Robotic milking system

By prioritizing teat attachment based on probability and storing false teat coordinates, the robotic milking system effectively addresses the challenge of differentiating between real and false teats, improving efficiency and capacity.

WO2025133833A1PCT designated stage expired Publication Date: 2025-06-26LELY PATENT NV
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
PCT/IB2024/062478
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-12-18
Filing Date
2024-12-11
Publication Date
2025-06-26

AI Technical Summary

Technical Problem

Existing robotic milking systems struggle to accurately differentiate between real teats and false teats, leading to disruptions in the milking process and reduced system capacity.

Method used

The robotic milking system improves teat identification by prioritizing attachment in strict order of probability, marking and storing false teat coordinates, and adjusting probabilities to zero for confirmed false teats, thereby optimizing the attachment process.

Benefits of technology

This approach enhances the system's ability to handle false teats, increasing efficiency and reliability by reducing attachment errors and allowing for a higher milking capacity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure IB2024062478_26062025_PF_FP_ABST
    Figure IB2024062478_26062025_PF_FP_ABST
Patent Text Reader

Abstract

A robotic milking system (1) has a milking stall, milking cups (2), a control device (12), a teat position-determining device (4) for determining teat candidate positions of the teats of the dairy animal, and a robot arm (3) for attaching the milking cups to the teats, based on the determined teat candidate positions. The teat position¬ determining device determines an animal reference position for said dairy animal, and the teat candidate positions by determining positions with respect to the animal reference position of all objects which meet a teat criterion, and a respective probability that it is a teat for each of the objects. The control device repeats the steps a) attaching a milking cup to an object at one of the teat candidate positions associated with the milking cup, in descending sequence of probability, b) milking with the milking cup, c) assessing whether the milking operation yields milk, d) if so, storing the object with the candidate teat position as a real teat with a real teat position, and e) if not, storing the object with the candidate teat position as a false teat with an associated false teat position, and detaching the milking cup until either an object is stored as an associated real teat with an associated real teat position for all milking cups, or either a real teat with a real teat position or a false teat with an associated false teat position is stored for all teat candidate positions.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] Robotic milking system

[0002] The present invention generally relates to a robotic milking system configured to extract milk from a dairy animal with teats in a milking operation.

[0003] Such a robotic milking system comprises milking means with milking cups, a control device for controlling the robotic milking system which is provided with a memory, a teat position-determining device for determining teat candidate positions for the teats of the dairy animal, and for passing on the determined teat candidate positions to the control device, a robot arm attaching the milking cups to the teats under the control of the control device based on the determined teat candidate positions.

[0004] Based on specific decision criteria, the control device then determines where it will direct the robot arm with a milking cup in order to attach the milking cup to a teat.

[0005] Such robotic milking systems are generally known on the market and they generally function satisfactorily. However, there is a desire to continuously improve the step of attaching the milking cup to the teat. The known teat position-determining devices which are often based on image processing regularly come up against limitations, with the robotic milking system not coping well with structures which resemble teats, but are not, so-called false teats. This limits the capacity of the known robotic milking systems.

[0006] It is therefore an object of the present invention to improve a robotic milking system of the type mentioned in the introduction in such a way that it is better able to handle false teats, and allows for a larger capacity, in particular by being quicker and / or more reliable at attaching the milking cups.

[0007] The present invention achieves this object, at least partly, by means of a robotic milking device according to Claim 1 . In this case, the inventor bases his invention on the following inventive idea. Normally, the most logical thing to do is to attach the milking cups, usually four in the case of dairy cows, to candidate teats in descending order of probability of the candidate teat being a real teat. In most cases, the candidates with the highest probability will also be real teats to be milked, and the milking cups will be attached and the teat will be milked. This number is usually identical to the number of teats to be milked, such as four in the case of healthy dairy cows.

[0008] However, in practice, it is possible for the number of candidate teats found to be higher than the number of teats to be milked. Or that, for example, four candidate teats are indeed found with a healthy cow, but that one or more candidates turn out not be real teats. For example, a real teat may be assigned a lower probability than a false teat. This may be caused by a variety of reasons, such as inadequate lighting, soiling by manure, an injury, but also by other teats covering up a real teat and, last but not least, due to the shape of the false teat. This shape may be very similar to a teat, for example a lump on the teat, a pronounced vein, a scar, etc. The control unit will then attempt to attach a milking cup to this false teat. However, it will not be possible to extract any milk from this false teat.

