In-Depth Analysis:

Approaching Birds with Drones

Objective of the Experiment:

How do birds react to drones approaching them in the wild?

How can a drone approach birds in different ways without causing them to fly away or get scared. This knowledge could help future researchers who want to use robots/drones in ornithology studies.

What is a Drone?

‘An unmanned aerial vehicle without any human pilot, crew, or passengers on board. A drone can navigate autonomously, without human control or beyond line of sight.’

In this experiment, a Phantom quadricopter drone designed by Cyleone was used. This drone has a diagonal length of 350 mm, a mass of 1030 g, and has a noise level of 60 dB.

(60 dB is approximately the sound level of two people having a normal conversation sitting at a distance of 1 meter away from each other.)

​

Robots are still marginal as tools in ecological research, yet they have a tremendous potential for biodiversity sampling, experimental biology, and behavioural studies.

1.

The Experiment

Introduction and Explanation

Overview

• Three different breeds of birds(mallard, wild flamingo, common greenshank) were approached with drones in both semi-captive and natural settings.
• Mallards were approached in a semi-captive setting in which they were capable of flying in and out of.
• Wild Flamingos and Common Greenshanks were approached in a natural setting.
• The birds were either floating on the water or standing in shallow water.
• All birds were non-breeding at the time of this study.
• Birds were not individually marked, and therefore some birds might have been approached more than once.

Wild Flamingo

There was an average of 35 Flamingos used in this experiment.

​

Mallard

There was an average of 5 Mallards used in this experiment.

Common Greenshank

There was an average of 19 Common Greenshanks used in this experiment.

​

How the experiment was performed:

• In March 2014, a drone equipped with a camera, which relayed images onto a screen, was sent out to approach different species of birds.
• There were three drones used in the colors: white, black, blue.
• The drone was launched at a minimum distance of 50 and 100 m from the birds in the semi-captive and the natural setting, respectively.
• One operator was in charge of the drone’s movements while another operator used binoculars to observe the birds.
• There was a multitude of attempts with the drone flying down to the birds from a variety of angles and with various speeds.
• Two minute breaks happened between each attempt.

Important Information

Type 1

No reaction

Type 2

Brief head/tail movements followed by either walking or swimming away from the drone

Type 3

Flying away

​

In this experiment, bird reactions were classified into the following groups:

Goal of the Study

Can a robot safely approach wild birds without startling or scaring them?

2.�Results and Discussion

What is the data from this study and did it achieve its initial goal?

Results

• A total of 204 approaches happened in 8 days: 48 on mallards and 156 on greenshanks and flamingos.
• In mallards, no reaction was recorded in 35 cases (72%), type 2 reactions in 9 cases, and type 3 reactions in 4 cases. These reactions took place when the drone was 4–8 m from the birds.
• In flamingos, no reaction was recorded in 75 cases (78%), type 2 reactions in 11 cases, and type 3 reactions in 10 cases; These reactions took place when the drone was 5–30 m from the birds.
• In greenshanks, no reaction was recorded in 53 cases (87%), type 2 reactions in 5 cases, and type 3 reactions in 2 cases; These reactions took place when the drone was 4–10 m from the birds.

Flamingos

Key:

= 2 ms^-1

= 4 ms^-1

= 6 ms^-1

= 8 ms^-1

​

​

​

Mallards

Key:

= 2 ms^-1

= 4 ms^-1

= 6 ms^-1

= 8 ms^-1

​

​

​

Greenshanks

Key:

= 2 ms^-1

= 4 ms^-1

= 6 ms^-1

= 8 ms^-1

​

​

​

Conclusions from Results

• Approach speed, drone color, and repeated approaches did not have any significant impact on the birds’ reactions.
• Bird behavior (resting/feeding) was a non-significant factor.
• Difference in angle approaches marked impact across all three species.
• All species had no reactions for drone approaches at 20°, 30°, and 60°, but had significant reactions at 90°.

New Hypothesis

A Phantom drone approaching a bird vertically is more disturbing, because it would be associated with a predator attack.

The study ended with a new hypothesis based off of results:

FINAL STATEMENT:

“When carefully flown, drones may be used in ornithology for a wide range of population censuses, measurements of biotic and abiotic variables, and recordings of bird behaviour. Those applications could be immensely useful, especially in inaccessible areas such as mountains or large wetlands.”

​

​

​

Discussion of Conclusion

The data supports the conclusion of this study because of how the birds only major reaction was towards approaches of 90°. Drone color, approach speed, and other angle approaches marked no reactions in all three species, meaning future researchers could use this to their advantage. The Phantom drone used had a rather low noise level (60dB), and was marked as an insignificant factor in the study, meaning other drones with similar noise levels can be used in various studies. Since the drone was able to approach birds up to 4 meters, this means bird behavior recordings from drones can be taken from closer than before without startling or making birds nervous. Overall, this study proved that some bird species are unbothered by drones and can be monitored or recorded by robots in the future.

3.�Critical Analysis

Confounding Variables, Improvements, Overall Critique and Questions

​

Confounding Variable: Bird Species in Study

• Differently sized birds will react differently to a large object approaching them.
• Mallards and greenshanks are medium-sized birds, while flamingos are quite large.
• A drone approaching a larger bird might seem less of a threat than a drone approaching a small songbird, similar to how a cat poses more of a threat to a rabbit rather than a wolf.
• If the study had one large, one medium, and one small bird species to test the drone approaches on, more accurate results could have come out of the experiments.

Confounding Variable: Difference of Drones

• Only variety in color was used in this study, and that factor proved to be insignificant.
• Different drone models and noise levels might have had different reactions.
• Although drone model might not affect bird reactions, it would be interesting to see, since some models might look like a predator bird, which would give a variety of reactions throughout the bird species.
• Birds like pigeons and owls have superb hearing, so having a drone with a lower or higher noise level would most likely end in different responses.

​

Possible Improvement

One element that I think might have improved results in this study would have been to spread out the approaches. In this experiment, the drone would approach a group of birds every two minutes until there was a reaction. This might lead to incorrect results if the birds choose to fly away since it would seem like they are frightened of the drone rather than annoyed or wary. If the study did one approach every hour or so, this would give the birds more time to ‘forget’ about the previous approach and give a more genuine reaction. Instead of performing 204 approaches in 8 days, I think it would have been better to perform these approaches over a 2 week period or more.

Valid….?

Generally, I think this study’s results are valid and achieve the goal of a robot approaching wild birds safely to gather data or footage. Although there are some variables that I would change or adjust, I think the overall study was performed well and the results make sense. It was interesting to see results for different approach speed and angle, even though they were not what I was expecting. After reading through the full experiment analysis and inspecting the results in detail, I confirmed that the study’s results appear logically accurate. Although it is impossible to say that this data applies for all bird species and types of drones, I think as a general base for future studies, these results are valid.

Follow-up Question

Will captive birds and wild birds react the same to a drone/robot approaching them?

Thank you!