Much more advanced sensors have been used to greatly promote intelligent production with the development of smart fishery. Harsh requirements can be found for the accuracy and biosafety of sensors in practical applications. Wearable electronic devices can be expected to serve as one of the core components with great application prospects. Among them, various sensors can pose an important impact on the design and function of future wearable electronic devices. However, most traditional sensors cannot fully meet the application requirements, due mainly to their rigid characteristics. For example, the invasive detection behavior of rigid sensors can cause fish damage, fish growth and activity, easily leading to data loss. Particularly, the conventional rigid sensors cannot be suitable for the modern fishery, such as large volume and weight, as well as biological incompatibility. Fortunately, flexible wearable sensors have attracted extensive attention and research interest in the fisheries field in recent years, due to the lightweight, portability, high integration, and excellent electrical performance. The scope of sensor applications has also been broadened in fisheries, due to their excellent stretchability and biocompatibility. In this review, a systematic summary was proposed of the materials, processes, energy supply, and communication systems required for the preparation of flexible sensors. Flexible materials were divided into the substrate, function and packaging, materials, according to their functions. These materials were needed to meet the particularity of the fishery application environment, in terms of biocompatibility, air permeability and light transmittance. A comparison was made between the preparation materials and properties of existing flexible sensors. The core process of sensor preparation was introduced, together with better energy supply and communication systems. The emerging nano-friction power generation technology was proposed for fishery flexible electronic equipment. And then, the application of flexible wearable sensing technology was reviewed on fish motion monitoring, water environment monitoring and aquatic product quality detection. The quality detection of aquatic products was divided into biogenic amine, antibacterial drug, and pH detection. Flexible electronic technology was dominated in production cost, measurement accuracy, biocompatibility and application range. At the same time, some recommendations were also given in the future. Fish motion monitoring was necessary to capture the sensing information under sensor application scenarios and biomarker monitoring. The much more flexible sensors can be expected to detect the ammonia nitrogen and dissolved oxygen in the water environment, in order to use environmentally friendly materials during preparation. The challenges and development trends of flexible wearable devices in fishery applications were as follows: 1) To improve the stability and reliability of wearable device sensing under complex conditions, in order to ensure the sensing function of fishery special application environment. 2) To design a micro-sensor system for the multi-parameter integrated detection for better sensing function and application range. 3) To develop reliable self-powered modules for wearable devices, such as nano-friction power generation with excellent biocompatibility and flexibility. 4) To promote the application scenarios of flexible sensors in fisheries, such as fish health management, fish disease prediction and research, and testing medical drugs.

Dissolved oxygen can play a key role in the production and living of aquatic ecosystems. New materials and artificial intelligence (AI) technologies can be expected to promote the process of smart fisheries in recent years. It is very necessary to equip it with light, small, soft, and biocompatible sensors. Flexible electronics and sensing technology can be combined to detect dissolved oxygen. In this study, a flexible dissolved oxygen sensor was prepared to measure the temperature function by magnetron sputtering. The inkjet was also dispensed on both sides of the flexible substrate material. A multi-layer structure of the planar electrode was then adopted for the vertical distribution of oxygen-permeable film to encapsulate the electrolyte and the planar electrode. The polarization voltage was determined for the dissolved oxygen sensor by the linear sweep voltammetry (LSV) scanning in the electrochemical workstation. The polarization time of the sensor was measured by the response experiment. In addition, a comparison was also made on the difference between the prepared and commercial electrodes. The surface morphology of dissolved oxygen and temperature sensors was characterized using optical microscopy. A series of tests were carried out on the linearity, sensitivity, response time, drift, stability, and mechanical bending properties of the sensor. Finally, the sensing circuit and intelligent processing were designed to verify the feasibility of the sensor in the detection of dissolved oxygen in aquaculture water. The experimental results showed that the optimal polarization voltage of the prepared dissolved oxygen sensor was −0.6 V and the polarization time was 42 s. There was less difference between the flexible sensor and commercial electrodes in CV scanning, indicating the better performance of ion diffusion. Both the dissolved oxygen sensor and the temperature sensor exhibited homogeneous and better surface morphology at high magnification. There was an excellent linear relationship between the collection current and the dissolved oxygen content (R²=0.994 5) at room temperature. The sensitivity of the sensor was −0.03 μA·L/mg, the response time was 16.8 s (the maximum difference of multiple measurements was 3.3 s), and the maximum difference within 7 days was 0.0195 μA. The resistance of the flexible temperature sensor shared a better linear relationship with temperature in the range of 0-150 ℃ and 0-30 ℃ ( R² were 0.994 9 and 0.997 6, respectively). The sensitivity of the sensor was −2.47 kΩ/℃, the response time was 3 s, and the hysteresis error was 2.17%. The flexible sensor maintained better performance for the measurement of dissolved oxygen and temperature in the range of 0-60°. The maximum error of the prepared sensor was less than 5% when detecting different content of dissolved oxygen in aquaculture water at various temperatures, compared with the commercial sensor. A better temperature compensation can be obtained to rapidly and accurately detect the content of dissolved oxygen and the temperature of the water body in fishery applications.