[0009] It will be clear that such false teats then disrupt the attachment process. According to the invention, the decision is then made to perform the attachment procedure in strict order of probability, to check if the candidate teat is a real teat and, if this is not the case, to mark the coordinates of the candidate teat which has been found to be a false teat as “false-teat coordinates”, and store these, at least for this milking operation. Although this may result in a slightly delayed milking operation the first time, it may be advantageous if a milking cup is kicked off or becomes detached from the teat in another way. After all, the milking cup has to be reattached and in that case it is advantageous if the control unit already knows that there is a candidate teat which belongs to a false teat, so that the control unit can ignore this candidate teat, which did indeed have a higher probability of being a real teat, during the attachment process.

[0010] In this case, the teat criterion is a criterion which indicates whether an object could be a teat. This teat criterion may be applied in any number of ways known in the prior art. For example, an object satisfies the teat criterion if it has certain minimum and / or maximum dimensions, such as a diameter of between 14 and 40 cm and a length of between 3 and 10 cm, although other dimensions are obviously certainly possible. Other criteria relate to the shape of the object, such as approximately cylindrical, etc. Incidentally, satisfying the teat criterion is a first step which has a "yes" or a "no" answer. The control unit then determines the probability whether the candidate teat is really a teat in accordance with techniques which, again, are known per se. For example, the chances are greater if the dimensions are in the centre of the range than if they are on the edges of the range, or if the shape is not cylindrical, but is, for example, angular or bent, etc.

[0011] The animal reference position may in this case be fulfilled in many different ways. Thus, it is possible to select a position with respect to the teat position-determining device itself as reference position. This may be useful, for example, in case the relevant time period is short, so that the dairy animal and the robot arm have not, or hardly, moved at all with respect to each other due to displacement of said dairy animal or the robot arm. Other options are described below in more detail.

[0012] In addition, the phrase "one of the teat candidate positions associated with the milking cup" is understood to mean that the milking cup is also attached to the teat for which the milking cup was intended, such as "left-hand front teat" or the like. After all, it may be the case that a different milking setting applies, such as stimulation time or milking vacuum, for each teat, for example four teats in a healthy dairy cow. Also, with some robotic milking systems, it may be the case that all milking cups are connected and remain connected to the robot arm by means of a connecting means, such as a wire or at least a milking hose. This is the case, for example, with robotic milking devices made by Lely and GEA. In this case, each milking cup is associated with one of the teats of the animal to be milked. Should a milking cup intended for the left-hand front teat be attached to a different teat, then this disrupts the attachment process, or sometimes leads to the connecting means literally becoming entangled. Nevertheless, with other robotic milking systems, it is possible to attach each milking cup to any desired teat, in which case it has to be clear to the control unit to which of the teats each milking cup is attached, for example in order to set the milking settings for each teat correctly.

[0013] If a candidate teat is identified as a false teat, and is stored as such, then the control unit automatically adjusts the probability of this object being a real teat to zero, or at least to a value which is so low that this object can be ruled out for attachment purposes. This adjustment to zero may be temporary, such as for the current milking operation, or for a relatively long time, such as permanent.

[0014] Specific embodiments of the invention are mentioned in the dependent claims, as well as in the following part of the description,

[0015] In embodiments, the control device is furthermore configured to, if a respective object is stored as an associated real teat with an associated real teat position for all milking cups, store at least one, and in particular each, of the other objects in the memory as a false teat with an associated false teat position. By the control device thus also storing the objects which have been identified as false teats, the control is able to prevent these objects, i.e. the false teats, from disrupting the process of attaching milking cups during a subsequent procedure. This may refer to, for example, re-attaching a milking cup during an ongoing milking operation or attaching milking cups during a subsequent milking operation.

