Application of artificial intelligence in the aquaculture industry is still in its infancy, but its potential to improve efficiencies in RAS farms is increasingly evident.
September 8, 2020 By Christian Pérez-Mallea
In the early days, only a few individuals would work within any given fish farm. All variables were controlled by humans, and tasks such as feeding, treatments, samplings and harvesting were performed manually. The focus was set on the biology of the fish.
Subsequently and due to the success of aquaculture, farming volumes were increased, and automation was needed. Routine machine operations were established and different robots were included in the production chain. The focus, this time, was set on the economics of the farm.
Lately, the sole presence of machines and routines has proven itself insufficient to control different variables simultaneously. Its dependence on human control combined with the probable need to refocus more efforts on biological aspects have fostered the arrival of digital applications to aquaculture.
Go with the flow
Computational fluid dynamics (CFD) can be described as a set of mathematical models used to simulate how gas and liquids flow. For fish farming purposes, this might be especially useful in land-based facilities where it allows users to predict the behavior of water flow in the tank.
Jorge Contreras, project manager at Scale AQ Chile, say they use this combination of applied mathematics, physics and computational software to simulate a new project. “It allows us to predict the behavior of the water flow in the farming tank and the quality of the water as well. The water quality is controlled by quickly extracting the unconsumed feed and the feces.”
The focus on water quality is the core of Scale AQ’s Optifarm concept, he explains. That is the mixture of their exclusive designs of ponds (Optitank), filters (Optitrap), and self-cleaning systems (Optiflow).
Beyond just extraordinarily complex algorithms and machine-to-machine communication, there is the Industrial Internet of Things (IIoT) which could be distinguished from domestic-use IoT, thanks to the intersection of information technology and operational technology. Meanwhile, blockchain uses data structure or chain of records in the form of blocks, that ensures security, transparency and decentralization.
Breeding and genetics company Benchmark Genetics is implementing these new technologies. “We use IoT technology to register different environmental factors and to monitor animal welfare and security factors on our farms. The implementation of blockchain technology is on our drawing table and we are working closely with one of our customers to establish a first version. We will make a formal announcement when we can offer our customers this service,” says the head of strategic business systems, Bára Gunnlaugsdóttir.
She explains that using IoT to register data eliminates human error and allows them to use manpower in a more productive way. Likewise, using this technology to monitor operations speeds up both the reaction time and decision process.
“Blockchain technology will reinforce the trust of our production and give our product traceability a higher credibility in the ‘One World, One Market’. Consumer requirements to make an informed purchase is only getting stronger, thus we need to have a technology in place that can support and adapt to those requirements,” she adds.
Visual perception and decision-making are AI tasks already helping aquaculture to grow stronger.
CEO and co-founder Valerie Robitaille explains that XpertSea has developed a platform that gathers data about health and quality of the fish from a distance. It can predict growth and simulate different farming and feeding strategies. Most of her customers are shrimp farmers in South America and Southeast Asia.
“We start early in the cycle, so we give a solution that counts, measures and weighs the shrimp larvae. And then, throughout the growth cycle, we keep gathering some information using AI, so we collect pictures of these organisms and then we use machine-learning to extract all types of valuable information about the shrimp,” she explains.
Her company is also working on an application that helps to detect any type of sanitary issues or diseases in shrimp. For example, they are already able to detect changes in pigmentation, necrosis and deformities.
Gunnlaugsdóttir explains that to be able to utilize AI you need to have access to vast amounts of relevant structured data. “We have already re-structured our data and databases to accommodate the implementation of AI. We have also implemented business intelligence tools to give us real-time access to analysed data. The plan is to incorporate AI information into some of our decision-making processes within the end of 2021,” she says.
Implementing AI on top of Benchmark Genetics’ business intelligence tools should improve their analytical capabilities and possible scenario predictions, resulting in faster learning along with swifter and more informed decision making, she says.
Where is the limit?
Several companies, such as Scale AQ, CageEye, 3SE, Observe Technologies and AquaByte are already offering different AI solutions for fish farms at the fattening stage. They cover either feeding or sanitary aspects, and even both. Yet, they do not have any digital application for freshwater. For example, Bryton Shang, CEO of Aquabyte, explains that his cameras typically capture the fish half a meter underwater so they could be used in a freshwater tank. “However, our focus is on the sea cages right now,” he says.
AI can potentially be used to help land-based aquaculture operate more cost-effectively. Gunnlaugsdóttir comments that the highest expense for freshwater land-based farms (incubation centres, hatcheries and smolt farms) is usually the cost of energy needed to operate.
“Implementing AI on energy usage is probably some years away but for now, the biggest advantages would be in optimizing production plans, management processes and predicting challenges before they happen,” she says.
The founder and CEO of 3SE, Víctor Valerio, believes that AI is becoming cheaper and more accessible every day, as more professionals develop more simple algorithms and better cameras and sensors are used. However to date, AI is not still not affordable.
For freshwater land-based farms, he says the required investments might be worthwhile especially in the fattening stage. This technology could also help in the physiological and morphological changes in the fish. Smoltification and parr marks in salmon could be noticed with artificial vision and image processing, while sexing could be performed with a laser during the vaccination process.
“We could also register the mucus cell counts of those fish going to the sea. When you transfer fish highly advanced into smoltification, probably your first barrier (mucus) is severely affected and, from an immune point of view, they are much more susceptible to diseases,” he explains.
In a similar thought, XpertSea’s Robitaille comments that “no matter what you do, no matter how much data or treatments you have, even if you have the best feed in the world, if you do not have good quality fry or larvae then you do not get good results.”
Evidently, the digital transformation in the aquaculture industry still represents a major economic undertaking that must be contrasted with all its benefits and potential advantages. It should also be considered that the growing volumes and sometimes remoteness of fish farms seem to be an ideal application of all these technological advances.
Print this page