Facial recognition for cows: Drones on the farm

Researchers at the University of Kentucky are developing an autonomous drone system to monitor cattle health in pasture. The drones will capture indicators like heart rate, body temperature and weight. 

In the basement of the mechanical engineering building on campus, a team is testing automated drones flying in formation around a model cow (named Chuck). 

The drones work in a set of four. Three worker drones hover around the cow, while a fourth observer drone uses its cameras to locate each worker drone. 

In the lab, the worker drone is simulated by cameras perched at the top of the wall. Those cameras triangulate cow and drone locations via retro-reflective markers. 

These modified 3DR Solo worker drones are labeled and equipped with gray markers so the observer cameras can track them. 

A controller keys in simple commands to initiate drone flight phases. Other than that, all the flying is autonomous. 

In order to recognize individual cattle in the field, the team needs to train the drone software with 3D models built from photos of real cows. 

This special pen is equipped with 40 cameras that take a simultaneous photo from 40 different angles to build a 3D model. 

The next phase of testing is flying drones around cattle in the field, to gauge cattle reactions and any possible stressors. 

This observer drone flies above the worker drones to capture location and image data. 

The drones in these field tests are controlled manually and flown around the cattle for 5-10 minutes per flight test.

A worker drone prepares to fly into the field to observe a small herd.

The worker drone lands on a special landing pad at command center. 

Autonomous health monitoring is just one way technology like drones can improve the efficiency and quality of cattle health monitoring.