[0016] In embodiments, the control device is configured to determine, in particular furthermore to memorize, the false teat positions with respect to one or more of the determined real teat positions. In some cases, it may be useful to determine and memorize, i.e. store them in the memory, the false teat positions with respect to an animal reference position. One could say that the one or more real teat positions here are the animal reference positions, but it is also possible to store additional spatial relations, with one or more false teat positions being determined and stored in relation to the animal reference position as well as in relation to the real teat positions. Each such relation may be useful when assessing a candidate teat in a subsequent attachment procedure for a milking cup.

[0017] In particular, the control unit is furthermore configured to store each false teat position as a fixed teat reference position with respect to one or more of the stored real teat positions. These embodiments allow for the possibility that the relative positions of objects on, in particular, the udder, such as real teats and false teats, may vary, depending on one or more variables. After all, an animal may not only grow over the course of time, but moreover, the shape of the udder, and thus these relative positions, may change on account of the milk production. The fixed teat reference position is then, for example, a false teat position with respect to a determined fixed value of the parameter or also, for example, an average, such as a running average, of the respective fixed teat positions in the preceding milking operations, or at least attachment procedures.

[0018] The control device is configured, for example, to correct each false teat position based on at least one predetermined adjustment parameter, in particular a time since a last milking operation of the dairy animal and / or a lactation number of the dairy animal. In practice, the position of an object on the udder has been found to correlate well with the milking interval, i.e. the time between the current milking operation and the preceding milking operation. This milking interval is thus a useful parameter for adjusting the various expected positions, both expected fixed teat positions and real teat positions. In addition, the lactation number may be useful because it gives an indication of how the milk production changes over time, and thus forms a useful second parameter.

[0019] In embodiments, the control device is configured to attach a milking cup to a real teat of the dairy animal a subsequent time, such as in a subsequent milking operation after a milking cup has been kicked off, ignoring any candidate teat position which satisfies a predetermined position criterion with regard to its position with respect to every stored false teat position. In this case, the control device is thus configured to ignore the respective object in advance, i.e. in effect irrespective of the calculated probability of an object or candidate teat being a real teat, that is to say to set said probability to zero, or at least to a low value, if the object / candidate teat is situated at a position which, according to the position criterion, corresponds to a stored, previously determined fixed teat position.

[0020] Said position criterion may take many forms, but in embodiments said position criterion comprises that the distance between the candidate teat position and a stored false teat position or false teat reference position is smaller than a predetermined threshold distance. It will be clear that, if the candidate teat position is equal to a previously determined false teat position, or at least differs from the latter by less than a threshold distance, it is highly likely that the respective object is a false teat, and can thus be ignored.

[0021] It has already been indicated above that the animal reference position may, in principle, even be the teat position-determining device. However, in some embodiments, the teat position-determining device is configured to determine the animal reference position as a position with respect to the teat-determining device and / or with respect to the milking parlour, or a predetermined animal part of the dairy animal. Since the dairy animal and the robot arm will be able to move during attachment procedures, it is advantageous to select the animal reference position with respect to the animal itself. Then the effect of a movement of the robot arm has at least been taken into account, while the animal reference position can be determined using the teat position-determining device itself. In particular, the animal part relates to an udder or hind leg of the dairy animal. In the case of "the" position of the udder, this may, for example, be the position of the centre of gravity of the udder, or the position of a recognisable reference point of the udder, such as a specific marking or even a readily recognisable false teat, for example a vein or the like. In the case of the position of "the" hind leg, this may be the average position of both hind legs, for example at the location of a joint. Nevertheless, other animal parts, as well as ways of determining or taking into account the position thereof, are also possible. In this case, it remains advantageous that the animal positiondetermining device which, due to the nature of its function is usually situated on the robot arm, will be able to determine this animal reference position itself, so that the overall complexity of the robotic milking device remains limited.

