Robots on the reef

| April 25, 2024

QUT researchers have developed a robot to capture images of baby tank-grown corals destined for the Great Barrier Reef.

The system will help keep the growing corals happy and healthy before they are deployed and save researchers thousands of hours of coral counting time.

Associate Investigator at the QUT Centre for Robotics and Research Fellow for the QUT School of Electrical Engineering and Robotics, Dr Dorian Tsai said the Coral Growout Robotic Assessment System (CGRAS) project was a collaboration involving the Australian Institute for Marine Science (AIMS), Reef Restoration & Adaptation Program (RRAP), and the QUT Research Engineering Facilities (REF) team.

“AIMS researchers have knowledge of the biological processes involved and will be the eventual operators of the technology that we develop here in the CGRAS project,” Dr Tsai said.

“One of the main interventions for RRAP is about developing mass coral aquaculture for coral conservation, adaptation and restoration.”

Dr Tsai said most other coral restoration techniques were primarily focused on fragmenting existing corals – a technique known as coral fragmentation.

“This method is typically limited to approximately 50 fragmented corals per origin or donor colony, which has limited scalability,” Dr Tsai said.

“We’re aiming to mass-reproduce corals via sexual reproduction, where parent colonies can produce over a million corals each, which allows us to ensure the corals can diversify and adapt to changing environments and has much better scalability.


The QUT CGRAS team (from left) Karen Jackel, Riki Lamont, Dr Andrew Lui, Dr Dorian Tsai, Timothy Morris, Garima Samvedi and Joshua Esplin.

“The difficulty is that we have to then be able to grow and monitor these baby corals in order to keep the corals happy and healthy, which CGRAS aims to solve.

“The challenge is that when we are growing corals en masse during their growout phase inside tanks on flat tiles, the corals are tiny, about 1mm in diameter.

“There are hundreds of thousands of these corals, and it takes on average 45 minutes for a trained expert to count a single coral tile, which is 28x28cm square. Eventually there will be over 9600 such tiles.”

Dr Tsai said that would amount to over 7200 hours of coral counting time per week, which would be highly repetitive and completely infeasible to complete manually.

“To do it manually, this would end up costing around $500k per week in labour, which amounts to easily $6 million/year over a 12-week growout period for just counting corals,” Dr Tsai said.

“That’s why we’ve developed a robot prototype to capture images of the baby corals in the tanks as they are growing in a repeatable, precise and flexible manner.

“We’re also leveraging state-of-the-art artificial intelligence algorithms to automatically detect and count these coral babies and track their growth over time.

“This will allow scientists to have a much clearer correlation analysis between their treatment strategies and coral growth, leading to higher coral survival rates, which is higher coral yield, which will allow RRAP to reach their large-scale deployment scenarios of over a million corals out to the Great Barrier Reef each year.”

Dr Tsai is assisting a QUT research team that includes Senior Project Specialist in Industry Development, Karen Jackel, Research Engineering Facilities (REF) Team Principal Engineer Garima Samvedi, REF Senior Research Engineer, Riki Lamont, REF Senior Research Engineer, Timothy Morris, REF Senior Research Engineer, Dr Andrew Lui, and REF Research Engineer, Joshua Esplin.

Dr Tsai said he was excited about the robot’s potential to restore reefs damaged by rising sea temperatures.

Ms Jackel said mass coral bleaching events had occurred more regularly.

“That is why we are trying to grow coral that will adapt and be more resilient to climate change,” Ms Jackel said.

Dr Tsai said there was currently a bottleneck in the overall process because it was done manually, and that is what CGRAS aimed to resolve.

Mr Lamont said the current design of the robot prototype (main image) housed a high-resolution submersible camera with a macroscopic lens – basically a waterproof microscope – mounted at the end of a robotic arm.

“The arm is wrapped up to prevent corrosion in the salt-water environment that is the aquaculture facility,” Mr Lamont said.

“The whole thing is mounted atop a mobile platform that moves manually and houses all of the computers.”

Dr Tsai said although an ideal approach could be to build a fully autonomous robot that could drive itself through the aquaculture facility, that was not a cost-effective or necessary solution.

“Moving the robot manually between tanks only costs a fraction of the time compared to imaging, interpreting the images and counting the corals,” Dr Tsai said.

“Furthermore, the costs, challenges and risks of a fullyautomated system for a still-developing coral conservation process are orders of magnitude more expensive than we currently have budget for.

“Instead, we aim to hit the sweet spot of what is achievable with a compliment of robotics and humans together to maximise operational and developmental costs for coral yield.

“By doing this, we can bring the camera to the corals repeatedly and accurately.

“The design is a prototype to capture images and serve as a tool for developing the software pipeline.”

Dr Tsai said the final design of the robot could look significantly different.

“But for now our current design allows us to move forward with development as the scientists and researchers continue to develop and uncover the underlying biological process of mass coral aquaculture,” Dr Tsai said.

Dr Tsai said this program was different from the Coral Spawn and Larvae Imaging Camera Systems (CSLICS) or “nanny-cams” project.

“That system was for the coral spawn when they were still eggs and embryos developing into coral larvae,” Dr Tsai said.

“This CGRAS project focuses on a later stage in coral development, where the coral larvae have settled onto the bottom (pictured above) and have started to grow into coral polyps.”

Dr Tsai said CGRAS was actively deployed in the National Sea Simulator (SeaSim) at AIMS near Townsville early this year, and its next iteration will be deployed for the next coral spawning in December 2025.

AIMS Research Program Director of Reef Recovery, Adaptation and Restoration, Dr Line Bay said AIMS scientists are developing methods to assist the Great Barrier Reef better cope with the effects of climate change.

“Projects like CGRAS are essential parts of the technology toolkit needed to deliver scale in this work,” Dr Bay said.

“We clearly need strong emissions reduction to provide the best future for coral reefs. But with a fifth mass coral bleaching event since 2016 unfolding on the Reef research into restoration and adaptation is also important.

“Projects like this will turbo-charge R&D into conservation aquaculture methods that aim to help the Reef cope with increasing temperatures.”

SHARE WITH: