Vision-language-action models must enable agents to execute long-horizon tasks under partial observability. However, most existing approaches remain observation-driven, relying on short context windows or repeated queries to vision-language models (VLMs). This leads to loss of task progress, action repetition under perceptual aliasing, and high inference latency. While semantic grounding is important, long-horizon manipulation fundamentally requires persistent, action-conditioned state representations. Current VLAs lack such representations and exhibit limited temporal and physical reasoning, making them ill-suited for multi-stage control. This paper introduces RB-VLA, a belief-centric architecture trained with self-supervised world-model objectives that maintains a compact latent state encoding task-relevant history, dynamics, and object interactions. Queried once per task, the VLM provides high-level intent, while the belief tracks task progress and enables phase-aware, causally grounded control under partial observability without storing raw observations or scaling memory with time. The belief and intent jointly condition a diffusion policy for robust closed-loop execution. RB-VLA outperforms prior VLAs on long-horizon benchmarks, achieving 52.5 percent and 37.5 percent higher success rates on multi-stage pick-and-place and stacking tasks, respectively, compared to pi_0. It also reduces inference latency by up to five times relative to baselines and eliminates memory growth across timesteps observed in existing VLAs. Ablations show the belief module is the primary driver of performance, increasing success rates from 32.5 percent without belief to 77.5 percent with belief.
Recursive CLIP: Vision-Language Distillation with Looped Transformer Blocks
Samsung AI Center Warsaw(三星人工智能中心华沙分部);AGH University of Krakow(克拉科夫AGH大学);ACC Cyfronet AGH(AGH Cyfronet计算中心);Institute of Fundamental Technological Research, Polish Academy of Sciences(波兰科学院基础技术研究所)
Carnegie Mellon University(卡内基梅隆大学);Hong Kong University of Science and Technology(香港科技大学);Mohamed bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学);Reality Labs, Meta Inc.(Meta公司Reality Labs)
Designing Ecosystems of Intelligence from First Principles
VERSES AI Research Lab(VERSES人工智能研究实验室);University College London(伦敦大学学院);Monash University(莫纳什大学);University of Sussex(萨塞克斯大学);Université du Québec à Montréal(魁北克大学蒙特利尔分校);University of Oxford(牛津大学);Max Planck Institute of Animal Behavior(马克斯·普朗克动物行为研究所);University of Konstanz(康斯坦茨大学);Network Science Institute(网络科学研究所)