Periodic aerial surveys provide visual data that can serve as the foundation for studying marine ecology, especially the population dynamics of marine life. 

“The power of these aerial surveys is that in practically every image, we observe a wide diversity of life forms and even marine debris,” explained Senior Scientist Steven Haddock. “As our capabilities grow, we plan to survey across the entire Monterey Bay regularly. The imagery we collect can support a number of science and conservation projects, from tracking the movements of iconic marine wildlife to monitoring the health of coastal ecosystems.”

MBARI’s UAVs carry high-resolution cameras that photograph objects, some as small as a paperclip, in remarkable detail. As the vehicle flies 60 meters (approximately 200 feet) above the ocean’s surface, it takes an image every two seconds. During a single 20-minute survey, the aerial vehicle takes approximately 400 photos. MBARI engineers combine these images to create a photomosaic of the surveyed area that can be processed using machine learning models. Although the camera on the UAV can capture high-resolution images with centimeter-level precision, it is challenging and laborious to differentiate the many organisms and phenomena embedded in the thousands of images. 

“AI has allowed us to assess and analyze the array of imaging data collected by MBARI’s fleet of uncrewed aerial vehicles more effectively,” said Senior Software Engineer Danelle Cline. “The machine learning models group images with similar objects, then our human experts look at the outliers.”

An AI program called Slicing Aided Hyper Inference (SAHI) helps detect and classify objects. MBARI researchers group similar organisms and phenomena together based on shared visual features, allowing the SAHI model to examine them quickly. This approach removes the need for pre-labeled data to train an AI model, optimizing limited human resources. The more these models are used, the better they will get at classifying marine life and ocean phenomena.

Whether flying through the air, riding waves at the ocean’s surface, or gliding underwater, advanced technology provides a well-rounded understanding of ocean health. 

Satellites have long been used to monitor the ocean on a global scale. From space, satellite camera resolution is typically hundreds of meters for a single pixel of the ocean surface. Clouds can obscure parts of the ocean. The satellite’s orbit and sensor characteristics can also result in time gaps of days or longer for any given patch of ocean. UAVs are not subject to these constraints. They fly less than 100 meters (328 feet) above the ocean and beneath clouds, resulting in a camera resolution of approximately one centimeter (less than half an inch) per pixel. They can be quickly deployed in response to short-lived events of scientific interest, although UAV flight operations can be limited by high winds or rain.

Aerial vehicles can support a variety of science applications. For example, by examining high-resolution images of the ocean’s surface, researchers can identify regions where upwelling brings cool, nutrient-rich water to the surface, leading to a bloom of plankton and marine life. Researchers can identify targets of interest in images acquired by UAV cameras and deploy in-water platforms for more detailed sampling. Aerial drones can provide a two-dimensional map of ocean color fronts from the air faster than an underwater or surface vehicle. Once they find a location of interest, our engineers can then program a long-range autonomous underwater vehicle (LRAUV) or Wave Glider to take a closer look.

Coordinated sampling with an aerial vehicle operating above MBARI’s advanced underwater robots enables a more thorough investigation of the physical, chemical, and biological processes that occur in the ocean. Together, they can produce a three-dimensional look at a research site. Last year, MBARI researchers deployed a UAV alongside an LRAUV carrying the Piscivore camera system. By scouting ocean conditions with a drone, researchers are able to deploy Piscivore where it would have the greatest chance of observing marine predators. 

MBARI engineers are finding innovative ways to grow our UAV program further.

New onboard tech will allow the team to secure exceptions to regulations that currently limit operations to a one-kilometer line-of-sight, enabling operations farther offshore. MBARI engineers are also developing additional payloads that can be deployed on an aerial drone. In the future, aerial vehicles could be equipped with sensors to measure atmospheric particles and gases, which are crucial for studying marine ecology and understanding the ocean-climate connection. Incorporating onboard AI will also allow the vehicle to autonomously detect and respond to significant events in real time, like tracking whale pods or surveying thermal fronts.

At MBARI, we believe that collaboration is essential to truly maximizing the potential of science and technology to help us understand the ocean. The UAV Team is also exploring opportunities to collaborate with the Monterey Bay National Marine Sanctuary and other oceanographic research groups across Monterey Bay to develop more research applications for aerial vehicles.

“As MBARI’s uncrewed aerial vehicle program continues to grow, we’ll export our data and technology to scientists worldwide to facilitate their research and engineering efforts. We’re excited to help contribute to the research programs of various scientists and conservationists, whether they are interested in jellies, sharks, mammals, or kelp,” said Haddock.

Story by Science Communication Fellow Marike Pinsonneault

For additional information or images relating to this article, please email pressroom@mbari.org.