Low-Power System for Spatial Tactile Detection Using a Self-Sensing Nanocomposite With On-Device Adaptive Regularization

Carbon nanostructured materials offer unique properties for spatial tactile sensors, but their inherent heterogeneous nature also presents challenges in signal processing to reduce hysteresis, enhance pressure-sensitive performance, and long-term stability. To address these challenges, a robust regularization technique using electrical mapping for implementation and evaluation within an embedded system is proposed. A low-power wireless sensing node is proposed to detect and classify the electrical response of a two-dimensional multiwall carbon nanotube/polypropylene (MWCNT/PP) tactile sensor using robust adaptive regularization and weighted order statistics methods. The location-dependent piezoresistive response of a MWCNT/PP square membrane is obtained from boundary electrical measurements and electrical mapping. The proposed system addresses the ill-posed inverse problem by implementing spatial tactile detection in a low-power microcontroller, thereby enabling the development of new Internet-of-Things (IoT) self-sensing integrated systems. Experimental results show a high accuracy of 93.94% for the spatial tactile detection, achieved with a low power consumption of 0.79 mW. The robust regularization coupled with the weighted order statistical classification and their implementation in a low-power microcontroller offers a promising approach to accurately detect spatial tactile responses.