Aquaculture Development: Smart Cage Net Farming System Lowers System Costs and Boosts Management Efficiency
As the global population grows, demand for seafood consumption is also rising; however, because of the escalating land costs and water quality challenges of land-based farming, offshore cage nets have become an important development trend in aquaculture abroad. Taiwan has been developing offshore cage net farming for over 30 years, but because of the many typhoons in summer, the management and facility wear-and-tear costs of cage net farming remain high. Therefore, National Taiwan Ocean University and the Artificial Intelligence Biotechnology and Medical Innovation Research Center of National Cheng Kung University have teamed up to build a "Smart Cage Net Farming Management System," hoping to reduce facility costs and improve farming management efficiency. Part of the hardware development and IoT system integration has already been completed, and this summer the team will go to Pingtung to collect data and run trials at offshore cage nets.
In countries that have been developing cage net farming for longer, such as Norway and Japan, automated management methods have already been introduced, said project principal investigator Zhang Zhongcheng, a professor in the Department of Electrical Engineering at National Taiwan Ocean University, but the two countries' integration of AI recognition and big data technology is not yet mature, and this is Taiwan's opportunity. This project brings together research teams in information and electrical engineering, aquaculture, and AI artificial intelligence, and also works with cage net farming operators, Chunghwa Telecom, and surveillance system operators, and the team is confident that within the next two and a half years it can design a system that meets industry needs.
System concept diagram.
Hsu Te-hua, an assistant professor in the Department of Aquaculture at National Taiwan Ocean University, explained that the biggest difference between offshore cage net farming and land-based farming is the "distance" in management. Depending on the fish species being farmed, if the stocking site is in sea areas 2 to 3 km from shore, farmers cannot patrol the farming area every day, so their ability to grasp the growth status of the farmed organisms is limited. If there is enough data at the farming site to assist decision-making, the survival rate of the farmed organisms can be improved. In addition, Taiwan has many typhoons, so the test on underwater cameras and sensing instruments is also more frequent. Foreign equipment is relatively expensive, and for farmers, a high replacement rate would mean a large expense, so how to maintain detection accuracy while also meeting cost requirements is also a problem this project hopes to solve.
The project has been underway for about a year and a half. In the indoor farming ponds at the Gongliao Marine Biology Research Center, the research team has successfully integrated data from surface and underwater cameras and sensing instruments, enabling AI to control the feeding system based on ripples, as well as recognize fish body length and perform other functions. What is more unusual is that, in order to reduce the cost of deploying monitoring instruments at multiple points on the cage nets, the research team built its own low-cost IoT system, including sets such as temperature sensors, power supply systems, and sensor control devices, as well as an anti-theft rotary-wing aircraft that can take off on schedule for patrols. As long as an unidentified vessel enters the farming area, it will immediately notify the farmer, reducing the chance of fish being stolen offshore.
Next, the research team will go to a cage net farm in Pingtung for on-site operation, collect farming data, and design a cage net floating and sinking device; in addition, it will also strengthen the underwater camera's technology for identifying fish amid wind and waves as well as its anti-fouling capability, and does not rule out combining sonar and other ultrasonic technology to identify the physiological characteristics of fish schools, in order to increase the accuracy of decision-making.
