© University of Canterbury
The programme uses AI to navigate autonomous underwater vehicles (AUVs) to take high-quality images, collect samples, and reconstruct 3D images of species in strong water currents. Led by Professor Richard Green, head of the Computer Science and Software Engineering department, the research programme focuses on using the technology to improve efficiencies in farming shellfish, ocean caged fin fish (salmon), and seaweed. It could also improve biosecurity monitoring, for instance wharf pylons could be checked for invasive species more frequently at cheaper costs.
Professor Green said the work could transform the country’s future high-tech aquaculture sector. “To enable the expansion of our aquaculture sector, we need this technology. The natural progression is to figure out how we can improve world food security. By enabling more automation of the farming, like we are already doing, we could expand farming without it being prohibitively expensive,” he commented in a press release.
The New Zealand Government’s aquaculture strategy set the goal of a $3 billion dollar aquaculture industry by 2035, with a $1 billion target for the Greenshell mussel industry. To achieve this, Professor Green says Aotearoa’s aquaculture must increase innovation, sustainability and technological capability.
However, the sector has faced challenges collecting accurate data and samples due to the difficulty of capturing images of moving species in ocean currents. Ten years of research has gone into Professor Green’s solution to these problems, including developing and testing AUV prototypes with visual recognition systems that can operate in changing underwater environments. The work means accurate digital 3D images can be taken of underwater species to monitor growth, and if marine pests and diseases are detected, they can be instantly removed.
“Our system will help to drastically improve yield prediction and accuracy and save operating costs for aquaculture industries,” Professor Green says. “Removing invasive marine organisms will not only save costs but support adaptation to climate change by supporting our battle against the growing influx of invasive species driven by climate change.”
A particular challenge has been adding AI into the navigation system to capture high-quality images of mussel ropes that are constantly moving in swells and currents. The team’s solution has been to predict the 3D location of movement about a second into the future. They also propose the technology could be used with GPS to scan scallop beds, providing surveys of the seabed as an alternative to net dragging.
“Mechanical engineers have done wonderful work over the last century, but we have hit a level we can’t just automate with mechanical engineering, we now need AI and algorithms as well,” Professor Green concludes.
