Embodied AI Glossary中文

GelSlim

Advanced

A slim, finger-shaped vision-based tactile sensor from MIT that can be mounted on a gripper for grasping.

GelSlim is a series of finger-shaped vision-based tactile sensors from Alberto Rodriguez’s group (the MCube Lab) at MIT, developed with Edward Adelson and others, part of the GelSight family. The original GelSight has a long optical path and a bulky body, making it hard to fit into cluttered environments; the first GelSlim (Donlon and colleagues, 2018) redesigned the optical path with mirrors and light guides to make the sensor a slim finger, swapped in a more wear-resistant gel and fabric skin, and used calibration to keep imaging stable. GelSlim 3.0 (Taylor, Dong, Rodriguez, 2021) shrank it further, measuring contact shape in real time, estimating a 3D distributed contact-force field, and detecting incipient slip; its fingertip module is swappable, and the design and software are open source.

ExampleMounting GelSlim 3.0 on a small parallel gripper for bin picking uses the tactile image to estimate the object’s position in the hand and detect whether it’s starting to slip.

Also called
GelSlim 3.0
Related
GelSight · Vision-Based Tactile Sensor · Slip Detection · Bin Picking · Parallel Jaw Gripper · GelSight Mini
Sources
GelSlim: A High-Resolution, Compact, Robust, and Calibrated Tactile-sensing Finger (arXiv 1803.00628)
GelSlim3.0: High-Resolution Measurement of Shape, Force and Slip in a Compact Tactile-Sensing Finger (arXiv 2103.12269)
As of
2021-03

See it in the full glossary →