See global airing times

The drones watching over cattle where cowboys cannot reach


Researchers in Kentucky and Colorado are training drones to find cattle and assess their health from the air.

For many cattle producers in Kentucky, 5pm isn’t the end of the workday – rather, it marks the transition from their non-farming day jobs to working with their livestock. But putting out hay, checking water tank levels, locating cattle and assessing herd health becomes more difficult after the Sun sets, and sick or calving cows make the chores trickier as they often hide themselves in secluded areas.

Time to call in the drones.

A team of researchers at the University of Kentucky are testing the feasibility of using unmanned aerial vehicles (UAVs) to remotely handle those assessments for small-herd cattle producers. “The idea is to make it easier for them to have this second form of income, this second job, by automating some of it and helping keep an eye on their cattle without them having to be there to do it all the time,” says Jesse Hoagg, a professor of mechanical engineering at the University of Kentucky.

Cattle production is “the most important agricultural industry” in the country, contributing $66bn (£50bn) to the US economy, according to the United States Department of Agriculture (USDA). But it’s been challenged by climate change, a declining labour supply and land development. Making the ranching industry more efficient through smart technology can help counteract many of those difficulties.

A drone-eye view of a farmyard (Credit: Reuters)

The weather in Colorado makes finding cattle in winter an unpleasant job (Credit: Reuters)

The ability to accurately and quickly monitor cattle location and wellbeing is also a time-saver for busy ranchers – one with important financial implications. Every year, more than 2.5 million cows in the US die from health problems, costing the cattle industry $1.5bn (£1.1bn). Sick cattle also eat and move less, stunting calves’ growth, impacting their ability to reproduce and reducing their value over time.

Hoagg and his team received a three-year, nearly $900,000 (£670,000) grant from the USDA National Institute of Food and Agriculture in February 2018 to pursue its drone project. The multidisciplinary effort is hoping to solve multiple pain points for producers – patrolling pastures to locate cattle, and monitoring health indicators such as weight and visible ailments like pink eye. To make this possible, they need to find a way to identify and track specific cows. Hoagg is turning to cow facial recognition.

The team is testing computer algorithms to make that last feat possible, as current facial identification technology is built to recognise human faces, not bovines. Features like noses, for example, can serve as cow fingerprints, though something as basic as dirt can throw off the software.

“A group of UAVs will collaboratively take images of cattle from different angles all at once,” says Hoagg. “We can stitch together a 3D model to estimate volume and weight, and track weight gain or loss. They also take videos to monitor physical activity levels.”

Ear tags (the analogue method of identifying specific animals) are unreliable as they tend to fall off, though the team is looking to them as a fallback. Another option, electromagnetic RFID tags, are costly and some require the drones getting closer to the animals than is safe.

Current regulations require drone operators to have visual lines of sight to the machines, but the University of Kentucky team’s technology will be capable of operating autonomously, with one observer UAV hovering above the field and determining which objects are cows and transmitting their locations to a number of “worker” UAVs. These worker UAVs are fitted with cameras that collect images and data as they fly. The eventual plan is for the machines to analyse the information they collect and send real-time alerts to producers when something looks off, like spotting a sickly cow or a downed fence.

A drone hovers a few feet away from a cow (Credit: Hoagg/Jackson/Sama/Yang)

The cows seem unfazed by the drones, but research is ongoing (Credit: Hoagg/Jackson/Sama/Yang)

To ensure the system is safe for the cattle (unusual sights and noises can stress cows and distract them from eating), the researchers are studying the animals’ physiological and behavioural responses to the drones by testing various flight patterns and heights at the university’s 1,484-acre (6 sq km) farm. The goal is to capture the highest quality images while not upsetting the cattle, so the team are using heart rate monitors and GPS collars to measure the animals' responses to the drones.

“It’s pretty promising. We’re able to get closer than I’d expect,” says Michael Sama, a biosystems engineering professor at the University of Kentucky. “[We can be] 20 or 30 feet away and it doesn’t affect them in any way whatsoever.” The animals also appear to become acclimated to machines flying at these altitudes over time. But at a height of around 8-12ft (2.4-3.7m), the UAVs are able to compel the cows to move so they can be more easily rounded up, another facet the team is beginning to study.

In order to accomplish this ambitious work, the researchers need a lot of data, which is difficult to collect. “You can’t just go online and say, ‘Give me a million pictures of cows.’ We have to generate that dataset,” Sama says. To build up that catalogue, they’ve crowdsourced facial images and installed dozens of cameras at a university farm that can capture up to 100 photos each time a cow runs by. “We can take images of cattle from all different directions simultaneously,” Hoagg explains. “It helps us train the algorithm that will reconstruct the 3D model of the cow from a relatively sparse set of images.”