[0022] Alternatively or additionally, the teat position-determining device comprises an additional measuring device which is configured to determine the animal reference position as a position with respect to the milking stall of a predetermined point on the dairy animal, in particular a rear end thereof. Although other parts of an animal are possible, such as a spine or hip joint, in particular the rear end of the dairy animal, in fact the centre of said rear end, and also referred to as the base of the tail, forms a useful animal reference position, for example for the following reasons. The shape of, in particular, dairy cows has a largely flat back section, apart from structures at the hip, and virtually perpendicularly drooping hindquarters. The position of the rear end, or the base of the tail, can therefore be determined relatively easily. In addition, the udder is situated near this point, so that any change in posture and other movements of the dairy animal will have relatively little effect on the relation between reference position and (real / false) teat position. In practice, the position of this rear end is often determined by means of a separate device, such as a stereo or time-of-flight 3D camera, although mechanical parts, such as a butt pan connected to the robot arm or the like, are possible.

[0023] The invention will now be explained in more detail with reference to a few exemplary embodiments and the drawing, in which:

[0024] - Figure 1 highly diagrammatically shows a side view of a robotic milking system 1 according to the invention,

[0025] - Figure 2 diagrammatically shows a possible image of the teat positiondetermining device,

[0026] - Figure 3 diagrammatically shows a practical use of the robotic milking system, and

[0027] - Figure 4 diagrammatically shows an effect of milking intervals.

[0028] Figure 1 highly diagrammatically shows a side view of a robotic milking system 1 according to the invention. The robotic milking system 1 comprises milking cups 2 on a robot arm 3, a teat position-detecting device 4, an animal position-determining device 5 and 5' on a bar 6 on a trolley 7 on the milking stall 8. In the latter, a dairy cow 9 with teats 10 on an udder 11 , and a base of the tail S, is situated. Reference numeral 12 denotes a control device.

[0029] In this case, the illustrated robotic milking system 1 is a single-box milking robot, intended to milk a single dairy animal each time. However, a multi-box system or a rotary milking system is also possible, with two or more milking stalls being operated by means of a single robot arm in order to attach milking cups, respectively a rotating platform with up to a few tens of milking stalls having been provided, with two or more robot arms attaching the milking cups for each milking stall. The remainder of this description is based on a single-box system.

[0030] In this case, the illustrated milking cups 2 are provided on the robot arm 3. Alternatively, the milking cups may also be provided in a magazine, from which a robot arm in each case grasps a milking cup in order to connect it to a teat. The invention is suitable for both systems.

[0031] The robot arm 3 is provided with a teat position-determining device 4, here a 3D camera, such as a time-of-flight camera. Alternatively, it is also possible to use a laser scanner, such as in the Lely Astronaut® system, or a 2D camera which is movable transversely, etc. The device 4 captures images and recognizes objects therein by means of image-processing algorithms. Incidentally, processing may take place in the device 4, but also completely or partly in the control device 12. Subsequently, it is determined whether the recognized objects could be teats, in other words the candidate teats are determined, again by means of algorithms which are known per se. The latter use, for example, algorithms to recognize objects based on distances, shapes (a teat is often cylindrical and has a rounded tip) and / or dimensions (a teat often has a length of between 2 cm and 10 cm, and a diameter of between 2 and 4 cm). In this way, a number of candidate teats are determined, together with their respective positions.

[0032] For these candidate teats, the control device 12 and / or the teat positiondetermining device 4 calculates the probability that the respective candidate teat is actually a teat, for example by assessing how closely the shape of the candidate teat resembles the expected shape of the teat, how big the candidate teat is compared to expectations, etc. It should be noted that the expectations may be based on general animal knowledge, specific knowledge of the breed of the dairy animal or the individual animal, as well as historical data.

[0033] In this way, a list of candidate teats with an associated probability and position is obtained. The processing of this list will be explained in more detail below. For now, it suffices to mention the fact that the positions of the candidate teats can be determined by the device 4 with respect to the device 4 itself, but in particular with respect to an animal reference position.

[0034] Here, the animal position-determining device 5 determines this animal reference position, in this case of the base of the tail S. This is a point which can be recognized relatively easily, due to the specific shape of a dairy cow. In case of other dairy animals, a different animal reference position might be desirable. As the animal position-determining device 5 is fixedly connected to the milking stall 8, the position of the base of the tail S determined by the device 5 is a position relative to this milking stall 8. Alternatively, the animal reference position, such as that of the base of the tail S, may also be determined by means of an animal-positioning device 5' which is in this case connected to a trolley 7 which is displaceable along the milking stall 8. Although the device 5' then determines the animal reference position of the base of the tail S as well as a relative position, here the position of the trolley 7 with respect to the milking stall 8 is added. The latter is monitored by the control unit 12, so that it can be taken into account, obviously in particular when the robot arm 3 is displaced by the control device 12.

