week NINE: Robotic Milking: Current Situation

week NINE: Robotic Milking: Current Situation

Douglas J. Reinemann
University of Wisconsin
Madison, Wisconsin, USA


The population of farms using automatic milking systems (AMS) has grown from the first
installation a dairy farm in the Netherlands in 1992 to more than 8000 milking units on more
than 2400 farms today. The vast majority of these farms are in north-western Europe, with the
Netherlands having the largest installed base and Scandinavia showing the fastest growth rate in
the past few years (Lind, 2007).

Regulation and Standards

The International Standards Organization (ISO) created a standard for automatic milking
installations (AMI), which was approved in 2007. The machine milking group of the
International Dairy Federation (IDF) suggested that, while many aspects of AMI are different
than conventional milking machines, many aspects are also shared. It was decided to
simultaneously create a new standard for AMI containing those aspects unique to this new
technology and to revise the milking machine standards to update them and revise them to allow
seamless reference from the AMI standard. This International standard specifies requirements for
construction including specific safety and hygiene aspects and minimum performance
requirements and testing for AMI. It does not contain requirements for the design of the building
in which the milking installation is installed.

Appendix Q to the Pasteurized Milk Ordinance (PMO) governing the Operation of Automatic
Milking Installations (AMI) for the Production of Grade “A” Raw Milk for Pasteurization was
approved at the 2003 meeting of the National Conference on Interstate Milk Shipments.
Highlights of this apprendix are as follows:

• AMI shall have the capability to identify and discard milk from animals that are
producing milk with abnormalities.
• Odor is currently evaluated on a bulk tank basis and is not different for a herd using AMI
• Animals producing milk with abnormalities milked immediately prior to system cleaning,
or the points of contact with abnormal milk will be cleaned and sanitized after milking
such an animal.
• The AMI milker box shall be kept as clean as any milking and equipment cleaning area.
• All ventilation air must come from outside the cattle housing area.
• AMIs shall be shut down to clean at an interval sufficient to prevent the milking system
from building up with soils (8 hrs recommended).
• AMI manufacturers shall submit data to FDA to show that “Teats shall be treated with a
sanitizing solution just prior to the time of milking and shall be dry before milking.”
• The teat cups of the milking cluster need to be adequately shielded during the udder
prepping process to assure that contaminants may not enter through the teat cup and get
into the milk.
• Adequate separation of milk and CIP solution shall be provided to minimize the risk of
cross contamination of milk with cleaning and sanitizing solutions.
• Milk shall be cooled to 10°C (50°F) within four hours or less after starting the milking
operation and the milk shall be cooled within two more hours to 7°C (45°F).

AM has been generally accepted by regulatory agencies although in some countries they are
subject to a higher degree of surveillance as a new technology and several countries have
developed guidelines or good management practices (GMP) for application of AM technology
while specific of regulations are being developed. These GMP are an invaluable aid to producers
considering, as well as those already using, AM machines to identify the critical management
tasks required to produce quality milk. They also serve as an educational tool for the variety of
farm advisors helping solve milk quality problems. GMPs have been used with success in other
areas of agricultural production systems and strike a workable compromise between rigid
regulations and lack of guidelines in areas of rapidly evolving practice. A GMP for automatic
milking was developed as an aid in providing quality assurance in milk production as an
extension on EU and national regulations and codes (Jepson et al, 2001). The basic principle of
the code is that AMI shall provide the same minimum guarantees of quality milk production as
traditional milking systems.

Milk Quality and Sensors

The overall goal of the rules and regulations regarding milk quality and safety are to ensure that
‘abnormal milk’ does not enter the raw milk supply system. The normal screen for abnormal
milk is visual inspection of the cow and/or the foremilk by a human being while performing the
tasks of udder and teat preparation and milking unit attachment. Automatic milking systems
typically rely on some form of sensor to measure various aspects of milk quality. This has
created the need for a better definition of ‘abnormal milk’.

Visual inspection is capable of detecting gross abnormalities in milk composition (clots, flakes,
or ‘wateriness’) and some substantial change in color due to blood in the milk or other gross
changes due to changes in lactation physiology. The primary emphasis in the development of
AM systems has been on mastitis detection. The practical implementation in the field results in a
cow being ‘flagged’ at one milking using a combination of data from milk quality sensors and
deviations in yield and behavior. Human inspection of the cow, foremilk, or milk quality data is
generally required to make the final decision to divert the milk from this cow at the next milking
and until the milk quality problem is resolved.

