arXiv 2605.278862026-05-27cs.RO

Tabero: 通过视觉、触觉和语言闭环力反馈学习轻柔操作

Tabero: Learning Gentle Manipulation with Closed-Loop Force Feedback from Vision, Touch, and Language

Hong Kong University of Science and Technology (Guangzhou)（香港科学与技术大学（广州））；Nvidia, Beijing, China（英伟达（北京，中国））

Qiwei Wu, Rui Zhang, Xin Xiang, Tao Li, Weihua Zhang, Junjie Lai, Renjing Xu

备注：: Code:https://github.com/NathanWu7/Tabero

AI总结：

针对现有视觉-语言-动作模型缺乏触觉反馈导致无法实现轻柔操作的问题，提出Tabero基准和模型套件，通过数据高效管道生成视觉-触觉-语言任务，并采用解耦力-位置命令接口的Tabero-VTLA架构，在保持高任务成功率的同时将轻柔指令下的平均夹持力降低70%以上。

摘要：

触觉感知对于机器人实现类人轻柔操作至关重要。然而，现有的视觉-语言-动作（VLA）模型由于缺乏对齐的视觉-触觉-语言数据以及有效的闭环力反馈机制，难以利用触觉反馈进行轻柔操作。为解决这些挑战，我们引入了Tabero，一个用于轻柔、语言条件化机器人操作的基准和模型套件，该操作要求细粒度的接触力感知。首先，Tabero基准通过提出一种数据高效的管道来解决触觉数据稀缺问题，该管道重新利用开源机器人操作轨迹生成多样化的视觉-触觉-语言任务，并建立了一个多维评估协议，同时衡量任务成功率和物理交互质量。其次，我们提出了Tabero-VTLA，一种具有解耦力-位置命令接口的架构；生成的力-位置命令由固定的混合控制器执行，以实现实时的力感知操作。在Tabero上评估，我们的模型在保持高任务成功率的同时，在轻柔指令下将平均夹持力降低了70%以上，展示了其基于多模态经验调节交互力的能力。我们的代码公开在 https://github.com/NathanWu7/Tabero。

英文摘要：

Tactile sensing is essential for robots to achieve human-like gentle manipulation. However, existing Vision-Language-Action (VLA) models struggle to exploit tactile feedback for gentle manipulation due to scarce aligned vision-tactile-language data and the lack of effective closed-loop force feedback mechanisms. To address these challenges, we introduce Tabero, a benchmark and model suite for gentle, language-conditioned robotic manipulation that demands fine-grained contact force perception. First, the Tabero benchmark addresses the scarcity of tactile data by presenting a data-efficient pipeline that repurposes open-source robot manipulation trajectories to generate diverse vision-tactile-language tasks, and establishes a multidimensional evaluation protocol that measures task success alongside physical interaction quality. Second, we propose Tabero-VTLA, an architecture with a decoupled force-position command interface; the resulting force-position commands are executed by a fixed hybrid controller to enable real-time, force-aware manipulation. Evaluated on Tabero, our model maintains high task success while reducing average grip force by over 70\% under gentle instructions, demonstrating its ability to modulate interaction forces based on multimodal experience. Our code is publicly available at https://github.com/NathanWu7/Tabero.

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。