Radar Can Tell the Difference Between Insect Species

Overview

European researchers built a radar system that can tell the difference between insect species, offering a possible cost-effective, noninvasive way to monitor pollinators such as bees. The system uses millimeter-wave radar to capture micro-Doppler signatures from insect wingbeats, then a machine learning model classifies the species. In tests, the model identified five pollinator species to the species level with 85 percent accuracy.

Key Takeaways

Researchers developed a radar system that could offer a cost-effective, noninvasive way to track pollinators.
Traditional identification often requires capturing and killing insects, and vision systems struggle with lighting, weather, and cluttered backgrounds.
The system uses micro-Doppler signatures from insect wingbeats, captured with millimeter-wave radar.
A machine learning model analyzed more than 70 features and classified five species at 85 percent accuracy.
Distinguishing the four bee species from the one wasp species reached 96 percent accuracy.

Stats & Key Facts

#The model classified five insect species to the species level with 85 percent accuracy.
#It distinguished the four bee species from the one wasp species with 96 percent accuracy.
#The model analyzed more than 70 features of the radar reflections.
#The model was trained on five species of pollinator insects.
#Findings were detailed on 28 April in the journal PNAS Nexus.

Radar Can Tell the Difference Between Insect Species

Why monitoring pollinators is hard

Existing methods are slow or limited.

›Bees and other pollinating insects play vital roles in food webs and crop pollination, yet monitoring them has proved difficult.
›Identifying pollinators traditionally requires capturing and killing insects to get a close look.
›Vision systems using machine learning face limits from variable lighting, poor weather, and cluttered backgrounds, and many insects fly away when approached.

Using radar instead of cameras

The team turned to radar to monitor lone insects near the ground.

›Scientists have used radar for decades to study migratory insects, but mostly insects flying in large numbers at high altitudes.
›This work focused on lone insects flying near the ground, as pollinators do when visiting flowers.
›The team focused on how insect wingbeats generate micro-Doppler signatures, distinctive time-varying patterns in radar reflections.

Adam Narbudowicz, an associate professor of space research and technology at the Technological University of Denmark, said the radar reflection from single insects is very weak and probably impossible to detect at a single point in time, so the team integrated signals over longer durations. Micro-Doppler signatures also help radar distinguish subtle differences between objects, such as birds versus drones.

The millimeter-wave radar setup

The team chose millimeter waves to match insect sizes.

›Millimeter waves better match insect sizes than other parts of the radio-frequency spectrum.
›Millimeter waves have also found use in recent generations of cellular networks.
›Insects, including honeybees and common wasps, were captured on the campus of Trinity College Dublin.

Several members of each species were individually placed in small plastic cylinders on top of a millimeter-wave antenna, which recorded their radar signatures, after which the insects were released.

What the model learned

Machine learning pulled subtle detail from weak signals.

›The model analyzed more than 70 features of the radar reflections.
›Features included wingbeat frequencies, the speed at which wing movements changed, and the strength of the reflections.
›Species-identification accuracy improved the longer the insects remained within the radar beam.

Narbudowicz said different species use their wings in different ways and that this is observable in radar signals, and that with sufficient machine learning the system can capture subtle details that are difficult to see in raw signals.

Results

The model reached high accuracy across species.

›The model classified the five kinds of insects to the species level with 85 percent accuracy.
›It distinguished the four bee species from the one wasp species, two different families, with 96 percent accuracy.
›The researchers detailed their findings on 28 April in the journal PNAS Nexus.

Frequently Asked Questions

What does the radar system do?

It distinguishes between insect species by analyzing micro-Doppler signatures from their wingbeats, offering a possible cost-effective, noninvasive way to monitor pollinators.

How accurate is the system?

It classified five pollinator species to the species level with 85 percent accuracy and distinguished the four bee species from the one wasp species with 96 percent accuracy.

Why use millimeter-wave radar?

Millimeter wavelengths better match insect sizes than other parts of the radio-frequency spectrum, and they are also used in recent cellular networks.

How was the data collected?

Insects captured at Trinity College Dublin were placed individually in small plastic cylinders on top of a millimeter-wave antenna that recorded their radar signatures, then released.

Where and when were the findings published?

The researchers detailed their findings on 28 April in the journal PNAS Nexus.

By reading the faint radar echoes of beating wings, the system points toward monitoring pollinators without capturing or harming them.