When you're watching a 3D movie, you can thank stereo vision for knowing exactly when that dinosaur's head is close enough for you to reach out and "touch." Stereo vision gives people the ability to see where objects are located in space, and how far away they are.
PNNL researchers have come up with a novel way to integrate that stereo vision feature into software to better "see" the flight patterns of bird and bats. This new, real-time capability will enable scientists to better identify the animal species and their flight patterns around offshore wind turbines.
The energy from wind turbines located in the ocean or other water bodies is more consistent, abundant, and reliable than that from land-based wind farms. But as offshore wind farms are being built, developers and regulators want to better understand how the birds and bats found offshore respond to the presence of turbines. Potential risks are the possibility of animals colliding with the turbines or being forced to detour around them. Understanding animal behavior around offshore turbines will help quantify those risks—especially for protected or sensitive species.
That's where PNNL's ThermalTracker comes in. The software extracts flight tracks of birds and bats from thermal video recordings and quantifies them by time of day or night, direction of travel, and more. This information, combined with other characteristics inferred from the software, can be used to determine which species of animals live offshore and where they're flying. PNNL researchers made an open-source version of ThermalTracker available on GitHub in 2016.
To make the tool even more valuable, the PNNL research team added stereo vision. The new version, called ThermalTracker2, processes the streams from two thermal cameras simultaneously. Going from one camera (two dimensions) to two cameras that have different perspectives (three dimensions) enables researchers to see the exact position of the birds and bats in space.
"The resulting tracking data show flight height and depth in the area where the turbine blades would be turning, typically 50 to 250 meters above the water surface," said PNNL engineer Shari Matzner. "Stereo vision also lets us estimate the body size and wing span of the flying animals, which helps us identify the species."
ThermalTracker2 runs in real time, with onboard processing that dramatically reduces the volume of data to be stored and transmitted. "If only the images that contain flight tracks are saved, it reduces the amount of data up to 300 percent or more," Matzner said. "This means users can put the cameras in a remote location and let them record for weeks without worrying about data storage running out."
Soaring Toward Commercialization
The Biodiversity Research Institute, which conducts wildlife research worldwide, successfully tested the prototype system using two thermal cameras and ThermalTracker2 in Portland, Maine in the summer of 2017.
After streamlining the stereo processing feature, the PNNL team plans to ruggedize the system to withstand a marine environment and conduct a field study at an offshore wind site in 2019. The U.S. Department of Energy's Wind Energy Technologies Office funds the research.
Organizations interested in licensing or collaboration opportunities for ThermalTracker2 should contact PNNL commercialization manager Sara Hunt.