Animal behaviour analysis with GPS and 3D accelerometers

A herd of dairy cows were equipped with GPS tracking collars and at the same time, their behaviour was manually scored with Pocket Observer software. TrackLab was used to visualize the data. The manually scored behaviours were used to classify the GPS data, and for foraging, resting and walking, the GPS data had a very high predictive value for the behaviours. Although ruminating and standing could not be distinguished on the basis of GPS data alone, a further experiment on Canada Geese indicated that the addition of accelerometer data to the GPS tags showed very promising results with respect to distinguishing more behaviours than could be classified using GPS alone. This opens up a spectrum of possibilities for farm mangers including automatic detection oestrus in cattle and geofencing applications.