The running kinematics of free-roaming giraffes, measured using a low cost unmanned aerial vehicle (UAV)

1. What they did

Studying how large wild animals move is hard: they cannot be brought into a laboratory, and fitting them with sensors means catching or darting them. The authors tested whether a low-cost drone (a DJI Phantom 4 unmanned aerial vehicle, UAV, costing under £1,000) could film free-roaming giraffes from the side and measure their running gait without any physical contact.

• Filming (October 2016). At three sites in the Free State Province, South Africa, they filmed giraffes at 120 frames per second. At site 1, two tame giraffes followed a vehicle along a 200 m track marked with white paint every 2 m, and the vehicle’s speed was varied. At sites 2 and 3, wild giraffes were set running for short distances by the sound and closeness of the drone. Each giraffe was made to run no more than twice a day, for under a minute at a time, in cool parts of the day.
• Scaling the video. Because a flying camera’s distance from the animal keeps changing, every video frame had to be scaled using a body part of known length: the length from the fetlock (metatarsophalangeal) joint to the point of the heel (calcaneus) in the hind leg. They compared three ways of measuring it: (A) close-range photographs with a laser rangefinder; (B) the 2 m paint marks on the track; and (C) a 3D model of the terrain built from drone photographs, using ground features visible in both the model and the video.
• Checking accuracy. They filmed a person running on a treadmill, holding the drone 4 m or 8 m away, with the runner either in the centre or at the side of the picture, and compared the drone’s speed measurements with the treadmill’s.
• Analysing the gait. They digitised the giraffes’ legs, withers, neck and body by hand, frame by frame, using the coat pattern as natural markers, and measured speed, stride length, stride frequency, foot contact times, the timing between left and right feet, neck angle and body pitch. They compared the results with what is predicted for other four-legged mammals.

2. What they found

The method

• Close-range photography (method A) gave the most consistent measurements. The track markers (method B) and the terrain model with artificial markers each differed from it by 3.7%; the terrain model with natural ground features differed by about 5%, with more variation.
• Method A was used for site 1, and the terrain model with natural features for sites 2 and 3.
• On the treadmill, the drone always measured lower speeds than the treadmill: 13% lower on average. The error was smallest at 8 m (8%) and largest at 4 m with the runner off centre (17%). Distance had the biggest effect.

How giraffes run

• Over 50 stride sequences were filmed; 25 strides from four similarly sized adults were analysed.
• Running speeds ranged from 3.4 to 6.9 m/s (about 12 to 25 km/h), with a mean of 5 m/s. Below this range, giraffes walked.
• The adults always used a rotary gallop and never trotted, paced or cantered. Their gallop was “grounded”: at least one foot was always on the ground. Brief moments with all feet in the air were seen only in young giraffes, which were not analysed.
• To go faster, giraffes mainly lengthened their stride (by about 0.5 m for every 1 m/s faster), while stride frequency rose only a little (0.05 strides per second for every 1 m/s).
• The forefeet stayed on the ground longer than the hind feet. Ground contact time fell with speed, but the time each foot spent swinging through the air did not change.
• Giraffes kept their feet on the ground for a smaller share of each stride than predicted for other mammals running at a comparable speed, though their relative stride length matched the prediction.
• As speed increased, the left and right feet of each pair landed closer together in time.
• The body pitched up and down once per stride in a smooth wave. The neck also swung once per stride but more irregularly, and its movement was often out of step with the body (mean congruity 70%).

3. What they concluded

• A low-cost drone can measure the running gait of free-roaming animals without touching them. To reduce error, users should fly as far away as practical and keep the animal in the centre box of a three by three grid over the picture.
• A terrain model can be used to scale video when photographs or ground markers are not possible, but adds about 5% error.
• Giraffes are “slow gallopers”. They go straight from a walk to a gallop at about 3.4 m/s, close to the speed at which other mammals of their size would start to trot.
• Their short foot contact times suggest giraffes might face high forces on their legs when galloping. The authors speculate that landing left and right feet closer together at higher speeds, and a modest top speed, may keep the stresses on their slender legs within safe limits.
• The neck may be held steady in space and partly separated from the pitching of the body, possibly because of the great weight of the head and neck.
• These ideas need testing with further field studies and computer models.

6. Other technologies, skills or innovations

• DJI Phantom 4 drone filming at 120 frames per second (1920 × 1080 pixels), with a three-axis stabilised camera mount.
• Close-range photographs (Canon EOS 500D) with a laser rangefinder (Leica Disto D5) to measure the giraffes’ leg bones.
• 3D textured terrain models built from GPS-stamped drone photographs taken at 40 m height (Pix4D software).
• Hand digitising of video in DLTDV6 for Matlab, using the giraffes’ own coat patterns as markers; filtering and analysis in Matlab.
• WebPlotDigitizer to read published data from earlier figures.

7. Implications for conservation (from the paper)

• Because the giraffes were filmed in their natural habitat, the results can be applied with more confidence to wild giraffes than results from a laboratory.
• Filming from a longer distance is also preferred because it reduces potential stress to the animals.
• Terrain models from the same drone could in future be used to study how slope, ground type and sharp turns affect giraffes’ athletic ability and energy costs.