· Research & technology· 日本語で読む
Fei-Fei Li’s Team Enables Precise Robot Manipulation With Tactile Feedback
Fei-Fei Li’s team used robots equipped with tactile feedback to successfully perform precise tasks such as turning book pages and squeezing toothpaste.
What happened
A study involving computer vision researcher Fei-Fei Li found that a robot equipped with high-frequency tactile feedback completed 12 tasks—including turning book pages, moving an egg, squeezing toothpaste and feeling a mahjong tile—with an average success rate of 65%, BigGo Finance reported. The study was published in June and uses a dual-rate architecture that processes vision and language at 30Hz and tactile data at 300Hz.
Why it matters
For developers who want robots to handle fine manual tasks at home or in the workplace, a key question is how well robots can manage force and slipping after touching an object. The study reported an average success rate of 65% across 12 tasks, including moving an egg, pulling out a thin card and screwing in a light bulb, where changes in contact determine success. The research team showed that tactile sensing helps robots understand an object’s state after contact, rather than recognize its position.
Sources
Translated from the Japanese original: /news/fei-fei-li-shi-robotto-ni-maajan-pai-ya-hamigakiko-no-shokka-8afbeb. The Japanese version is authoritative if they differ.