Lu Xiao, Wang Tianhong, Zhong Songyi, Cao Tianqi, Zhou Chenghao, Li Long, Zhang Quan, Tian Shiwei, Jin Tao, Yue Tao, Xie Shaorong
School of Computer Engineering and Science, Shanghai University, Shanghai 200444, China.
Shanghai Key Laboratory of Intelligent Manufacturing and Robotics, School of Mechatronics Engineering and Automation, Shanghai University, Shanghai 200444, China.
Cyborg Bionic Syst. 2025 Jul 2;6:0320. doi: 10.34133/cbsystems.0320. eCollection 2025.
Object property perception, as a core component of tactile sensing technology, faces severe challenges due to its inherent complexity and diversity, particularly under the constraints of decoupling difficulty and limited precision. Herein, this paper introduces an innovative approach to object property perception utilizing triboelectric-magnetoelastic sensing. This technology integrates triboelectricity and magnetoelasticity, achieving a self-powered sensing mechanism that requires no external power source for sensing signal generation. Moreover, by deploying a triboelectric array, it comprehensively captures multi-dimensional information of objects. Concurrently, in conjunction with magnetoelastic sensing technology, it provides stable and reliable mechanical information, ensuring that the system can accurately decouple key characteristics of objects, such as material properties, softness, and roughness, even in open environments where temperature, humidity, and mechanical conditions change in real time. Furthermore, by combining deep learning algorithms, it achieves exceptionally high recognition accuracy for object properties (material recognition accuracy: 99%, softness recognition accuracy: 100%, roughness recognition accuracy: 95%). Even in complex scenarios with intertwined multiple properties, the overall recognition accuracy remains consistently above 95%. The multimodal tactile sensing technology proposed in this paper provides robust technical support and theoretical foundation for the intelligent development of robots and the enhancement of real-time tactile perception capabilities.
物体特性感知作为触觉传感技术的核心组成部分，因其固有的复杂性和多样性而面临严峻挑战，尤其是在解耦困难和精度有限的限制下。在此，本文介绍了一种利用摩擦电-磁弹性传感的创新物体特性感知方法。该技术将摩擦电与磁弹性相结合，实现了一种无需外部电源来产生传感信号的自供电传感机制。此外，通过部署摩擦电阵列，它全面捕获物体的多维信息。同时，结合磁弹性传感技术，它提供稳定可靠的力学信息，确保系统即使在温度、湿度和机械条件实时变化的开放环境中，也能准确解耦物体的关键特性，如材料特性、柔软度和粗糙度。此外，通过结合深度学习算法，它对物体特性实现了极高的识别准确率（材料识别准确率：99%，柔软度识别准确率：100%，粗糙度识别准确率：95%）。即使在多种特性交织的复杂场景中，整体识别准确率也始终保持在95%以上。本文提出的多模态触觉传感技术为机器人的智能发展和实时触觉感知能力的提升提供了有力的技术支持和理论基础。