Bio-sensing systems, in general, respond to some change in the chemical composition, or
changes in the visible or non-visible light transmission or reflection of the milk. The basis for
detecting abnormal milk with biosensors is thus quite different than for visual inspection.
Biosensors have the ability to detect smaller changes in the visible light spectrum than the human
eye. Visual inspection cannot be used to detect changes in chemical composition or the nonvisible
light spectra. Biosensors thus have the capacity to be a much more sensitive detection
system for milk quality than human visual inspection. At present we do not have a welldeveloped
set of criteria for identifying ‘abnormal’ milk using biosensors.

A review of the state-of-the-art sensing technology for AMS suggests that while the sensing
systems currently being used lack the sensitivity for automated diversion of ‘abnormal’ milk,
they do provide sufficient information for motivated dairy producers to achieve milk quality that
meets or exceeds national averages. Electrical conductivity and milk color are the most widely
used on-line milk sensing methods and deviation in milk yield and milking interval are widely
used supporting diagnostic techniques. A number of other methods using visible and other light
spectra have shown promise in detecting milk abnormalities and measuring various components
of milk. Several methods of measuring the somatic cell count (SCC) of milk at cow-side are
being applied to AM. Developments in AMS milk sensing systems point toward the use of
inputs from a number of sources including milk composition, animal behaviors, and milking
characteristics combined and analyzed by centralized ‘smart’ system to improve diagnostic
accuracy. (Reinemann and Helgren, 2004)

Following the International Symposium on Automatic Milking, Koning & Meijering (2004)
concluded based upon results from commercial farms that milk quality is somewhat negatively
affected after introduction to automatic milking. There were small significant increases in
bacterial counts, (TPC), somatic cell counts, freezing point and free fatty acids. The highest
increase was for TPC and somatic cells in the first six months after introduction. After this
period it improved and stabilized around the level of conventional farms (Koning et al., 2004).
German, Danish and Dutch farms (262) were included in the study.

Helgren & Reinemann (2006) studied milk quality of 12 AM farms in the USA for three years as
part of a pilot study of AM technology in the USA. Daily records of bulk tank somatic cell count
(SCC) and total bacterial count (TBC) data were analyzed and compared to corresponding data
from a cohort conventional farms in Wisconsin as well as data from European AM installations.
The geometric means for all farms were 268,000 cells/mL SCC and 13,300 cfu/mL TBC. There
was no significant difference in SCC between AM farms and the cohort of conventional farms,
but a clear and significant seasonal effect was evident for SCC for both farm types, with higher
values observed during the summer months (July, August, and September). The TBC of milk
from AM farms was lower than that for a cohort of conventional farms, and there was some
evidence of a seasonal effect on TBC for both types of farms. Both SCC and TBC decreased as
the amount of time that a farm utilized AM increased.

Herd Management and Views of AM Users

A survey was conducted of 10 farms in the USA and 15 farms in Canada using automatic (or
robotic) milking systems (AMS) to determine how AMS facilities were being designed and
managed in the North American setting. Surveys were conducted in-person with the farm
manager during visits to the farm. All of the AMS users surveyed indicated that, overall, they
were satisfied to very satisfied with their AMS. Most users indicated that AMS has allowed
them more time for managerial tasks, and more importantly, more time for themselves and their
families and has decreased stress levels for both cows and themselves (deJong, et al., 2003).
Karttunen (2003) carried out an interview study in Finland among AMS users. In general the
users were satisfied with the system: flexibility of working time and reduction of physical work
load were among the most appreciated advantages. Other mentioned advantages were health of personnel, improved safety for the personnel, and quality of milk extraction. Mentioned
disadvantages were dependence of electricity, high investment costs and higher consumption of

AM combined with grazing has been successfully demonstrated in several countries (Jagtenberg
& Dooren, 2001; Karlsson, 2001; Jago et al., 2002). Concentrates are typically fed in the AM
station and milking frequencies as high as 2.3/cow/day have been obtained using special
motivational strategies to encourage visits to the automatic milking installations.

