Automatic milking systems (AMS), one of the earliest precision livestock farming developments, have revolutionized dairy farming around the world. While robots control the milking process, there have also been numerous changes to how the whole farm system is managed. Milking is no longer performed in defined sessions; rather, the cow can now choose when to be milked in AMS, allowing milking to be distributed throughout a 24 h period. Despite this ability, there has been little attention given to milking robot utilization across 24 h. In order to formulate relevant research questions and improve farm AMS management there is a need to determine the current knowledge gaps regarding the distribution of robot utilization. Feed, animal and management factors and their interplay on levels of milking robot utilization across 24 h for both indoor and pasture-based systems are here reviewed. The impact of the timing, type and quantity of feed offered and their interaction with the distance of feed from the parlour; herd social dynamics, climate and various other management factors on robot utilization through 24 h are provided. This novel review draws together both the opportunities and challenges that exist for farm management to use these factors to improved system efficiency and those that exist for further research.Keywords: automatic milking system, feeding behaviour, robot idle time, grazing, PLF (precision livestock farming) ImplicationsMilking robots have revolutionized the dairy industry with farmers achieving high levels of robot utilization obtaining greater returns on asset. The first installations were typically associated with 'indoor' systems and nearby grazing fields. Nowadays there is an increasing interest regarding the integration of robots into larger scale pasture-based dairy systems. This review explores the published literature on both 'indoor' and 'pasture-based' dairy systems in relation to milking robot utilization. IntroductionRobotic milking systems have revolutionized the dairy industry. The first dairy cow was milked, more or less without traditional human involvement, in 1986 with a robotic milking box at the experimental farm de Waiboerhoeve, Lelystad, the Netherlands by Gascoigne Melotte, following the US Patent 4010714A (Notsuki and Ueno, 1977). A system from the company Prolion was installed on the experimental farm IMAG-DLO Duiven, the Netherlands, in 1990 and on a commercial dairy in 1992. More institutes and companies became active in the development of robotic milking systems in the nineties as described by Kuipers and Rossing (1996). Since that time until 2011, automatic milking systems (AMS) have been installed on over 10 000 farms worldwide (de Koning, 2011). These installations are predominantly for 'indoor' systems where cows are generally 'housed' in barns and offered a partial mixed ration (PMR) in the feeding alley and grain-based concentrate supplement either in the milking unit or in a nearby concentrate self-feeder. While there have been numerous AMS installations in indoo...

Hot weather is known to negatively affect cow performance primarily through reduced feed intake and milk yield. However, little information is available on how it affects cow milk yield and milking frequency in automatic milking systems (AMS). Milking data were collected from 6 pasture-based AMS farms in Australia to assess the effect of temperature-humidity index (THI) on milk yield and milking frequency. Daily measures of average milk yield per cow and average milking frequency per cow during December to February (Australian summer) were assessed for associations with maximum, minimum, and average THI from d 0, -1, -2, and -3 in relation to the milking data. Average daily milk yield per cow was negatively associated with an increasing maximum, minimum, and average THI (-0.11, -0.08, and -0.15 kg/THI unit increase, respectively) on the collection day and up to 3 d prior. The average daily milking frequency was negatively associated with maximum THI on 1 d (-0.003/THI unit increase) and 2 d (-0.003/THI unit increase) before collection. Our results show that high THI conditions were negatively associated with milking frequency and milk yield in a pasture-based AMS and that research into management and infrastructure (cow cooling) in these systems is warranted to reduce production losses.

The diurnal variation in pasture nutritive value adds a confounding factor to studies elucidating the effect of time of day on behavior. Our work separates the effect of time of day on both feeding and lying patterns for cows outdoors to enable the alignment of feeding behavior with feed management. We determined the diurnal intake patterns and behavior of dairy cows when the nutritive value of feed remained constant throughout 24 h in an outdoor environment, and when feed access was restricted. Nine nonlactating Holstein-Friesian cows (live weight 626 ± 53 kg, age 96 ± 33 mo; mean ± SD) were split into 3 groups of 3 and offered lucerne hay cubes (cube volume 32 mm) ad libitum according to 3 treatments: full access (FA, feed access 24 h), day access (DA, feed access between 0600 and 1800 h), and night access (NA, feed access between 1800 and 0600 h). Treatments were applied to individual cows in a crossover design with 7-d periods. During the last 4 d of each period, data were collected on feed intake, as well as feeding and lying behaviors. Total daily intake was greater for cows on the FA treatment (3.5% of BW) compared with the DA and NA treatments at 3.1 and 2.9% of BW, respectively. The cows with FA consumed 69% of their total intake during the day (0600-1800 h), with the greatest intake (39%) occurring during 1200 to 1800 h and only 12% of intake occurring during 2400 to 0600 h. Cows with DA consumed 56% of feed during 0600 to 1200 h and 44% during 1200 to 1800 h. In contrast, NA cows consumed more feed (74%) during the first 6 h period (1800-2400 h), thus maximizing lying time between 2400 and 0600 h. The time spent lying throughout daylight periods varied between treatments; however, total daily lying time was similar across the 3 treatments. This experiment shows the feeding and lying behaviors of cows when feed quality remains constant throughout 24 h, which will assist the formulation of variable feed allocation strategies for future testing in both robotic and conventional milking systems. Varying the quantity of feed offered throughout 24 h may benefit robot utilization at night in automatic milking systems through increased feeding activity, and as we observed, is likely to have little effect on lying time or DMI, with cows readily adapting to changes in feed management. Conversely, aligning feed on offer with preferred feeding time in conventional milking systems may increase the intake of high quality pasture.

