跳到正文
arXiv cs.AI· Vaidehi Bagaria, Bijo Sebastian, Nirav Kumar Patel·· 4 小时前AI 评分43

递归信念视觉-语言-动作模型 RB-VLA

Recursive Belief Vision Language Action Models

AI 导读

论文提出 RB-VLA,一种以信念为中心的视觉-语言-动作架构,通过自监督世界模型目标维护紧凑的潜在状态,VLM 每个任务仅查询一次。相比 pi_0,RB-VLA 在多阶段抓放和堆叠任务上分别取得 52.5% 和 37.5% 的更高成功率,推理延迟最多降低五倍,并消除内存随时间步增长。消融实验显示信念模块是性能主要来源,成功率从无信念的 32.5% 提升至 77.5%。

正文 · 原文

View PDF HTML (experimental)

Abstract: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.
Comments: Accepted at the 2026 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2602.20659 [cs.AI]
  (or arXiv:2602.20659v3 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2602.20659

arXiv-issued DOI via DataCite

Submission history

From: Vaidehi Bagaria [view email]
[v1] Tue, 24 Feb 2026 08:02:16 UTC (14,162 KB)
[v2] Wed, 25 Feb 2026 17:38:24 UTC (14,166 KB)
[v3] Fri, 13 Nov 2026 10:29:15 UTC (14,166 KB)

来源:arXiv cs.AI · arxiv.org