Robotic Milking of Dairy Cows

Here are some of the most asked questions about robotic milking systems. Answers are based on formal interviews with Canadian robot owners, Ontario milk quality records and European research studies. 

Ontario farmers reported 0-3 extra culls from an average herd size of 94 cows. These were typically cows with very close teat placement, where 2 rear teats are touching and seen as one by the sensor. Very high rear udders, that make rear teats hard to see in a horizontal plane, are also a problem. Other udder shapes that don't work with robots would be culls in a parlor situation as well, and were excluded from this estimate.

The more frustrating cows are those that attach readily but refuse to go to the robot stall for voluntary milking. Based on Canadian experience, this includes 5%-15% of the herd, depending on training and management practices. Most farmers feel this is too many cows to cull, and consequently, they make moving these cows to the robot twice daily a part of the management routine. With time and appropriate management it should be possible to get this number down to 5%. At this level most cows in this group will be there for a health-related reason, such as sore feet or clinical mastitis. 

Typically, when an existing herd is switched to robotic milking it takes 3-4 weeks to reach a point where 85%-90% of cows use the system voluntarily. Younger cows and more aggressive cows seem to adapt most easily. Current philosophy in training is to work cows aggressively for 2-3 weeks by chasing every cow to the robot whenever her milking interval exceeds 8 hours. This training period assures cows have repeated exposure to the milking stall and understand that frequent visits are possible and rewarded with small allotments of grain fed during milking. After this period cows are left to fend for themselves and only moved by hand if milking intervals exceed 16 hours.

Most new heifers learn the routine in 2 or 3 days, but some individuals take longer. A fresh cow may need individual attention for 1 or 2 days to assist with initial teat location.

Tools used to encourage cows to attend for voluntary milking include grain fed in the robot stall. The 15 Ontario herds reported average grain feeding of 1-6 kg/day in the milking stall, with an overall average of 2.5 kg. In the survey, voluntary use was slightly better for cows fed more grain in the stall. To encourage good cow flow, robot-milking stalls are located in the path from the freestall "resting area" to the feed manger. There is usually a gate in the crossover, which directs cows through the milking stall.

With many of the Lely systems in Ontario, this is the only restriction to normal cow traffic. Several 3 and 6 row layouts and other barns with unrestricted cow movement demonstrate that the "free cow traffic" approach is working. This system is favoured because "one way cow flow" may restrict feed intake for some cows. With one way cow flow, one way gates at all crossovers ensure cows must pass by the robot on each trip from the resting area to the feed manger. This is more common with multi-box systems. Eight out of 15 Ontario farms have opted for free cow traffic, but 6 of these still use a small holding area with a one way gate in front of the milking stall. This permits them to chase the "long interval' cows into this area twice daily while still encouraging them to enter the stall voluntarily. 

The milking system identifies cows with problems in several ways. One of the best tools is voluntary attendance records. If a new cow appears on the long interval list, she is most likely either lame or has a sore swollen quarter and would prefer not to be milked. The computer also records either whole udder or quarter milk yield and can flag cows that deviate from a "predicted yield" based on milking interval.

Electrical conductivity of the milk is measured for each quarter. Large changes in conductivity are helpful in identifying new infections, which often present themselves as clinical cases. There are a high number of false alarms, so do not rely on conductivity alone. The best tool for detecting abnormal milk and clinical mastitis is probably the recently introduced milk colour sensor. Based on European studies this device does an excellent job of identifying bloody milk. Preliminary reports also suggest it can identify yellow or watery secretions associated with clinical mastitis.

Treated cows are identified in the computer and their milk is diverted to waste milk storage or the manure pit. The systems are programmed to rinse all milk contact surfaces thoroughly after a treated cow is milked. 

The routine of the 15 herds in Ontario includes a morning and afternoon trip to the barn to identify "long interval cows" and "cows not attached". These are brought to the robot, examined on the way for possible lameness or udder problems, and presented for automatic attachment. While these cows are milked you can perform other chores such as feeding, breeding and stall maintenance. Fresh cows and clinical mastitis cases may need help attaching.

Robotic milkers include sophisticated warning systems that alert the farmer by mobile phone or beeper when problems occur. In a survey, the frequency of these alerts varied from 2 per week to 1 in two months. Farms with frequent alerts reported the common cause was "no milk because a milked cow refused to leave the stall." Where alerts were less frequent, they usually involved twisted or damaged inflations. On average, cows were milked 2.7 times per day on these farms. Dutch studies report a reduction of labour by 10% on farms that previously milked twice per day. Most Ontario robot farms estimate labor savings at 20%-30% compared to 3-times milking. 

The answer to this depends on the size of the herd and the cost of labour. Although robotic milking systems have a high capital cost they can compare favourably to parlour milking in 60-180 cow herds. Using traditional parlour milking technology, these herds face a "trade off" between investing too much capital in a labour efficient parlour that is under utilized, or spending too many hours milking with low cost but inefficient equipment. Doug Reinemann of the University of Wisconsin recently estimated the cost of milking in 70-140 cow herds at $2.15-$3.65 per hundredweight, 3 times as much as for a 400 cow herd. He estimates milking costs with robots in this same herd size at $1.30-$2.00 per hundredweight.

