From World Action Models to Embodied Brains: A Roadmap for Open-World Physical Intelligence

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Yuanzhi Liang, Xufeng Zhan, Haibin Huang, Chi Zhang, Xuelong Li
• Venue: arXiv: cs.RO
• Published: 2026-07-13
• Collected: 2026-07-14
• URL: http://arxiv.org/abs/2607.11689v1
• PDF: https://arxiv.org/pdf/2607.11689v1
• DOI: UNKNOWN

Plain-Language Brief

The paper discusses the development of artificial intelligence that can act in the physical world. It highlights the importance of World Action Models (WAMs) and identifies three gaps in current research: model roles, objectives, and system composition. The proposed solution is an 'embodied brain' model that integrates various contexts, compares interventions, and issues state-transition or capability requests. WAMs and physical harnesses are proposed for its predictive functions and grounding outputs, respectively.

What the Document Says

• Artificial general intelligence requires agents to reason and act in the physical world
• Action models, vision-language-action policies, and world models have contributed to this goal
• World Action Models (WAMs) are promising because they connect interventions with consequences
• Progress is fragmented due to incompatible action spaces, prediction targets, different conventions for datasets and tasks, and limited interfaces for reuse and evaluation
• The paper identifies three coupled gaps: model roles and representations, objectives and standardization, and system composition
• A proposed solution is the 'embodied brain', a long-term model target for integrating multimodal context, comparing candidate interventions, and issuing state-transition or capability requests
• WAMs are proposed for its predictive functions
• A physical harness grounds model outputs through tools, controllers, verification, and trace logging
• Shared contracts align heterogeneous models, data, tasks, and embodiments
• Closed-loop post-training converts verified interaction into reusable experience

Key Concepts Mentioned

• Artificial general intelligence
• World Action Models (WAMs)
• embodied brain
• action spaces
• prediction targets
• datasets
• tasks
• runtime systems

Claims Made by the Document

• Current progress in artificial general intelligence is fragmented
• The 'embodied brain' model can integrate multimodal context, compare candidate interventions, and issue state-transition or capability requests

Evidence Used

• Fragmentation identified due to inconsistencies in action spaces, prediction targets, datasets, tasks, and runtime systems
• Proposed 'embodied brain' model described as a solution to the identified gaps

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Artificial general intelligence ultimately requires agents that can reason and act in the physical world. Action models, vision-language-action policies, and world models have advanced this goal, while World Action Models (WAMs) are particularly promising because they connect candidate interventions with predicted consequences. However, progress remains fragmented: models use incompatible action spaces and prediction targets, datasets and tasks follow different conventions, and runtime systems expose limited interfaces for reuse and evaluation. We review the evolution toward WAMs and organize these limitations into three coupled gaps: model roles and representations, objectives and standardization, and system composition. Building on this analysis, we propose a co-evolution roadmap for physical intelligence centered on the \emph{embodied brain}, a long-term model target for integrating multimodal context, comparing candidate interventions, and issuing state-transition or capability requests rather than direct actuator commands. WAMs provide promising prototypes for its predictive functions, while a physical harness grounds model outputs through tools, controllers, verification, and trace logging. Shared contracts align heterogeneous models, data, tasks, and embodiments, and closed-loop post-training converts verified interaction into reusable experience. Together, these components define a modular physical-intelligence stack for adaptive and self-improving embodied agents.