The behaviour of dairy cows in an automatic milking system where selection for milking takes place in the milking stalls

An automatic milking system (AMS) enables cows to be milked by a milking robot at fairly regular intervals, more or less voluntarily, and without human supervision. To reduce the cost and simplify the operation of an AMS for dairy farmers, the selection stall, where the system's computer decides whether cows are to be milked or not, was removed and the selection process was transferred to the milking stalls. The behaviour of 18 dairy cows in study 1 and 24 dairy cows in study 2 was investigated in a section of a loose housing system in which such an AMS was installed. In study 1, 18 cows had free access to the AMS from 0500 to 2300 hours. Per day, each cow paid on average 2.8 (SD=0.2) milking visits (cow was milked), 2.0 (SD=1.4) non-milking visits (cow was not milked), and 0.3 (SD=0.4) milking-failure visits (robot failed to attach milking cluster) to the AMS. In study 2, 24 cows had free access to the AMS 24 h a day. Each cow paid on average 2.8 (SD=0.4) milking visits, 2.5 (SD=2.0) non-milking visits, and 0.7 (SD=1.7) milking-failure visits to the AMS daily. In both studies, the type of visit a cow paid to the AMS determined the time interval until her next visit. In study 1, time intervals were: 321 min (SEM=12) after milking visits, 145 min (SEM=16) after non-milking visits, 115 min (SEM=10) after milking-failure visits. In study 2, time intervals were: 337 min (SEM=14) after milking visits, 117 min (SEM=17) after non-milking visits, 164 min (SEM=13) after milking-failure visits. In both studies, cows spent the longest time in the entry area of the AMS during non-milking visits (29 s (SEM=5) study 1; 33 s (SEM=3) study 2). Cows left the milking stalls slowest during non-milking visits in study 1 (10 s, SEM=1), and during milking-failure visits in study 2 (18 s, SEM=5). They spent the longest time in the exit area of the AMS after non-milking visits in study 1 (202 s, SEM=23), and after milking-failure visits in study 2 (141 s, SEM=28). In both studies, they defecated and urinated in the exit area. It was concluded that cows that do not need to be milked will enter the AMS if there is no selection stall to prevent unwarranted visits, and that cow traffic inside the system is slow. The combination of non-milking visits and slow cow traffic will lower the milking efficiency of the AMS.