SmartLivestock Welfare Framework for Continuous Behavioral and Health Monitoring of Farm Animals

Abstract

Continuous welfare assessment on commercial farms is still limited by infrequent and labour-intensive inspection times, which miss the dynamic changes in behaviour and physiology leading up to clinical signs of disease or distress. The SmartLivestock Welfare Framework (SLWF) is introduced which is a four-layer architecture, where a wearable and ambient sensor is combined with an edge device for feature extraction, cloud-based behaviour classification and a decision support system for farmers to continuously monitor the wellbeing status of cattle, sheep and swine without being invasive. These streams of information (tri-axial accelerometer, RFID, acoustics, and thermal imaging) are combined in a Welfare Index to indicate deviations from normal lying behaviour, feeding time, locomotion, and vocalisations that relate to lameness, mastitis, heat stress and parturition. The data flow, latency budgets and alerting logic throughout the architecture is illustrated using a prototype deployment scenario. The framework is placed into the context of the current precision livestock farming (PLF) literature and some barriers are identified, such as the need for validation, the cost of the sensors and the need to standardise the data before it can be adopted on a farm. It proposes a repeatable reference design that can be used by research groups and technology suppliers interested in bridging the gap between research and deployable welfare-monitoring systems.