It's been a dozen years since the "Hudson Bay Miracle" made history with an emergency landing of a passenger plane after it collided with a flock of Canada geese during takeoff. To prevent such hazardous incidents, airport biologists strive to mitigate aircraft-wildlife strikes. Knowing what is inhabiting the landscape and monitoring that wildlife in a common way among airports is essential for informing mitigation efforts.
Traditionally, biologists have surveyed wildlife at airports, taking visual counts of birds and mammals. But, as a team of MSU and USDA researchers have shown, modern technology can make that work easier and produce a more accurate scope of wildlife activity on airport properties.
Since May of 2020, Dr. Raymond Iglay, principal investigator and assistant professor of wildlife ecology, and his team have investigated small Unmanned Aircraft Systems, more commonly known as sUAS or drones, to improve the methodology of monitoring wildlife in airport environments.
"As biologists, we have to count animals that are often many in number, hidden or camouflaged, and moving around us. So, it's hard to get an accurate count. This leads to something called observer or sampling bias," Iglay said. "But, we're focusing on learning how and where to fly a drone to count animals, run the collected images though computer software designed to detect these animals, know exactly what animals are in the picture, where they are within the landscape, and their number. Essentially, we hope to reduce bias to improve accuracy."
Improving the accuracy of wildlife surveys at airports can translate to saved lives and reduced damage to aircraft.
The team's goal is to standardize the methodology and develop best practices for flying the drones, capturing images, and identifying animals in the images without disturbing animals.
Currently, the team is developing selection criteria for the drone models and sensors most appropriate for wildlife species hazardous to aircraft. They are also running field experiments to investigate associated human observer bias with counting animals including detecting animals in drone-captured images and accurately identifying them compared to using computers for animal detection and identification. The project focuses on sampling birds and mammals during the daytime using cameras and at night using thermal sensors.
To train computers to correctly identify species, the team will compare computer and human observations of captured images from a series of decoy set ups. They have run tests at the H. H. Leveck Animal Research Center and other MSU properties and hope to test at Bryan Airport in Starkville and Golden Triangle Regional Airport, if approved.
The team has spent much effort to ensure the captured images are identified accurately.
"We don't have an image repository to ensure accuracy, and this presents a challenge in teaching the computer to identify species by their characteristics," said Iglay. "To keep up with rapidly changing technologies, it's essential to constantly update computer models with new images to improve automated classification of animals from imagery."
Dr. Kristine Evans, assistant professor of conservation biology, said standardizing methods with dynamic species can be challenging.
"The big challenge is that animals move whether naturally or in response to the sound or sight of the drone," she said. "That can be an issue when we try to monitor the animals and get accurate species identification and counts."
Evans said that collaboration is key to solving this noting the work is a collaborative effort on three fronts. First, this effort is part of a larger research collaboration among MSU, the Ohio Field Station of USDA/APHIS Wildlife Services National Wildlife Research Center (USDA/APHIS WS NWRC), and the Federal Aviation Administration, or FAA. Specifically, Dr. Bradley Blackwell, research wildlife biologist, field station and project leader, and Dr. Morgan Drabik-Hamshare, research biologist of USDA/APHIS WS NWRC, work closely with MSU team members regarding project synthesis and development, related research endeavors, and coordinating research outputs and outcomes. Second, Dr. Wesley Major, airport safety specialist, and Mike DiPilato, airport research specialist of FAA provide further guidance regarding sUAS, related research, policy guidelines, and other airport operations and safety perspectives. Internally, MSU team members represent different departments across campus.
"This is a perfect example of interdisciplinary work. We biologists are working side by side with computer engineers," said Evans. "We draw from each other's strengths, and I don't think we could be doing what we're doing without that collaboration."
While the aviation industry stands to gain direct benefits from the study, the research has broader benefits as well.
"We aim to provide FAA with deliverables that will allow them to monitor wildlife using sUAS and inform mitigation efforts regarding aircraft-wildlife strikes," said Iglay. "But our methods have broader applications, to government agencies, academic researchers, and industry."
Dr. Sathisth Samiappan, assistant research professor in the Department of Electrical and Computer Engineering, Dr. Santahan Krishnan Boopalan, postdoctoral associate, as well as Dr. Landon Jones and Dr. Jared Elmore, both research associates in the Department of Wildlife, Fisheries and Aquaculture, are collaborators on the research.
This research is funded by two cooperative agreements with the USDA/APHIS Wildlife Services National Wildlife Research Center Ohio Field Station, who has an interagency agreement with the FAA. This research is complemented by a third cooperative agreement with Dr. Samiappan as lead investigator and one cooperative agreement pending under the direction of Dr. Evans.