Combining milking machine and accelerometer data to indicate the level of cow comfort on different milking machine settings.


Increased data routinely collected by modern milking machine sensors and software has occurred together with increased cow attached sensor data, usually applied for fertility and health monitoring. To increase milking efficiency and gentle, quick and complete milking, it makes sense to investigate how various milking machine settings will impact on gentleness of milking through a proxy measurement of cow comfort during milking. Therefore, the use of leg-mounted accelerometers was investigated by the authors cited, as a non-invasive labour efficient means of estimating cow comfort on different automatic cluster remover (ACR) milk flow-rate switch-point settings.

Forty-four cows were included in the study. They were milked with two different ACR milk flow-rate switch-points which were pre-programmed into the milking software, so that when the cows entered the milking parlour, the designated ACR setting automatically kicked in. Half of the cows were milked with an ACR setting of 0.2 kg per minute and the other half 0.8 kg per minute. After two weeks, cows were crossed over to the other ACR setting for a further two weeks. Each cow had a three-dimensional accelerometer attached to a rear leg for the duration of the experiment. This measured leg movement and included a step count feature that could detect leg movements that did not necessarily involve forward movement. The time of each cluster attachment and the duration of milking was logged by the milking equipment. This enabled leg movement data from the relevant time period to be extracted from the downloaded accelerometer data.

The results showed that significantly more rear leg stepping, as indicator of cow comfort, occurred during daily milking when the ACR activated at 0.2 kg per minute (11.7 steps) compared with 0.8 kg per minute (10.1 steps). With a 16 to 8 hour AM to PM milking interval, the shorter milking interval resulted in lower udder fill, which resulted in reduced rear leg movement when the milk flow-rate switch-point for cluster removal increased from 0.2 kg per minute (5.75 steps) to 0.8 kg per minute (4.96 steps). There was however no significant difference in stepping rates between cluster removal settings during AM milkings.

The 0.2 kg per minute ACR setting which resulted in greater leg movement, extended total milking time by 70 seconds, and resulted in lower mean flow-rates while producing similar milk yield and SCC. The lower flow-rates corresponded with higher vacuum levels at the teat-end recorded on this milking setting, potentially impacting cow comfort. This provides further incentive to consider cluster removal settings above 0.2 kg per minute.