[0035] Figure 2 diagrammatically shows a possible image of the teat position- determining device. The image shows the udder 11 with a left-hand front teat 10-LF, a left-hand rear teat 10-LB, a right-hand front teat 10-RF and a right-hand rear teat 10-B, as well as a left-hand hind leg 13-L and a right-hand hind leg 13-1 . In addition, a leg lump 14, a tail tip 15 and a vein 16 can be seen.

[0036] The illustrated image is typical insofar as many problems which occur when interpreting the image and trying to find the teats have been combined. However, this is an extreme situation as the problems never, or rarely, occur all at once.

[0037] In principle, it is relatively simple to recognize the hind legs 13-L and 13-R in the image. They are much wider than a typical teat. In this case, however, a leg lump 14 is situated on the left-hand hind leg 13-L, and another one at approximately the position of the other teats 10. The shape and the dimensions of the leg lump 14 do not differ much from a typical teat. If the left-hand hind leg 13-L is sufficiently close to the udder 11 and the teats 1 , it is conceivable that the leg lump 14 could be seen as a candidate teat.

[0038] Also visible is the tail tip 15 which seemingly hangs between the teats. Again, the shape and dimensions here also do not differ that much from those of a typical teat. If the tail tip does not move, or hardly moves, while the images are being captured and processed, the tail tip 15 may also be seen as a candidate teat, in particular if the distance at which it is being observed is of minor importance when determining the candidate teats.

[0039] Furthermore, a thick vein 16 is present on the udder 11 which protrudes in such a way from the surface of the udder that this vein 16 in the image likewise resembles a teat. Thus, this vein 16 is also a candidate teat.

[0040] Lastly, teats can be seen in the image, but not in an ideal way. On the one hand, the left-hand and right-hand front teats 10-LF and 10-RF are clearly visible and have typical dimensions. These candidate teats will therefore clearly fulfil a predetermined teat criterion which may, for example, be that the object has to have a length of between 2 and 10 cm and a diameter of between 2 and 4 cm, etc. In the control device, these objects / candidate teats will thus be given a high probability of actually being a teat. By contrast, the right-hand rear teat 10-RB is largely covered by the right-hand front teat 10- RF. This may mean, for example, that this object is either ignored because it does not meet the teat criterion, or that it is indeed seen as a candidate teat, but the control device deems the probability that it is actually a teat to be too low, compared to, for example, the teat tip 15 which is now situated adjacent thereto.

[0041] The left-hand rear teat 10-LB no longer stands out clearly from its environment, because the vein 16 seems to be virtually the same size, and is situated immediately next to the left-hand rear teat 10-LB in the image. Expectations are that both objects will satisfy the teat criterion and thus be a candidate teat, and the calculated probability for both candidate teats will be similar.

[0042] In this case, it is important to note that with all the calculations for the teat criterion and the probability that an object is a teat, the relative positions have not yet been taken into account, including the historical data with regard thereto. This will be explained in more detail below. The illustrated image and the problems which are associated therewith are particularly important for new dairy animals for which no such data are available at all, although the milking of dairy animals which are known per se may benefit from the present invention as well.

[0043] Figure 3 diagrammatically shows a practical use of the robotic milking system. In this case, identical parts again have the same reference numerals.

[0044] Here, the dairy animal 9 is again shown with a tail 17 with the tail tip 15 and base of the tail S, as well as a diagrammatical hip lump 18. Furthermore, it has an udder 11 with the left-hand front teat 10-LF and the right-hand rear teat 10-LB shown here, as well as the vein 16. Reference numeral 20 denotes the centre of gravity of the udder.

[0045] Here, only the robot arm 3 with a milking cup 2, and the camera 4 on the robot arm and the camera 5 are shown of the robotic milking system.

