Manta ID
World’s First Underwater Camera Trap
To understand the lives of the elusive manta rays, we designed and built an underwater camera trap
I moved to Praia Do Tofo in Mozambique in early 2013 on a recommendation from Dr. Wallace J Nichols, to help study manta rays. Marine Megafauna Foundation has been conducting groundbreaking work in the research of manta rays and whale sharks, and they needed technical assistance in designing and building a device that could automatically identify manta rays.
Collaborator : Daan van Duinkerken
We collaborated on many aspects. Daan had developed an initial prototype as part of his thesis work, and was the expert on underwater housings and reef manta rays. I brought the hardware and firmware expertise, helping design the next iteration.
Organization : Marine Megafauna Foundation
Manta Rays are mysterious creatures, difficult to track and study. However, they bear natural spot markings which are unique to individuals and can be used to catalog specimens or monitor individuals within populations, in addition to understanding geographic range, with the help of global photo identification databases.
Marine Megafauna’s work in understanding the lives, habits and migration patters directly led to the discovery of a new species of manta rays and to the listing of manta rays and whale sharks on the IUCN Red List of Threatened Species. This allowed higher levels of protection for these amazing animals. The long term goal was to have an automated pipeline that would capture these images and feed them into Manta Matcher.
Design Requirements
Borrowing from Terrestrial Camera Traps
Terrestrial Traps use a beam of infrared light to detect motion. Using it underwater wouldn’t work because it is absorbed within a few feet underwater.
There were no known techniques to trigger camera traps underwater, so a large part of our research and design was targeted towards a trigger solution. We kept coming back to this phase, trying out new mechanisms.
Specifications
Needs to be installed at depths of 20 - 30 meters in reefs with strong swell
Must capture images of the ventral side of Manta Rays. Images of other large marine invertebrates, like sharks would be valued
Must not require frequent servicing and retrieval
Design Process
Shadow Detection : From both Daan's and my earlier experiments hacking a GoPro and used an Arduino as the controller, triggering the camera whenever a ‘dip’ in light levels was flawed. In theory image recognition could have been used, but conditions would worsen underwater, where light is more diffuse, hence shadows ill defined. Moreover, lights levels in the reefs we were studying, were prone to fluctuation due to changing tides, swell and plankton blooms.
Laser Detection : Daan had come to the conclusiond thar lasers were the way to go underwater. We designed a new system and triggering algorithm, using an Arduino-friendly low resolution grayscale camera, the ArduEye. The idea was to pulse laser light, and if reflections were received, there was a creature overhead. We postulated that using multiple lasers spread over an area, would allow triggering to be more sensitive to the size of the subject.
We prototyped the application and were able to fine tune the detection parameters. In the final design, a smaller arduino and smaller GoPro-like camera, the HackHD, were used to minimize footprint, for the smallest possible housing. Dry tests were successful. Our method of using 2 lasers to detect creatures atleast as wide as the inter-laser distance came in handy.
Reducing Footprint : A smaller housing means less chance of leaking and more space for batteries = longer life between retrievals. The final circuit design & physical sizing was done using Fritzing, and implemented on a perfboard, which included an LDR to detect darkness levels, below which the primary camera would not provide useful information, thus conserving battery at night.
The final housing design
Key Insight
We tested the trigger algorithm and circuitry with a red laser underwater. However, the wavelength of light dictates how far light will go before getting absorbed in water. From a distance perspective red is the worst choice, and blue the best. However, at our installation depth range of 15-30m, blue light is prevalent, so distinguishing between ambient and reflected light would be difficult, but a blue-green laser worked well in Daan’s final tests.
Testing
Ando testing the camera trap trigger with red lasers visible in the deep pool at about 5m
The original inspiration for a laser-based system is the method used by manta ray researchers to size individuals. This is done by using 2 blue-green lasers mounted on the camera rig. The laser dots in the image then provide the data to calculate size
Trap deployed for testing at 25m
Daan waterproofing the electronics
Did it work?
Yes and no. Turns out, electronics are not a big fan of a salty air environment or being short-circuited, or flooded with sea water at 25 meters below. All of those did happen, and getting replacement parts in rural Mozambique in 2013 was quite a challenge. Daan once had to wait more than 6 months for a camera module part. Every time we had a working prototype, something would go wrong, and we’d have to build another, usually from scratch. That being said, it did work for a short period. Daan continued to work on this project after I left, but unfortunately he eventually abandoned it.
It was quite the adventure, and the time I spent there above and below the waves, led to something incredible that I could not have imagined otherwise : Underwater VR
#conservation #conservationtech #underwater #marinebiology #design #hardware #diving #mantaray