外观
论文arXiv 2602.081673 个月前
动作预测型具身推理的自监督自举
R&B-EnCoRe将推理作为潜变量自监督精炼，在多类具身任务和VLA架构上验证动作预测推理。
Franka FR3原文链接：Self-Supervised Bootstrapping of Action-Predictive Embodied Reasoning（arXiv 2602.08167 · 发布日期：2026-07-13）
Self-Supervised Bootstrapping of Action-Predictive Embodied Reasoning
R&B-EnCoRe将推理作为潜变量自监督精炼，在多类具身任务和VLA架构上验证动作预测推理。
核心要点
- 现有Embodied CoT依赖刚性模板指定推理基元，可能引入干扰动作预测的无关信息。
- R&B-EnCoRe用重要性加权变分推断，将推理作为潜变量进行自监督精炼。
- 该方法无需外部奖励、验证器或人工标注，生成并蒸馏具身特定推理训练数据。
- 在操作、导航和自动驾驶中，相比全基元推理取得成功率、得分和碰撞率改进。
验证范围与局限：验证覆盖Franka Panda仿真、WidowX真机操作，以及多类导航和自动驾驶具身。
英文摘要 (Abstract)
Embodied Chain-of-Thought (CoT) reasoning has significantly enhanced Vision-Language-Action (VLA) models, yet current methods rely on rigid templates to specify reasoning primitives (e.g., objects in the scene, high-level plans, structural affordances).These templates can force policies to process irrelevant information that distracts from critical action-prediction signals.This creates a bottleneck: without successful policies, we cannot verify reasoning quality; without quality reasoning, we cannot build robust policies.We introduce R&B-EnCoRe, which enables models to bootstrap embodied reasoning from internetscale knowledge through self-supervised refinement.By treating reasoning as a latent variable within importance-weighted variational inference, models can generate and distill a refined reasoning training dataset of embodiment-specific strategies without external rewards, verifiers, or human annotation.We validate R&B-EnCoRe across manipulation (Franka Panda in simulation, WidowX in hardware), legged navigation (bipedal, wheeled, bicycle, quadruped), and autonomous driving embodiments using various VLA architectures with 1B, 4B, 7B, and 30B parameters.Our approach achieves 28% gains in manipulation success, 101% improvement in navigation scores, and 21% reduction in collision-rate metric over models that indiscriminately reason about all available primitives.R&B-EnCoRe enables models to distill reasoning that is predictive of successful control, bypassing manual annotation engineering while grounding internet-scale knowledge in physical execution.
作者：Milan Ganai, Katie Z Luo, Jonas Frey, Clark Barrett, Marco Pavone
来源
本页事实对应的公开页面，以原文为准。