VISTA: 通过视觉跟踪偏好优化增强视觉条件化在视觉-语言-动作模型中
VISTA: Enhancing Visual Conditioning via Track-Following Preference Optimization in Vision-Language-Action Models
AI总结：
VISTA通过视觉跟踪偏好优化提升视觉条件化，增强VLA模型在视觉-动作对齐和任务性能上的表现。
摘要：
视觉-语言-动作（VLA）模型在广泛的机器人操作任务中表现出强大的性能。尽管取得了成功，但将大型预训练视觉-语言模型（VLMs）扩展到动作空间可能会导致视觉-动作不一致，其中动作预测表现出对当前视觉状态的弱依赖性，导致不可靠的动作输出。在本文中，我们通过视觉条件化的角度研究VLA模型，并通过实验证明，成功的 rollout 一致表现出比失败的更强的视觉依赖性。受这一观察的启发，我们提出了一种训练框架，明确加强VLA模型中的视觉条件化。我们的方法首先通过偏好优化在跟踪跟随的替代任务上对齐动作预测与视觉输入，然后通过在监督微调期间通过潜在空间蒸馏将增强的对齐性转移到遵循指示的任务中。在不引入架构修改或额外数据收集的情况下，我们的方法提高了离散OpenVLA的视觉条件化和任务性能，并进一步在扩展到连续OpenVLA-OFT设置时产生一致的收益。项目网站：https://vista-vla.github.io/。
英文摘要：
Vision-Language-Action (VLA) models have demonstrated strong performance across a wide range of robotic manipulation tasks. Despite the success, extending large pretrained Vision-Language Models (VLMs) to the action space can induce vision-action misalignment, where action predictions exhibit weak dependence on the current visual state, leading to unreliable action outputs. In this work, we study VLA models through the lens of visual conditioning and empirically show that successful rollouts consistently exhibit stronger visual dependence than failed ones. Motivated by this observation, we propose a training framework that explicitly strengthens visual conditioning in VLA models. Our approach first aligns action prediction with visual input via preference optimization on a track-following surrogate task, and then transfers the enhanced alignment to instruction-following task through latent-space distillation during supervised finetuning. Without introducing architectural modifications or additional data collection, our method improves both visual conditioning and task performance for discrete OpenVLA, and further yields consistent gains when extended to the continuous OpenVLA-OFT setting. Project website: https://vista-vla.github.io/ .
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
