Eagle vs. Drone
What a drone can tell us about the force behind a wedge-tailed eagle strike
I almost came as a shark actually, but then I realized an eagle’s slightly better. — Jarrod speaking to Lily in Taika Waititi’s Eagle vs Shark
The wedge-tailed eagle is not a small or bashful bird. A station owner in Western Australia (WA) once told me that he stopped his vehicle in the road when a wedgie was feeding on a kangaroo and unwilling to get out of the way. When he attempted to shoo the bird, the wedgie hopped into the truck, thrashed around for a while, and then hopped back out and flew away. The station owner said every bit of fabric in the Land Cruiser was torn to shreds, the dashboard was broken, the steering wheel was bent, and the windshield was cracked.
Our eBee classic fixed wing UAV, on the other hand, is 0.69 kg of foam, held together with glue and tape, and flies at about 40 km/hr (Figure 2). Like Lily’s shark, was our drone up for a confrontation with an eagle? We have flown this drone in the mountains of the Canadian Rockies, islands of the Bahamas, desert-scapes ofNamibia, thin air of the Bolivian altiplano, and forests of the eastern USA, and the only danger is sticking the landing. But in Australia, wedgies turn out to be far more dangerous than landings.
We first learned to fear the wedgie while searching for Hadean zircons in the Maynard Hills of Western Australia. Our goal was to find individual zircon crystals in the sediments laid down by rivers 3 billion years ago, and now exposed as layers of rock in low hills deep in the West Australian outback. Could tiny magnetite grains in these ancient zircon crystals tell us the strength of Earth’s magnetic field during the first few hundred million years of history?
When conducting geological field work of this flavor, we like to start by flying a number of 30-minute drone missions to build both a three-dimensional model of the landscape and a multispectral photomosaic, each tied to the same geographic coordinate system. Then, using a combination of GPS and iPads, we can constantly compare scales: (1) what the drone sees from above, (2) what our eye sees on the ground, and (3) what the magnifying lens can tell us about the insides of rocks.
Get Adam Maloof’s stories in your inbox
Join Medium for free to get updates from this writer.
By the time we launched the drone on August 04, 2017, this mapping approach was part of our routine. My student ran the drone operation, while I worked on the map, both of us watching as the drone flew its grid, and communicating by walkie talkie. But little over a minute into our fourth flight this day, while the drone was right above me, I saw a wedge-tailed eagle in the distance. As the wedgie got closer, apparently my walkie talkie comms got increasingly colorful/panicked, but I don’t remember what I said — I must have blacked out. 2.5 minutes into the flight, the wedgie dove and raked its talon across the drone (Figure 3). A chunk of foam got caught on the eagle’s talon, the battery cover popped off and floated down to Earth (eventually found later that day by two other members of our team), one of the wings dislodged but did not completely detach, and the propeller disconnected from the axle. The drone started fluttering and wobbling in an injured glide back toward the launch site (Figure 3), while the wedgie carved a tight upward spiral and turned to begin a second approach. The wedgie’s second dive looked like it would be the end of our drone, but the eagle pulled-up just short, perhaps satisfied that the drone was wounded and about to crash (which is exactly what happened next).
One cool thing did come out of this eagle strike. The drone has an onboard accelerometer, pitot tube (for relative air speed), barometer, and GPS to help it navigate. We sent the ‘blackbox’ binary flight log to the makers of the drone, and after a bit of negotiating, they sent us back all the on board sensor data. Can we calculate what it would feel like to be struck by a wedgie (Figure 4)? Could the escalating face-off between drones and wedgies in Australia (Figure 5) be a rich new source of raptor behavior data for ornithologists?
Impulse is the total shove an object receives (force times how long the force acts ) and impulse equals the change in momentum (mass × velocity). To measure the eagle’s shove, we compare the drone’s velocity just before and just after contact: Δv = 0.8 ± 3.7 m/s (Fig. 4B). The wide error bar is unavoidable: (1) the GPS itself is only accurate to a few meters, and (2) the GPS fixes arrive only three times per second, forcing us to average over ~0.7 s bracketing the hit, which is a window long enough that gravity, lift, and drag also might change the velocity. Multiplying that negligible Δv by the drone’s 0.69 kg mass gives a linear impulse |J| = 0.6 ± 2.6 N·s — indistinguishable from zero. The eagle imparted essentially no forward push.