Of the 15 herd owners interviewed in Ontario, 4 chose robots because when they considered the extra space needed for a parlour and the cost of the parlour equipment, robotic milking was clearly the cheaper option. Three times daily milking is difficult to schedule on smaller freestall operations. Nine of the Ontario robot herds listed the ability to milk more frequently without added labour as a key reason for choosing robots.

As a general guideline, and given current economic conditions, herds of 200 cows or more may be best off with up-to-date, efficient parlour technology. Herds under this size should, if labour is cheap and readily available, minimize their parlour investment and use low cost parlour options such as swing parlours and flat barns. If labour is in short supply, herds under 200 cows are an excellent fit for robotic milking. By eliminating milking labour and the need for the large parlour, robotic milking can potentially make the single-family farm of 100-120 cows operated with family labour a much more competitive and sustainable enterprise in the future. 

Most freestall barns can be readily adapted for robots. In Ontario half of the systems are in existing facilities that previously milked in parlours. With unrestricted cow traffic the layout of the barn can include 2 or 3 rows per group, in 3, 4 or 6 row barns. As with parlours, group size is linked to the capacity of the robot. With single box systems this means groups of 60 cows per stall. With multi-box systems, 60 cows for the first box and 30 per additional stall are recommended. 

Grain in the milking stall is one of the factors that motivate cows to visit and be milked 2-4 times daily. For herds set up for TMR, robotic milking does require some adjustment. Fourteen of the 15 robot herds in Ontario do feed a mixed ration in the bunk, but also feed a prepared grain ration in the milking stall. One herd feeds only forage in the bunk and has both milking stalls and computer feeders for grain feeding. Because these herds want to maintain the benefits of TMR feeding, several are experimenting with ways to minimize the amount of grain fed in the milking stall. Very palatable ingredients, presented in the best physical form possible (such as pelleted corn distillers grainswith no mineral added) may promote good cow traffic while taking as little out of the TMR as possible. 

Yes. European experience has taught us little about operating robots in uninsulated barns where freezing conditions occur in winter. Ontario experience indicates it is necessary to keep this equipment in a frost-free environment. In cold barns this can be done by enclosing the "clean side" of the milking stall in an insulated, heated and washable room ventilated separately from the rest of the barn.

Stray voltage, not found on European phase to phase electrical distribution, is also a potential concern. Robotic milking depends on voluntary attendance by cows. Cows avoid stray voltage and poor voluntary stall use can, in some situations, be attributed to stray voltage.

Thirdly, robotic milking systems involve frequent rinsing of milk contact surfaces. In Dutch studies this has been suggested as the reason for slightly higher freezing points in milk from robot farms. Studies from many areas of the U.S. and Canada indicate that well water is not always of potable quality. This should be of concern on all dairies, but may be especially important for robotic milking installations. 

Both in the U.S. and Canada, many of the current regulations for milking and milk handling do not provide for this new technology. As an example, Ontario requires that bulk tanks cool milk to a specified temperature by a specified time after the "end of milking." Since milking continues 24 hours per day on these farms it is impossible to apply this regulation. Because much of this technology originates in Europe, components do not always meet 3A standards. In many states and provinces robotic milking is being permitted under special permits limited to specific farms to permit field evaluation of the technology. Ontario has recently drafted guidelines specific to these systems. These guidelines include special provisions for buffer tanks used for temporary storage during and after milk pick up, washing and rinsing of continuous use milking systems, and design, location and ventilation of robotic milking "rooms". 

European research studies raise some red flags about milk quality with milking robots. Dutch studies showed that, especially in the first few years, farms experienced an increase in bacteria counts, freezing point, and free fatty acids when they switched to robotic milking.. Later reports from Holland suggest that "second and third generation robotic milking systems" had a somewhat better track record.

In Ontario, milk quality results have been mixed. On 123 tests for freezing point, 2 samples were in the penalty range, compared to an average of 1.3 per 1000 for non-robot farms. Milk from Ontario dairy farms is analyzed monthly for bacteria using the more sensitive "bactoscan" method. Last year, the average result for all farms was 38,000. The 13 single box robotic milking systems in use at that time averaged 41,000. The 2 multi-box systems did not fare as well, but recent changes in how they handle waste milk is producing more promising results now. Somatic cell counts of robot herds in Ontario have averaged 280,000 and are comparable to the 245,000 average for all herds last year. 

Undoubtedly there are many more questions than this to be answered in the coming months. New installations in Pennsylvania, Wisconsin, Quebec, Nova Scotia and several additional herds in Ontario, suggest that interest in this technology continues to grow. At least 5, and possibly more milking equipment companies, are either ready to look at the North American market or are already here. The questions above and the often partial answers, suggest that, like any new technology, there will be much to learn, but also much to gain as robotic milking systems find their niche in the business that is dairy farming.