FlowTouch: View-Invariant Visuo-Tactile Prediction
FlowTouch: View-Invariant Visuo-Tactile Prediction
This paper introduces FlowTouch, a novel model for view-invariant visuo-tactile prediction that leverages an object's local 3D mesh to encode rich information while abstracting away scene-dependent details. By integrating scene reconstruction and Flow Matching-based models for image generation, FlowTouch bridges the sim-to-real gap and generalizes to new sensor instances. The resulting tactile images can be used for downstream grasp stability prediction. Code, datasets, and videos are available at https://flowtouch.github.io/.
The authors present FlowTouch, a model that uses local 3D mesh encoding and flow matching to predict view-invariant tactile patterns from visual input, demonstrating generalization across sensor instances and downstream grasp stability prediction.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- code: Available Artifact link Open source S5
- dataset: Available Artifact link Open source S5
- weights: not_reported
- paper: not_reported
- code: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
further show that the resulting tactile images can be used for downstream grasp stability prediction. Our code, datasets and videos are available at https://flowtouch.github.io/
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org