Magnetic Tactile Sensing

Mixes magnetic particles into a soft gel and uses a magnetometer to sense contact from the magnetic-field change deformation causes.

Magnetic tactile sensing is one approach to building tactile sensors: magnetized particles or small magnets are embedded in a soft elastomer, with a Hall-effect sensor or magnetometer placed underneath. When the elastomer touches an object it deforms, the embedded magnetic source moves with it, and the magnetic-field reading changes accordingly; calibration or machine learning then converts this reading into contact location, pressure, and shear force (force along the surface). The advantage is that the soft gel layer is separate from the circuit board, so a worn gel pad can simply be swapped out, keeping cost low and shape easy to customize; the drawback is susceptibility to external magnetic fields and nearby motors, with spatial resolution usually lower than a vision-based tactile sensor’s. Landmark examples include ReSkin (CoRL 2021), a collaboration between Carnegie Mellon University and Meta, and Lerrel Pinto’s group’s AnySkin (2024), the latter built around the idea that swapping in a fresh skin lets an already-trained manipulation policy keep working without recalibration.

ExampleThe AnySkin paper mounts magnetic skin on gripper fingertips for slip detection and policy learning, and shows that after swapping in a different, fresh skin the policy still works directly.

Also called

Hall-Effect Tactile Sensing, Magnetic Tactile Skin

Related

Tactile Sensor · ReSkin / AnySkin · uSkin · Vision-Based Tactile Sensor · Electronic Skin · Slip Detection

Sources

ReSkin: versatile, replaceable, lasting tactile skins (CoRL 2021)
AnySkin: Plug-and-play Skin Sensing for Robotic Touch

As of

2024-09