Development of a multi-sensor and multi-application device for monitoring indoor and outdoor sheep behaviour
Résumé
In ruminant behavioural research such as the investigation of feeding and social
behaviours in the barn or at pasture, limits in data recording are based on (1) the
difficulties both in simultaneously recording several different variables on a given
animal and a great number of data for each one, (2) the interference between animal
behaviour and the presence of the observer and finally (3) the time-consuming aspect of
data acquisition and analysis.
Our objective was thus (1) to develop a recording device, either embedded on the
animal or installed in fixed positions in the barn, capable of synchronising multiple
sensors to characterise sheep behaviour and physiology parameters in their feeding and
social environments, and (2) to design algorithms to automatically process these signals.
In collaboration with the Effidence start-up specialised in robotics, we designed a
monitoring platform integrating a set of sensors able to record animal tracking (GPS,
rangefinder), feeding activity (accelerometer, gyroscope, video camera, microphone)
and emotional reactivity (heart rate sensor). Software was developed to allow for
synchronous visualisation and real-time processing of the signals, and to extract
characters of biological interest (e.g. time-budget) from automatically detectable states.
This upgradable and adaptable plug-and-play device, designed to allow for changes in
sensors according to research requirements, constitutes a promising experimental tool to
develop new PLF (Precision Livestock Farming) means/resources which now requires
further developments in terms of miniaturization and validation.