[0046] Here, the camera 5 determines the animal position with respect to the camera 5 itself. This is often fixed in the milking stall, see Figure 1 , so that every position is also a relative position with respect to the milking stall, at least can be converted to the latter by the control device. The animal position in the present invention serves as a possible animal reference position. In order to determine the animal reference position, the control device chooses a characteristic feature of the dairy animal 9, such as the base of the tail S or the hip lump 18. Since the udder 11 in principle has a fixed position with respect to said characteristic features, every position measured with respect to the udder 11 is in turn a relative position with respect to the milking stall. This fixed position may, for example, have been established from historical data. Because the control device knows and adjusts the position of the robot arm 3, and therefore of the milking cup 2, with respect to the milking stall, the control device can send the milking cup, for example, to the left-hand front teat 10-LF, based on the animal (reference) position, such as S or 18, as well as on the position, measured by the device / camera 4, of the object which has been recognised as the left-hand front teat 10-LF with respect to the udder 11 . The control unit does this, for example, by determining the position of the centre of gravity 20 of the udder 11 in the camera image, and by also determining the position of the object / teat 10- LF with respect to said centre of gravity 20. Combining these with the measured animal reference position and the known historical relative position of the centre of gravity 20 with respect to the animal reference position, the control unit calculates the required position of the teat 10-LF with respect to the milking stall.

[0047] The above-described technique is known per se for determining teat positions. In this case, it has however been assumed that the control unit knows which objects are teats and which objects are not. With the present invention, the teats, such as four teats in the case of a healthy dairy cow, can easily be found if only four objects meet the formulated teat criterion, and obviously if all four also turn out to be real teats. If the image contains more objects than the expected number of teats, the control unit has to make a decision as to which objects are teats. The control unit does this, for example, based on the probability of an object / candidate teat being a teat. In this case, the control unit applies various algorithms which are known per se and ultimately calculates a list of expected teats from the candidate teats. The required number of candidate teats with the highest probability, such as four with a healthy dairy cow, which example will now be described without excluding other dairy animals, is selected by the control unit to attach a milking cup to. The respective milking cups are guided to the positions of the four candidate teats as determined by the control unit with this highest probability.

[0048] If a milking cup has been attached and the milking cup actually milks milk, the respective candidate teat has been found to be an actual teat, and the position as determined beforehand can be memorized for subsequent use, e.g. when a milking cup has been kicked off, etc. but if the milking cup does not milk milk, the candidate teat has been found to be a false teat, and the milking cup will have to be attached again, but in that case to a real teat. In order to prevent the control unit from again attaching the milking cup to the false teat to which it was attached earlier, the control unit preferably memorizes the position of the false teat, more particularly as a combination of said object and the position with respect to the animal reference position. This object or false teat is then given a probability of zero as a candidate teat, or in any case a very low number, of being a real teat. Subsequently, the control unit determines which candidate teat now has the highest probability of being a real teat, and thus continues the attachment procedure for the milking cups. The control unit repeats this until all milking cups have been attached to a real teat. Every false teat, that is to say every candidate teat where an attachment procedure did not result in the extraction of milk and every candidate teat which no longer had to be tested for milking purposes because the four teats had already been found, can be stored in the memory in this way, as a list of false teat positions with respect to one of the animal reference positions. This list may be very useful in case the milking cup drops down or is kicked off during an ongoing milking operation, but also in case of a subsequent milking operation in which all milking cups still have to be attached. The greatest effect is to be expected with a dairy animal without any known historical data and with dairy animals which enter into lactation after having been dry for some time.

[0049] Figure 4 diagrammatically shows an effect of milking intervals. The dairy animal 9 again has a hip lump 18 and a base of the tail S, as well as an udder 11 with a left-hand front teat 10-LF, a left-hand rear teat 10-LB and a vein 16, all three of which have been drawn by means of a solid line for a time after the milking operation, and have been drawn by means of a dashed line for a time immediately before a milking operation after a slightly longer milking interval Ml, with the reference numerals being provided with an accent.

