Meeting Abstract
GPS measurement of position and speed is an important tool for animal locomotion and behaviour research in the field. A major limitation of frequent GPS localisation is the high power requirement of the receiver. Tracking collars for deployment on wild animals are typically designed to last months or years to avoid the need for frequent battery replacement. The tight power budget restricts the number of GPS measurements that can be made, giving only a limited picture of an animal’s movement, speed and locomotor repertoire. Accelerometers, gyroscopes based on MEMS (micro-electromechanical systems) technology, and magnetometers operate at much lower power so that measurements can be made more frequently. Here we investigated how data from MEMS sensors can be used to interpolate between infrequent GPS measurements, and hence give an effective improvement in temporal resolution from a collar system deployed on quadrupeds. This was assessed using data from custom-built collars deployed on carnivores in the wild, and over short test periods on dogs and horses in controlled conditions of high and low speed locomotion. Each collar recorded data from a triaxial accelerometer, triaxial gyroscope and triaxial magnetometer. Reference speeds and positions were obtained from GPS measurements, recorded at maximum rate in short tests. Different approaches were evaluated to estimate speed, distance travelled, heading and hence dead reckoned change in position from accelerometer and magnetometer data, for comparison with the reference data. Automatic algorithm tuning and smoothing between infrequent GPS measurements was also investigated and validated with high rate GPS observations.
