原文：Visuo-tactile Diffusion Policy for Robotic Manipulation with Auxiliary Tactile Dynamics Prediction
作者：Kaihong Ouyang, Zihao Su, Chengxuan Zheng, Weiyi Lai, Dong Zhou
来源：arXiv cs.RO（机器人）
正文
Computer Science > Robotics
arXiv:2611.02677v1 (cs)
[Submitted on 2 Nov 2026]
Title:Visuo-tactile Diffusion Policy for Robotic Manipulation with Auxiliary Tactile Dynamics Prediction
Authors:Kaihong Ouyang, Zihao Su, Chengxuan Zheng, Weiyi Lai, Dong Zhou
View a PDF of the paper titled Visuo-tactile Diffusion Policy for Robotic Manipulation with Auxiliary Tactile Dynamics Prediction, by Kaihong Ouyang and 4 other authors
View PDF
Abstract:Long-horizon robotic assembly requires reliable perception and action across grasping, alignment, and insertion. Vision captures global object pose and scene context, where as fingertip touch provides local contact cues that are especially useful as contact states change. We present a visuotactile diffusion policy that combines four-step histories from two camera views and two fingertip tactile images to condition action-sequence generation. A recurrent temporal representation integrates recent observations from both modalities. To encourage it to capture how contact evolves under action, an auxiliary objective predicts fingertip-image changes two steps ahead from the temporal representation and demonstrated actions. This objective is used only during training and adds no inference-time computation. Across three independent training seeds in MuJoCo assembly experiments, the full model achieves an average success rate of $59.0% \pm 2.6%$, improving on the visual-only baseline ($42.7% \pm 1.5%$) by 16.3 percentage points.
Comments:
7 pages, 2 figures. Presented at ICCSSE 2026
Subjects:
Robotics (cs.RO)
Cite as:
arXiv:2611.02677 [cs.RO]
(or
arXiv:2611.02677v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2611.02677
Focus to learn more
arXiv-issued DOI via DataCite (pending registration)
主题
机器人
由「前沿雷达」于 2026-11-02 采集。正文取自原文页面，已保留出处链接。标题与正文版权归原作者所有。
评论
加载中…