Milt Thomas, an animal breeding professor at Colorado State University (CSU) ran into a similar data collection problem – and also looked to drones for help.

Researchers in the US mountain states have spent years studying whether genetics affect the grazing patterns of cattle (short answer: they do) and if genetic selection could help better distribute the animals across the expansive landscapes that define the ranches of the American west. “There are some cows that are lazy and lay around the water and overgraze the grass around the water,” explains Thomas. “There are other cows that like to get up every morning and go for a long hike, go up the side of the mountains.” They refer to these groups as bottom dwellers and hilltoppers, respectively.

A 3D cow render using thousands of images taken by drones (Credit: Hoagg/Jackson/Sama/Yang)

This graphic is created by combining drone images of a cow to estimate its volume and weight (Credit: Hoagg/Jackson/Sama/Yang)

Often, these grazing patterns have been assessed using GPS collars, a time-consuming and labour-intensive method, particularly in far-reaching pastures. For genetic studies, researchers also need to be able to distinguish individual cows. “If we could come up with a way to count [and recognise] all the cows in a pasture at the same time, it would improve our ability to [identify the true genetic markers]. That’s why we’re so interested in drones,” says Derek Bailey, a professor of range sciences at New Mexico State University who has been tracking cows since 1998 and has worked with the CSU team. He used to ride through pastures in Montana on horseback – it would take him and two others about two hours to track 80% of the 180 cows. A UAV can fly over the same area multiple times a day and collect significantly more data, much faster.

Where and how much cows graze has a significant impact on the sustainability of the land. Overgrazing can lead to soil erosion, weed infestation, poor water quality and the introduction of invasive species, while undergrazing builds up fuel for wildfires. “[This technology will] improve the sustainability of grazing especially in sensitive areas up in the mountains. It’s a big deal in Colorado, Wyoming, New Mexico – all over the American west,” Bailey says. Grazing decisions also impact the animals, whose nutrition improves with a more diverse diet. And if ranchers can spread out their cattle and encourage them to graze more terrain, they can potentially add additional cows without hurting the land.

Cattle that move around to seek a more diverse diet “always last longer in our herd than cattle that do not”, says Nick Wamsley, assistant manager at Silver Spur Ranch in Wyoming. The CSU crew put GPS collars on some of Silver Spur’s cattle to gather data on grazing patterns. “Out here in the arid west, it can be very touch-and-go as to what the year may bring in terms of moisture, in terms of forage production,” Wamsley adds. “Ideally, we would have cows that would be able to get out, move around and find different forage sources, and do it efficiently.”

Though some of the research goals and questions are the same in Colorado and Kentucky, the end goal for CSU is different. They want to “develop a selection tool for breeders to produce cows that are more environmentally adapted”, or have less of an impact on the land, says Mark Enns, a professor of animal sciences. “It’s something we do for a lot of traits – meat quality and quantity, how fast animals grow, reproductive ability. This would be something added onto that…”

Cattle stand in a vast area of wilderness (Credit: Hoagg/Jackson/Sama/Yang

Ranchers have vast areas of land to cover in search of their cattle (Credit: Hoagg/Jackson/Sama/Yang)

Enns and his colleagues plan to wrap up some of this work in the next couple of years. The University of Kentucky’s crew is on a similar timeline. Questions around approach paths, flight formations and optimal angles for photos remain for them – some of which are being tested in a controlled, on-campus environment using a calf statue named Chuck – but the team is focused on integrating the various systems they’ve created over the coming months.

Josh Jackson, who raises a herd of about 70 cattle when he’s not working as the University of Kentucky’s livestock system extension specialist, expects much of that research to be published in the next two to three years. But the commercialisation of this technology may still be as long as a decade away, in part due to current drone regulations, costs and how quickly producers and ranchers come on board.

Still, the research looks promising. And the long-term vision certainly sounds appealing: Ranchers (or hired consultants) could, with the push of a button – a “big red button that says ‘fly’”, in Jackson’s mind – have cattle locations and urgent alerts sent directly to their phones or tablets, making second shifts for the 60% of small-herd producers who still work two jobs a whole lot easier.

This article is part of a multimedia series Follow the Food by BBC Future and BBC World News. Now in its second series, Follow the Food investigates how agriculture is responding to the profound challenges of climate change, environmental degradation, rapidly growing populations and the Covid-19 pandemic, which has brought new challenges to our global food supply chains. Follow the Food traces emerging answers to these problems – both high-tech and low-tech, local and global – from farmers, growers and researchers across six continents.