[0050] As can be seen, the udder 11 / 1 T is much smaller directly after milking than it is immediately before a milking operation after the milking interval Ml. In this situation, the various candidate teats 10'-LF, 1 O'-LB and 16' have a different position with respect to the animal reference position S or 18, and also with respect to each other. By collecting the positions of real teats and false teats during some milking intervals, it is possible, for example by means of interpolation, for the control unit to arrive at a satisfactory estimate of the position with respect to the animal reference point, which preferably does not depend on the milking interval. As a result thereof, the control unit is still able to take into account stored false teat positions in a reliable manner, even in case of a different milking interval, when assessing an image for the purpose of a current milking operation.

Claims

CLAIMS1. Robotic milking system configured to extract milk from a dairy animal with teats in a milking operation, comprising- a milking stall for receiving the dairy animal for the milking operation,- milking means with milking cups,- a control device for controlling the robotic milking system, and provided with a memory,- a teat position-determining device for determining teat candidate positions of the teats of the dairy animal, and for passing on the determined teat candidate positions to the control device,- a robot arm for attaching the milking cups to the teats based on the determined teat candidate positions under control of the control device, wherein the teat position-determining device is configured to determine an animal reference position for said dairy animal, and to determine said teat candidate positions by determining positions with respect to said animal reference position of all objects which meet a teat criterion, and for determining a respective probability that said object is a teat for each of the objects, wherein the control device is configured to repeat the following steps: a) attaching one of the milking cups to a respective one of said objects at one of the teat candidate positions corresponding to the milking cup, in descending order of said probability, b) performing a milking procedure on said object with said milking cup, c) assessing whether said milking procedure results in an extraction of milk from said object based on a milking criterion, d) if the milking procedure results in an extraction of milk from said object: storing, at least for said milking operation, said object with said candidate teat position as a real teat with a real teat position in the memory, and e) if the milking procedure does not result in an extraction of milk from said object: storing, at least for said milking operation, said object with said candidate teat position as a false teat with an associated false teat position in the memory, and detaching the milking cup from said object, until, at least for said milking operation, either a respective object is stored as an associated real teat with an associated real teat position for all milking cups, or else a real teat with a real teat position or a false teat with an associated false teat position is stored for all teat candidate positions.

2. Robotic milking system according to Claim 1 , wherein the control device is furthermore configured to store, if a respective object as an associated real teat with an associated real teat position has been stored for all milking cups, at least one, and in particular each, of the other objects in the memory as a false teat with an associated false teat position.

3. Robotic milking system according to one of the preceding claims, wherein the control device is configured to determine, in particular furthermore to memorize, the false teat positions with respect to one or more of the determined real teat positions.

4. Robotic milking system according to one of the preceding claims, wherein the control unit furthermore is configured to store each false teat position as a fixed teat reference position with respect to one or more of the stored real teat positions.

5. Robotic milking system according to one of the preceding claims in as far as it is dependent on Claim 3, wherein the control device is configured to, next time when attaching a milking cup to a real teat of the dairy animal, ignore each candidate teat position which meets a predetermined position criterion regarding its position with respect to every stored false teat position.

6. Robotic milking system according to Claim 5, wherein said position criterion comprises that the distance between the candidate teat position and a stored false teat position or false teat reference position is smaller than a predetermined threshold distance.

7. Robotic milking system according to any preceding claim, wherein the control device is configured to correct every false teat position based on at least one predetermined adjustment parameter, in particular a time since a last milking operation of the dairy animal and / or a lactation number of the dairy animal.

8. Robotic milking system according to one of the preceding claims, wherein the teat position-determining device is configured to determine the animal reference position as a position with respect to the teat-determining device and / or with respect to the milking stall, of a predetermined animal part of the dairy animal, in particular of an udder or hind leg of the dairy animal.

9. Robotic milking system according to Claim 8, wherein the teat positiondetermining device comprises an additional measuring device which is configured to determine the animal reference position as a position with respect to the milking stall of a predetermined point on the dairy animal, in particular a rear end thereof.

Citation Information

Patent Citations

  • Vision System for Teat Detection

    US20180049392A1

  • Method of performing an animal related action on an animal and a device for performing said method

    US7228815B2

  • Sensor for milking machines

    WO1999063806A1