Automatic monitoring of primate locomotor behaviour using accelerometers

Accelerometry data were transmitted by a radio collar attached to a hand-reared red-ruffed lemur housed in a large indoor/outdoor enclosure at Chester Zoo. An observer simultaneously recorded locomotor behaviour using a manually operated event recorder. Both data streams were recorded directly to hard disk to ensure accurate synchrony. Leaps were modelled using a y = x2 - x3 formulation for the take-off acceleration, to link peak acceleration to leap distance. Cyclic locomotor modes were analysed using power spectra and the modal frequency used to estimate stride periodicity. Comparison of the dual data shows that leaping behaviour can be recorded reliably, and acceleration magnitude provides accurate predictions of the distance travelled. Cyclic activities were less well characterised, but calibration should permit travel distance estimations equalling or bettering those from conventional techniques.