Skip to content

Event

FlowTouch: A Viewpoint-Independent Visuotactile Prediction Model

1 reports · 1 sources

Seongjin Bien and co-authors propose FlowTouch, a viewpoint-independent visual-tactile prediction model that encodes information using local 3D meshes of objects to predict tactile patterns, enabling robots to obtain tactile feedback even before contacting objects. The method combines scene reconstruction with an image generation model based on Flow Matching. The authors state that results show FlowTouch can narrow the sim-to-real gap and generalize to new sensor instances; the generated tactile images can also be used for downstream grasp stability prediction. Code, datasets, and videos have been made publicly available.

Report timeline

  1. FlowTouch: A View-Invariant Visuo-Tactile Prediction Model

    FlowTouch is a view-invariant visuo-tactile prediction model that uses an object's local 3D mesh to encode information for predicting tactile patterns, enabling tactile feedback even before the object is touched. The method combines scene reconstruction with a Flow Matching-based image generation model. Results show that it narrows the sim-to-real gap and generalizes to new sensor instances, and the generated tactile images can also be used for downstream grasp stability prediction. Code, dataset, and videos are publicly available.

    arXiv cs.RO ·

Related topics

Perception and localizationGrasping and manipulationRobot models and learningData and simulationResearch papers

Interest trend

Current
3
Peak
4Nov 4 01:00
Past 24 hours
–

The trend compares participants observed throughout the same period, so its range may be smaller than the current interest count. Move or tap to inspect each hour; use the arrow keys on a keyboard.