Innovative Aquaculture Startup Leverages AI to Enhance Efficiency and Sustainability

GoSmart, an active participant in the NVIDIA Metropolis and Inception programs, revolutionizes fish farming by utilizing autonomous systems powered by the NVIDIA Jetson platform for edge AI. Their cutting-edge technology enables farmers to efficiently manage fish feed and optimize production processes.

As a student of marine biology, Josef Melchner always harbored a dream of exploring the oceans, searching for dolphins, whales, and fish. However, he also had a desire to make a practical impact that would benefit the world. When the time came to choose a career path, he enthusiastically embraced aquaculture.

Today, Melchner serves as the CEO of GoSmart, an Israeli company that utilizes AI and machine learning to enhance the efficiency and sustainability of fish farming. GoSmart is a proud member of the NVIDIA Metropolis vision AI partner ecosystem and the NVIDIA Inception program, which supports innovative startups. The company offers fully autonomous, energy-efficient systems that are about the size of a soda bottle and can be attached to aquaculture cages, ponds, or tanks.

Powered by the NVIDIA Jetson platform for edge AI, these systems analyze various parameters within the fish farming environment, such as the average weight and population distribution of the fish, as well as temperature and oxygen levels. The gathered information is then provided to users through GoSmart’s software-as-a-service, enabling fish farmers to determine the optimal feeding and harvesting schedules for their fish more accurately and efficiently, all in real time.

“The parameters analyzed by GoSmart systems play a crucial role in managing the fish feeding regime,” Melchner explains. “Optimizing the right levels of fish feed not only saves money for farmers but also reduces organic waste and debris in the aquaculture environment.”

Skretting, one of the world’s largest fish feed producers, has deployed GoSmart systems as part of its sustainable expansion initiatives across eight countries in southern Europe. These systems provide personalized and digitized information to farmers, contributing to the industry’s advancement.

GoSmart, founded in 2020, focuses on fish farming due to its commitment to environmental sustainability. While the world faces a protein deficiency, marine protein is still predominantly sourced from wild capture through fishing nets and long lines, unlike other alternative protein sources such as cattle, pigs, and chicken, which are largely farmed. This overfishing approach negatively impacts the planet by disrupting the levels of algae in the ocean. Algae serves as a crucial carbon sink, absorbing carbon from the atmosphere and releasing oxygen.

Understanding the critical nature of this issue, Melchner dedicated his life’s work to aquaculture.

The GoSmart system incorporates lithium-ion batteries charged by solar panels and is equipped with power-management software, enabling autonomous operation modes such as sleep, shutdown, and wake-up. The system ensures efficient farming practices and productivity.

AI plays a significant role in boosting farming efficiency in GoSmart systems. These systems utilize sensors, cameras, and NVIDIA Jetson modules to enable edge AI analysis of environmental factors that affect fish feeding, growth, health, welfare, and the dispersion of excessive organic matter in the water due to inefficient operations.

Melchner explains, “We aimed to utilize the best AI processor with high performance in a compact, submersible, and affordable system for fish farmers, which is why we chose the Jetson series.”

Currently, GoSmart is training its systems to analyze fish behavior and indicators of diseases. This expansion builds upon the existing capabilities of determining fish weight, population distribution, temperature, and oxygen levels. Thanks to Jetson’s parallel processing capabilities, all these characteristics can be analyzed simultaneously and in real time.

Additionally, GoSmart is exploring the powerful Jetson Orin lineup of modules to further enhance their system capabilities.

To train their AI algorithms, the GoSmart team initially measured thousands of fish manually before deploying cameras to analyze millions more. “There was a lot of diving and numerous underwater experiments,” recalls Melchner.

For high-performance deep learning inference, GoSmart intends to leverage the NVIDIA TensorRT software development kit and the open-source NVIDIA Triton Inference Server software.

Being part of the NVIDIA Metropolis and Inception programs, GoSmart collaborates closely with NVIDIA engineers and stays at the forefront of the latest technologies. This collaboration ensures that GoSmart’s algorithms become faster and more efficient.

According to Melchner, GoSmart has the potential to help farmers reduce fish feed by up to 15%. Some customers have experienced a shortened fish growth time and accelerated time to market by an entire month through the implementation of GoSmart technology.

Melchner envisions a future where aquaculture undergoes a radical transformation. He aspires to see GoSmart systems deployed in every cage, pond, and tank worldwide, aiming to revolutionize the entire aquaculture industry. GoSmart’s future plans include integrating AI models that analyze fish behavior and disease and eventually incorporating their solution with an autonomous feeding barge, ensuring precise and timely feeding for fish.

Leave a Reply

Your email address will not be published. Required fields are marked *