Automatic multi-weigh-station for assessing sheep liveweight in small flocks

Abstract

Sheep liveweight is a vital physiological parameter to monitor as it is directly correlated with the animals internal energy balance. Increases in liveweight in mature sheep coincide with an increase in fat and protein tissue while weight loss denotes an inverse relationship. This relationship underscores appropriate pasture stocking rates, production outputs, the correct dosing of medicinal treatments for maintaining a healthy flock and finding target markets. The sheep production industry is yearning for an accurate platform to provide real-time and accurate liveweight measurements without extensive labour. The two main methods for measuring sheep liveweight include; static yard weighting, which while accurate is labour intensive, and walk-over-weighing — which has the advantage of autonomy but lacks accuracy and requires the flock to be trained. In this work we describe a custom platform that accurately measures the liveweight of up-to three sheep simultaneously while they drink or eat from a field trough. A plywood frame consisting of three separated static weigh platforms — each with two commercially available weigh bars — was constructed to fit along side a conventional 3 metre trough. These strain gauges plug directly into a custom circuit board with a Teensy 3.6 microcontroller connected to several Analog-to-Digital converters, SDI-12, I2C and OneWire interfaces for additional environmental/agricultural sensors — all powered by a commonly available 12 V lead–acid battery and commercial solar charger. The device can communicate over LoRaWAN — a long range, low-power wireless communication architecture — making it suitable for remote deployment on large farms where traditional 3/4G telecommunications are unavailable. Firmware on the device sends device status updates every 15 min and calculates/transmits liveweight parameters in real-time — as animals use the trough. Five untrained mature wethers were given access to the device over a period of 57 days. A total of 1149 weights were captured, of that 853 were automatically flagged as reliable using algorithms provided in this work — equating to ∼74% of the original dataset. This work aims to enable producers to a quantify the physiological state of their flock without extensive labour, acting to increase their available decision making information in regards to flock productivity, stocking rates and animal welfare.