An Underwater Near-Infrared Spectral Continuum Robot As a Tool for in Situ Detection and Classification

Yunfei Li,Fuzhou Shen,Ziyue Lang,Maojian Cai,Qingyi Zeng,Qian Liu,Fuhong Cai
DOI: https://doi.org/10.1016/j.measurement.2023.112913
IF: 5.6
2023-01-01
Measurement
Abstract:Near-infrared spectral data can be used as a critical indicator for the quantitative analysis of sample molecules. However, water can strongly absorb infrared radiation, and the infrared spectrum of underwater targets cannot be directly obtained. In this study, we propose an underwater in situ near-infrared spectroscopy (875 nm-2350 nm) detection system that utilizes the structure of an optical fiber probe housed in a manipulator to perform approaching observation for underwater samples and uses compressed air to blow away the water between the probe and the target. To demonstrate the system validity, we built a set of optical fiber spectrometers embedded in the robotic arm of the traction lines tendon that can provide an air-blowing function to detect the reflection spectra of 15 minerals underwater and constructed the corresponding spectral database. Based on the above spectral database, an automated spectral identification algorithm was used to classify substances, and the SVM classification accuracy could reach 97 %. The experimental results confirm that our system is stable and can be used for infrared spectral detection of underwater mineral reflection. The relevant spectral data can be used to realize high-precision automatic mineral identification. Due to the compact structure of the system, the system will be able to carry out marine mineral detection in the future.
What problem does this paper attempt to address?