MotuBrain: An Advanced World Action Model for Robot Control

arXiv 2026 60.2 method

TLDR

MotuBrain is a unified World Action Model that jointly models video and action for robot control, achieving high success rates and real-world deployability.

Reasoning

The paper presents a novel architecture combining video and action modeling with strong empirical results on benchmarks and real-world deployment. However, the abstract lacks detailed comparisons to prior work and does not fully disclose limitations or failure cases.

Read-first score

Read-first score 60.2, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 51.

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 18%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=baseline,benchmark,experiment,metric,result

Topical relevance 29%

72.9

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Citation impact 18%

66.5

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.66473274

Reproducibility 18%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

FrontierBridgeMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 375.

Keyword Scores

world model

generative world model

video world model

world dynamics prediction

interactive world model

world simulator

model-based reinforcement learning world model

Deep Analysis

Innovations

UniDiffuser formulation with a three-stream Mixture-of-Transformers architecture for joint video-action modeling
Unified multiview modeling
Independent text stream for stronger language-action coupling
Shared cross-embodiment action representation
Efficient post-training and deployment recipe for long-horizon real-world control
Inference stack combining step reduction, compilation, FP8 quantization, DiT caching, V2A-style action-only inference, and real-time chunked closed-loop execution achieving over 50x speedup and up to 11 Hz inference

Methodology

MotuBrain is a unified World Action Model built on a UniDiffuser formulation with a three-stream Mixture-of-Transformers architecture. It jointly models video and action, supporting policy learning, world modeling, video generation, inverse dynamics, and joint video-action prediction. The model scales to heterogeneous multimodal data including video-only, task-agnostic, and cross-embodiment robot data, and is evaluated on RoboTwin 2.0 and WorldArena benchmarks using success rate and EWMScore metrics.

Key Results

MotuBrain achieves 95.8% and 96.1% average success on RoboTwin 2.0 under clean and randomized settings, respectively, attains the strongest reported EWMScore in WorldArena, and adapts to new humanoid embodiments with only 50–100 trajectories, while achieving over 50x speedup over a naive baseline and up to 11 Hz inference.

Resources

Paper