In pasture-based automatic milking systems (AMS), a decrease in robot utilization (RU) often occurs in the early morning hours. Novel feeding strategies that encourage voluntary cow traffic throughout 24 h could help mitigate this problem. We determined the effect of 3 distinct pasture allocation methods on RU patterns throughout a 24-h period. The experiment was conducted at the University of Melbourne's Dookie research farm in northern Victoria, Australia. Three Lely Astronaut A3 robotic milking units (Lely, Maassluis, the Netherlands) milked 133 cows, grazing pasture, with concentrate offered at milking in the robots. The farm operated a system of 3-way grazing, with active access to each pasture allocation: 2030-0400 h (allocation A), 0400-1330 h (allocation B), and 1330-2030 h (allocation C). Treatments varied in the quantity of feed offered per hour of active access to each of the 3 pasture allocations. The control treatment offered the same proportion of feed (corrected for active access time) in all 3 pasture allocations (allocation A = 31.3%, B = 39.6%, and C = 29.2%). The day treatment offered the largest proportion of feed during the day (allocation A = 20%, B = 40%, and C = 40%), following the cows' diurnal pattern of feeding activity. The night treatment offered the largest proportion of feed at night (allocation A = 42%, B = 40%, and C = 18%). Due to the nature of pasture-based AMS, treatments could not be applied simultaneously. Therefore, treatments were applied to the entire herd and repeated twice over 42 d, lasting 7 d/treatment, with the first 3 d for habituation, followed by 4 d of data collection. Robot utilization (milkings/h) varied throughout 24 h between treatments, with the night treatment recording greater RU at 0800, 1800, and 1900 h and lower RU between 2100 to 0100 h, compared with the day treatment. The proportion of the herd milking between 0000 and 0600 h was greater for the control (43.3%) and day (45.3%) treatments compared with the night treatment (25.8%). Herd-average daily pasture intake was similar (10.5 kg of dry matter) for all treatments. This experiment is the first to demonstrate the manipulation of RU by varying the quantity of pasture offered. However, the use of variable allocation alone did not eliminate the decrease in RU between 0000 and 0600 h, with the timing of allocation also likely to play a role. We recommend a further research focus on combining both timing and quantity of pasture allocated to improve RU in pasture-based AMS.

The diurnal feeding patterns of dairy cows affects the 24 h robot utilisation of pasture-based automatic milking systems (AMS). A decline in robot utilisation between 2400 and 0600 h currently occurs in pasture-based AMS, as cow feeding activity is greatly reduced during this time. Here, we investigate the effect of a temporal variation in feed quality and quantity on cow feeding behaviour between 2400 and 0600 h as a potential tool to increase voluntary cow trafficking in an AMS at night. The day was allocated into four equal feeding periods (0600 to 1200, 1200 to 1800, 1800 to 2400 and 2400 to 0600 h). Lucerne hay cubes (CP = 19.1%, water soluble carbohydrate = 3.8%) and oat, ryegrass and clover hay cubes with 20% molasses (CP = 11.8%, water soluble carbohydrate = 10.7%) were offered as the ‘standard’ and ‘preferred’ (preference determined previously) feed types, respectively. The four treatments were (1) standard feed offered ad libitum (AL) throughout 24 h; (2) as per AL, with preferred feed replacing standard feed between 2400 and 0600 h (AL + P); (3) standard feed offered at a restricted rate, with quantity varying between each feeding period (20:10:30:60%, respectively) as a proportion of the (previously) measured daily ad libitum intake (VA); (4) as per VA, with preferred feed replacing standard feed between 2400 and 0600 h (VA + P). Eight non-lactating dairy cows were used in a 4 × 4 Latin square design. During each experimental period, treatment cows were fed for 7 days, including 3 days habituation and 4 days data collection. Total daily intake was approximately 8% greater (P < 0.001) for the AL and AL + P treatments (23.1 and 22.9 kg DM/cow) as compared with the VA and VA + P treatments (21.6 and 20.9 kg DM/cow). The AL + P and VA treatments had 21% and 90% greater (P < 0.001) dry matter intake (DMI) between 2400 and 0600 h, respectively, compared with the AL treatment. In contrast, the VA + P treatment had similar DMI to the VA treatment. Our experiment shows ability to increase cow feeding activity at night by varying feed type and quantity, though it is possible that a penalty to total DMI may occur using VA. Further research is required to determine if the implementation of variable feed allocation on pasture-based AMS farms is likely to improve milking robot utilisation by increasing cow feeding activity at night.