講演情報
[16p-M_B104-6]Spider Web-Inspired Pressure Sensor Array
〇(M1)Ketong Gao1, Haruki Nakamura1, Atsushi Nitta1, Kuniharu Takei1 (1.Hokkaido Univ.)
キーワード：
Flexible pressure sensor、Softness and force sensing、Machine learning
Flexible and stretchable pressure sensors are essential for tactile sensing on curved and compliant surfaces in applications such as human–machine interfaces, wearable electronics, and soft robotics. However, achieving stable conformal contact with soft or highly curved objects while maintaining high-resolution force and softness sensing remains challenging, often requiring complex electrode configurations. In this work, we propose a spiderweb-inspired pressure sensor architecture integrated with machine learning to enable efficient force, position, and softness sensing using a reduced number of electrodes. The sensor is fabricated by dispenser-printing a conductive composite ink composed of carbon black, PDMS, and Ecoflex into a spiderweb-like mesh structure, followed by thermal curing. Eight printed electrodes, including one ground and seven signal channels, provide multiple sensing pathways while minimizing system complexity. Under applied forces, mesh deformation induces distinct resistance changes across the output channels, generating rich multi-channel signals. Machine learning algorithms are employed to analyze these signals, achieving accurate force magnitude and spatial position estimation with an accuracy of 92.52% and a low mean square error of 0.0676. Furthermore, the sensor reliably discriminates material softness, reaching a classification accuracy of 0.9468. These results demonstrate that the proposed spiderweb-inspired sensor, combined with machine learning, offers a promising approach for high-performance tactile sensing on soft and curved surfaces, with strong potential for advanced soft robotic applications.