Accurate, non-destructive quantification of leaf nitrogen content (LNC) is crucial for monitoring crop health and growth. Traditional empirical methods require extensive in-situ data for training, while physically-based methods are limited by ill-posed inversion, and hybrid methods suffer from domain shift between in-situ and simulated data. To overcome these limitations, this study introduces DeepSpecN, a novel hybrid method for maize LNC estimation using leaf-scale hyperspectral bidirectional reflectance. Without requiring in-situ data for training, DeepSpecN combines four key components: continuous wavelet transform (CWT) for reducing specular reflection, PROSPECT-PRO for simulating training data, an improved Transformer model for feature learning, and a spectral similarity-based sample selection method for selecting more valuable training samples. DeepSpecN and other methods, including physically-based methods, non-parametric regression based hybrid methods, and parametric regression methods based on vegetation indices (VIs), were validated using bidirectional reflectance data from 1724 maize leaves. The results showed that, when trained on representative samples, DeepSpecN achieved the highest estimation accuracy among all the methods (RMSE ​= ​0.247 ​g/m², R² ​= ​0.665). The sample selection strategy mitigated the effects of domain shift by identifying representative training samples with high spectral similarity from the simulated database. Furthermore, the results showed that the Chlorophyll (Chl)-based empirical formulas estimated maize LNC more accurately than those based on leaf protein content. Moreover, the validation results on four different crop species confirm the generalizability of DeepSpecN. Our findings demonstrate the potential of hybrid methods that utilize bidirectional reflectance spectra, developed by addressing the domain shift issue, to improve the LNC estimation accuracy.

Land inspection robots, endowed with autonomous navigation, obstacle avoidance, and mobility capabilities, are increasingly showcasing their unique advantages and practical utility across various sectors, particularly in fisheries and aquaculture. This paper delves into the critical technologies underpinning land inspection robots and explores their potential application in modernizing and adding intelligence to the aquaculture industry. The paper comprehensively analyzes the trifecta of land inspection robots: sensor technology, mechanical design, and control systems. Sensors play a pivotal role in capturing an array of environmental parameters, including temperature, humidity, light, pressure, sound, and imagery, providing a comprehensive dataset for decision-making. The mechanical structure serves as the backbone, enabling the robots to traverse challenging terrains, negotiate obstacles, and perform a diverse range of inspection tasks. The control technology, meanwhile, governs autonomous movement, task execution, and decision-making, encompassing motion control, path planning, and task scheduling. In the context of aquaculture, land inspection robots hold immense promise. They facilitate real-time monitoring of critical water quality parameters such as dissolved oxygen, pH, and ammonia, ensuring optimal conditions for fish health and enhancing production efficiency and product quality. The robots' ability to detect fish abnormalities early on enables proactive management, reducing risks and improving disease control. Furthermore, their remote observation and control capabilities streamline intelligent management of fish breeding facilities, lowering labor costs and enhancing operational efficiency. However, the application of land inspection robots in aquaculture faces unique challenges stemming from the complex and variable nature of the environment. The robots must exhibit a high degree of adaptability, interference resistance, and self-learning capabilities to accommodate diverse fishery scenarios and species. Additionally, the stringent requirements for information accuracy and real-time data necessitate robust stability and reliability in data gathering and transmission processes. Technical bottlenecks, such as limitations in environmental perception, image quality degradation due to water reflections and turbidity, and energy management in wet and corrosive environments, hinder widespread adoption. To overcome these challenges and unlock the full potential of land inspection robots in aquaculture, several development suggestions are proposed. Firstly, enhancing environmental adaptability designs, including waterproofing and corrosion resistance, is crucial for stable operation in harsh conditions. Secondly, advancing image processing technologies can improve image quality and fish recognition accuracy. Thirdly, establishing a comprehensive data processing platform, integrating cloud computing and big data analytics, will streamline data collection, storage, and analysis, enabling smarter decision-making. Lastly, optimizing energy management systems, including high-capacity battery technologies and autonomous charging capabilities, will prolong operational durations and reduce downtime. In conclusion, while the application of land inspection robots in aquaculture faces technical hurdles, their potential to revolutionize the industry remains substantial. With continued technological advancements and strategic development efforts, these robots are poised to play a pivotal role in driving the modernization and intelligent transformation of fisheries and aquaculture.