The eagle’s impact on the drone shows up mostly as spin. The twisting shove, or angular impulse, equals the drone’s rotational inertia times its change in spin rate. Rotational inertia is how stubbornly a body resists being spun (the spin-equivalent of mass); like a figure skater spreading their arms to slow down, rotational inertia grows as mass sits farther from the axis. Tallying the eBee’s parts — the light foam-and-carbon wing spread across its ~1 m span, the heavy battery, camera and autopilot bunched at the center — we estimate about 0.022, 0.004 and 0.027 kg·m² about the roll, pitch and yaw axes (pitch smallest, since little mass lies fore-and-aft of the center; roll and yaw larger, set by the long wings). With the measured spin-up, the eagle’s twisting impulse was 0.03–0.13 N·m·s (we can’t resolve whether the talon flick/grab lasted <0.1 s or up to 0.3 s).
The violent +466°/s pitch spike at +0.5 s (Fig. 4A) is the drone ringing from the collision. Hit hard, a stable flying wing over-corrects and oscillates in pitch, swinging ±35° several times a second before damping out as it fights to point back into the wind. From there, the flight was a wobbly injured glide. The speed sag in Figure 4B over the next half-second is not the eagle slowing the drone. With the propeller knocked off its drive, there was no thrust, and with a wing dislodged and a bite of foam gone, the lift was degraded and lopsided.
Surprisingly few predatory strikes have been measured from the instruments embedded in a victim (i.e., the drone), which is what makes this dataset unusual. Published animal strike numbers fall into three regimes, and our wedgie sits in the rarely-quantified third:
1. Static grip and bite forces: For raptors, these forces are measured on a force transducer as the predator clamps down. Small falcons and kestrels grip with only a few newtons or low tens of newtons (Sustaita & Hertel 2010). Rigorous in-vivo figures for eagles the size of a wedgie are scarce.
2. High-speed collisions: The aviation bird-strike literature frames the bird as the unlucky projectile slamming into a fast aircraft. Certification rules (EASA CS-25 / FAA Part 25) require airliners to survive a 1 kg bird at cruise speed, and simulations of such impacts report peak forces of tens of kilonewtons over a few milliseconds (e.g., Nurrohmad and Ihsan, 2023). A 1 kg bird at an 80 m/s closing speed carries ~80 kg·m/s of momentum — so a collision that stops the bird delivers ~80 N·s, more than twenty times the upper bound on the linear impulse our eagle transferred.
3. Active aerial predatory strikes. A close cousin to our wedgie is the peregrine falcon, which dives at up to 88 m/s (317 km/hr) and hits prey with its closed talons like a club in a high-momentum collision (Mills et al. 2018). Our wedge-tailed eagle did the opposite: they matched the drone’s speed and then grabbed/raked, applying a light torque (0.1–1 N·m) similar to what you would need to turn a door knob or open a water bottle, which caused the drone to spin about as fast as a vinyl record.
Just three months after our scrape with the wedgie, numerous reports from Australia described similar perils. At one Western Australian gold mine, eagles destroyed a dozen mapping drones in two years, resulting in ~US$210k in losses (Kesteloo, 2017). Figure 5 illustrates one such wedgie strike, which resulted in a massive blunt-force impact that destroyed the drone. In comparison, our wedgie gave our drone a love tap/twist! Clearly wedgies have more than one mode of drone inspection.
For all of our subsequent field work in Australia, we have posted sentinels with binoculars in the four corners of the survey area, and devoted our entire attention to the drone while it flies. We also try to fly in the morning on overcast days when their are fewer shadows, and when the hot columns of air (thermals) that rise from the warming ground in the afternoon are not yet available for the eagles to ride. The minute someone spies a wedgie, the walkie-talkies start crackling. If an eagle gets within one kilometer of us, we rush the drone home for an emergency landing and pause until the wedgie gets bored and moves on. Perhaps no continent is as dangerous to drones as Australia.