Future Challenges of AM

The experience of AM clearly indicates that it is possible to produce milk of the same or better
quality than conventional methods of milk harvesting. AM systems relieve the dairy farmer
from the physical labour of milking and also provide a wealth of information for herd
management. These systems use a higher level of technology than conventional milk harvesting
techniques and, therefore, will require a higher level of management skill to use this technology

The definition of abnormal milk will likely evolve as biosensor technology develops and offers
the possibility of on-line measurement of more aspects of milk quality. This is an area of rapid
technological development. These developments are fueled by the prospect of a commercially
viable product (providing the market with management information at a price justified by the
benefits) and are also highly influenced by the regulatory climate (what types of technology are
allowable and/or mandatory). The challenge to regulatory agencies will be to ensure the quality
and safety of the raw milk supplied from automatic milking systems while not stifling the
development of new technologies that could significantly improve milk quality and safety.

The types of AMS currently available on the market were developed primarily to meet the needs
and market and social conditions of single-family owner/operator dairy farms in Europe and the
traditional dairy states in the USA. These farms are generally located near population centers
where the urban pressures of higher land prices, higher price for labor and increasing
environmental regulations are significant factors in the future economic viability of these farms.
Health issues, unusual work hours, and working conditions have made obtaining reliable milking
labor a major concern of these dairy producers. AMS technology can provide an option for these
farms to reduce the labor requirements of milking and allow some of these farms to continue
dairy production and make them more attractive to new producers.

At present, AMS technology is more costly than other methods of harvesting milk. It is clear,
however, from the adoption rate in Europe and high degree of interest in the USA that this is not
the sole factor in the decision to purchase an AMS. It appears that many producers are willing to
pay a premium for the improved quality of life offered by AMS. The economics of milk
production using AM technology must be able to compare favorably to large farms to be viable
in the long term. The owner/operator will need to acquire higher-level management skills of
cows, business and technology to optimize the investment in automation.

There has been considerable interest in using AM technology on large farms. The most common
concept is to multiply the type of modules already in use on small farms. One large farm
operation in the USA has discontinued the use of AM technology while another has begun to
implement AM on a large scale. An intermediate option for AM technology, which may be
particularly attractive on large farms, could be as an assistant to a human operator. The way and
extent to which AM will be implemented on large farms remains a question and is an area of
research and development.

A major public education effort will be required to ensure that AMS users clearly understand the
management skills and economics required for its successful implementation and that legislative
bodies clearly understand AM so that the resulting rules and regulations achieve their desired
goals (Reinemann, et al., 2002).


de Jong, W., A. Finnema, D.J. Reinemann, 2003. Survey of Management Practices of Farms
Using Automatic Milking Systems in North America. ASAE Annual International Meeting
Technical Paper No. 033017

Helgren, J.M., D.J. Reinemann, 2006. Survey of Milk Quality On U.S. Dairy Farms Utilizing
Automatic Milking Systems. Transactions of the ASABE, 49(2)551- 556.

Jago, J., P. Copeman, K. Bright, D. McLean, I. Ohmstad, M. Woolford. An innovative farm
system combining automated milking with grazing. Proceedings of the New Zealand Society of
Animal Production 62, 115-119 (2002).

Jagtenberg, K., H.J. van Dooren. De melkrobot: weiden naar keuze. Praktijkonderzoek Rundvee
14, 23-25 (2001).

Jepsen, L., B. Everitt, C. Cook, R.H. Oost, H. Hogeveen, J.B. Rasmussen, 2001. Good
Management Practice Code for Milking with Automatic Milking Systems. Draft Documents
March 2001.

Karlsson, M. Betesdrift i kombination med automatisk mjölkning. En fältstudie på tre gårdar.
Examensarbete 150. Institutionen för husdjurens utfodring och vård. Swedish University of
Agricultural Sciences. Uppsala. 39 pp. (2001).

Karttunen, J. 2003. Robotic milking will change work routines. Teho No. 5, 4-6, 38-39.

Koning, K. de, A. Meijering. 2004. Excepts from main papers www.automaticmilking.nl

Koning, K. de, B. Slaghuis, Y. van der Vorst. 2004. Milk quality on farms with an automatic
milking system. www.automaticmilking.nl

Lind, O., 2007. Personal communication.

Reinemann, D.J., O. Lind, J. Rodenberg, 2002. A Global Perspective on Automatic Milking
Systems Rules and Regulations. Proc. First North American Conference on Robotic Milking,
March 20-21, 2002, Toronto, Ontario, Canada.

Reinemann, D.J., J.M. Helgren. 2004. Online Milk Sensing Issues for Automatic Milking. Paper
Number: 04-4191, Presented at the 2004 ASAE/CSAE Annual International Meeting, Ottawa,
Ontario, Canada