Presentation Information
[9a-PB4-19]Development of a Flexible Electronic Skin for Detecting Surface Roughness
〇Donghyun Kang1, Atsushi Nitta1, Haruki Nakamura1, Yijun Liu1, Kuniharu Takei1 (1.Hokkaido Univ.)
Keywords:
sensor,flexible sensor,flexible device
This study presents an electronic skin system that combines a flexible pressure sensor and machine learning to estimate surface roughness for robotic object identification. The sensor incorporates a microstructured polydimethylsiloxane (PDMS) layer on a PET film. When scanning an object, variations in surface roughness are detected as changes in electrical resistance. The resulting time-series data is analyzed using Reservoir Computing (RC). Following hyperparameter optimization, the RC algorithm successfully predicted the spacing and height of surface roughness with high accuracy, achieving Normalized Root Mean Square Errors (NRMSE) of under 13% and 7%, respectively. These results confirm the device's ability to detect tactile information similar to a human fingertip. Future work will explore measuring multiple physical quantities, including applied pressure, to further advance electronic skin applications in the robotics field.