Effects of continuous drone herding on behavioral response and spatial distribution of grazing cattle

Abstract

Robotic herding is expected to become a new technology for the management of grazing animals. We investigated the behavioral responses of cows to drone herding on consecutive days and the potential of this technology to manipulate grazing distribution in a pasture. A herd of ∼30 cows was stocked in a 1.1-ha pasture for 5 consecutive days each month from May to October 2022. The cows were herded using a drone for 10 days during the grazing period in August (days 1–5) and September (days 6–10). The pasture was divided into nine plots, with two and three plots assigned as the ‘herding area’ in August and September, respectively. When the cows were grazing in the herding area, the operator maneuvered the drone to move the animals out of the area. The drone first approached at an altitude of 10 m; if the cows did not move away, the altitude was gradually lowered to 3 m. The behavioral responses to the drone, success or failure of herding (whether the cows left the area or not), and drone altitude were classified based on videos recorded during herding. The behavior and location of the cows in the pasture were observed during this period. The utilization rates of the plots (percentage of grazing time) were calculated to evaluate the effects of herding and the number of herding days. The cows responded in 59% and 46% of the cases on days 1 and 2, respectively. Startle behavior was recorded in 23% of the responses on day 1. Herding was successful in 51–75% of the cases on days 1–3. However, these percentages decreased thereafter. The utilization rate of the herding area on day 1 was approximately half that on days without herding. It increased from day 2, and almost no effect on grazing distribution was observed on or after day 3. These findings suggest that the cows exhibited little fear of drones after 1-day herding, and responses to the drones decreased as the animals subsequently became habituated to frequently repeated stimuli. Even after a 24-day interval, responsiveness to drone herding barely recovered and then declined. Therefore, drone herding was hardly effective in manipulating grazing distribution. Further investigation of stimulation that persistently elicits desirable responses is needed for future applications in technologies for the autonomous control of animal movement.