Sophisticated Smart Technology Triggering a Dynamic New Era for Aquaculture

Whether land-based or open ocean, aquaculture has a promising future due to the recent development of low-cost, high-resolution sensors and the push to apply machine learning and artificial intelligence to fish production.

This ability to look at what’s happening underwater, listen to new streams of data and then learn from this wealth of new information is going to be the catalyst that drives aquaculture to a more prominent role in feeding the planet’s growing population in the coming decades.

The industry can be proud of its 50-year evolution. In the 1960s and ‘70s, nascent production sites yielded at most hundreds of tons of fish and required a large staff that was fairly hands-on. Fast forward to the early years of the 21^st century and we saw large-scale commercial farming with individual sites producing more than 3,000 tons.

Despite the impressive production numbers, critical decisions such as the appropriate feed amounts and target transfer/harvest dates were made using hand-collected, statistically small data sets. This data would then have to be manually entered into farm management systems where human error or missing information was quite common.

At today’s land-based fish farms and open ocean production sites, we are getting a glimpse of how our industry can benefit from the recent surge in smart equipment and operations.

It Starts with the Sensors

Temperature, dissolved oxygen, carbon dioxide and other sensors and alarms have been used in land-based aquaculture for many years. However, not until recently have these devices been connected to tablets and smartphones through the cloud. Managers are now alerted to any variances regardless of where they happen to be.

Most of the alerts are managed by software that adjust settings on-site to address an issue before it becomes problematic. This closed-loop capability has enabled land-based projects to scale significantly in recent years. It is unfathomable to imagine a 5,000 ton operation without smart capabilities to address the constant need to manage tank chemistry.

Sensors have also improved. Until recently, galvanic sensors were required to measure dissolved oxygen. Because the sensor itself consumed oxygen while measuring, if it was not in a tank with moving water or water actively agitated by the technician, readings could be significantly inaccurate. Optical sensors now measure DO passively, not consuming oxygen and producing appreciably more accurate readings.

We see similar benefits in non-invasive data collection, wireless data transmission and cloud-based analytics in the open ocean environment. Small affordable sensors monitoring temperature, depth, dissolved oxygen and other parameters are now placed throughout the grid system and individual pens.

Data is then transmitted wirelessly underwater in real-time to a submerged hub (hydrophone) and then transmitted wirelessly to shore. It is then sent to the cloud where farm staff can view the current conditions throughout the production site. This data is also automatically stored for subsequent analysis to identify problems or areas of opportunity.

In addition, load cells placed strategically throughout the grid system measure line tension. As with the environmental sensors, data is transmitted acoustically to the submerged hub and ultimately to the cloud. Alerts are sent to the offshore manager if any anomalies are detected.

Cameras and AI

Innovasea is pioneering much of this technology with high-resolution underwater cameras that capture images for efficient feed management and biomass estimation. Feed managers use Innovasea’s IP cameras and subsea network to view fish during feeding. This is buttressed by pellet detection and satiation software that advises operators when to increase, decrease and stop feeding based on the number of pellets not being eaten and species-specific fish behavior.

At the same time, stereoscopic cameras instantly and constantly assess fish weight based on measured dimensions collected and growth rate algorithms. In near real-time, farm operators receive a biomass estimation report accurate to within 3 percent. Additional algorithms then determine the appropriate amount of feed for the immediate future.

This capability is used with both land-based and open ocean systems. Gone are the days where every tank or pen required physical sampling of a small set of fish to determine biomass. The biomass estimation system constantly measures tens of thousands of fish, providing a much more accurate assessment than catching and weighing a statistically insignificant sample set. Importantly, these images are captured passively and thus do not affect the fish, preventing stress or the risk of putting them off feed.

In aggregate, modern aquaculture technology and smart systems provide a crucial set of advantages that enable the producer to take farming to the next level:

• Real-time alerts and closed-loop decisions mitigate risk, reducing loss and potentially lowering insurance costs.
• Passive data collection minimizes stress on stocks, improving feed conversion ratios. It also relieves the staff of the responsibility of collecting data, thus making it more likely to be collected, properly recorded and uploaded to the appropriate location.
• Wireless data transmission reduces the chance of cable failures, a common problem on both land and in the ocean.
• It also enables centralized command and control. This feature, coupled with AI, enables staff to oversee a greater range of responsibilities in a comfortable environment, improving efficiency and accuracy.

Collectively this new ability to look, listen and learn will enable scaled-up operations and improved bottom lines, which will be imperative as the industry grows over the next few decades to sustainably feed the globe’s growing middle class.