01
SlipSense: from pressure and vibration to a timely regrasp
Pressure and vibration improve slip recognition together. Independent motion labels and five failed recoveries explain why detector quality and controller quality need separate tests.
Research / RSSView raw RSS (XML)
Follow research updates
Get source-backed robot skin, tactile AI, and electronic skin news in your RSS reader. Each update links to the full article on RoboSkin.ai.
Copy this address into your RSS reader.
Read updates your way
RSS brings new articles into a reading app, so you can follow updates without checking the website each time.
- Open an RSS reader, such as Feedly or Inoreader.
- Choose the option to add a feed or follow a website, then paste the feed address.
- Confirm the subscription in your reader. New articles will appear as the reader checks for updates.
Latest feed updates
Browse researchRead article
02
Touch2Trace: what makes a tactile cable-tracing policy work
A 60 Hz tactile policy traces farther than a joint-only baseline, but the useful lesson is how sensor detail, pretraining, temporal context and action timing interact.
03
Visible Touch: how contact overlays reach a visual robot policy
Rendering contact as image arrows reuses a visual policy’s input pathway. The hardware and ablations show why spatial alignment, signal normalization and training still matter.
04
UniVTAC separates tactile simulation, representation learning, and policy evaluation
Choose the matching code branch and dataset before using UniVTAC. This guide separates its tactile encoder, simulation benchmark, public downloads and paper-reported physical results.
05
DexTouch-WM learns robot contact dynamics from human touch
Human tactile pretraining improves the paper’s contact predictions, but synthetic training data does not consistently improve real-robot policy scores. Here is what each experiment measures.
06
Bench2Dex benchmarks bimanual robots with simulated touch
Bench2Dex connects teleoperation, replay, tactile maps, and evaluation across 12 hand embodiments. We examine the data contract and distinguish simulation access from physical sensor validation.