Zhang Tong, Zhao Minghui, Zhai Mingxuan, Wang Lisha, Ma Xingyu, Liao Shengmei, Wang Xiaona, Liu Yijian, Chen Da
College of Electronic and Information Engineering, Shandong University of Science and Technology, 266590 Qingdao, China.
Laser Institute, Qilu University of Technology (Shandong Academy of Sciences), Qingdao, Shandong 266000, China.
ACS Appl Mater Interfaces. 2024 Feb 28;16(8):11013-11025. doi: 10.1021/acsami.3c17880. Epub 2024 Feb 14.
Industrial robots are the main piece of equipment of intelligent manufacturing, and array-type tactile sensors are considered to be the core devices for their active sensing and understanding of the production environment. A great challenge for existing array-type tactile sensors is the wiring of sensing units in a limited area, the contradiction between a small number of sensing units and high resolution, and the deviation of the overall output pattern due to the difference in the performance of each sensing unit itself. Inspired by the human somatosensory processing hierarchy, we combine tactile sensors with artificial intelligence algorithms to simplify the sensor architecture while achieving tactile resolution capabilities far greater than the number of signal channels. The prepared 8-electrode carbon-based conductive network achieves high-precision identification of 32 regions with 97% classification accuracy assisted by a quadratic discriminant analysis algorithm. Notably, the output of the sensor remains unchanged after 13,000 cycles at 60 kPa, indicating its excellent durability performance. Moreover, the large-area skin-like continuous conductive network is simple to fabricate, cost-effective, and can be easily scaled up/down depending on the application. This work may address the increasing need for simple fabrication, rapid integration, and adaptable geometry tactile sensors for use in industrial robots.
工业机器人是智能制造的主要设备，而阵列式触觉传感器被认为是其主动感知和理解生产环境的核心器件。现有阵列式触觉传感器面临的一个巨大挑战是在有限区域内传感单元的布线、少量传感单元与高分辨率之间的矛盾，以及由于每个传感单元自身性能差异导致的整体输出模式偏差。受人类体感处理层次结构的启发，我们将触觉传感器与人工智能算法相结合，在简化传感器架构的同时，实现了远超信号通道数量的触觉分辨率能力。制备的8电极碳基导电网络在二次判别分析算法的辅助下，实现了对32个区域的高精度识别，分类准确率达97%。值得注意的是，该传感器在60 kPa压力下经过13000次循环后输出保持不变，表明其具有出色的耐久性性能。此外，大面积皮肤状连续导电网络易于制造、成本效益高，并且可以根据应用轻松进行放大/缩小。这项工作可能满足工业机器人对简单制造、快速集成和几何形状可适配的触觉传感器日益增长的需求。
