A robot at the Australian Institute of Marine Science is simplifying a demanding part of coral restoration: checking tiny animals in nursery tanks. CGRAS photographs settled corals at the institute’s National Sea Simulator and uses AI to count and monitor them.[2]
A preprint released on 30 September 2026 reports 22.2 hours of robotic imaging and analysis for a workload yielding 160 successfully processed tiles. The researchers estimate equivalent manual work at 213.5 hours: about 9.6 times as long.[1]
A young coral is a small target
After a free-swimming larva settles, the young coral lives attached to a surface. In the AIMS nursery, those surfaces are prepared tiles. The animals initially measure around a millimetre across. A robotic arm carries a waterproof camera with a macro lens across two tanks holding 25 tiles each.[2]
Developed with Queensland University of Technology, CGRAS entered full-time use during the 2025–26 spawning season. It revisits the tiles to record their changing condition.[2]
A nursery needs that record because producing larvae is only one stage of raising a coral. AIMS’ aquaculture programme also investigates rearing density, the arrangement of holding tanks and how water flows around the animals. Its grow-out work follows multiple species for periods of up to 12 weeks before their move to the reef.[3]
The study’s 2025 monitoring workload
| Stage | Time or result |
|---|---|
| Robotic image capture | 14.6 hoursSource 1 |
| Image analysis | 7.6 hoursSource 1 |
| Combined operation | 22.2 hoursSource 1 |
| Equivalent manual work | 213.5 hours estimatedSource 1 |
| Tile images successfully processed | 160 of 182Source 1 |
Twenty-two tile images failed processing, mainly because of image-stitching problems or obstructed positioning markers. The successful images become maps of where living corals are concentrated, helping staff select parts of a tile for later deployment.[1]
Monitoring changes what a nursery can learn
The engineering is useful for more than replacing a count. With repeated observations, a researcher can examine when a batch begins to decline, whether corals grow differently across a tank and what changes after a rearing adjustment. The same images can connect the amount produced with the conditions under which it survived.
That creates a more informative experiment. An end-point count can show that one treatment produced fewer animals. A sequence of counts can help locate when the difference emerged. More frequent observations give scientists additional opportunities to investigate a problem while the remaining corals are still in their care.
CGRAS fits beside another camera system, CSLICS, which monitors the earlier free-floating stage. Its researchers describe modular cameras and object detectors trained with expert-reviewed labels to count coral spawn in rearing tanks. The updated study was accepted for presentation at the 2026 IEEE International Conference on Robotics and Automation.[4]
These systems watch different stages of the same biological process. Floating larvae need observation within the water; settled recruits need a detailed view of a fixed surface. Linking those stages gives an aquaculture team a better account of what happens between fertilisation and the point when corals are ready to leave the nursery.
The nursery is connected to work on the reef
The wider deployment programme has already moved beyond tanks. In February 2026, AIMS reported that its partners placed 44,608 coral seeding devices on three Great Barrier Reef sites during the 2025–26 season. Thirteen vessels from marine industries took part, alongside researchers and Indigenous Rangers.[5]
That is the scale of practical work nursery tools are intended to support. Coral production, handling, transport and placement have to fit together. In the programme’s first year, about 130 people received training in aspects of coral-spawn collection and release or conservation aquaculture. Trial reefs will be monitored for survival, growth and responses to environmental pressures.[5]
A successful nursery must know what it is sending out. By making repeated close-up observation less laborious, AI and robotics give coral growers a stronger basis for choosing rearing conditions and selecting animals for deployment. The gain is an improved capacity to watch, learn and raise the next batch.
Sources & further reading
- Coral Grow-out Robotic Assessment System (CGRAS): Scaling Coral Recruit Monitoring Through Robotics and Computer Vision QUT and AIMS researchers / arXiv · 30 Sep 2026 · Accessed 11 Oct 2026
- Five tech innovations turbocharging AIMS coral aquaculture into the future Australian Institute of Marine Science · 15 Jan 2026 · Accessed 11 Oct 2026
- Upscaling coral aquaculture for reef restoration Australian Institute of Marine Science · Accessed 11 Oct 2026
- Automated Coral Spawn Monitoring for Reef Restoration: The Coral Spawn and Larvae Imaging Camera System (CSLICS) QUT and AIMS researchers / arXiv · 22 Sep 2025 · Accessed 11 Oct 2026
- AIMS takes first shot at largescale reef restoration with marine industry Australian Institute of Marine Science · 26 Feb 2026 · Accessed 11 Oct 2026