Meta is advancing the science of embodied AI by allowing robots to touch and feel.
The tech giant clarified in a blog post that the new devices are not intended for consumers.
Rather, they are intended for scientists. Meta believes they will be used to further research on AI that can “learn about the world in richer detail” and “better understand and model the physical world.”
Meta’s cooperation with GelSight will provide Digit 360 to the global research community in the following year.
The device’s applications go beyond robotics; it could be useful in disciplines such as medicine, prosthetics, virtual reality, and telepresence.
For example, in virtual worlds, Digit 360 may let users engage with objects not only visually but also through a digitised sense of touch, thereby rooting their interactions in a genuine depiction of object attributes.
Digit 360, the successor to Meta’s Digit sensor, digitises touch signals and detects changes in its surroundings using an on-device AI chip and around 18 “sensing features”.
“We developed a touch-perception-specific optical system with a wide field of view … for capturing omnidirectional deformations on the fingertip surface,” Meta explained in a blog post.
“Additionally, we equipped the sensor with many sensing modalities, since each touch interaction with the environment has a unique profile produced by the mechanical, geometrical, and chemical properties of a surface to perceive vibrations, sense heat, and even smell odour.”
Wonik Robotics, a South Korean robotics company, collaborated with Meta to produce the next iteration of the Allegro Hand, which is equipped with touch sensors based on Meta’s Digit Plexus.
The Allegro Hand, which is also slated to be released next year, will give an enhanced tool for researchers seeking to make robots more sensitive and adept at delicate jobs ranging from intricate component assembly to potentially lifesaving healthcare operations.
We earlier reported that Meta has released an “open” implementation of the viral generate-a-podcast tool in Google NotebookLM.
