Gel Biter: food texture discriminator based on physical reservoir computing with multiple soft materials

The human oral structure contains organs with distinctly different physical properties, such as teeth, gums, and tongues. When food enters the oral cavity, we can recognize the tactile sensation and shape of the object from multiple perspectives through the texture of the teeth and tongue. Therefore...

Full description

Saved in:
Bibliographic Details
Published in:Artificial life and robotics 2022-11, Vol.27 (4), p.674-683
Main Authors: Hirose, Kosuke, Sudo, Ikuma, Ogawa, Jun, Watanabe, Yosuke, Shiblee, M. D. Nahin Islam, Khosla, Ajit, Kawakami, Masaru, Furukawa, Hidemitsu
Format: Article
Language:English
Subjects:
Artificial Intelligence
Computation by Abstract Devices
Computer Science
Control
End effectors
Food
Machine learning
Mechatronics
Object recognition
Organs
Original Article
Physical properties
Piezoelectric films
Robot arms
Robotics
Tactile sensors (robotics)
Texture
Touch
Waveforms
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The human oral structure contains organs with distinctly different physical properties, such as teeth, gums, and tongues. When food enters the oral cavity, we can recognize the tactile sensation and shape of the object from multiple perspectives through the texture of the teeth and tongue. Therefore, it is possible to regard oral structures as a group of tactile sensors based on these functions. In this study, we developed a soft-matter artificial mouth that can accurately detect subtle differences in texture by creating and combining oral structural organs using polymer materials with different physical properties and mounting them as end-effectors for a robot arm. The same piezoelectric film sensor was embedded inside each organ, making it possible to acquire tactile sensations from the same object as completely different signal waveforms. We tested whether the sensor data obtained from each soft-matter material could be used for excellent object recognition by applying various machine learning methods. In an actual experiment, we learned the waveform data obtained from chewing sweets and snacks, such as rice crackers, and applied machine learning to classify the data, which led to an accuracy rate of over 90%.
ISSN:1433-5298
1614-7456
DOI:10.1007/s10015-022-00814-2