Cornell researchers are developing intelligent underwater recorders that distinguish marine mammal sounds from background noise and are creating a mobile recording device that uses wave motion to propel itself.

The work in the Cornell Lab of Ornithology's Bioacoustics Research Program (BRP) is made possible by two three-year grants totaling more than $2 million from the Office of Naval Research administered through the National Oceanographic Partnership Program. Christopher Clark, BRP's director, is the principal investigator (PI) on both projects.

Clark and BRP promote marine mammal conservation through research to document ocean noise and its repercussions while also tracking such endangered species as the North Atlantic right whale.

"The overall reason for doing this is to try to reduce the effects that man-made noises have on marine life," said Harold Cheyne, BRP's technology director.

One grant, for $350,000 per year, will support the research and development of smart recording devices and analysis systems that can detect marine mammals, classify the type of animal and, with multiple microphones in the water, estimate the animals' precise locations.

Teaming up with Yann LeCun, the Silver Professor of Computer Science and Neural Science at New York University, the team will use "deep learning algorithms" to create software for detecting and classifying marine mammal sounds.

"These algorithms will be used to identify vocalizations from marine mammals, like whales, but the biggest challenge is to ensure that the computer system will provide us a high degree of confidence to detect, classify and locate marine mammals," said Peter Dugan, BRP director of applied science and engineering and the project's co-PI. After the devices record underwater animal sounds, "the deep learning algorithms monitor the sound stream and use multiple layers of computer processing to make decisions as to whether the sound originates from an animal or from other sources," said Dugan.

Once developed, the new system will be the first to integrate deep learning algorithms into passive acoustic monitoring and provide a recording device for locating species, Dugan added. The system will also be tested under a variety of field conditions.

The other grant, also for close to $350,000 per year, enables BRP researchers to collaborate with engineers at the Science Applications International Corp. (SAIC) to develop a mobile marine acoustic monitoring and recording device powered by waves and solar energy.

The autonomous maritime vehicle consists of a square meter-sized floating platform, an underwater glider wing and an electromechanical tether that connects the platform to the glider wing. The glider wing converts the differential in-wave motion into thrust, and tows the platform behind it. The entire system can be deployed from a small ship or from a nearby port or harbor for missions.

BRP currently employs stationary recording units that are anchored to the sea floor, but the new device would allow the recorders to move into areas of interest or in the direction of mammal or other sounds. The recorders would be housed within the floating platform, which uses solar panels to power the electronics.

"With the underwater stationary recorders, it's costly to handle and deploy them, and then it takes time to get data because you have to wait a few months to retrieve them," said Cheyne, the project's co-PI.

SAIC also has proprietary satellite communication technologies that will allow researchers for the first time to use a satellite relay to transmit real-time data streaming from an autonomous surface craft to an on-ship data collection system.

The project also teams with Rose Paradis from Lockheed Martin Co., and engineers from the Pacific Northwest National Laboratory and Hydroscience Technologies Inc.