Paper page - ABot-M0.5: Unified Mobility-and-Manipulation World Action Model

Source: https://huggingface.co/papers/2607.00678 Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

ABot-M0.5 is a World Action Model for mobile manipulation that improves performance through temporal granularity alignment, action space disentanglement, and train-test consistency in autoregressive prediction.

Mobile manipulationis a key capability for general-purpose robots, yet remains challenging for current embodied learning methods. VLA policies are typically reactive and lack explicit world modeling, while existingWorld Action Models(WAMs) are still poorly aligned with the structure ofmobile manipulation: they operate on coarse video chunks, model entangled navigation-manipulation actions, and traininverse dynamicsunder supervision that does not match autoregressive inference. As a result, they often miss fine-grained contact dynamics, suffer from action-distribution conflicts, and accumulate errors over long-horizon rollouts. We propose ABot-M0.5, a new WAM built on the insight thatmobile manipulationrequires alignment at three levels:temporal granularity,action space, and train-test consistency. To aligntemporal granularity, we introduce intermediate latent actions that capture local visual state transitions and serve as an bridgingaction spacebetween video latents and embodiment-specific controls. To alignaction space, we design a dual-levelMixture-of-Transformersarchitecture that disentangles both modality representations and heterogeneous action subspaces such as base movement and arm manipulation. To align inference conditions, we propose thedream-forcingtraining strategy that progressively trainsinverse dynamicson model-predicted videos, improving train-test alignment and robustness duringautoregressive prediction. Experiments on challenging mobile and fine-grained manipulation benchmarks demonstrate that ABot-M0.5 achieves state-of-the-art performance in both long-horizon task success and finegrained control accuracy. These results highlight the critical importance of granularity-aligned, action-disentangled, and inference-consistent world-action modeling.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2607\.00678

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2607.00678 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2607.00678 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.00678 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.