Deep-learning-assisted triboelectric whisker for near field perception and online state estimation of underwater vehicle

• Published in: Nano energy Vol. 129; p. 110011
• Main Authors: Xu, Peng, Liu, Jianhua, Liu, Bo, Li, Yuanzheng, Jin, Hao, Mu, Zhaoyang, Guan, Tangzhen, Xie, Guangming, Wang, Hao, Xu, Minyi
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.10.2024
• Subjects: Deep learning; Self-powered; Triboelectric nanogenerators; Underwater perception; Whisker sensor; Deep learning; Triboelectric nanogenerators; Self-powered; Underwater perception; Whisker sensor
• DOI: 10.1016/j.nanoen.2024.110011
• ISSN: 2211-2855

Agile near field perception remains a challenge for underwater vehicles, which could significantly enhance their capability of online state estimation. Harbor seals have evolved a whisker array that can accurately measure and identify environmental information in their surroundings. Inspired by the "smart" whiskers of harbor seals, the study has designed a deep learning-assisted bionic underwater triboelectric whisker sensor (UTWS) to passively perceive diverse hydrodynamic flow fields, including omni-directional steady flow and wakes induced by bluff body upstream. The device mainly comprised an elliptical whisker shaft with a high-aspect-ratio of 0.403, four flexible triboelectric sensing units mimicking the nerve in the follicular sinus complex, and a flexible corrugated joint imitating the surface skin of marine organisms' cheeks. The UTWS demonstrated impressive advantages of rapid response times of 21 ms, high sensitivity of 1.16 V/m.s−1, high signal-to-noise-ratio of 61.66 dB. By implementing deep learning analytics to process the multi-channel signals, the underwater vehicle equipped with the UTWS can accomplish online velocity estimation proficiently, with an approximate root mean square error of approximately 0.093 in the verification case. Thus, this UTWS-based, deep-learning-assisted perception could become a promising tool for integration with underwater vehicles in the local navigation tasks. [Display omitted] •A deep learning-assisted bionic underwater triboelectric whisker was designed to passively perceive diverse flow fields.•The double-layer air chamber shielding technique has been employed to minimize signal interference from ions in water.•The UTWS demonstrated impressive advantages of response times of 21 ms, sensitivity of 1.16 V/m.s−1, SNR of 61.66 dB.•The underwater vehicle equipped with the UTWS can accomplish online velocity estimation proficiently with a RMSE of 0.093.