved by shared parameters, while their distinct output structures are preserved. The same parameterization supports multiple functions, including joint video-action generation, world prediction, policy inference, action-conditioned generation, video-to-action generation, and inverse dynamics. To examine whether the shared representation captures action-related information, we introduce a lightweight motion probe that shows the frozen visual representation contains recoverable motion information. Thinker-WM is trained on a heterogeneous corpus comprising approximately 20K hours of egocentric video, 10K hours of real-robot data, and 1K hours of simulation data, including approximately 4K hours of real-robot data with aligned depth supervision. Extensive experiments demonstrate strong performance across embodied world-model evaluations. Thinker-WM ranks first on EWMBench, WorldOlympiad and WorldArena, while controlled studies further demonstrate the benefits of action conditioning, geometric interaction, and multi-function training within the shared representation.

関連論文

PR本紙発行元 EmplifAI