Summary
DexSkin, presented as an Oral paper at CoRL 2025, is a soft, wearable capacitive electronic skin that covers nearly the entire surface of a robotic gripper finger, rather than just a small fingertip patch like conventional tactile sensors. This high-coverage sensing lets robots detect not only fingertip contact but also where an object touches the sides of the fingers, which is critical for in-hand manipulation tasks such as rotating objects. A second key contribution is cross-sensor-instance calibration: because manufacturing tolerances cause readings to differ between individual DexSkin units, the paper proposes a calibration scheme that enables models trained on one sensor to transfer to another. This allows data-driven approaches such as imitation learning and reinforcement learning to be trained on one skin and deployed on another, validated through online RL experiments. The authors argue that the biggest bottleneck in robotic manipulation is not the 'brain' but the 'skin' — robots need human-like, body-wide tactile perception.
Human skin provides tactile sensing over the entire body, whereas robotic tactile sensors typically cover only a small fingertip area. DexSkin — an Oral paper at CoRL 2025 — aims to change this: a soft, wearable capacitive electronic skin that can cover almost the entire surface of gripper fingers.
This means a robot can sense not only fingertip contact, but also exactly where an object touches the sides of its fingers — crucial for tasks like in-hand object rotation. Conventional fingertip tactile sensing is like feeling the world blindfolded with only your fingertips: you can touch things, but you don't know where they are on your palm.
Another key contribution is cross-sensor-instance calibration — different DexSkin instances produce different readings due to manufacturing tolerances. The paper proposes a calibration scheme that lets models transfer across sensors. This means data-driven methods (imitation learning, RL) can be trained on one sensor and deployed on another. Online reinforcement learning experiments also validate feasibility.
*Core insight*: The biggest bottleneck in robotic manipulation is not the 'brain' but the 'skin' — robots need human-like, body-wide tactile perception.
Reference: [DexSkin: High-Coverage Conformable Robotic Skin / CoRL 2025 Oral]
This page is an English static mirror generated for search and AI citation.
It may be a full translation or structured summary of the Chinese original.
Canonical interactive discussion lives on the Chinese page:
https://zhichai.net/topic/177619968