Milking or feeding system and method for ascertaining the occupancy of a milking or feeding stall of a milking or feeding system by a farm animal

Abstract

The invention relates to a method for ascertaining the occupancy of a milking stall of a milking system by a farm animal (10) or for ascertaining the occupancy of a feeding stall of a feeding system by a farm animal (10), comprising the following steps: - capturing at least one milking stall (4) or feeding stall in an image or an image sequence by means of at least one camera (7); - evaluating the image or the image sequence in order to identify the at least one milking stall (4) or feeding stall; - evaluating the image or the image sequence in order to identify a farm animal (10) which is located in the at least one milking stall (4) or feeding stall; - creating and storing an association between the at least one identified milking stall (4) or feeding stall and the identified farm animal (10). The invention further relates to a milking or feeding system having at least two milking stalls (4) or feeding stalls, at least one camera (7), in the capturing range of which the at least two milking stalls (4) or feeding stalls are located, and an evaluation device, the milking or feeding system being designed to carry out such a method.

Description

[0001] Milking or feeding system and method for determining whether a milking or feeding place of a milking or feeding system is occupied by a farm animal

[0002] The invention relates to a method for determining whether a farm animal is occupying a milking stall in a milking system or whether a farm animal is occupying a feeding station in a feeding system. The invention further relates to a milking or feeding system equipped for carrying out the method, comprising at least two milking or feeding stalls.

[0003] Before milking a dairy animal, such as a cow, in an automated milking system, the animal is typically identified. This identification allows the milking system to be operated with animal-specific parameters, and information collected during milking, such as the milk yield and / or milk quality, can be recorded in a database. This individualized control of the milking process allows for optimal adjustments to the animal's lactation cycle and takes into account individual characteristics, such as when attaching the milking unit. Furthermore, animal identification is necessary to discard milk, for example, in cases of known illnesses. Recording milk yield and / or milk quality on an individual basis also enables the early detection of existing or developing diseases.

[0004] Similarly, it is important to feed a farm animal individually in a feeding system, which is only possible if the animal and the feeding area it uses can be clearly assigned to each other.

[0005] In milking or feeding systems with a smaller number of milking or feeding stalls, for example, a single-digit number, it is common practice to equip each stall with a device for reading an animal-specific identification tag. For example, an RFID (Radio Frequency Identification) reader can be installed at each stall, which reads the corresponding RFID chip with which each animal is identified. In this way, a secure link between an animal's identification and its respective stall is established. However, in milking or feeding systems with a larger number of stalls, this requires a significant amount of equipment. For example, in rotary milking parlors, which often have tens to over a hundred milking stalls, a different approach is generally used: The animals pass through an identification zone in which, for example,The RF ID reader is located just before the milking stall enters the next available milking station as it swings into the entry area. While this method allows for unambiguous identification of the animal, it does not guarantee that the animal will actually occupy the next available stall. For example, the animal might enter the next stall but one, or even push into a stall that has already been partially swung past, if that stall was previously unoccupied. In such cases, the assumed assignment of animal to stall is likely to be incorrect.

[0006] To improve the reliability of uniquely identifying animals and milking stalls without requiring an RFID reader at each stall, a system is known from German patent application DE 100 33 706 A1. In this system, each animal and each milking stall is assigned an identifier, for example, in the form of an RFID chip. A stationary reader for animal identification and a stationary reader for marking the milking stall are located in the entrance area of ​​the milking carousel or at its perimeter. The readers are positioned close enough to each other to receive a unique signal from both the animal and the milking stall in which that animal is located. At the same time, the two readers are spaced far enough apart to prevent interference. This approach requires only two readers for the entire milking system, rather than an RFID reader at each milking stall.

[0007] An alternative system is known from German patent application DE 10 2005 009 741 A1, in which the milking stalls are not identified separately. Instead, the milking stall facing the entrance area, and therefore the next one to be used, is determined based on the rotation of the milking carousel. Animal identification is achieved using a combination of an individual RFID chip and animal recognition technology at the entrance of the milking carousel. Additionally, the animal's movement from the entrance area towards the milking stall is recorded, for example, optically using a camera, to verify that the previously identified animal actually enters the next designated milking stall.

[0008] The first method reduces the equipment required; however, two RFID readers are still necessary, and RFID chips are needed as markers for each milking station. The second method reduces the probability of an incorrect animal-station assignment, but misassignment cannot be completely ruled out, as unambiguous station identification is not achieved. While this method reduces the "information gap" between identification within the detection area and actual entry into a milking station, it cannot eliminate it entirely.

[0009] It is an object of the present invention to provide a method of the type mentioned at the outset in which, with minimal equipment, reliable statements can be made about the assignment of an animal to a feeding station, even for milking or feeding systems with multiple stations. It is a further object to describe a milking or feeding system designed to carry out such a method.

[0010] This task is solved by a method or a milking or feeding system with the features of the respective independent claim. Advantageous embodiments and further developments are the subject of the dependent claims.

[0011] A method according to the invention for determining whether a farm animal is occupying a milking stall in a milking system or whether a farm animal is occupying a feeding station in a feeding system comprises the following steps: At least one milking stall or feeding station is captured in an image or image sequence by at least one camera, and the image or image sequence is evaluated to identify the at least one milking stall or feeding station. Furthermore, the image or image sequence is evaluated to identify a farm animal located in the at least one milking stall or feeding station. Finally, a mapping between the at least one identified milking stall or feeding station and the identified farm animal is created and stored.

[0012] This method uses a single technology, specifically camera image capture, to identify both the milking or feeding area and the livestock standing there. A unique assignment is established using only one type of sensor, thus requiring minimal equipment. The resulting assignment can be transferred to and stored in a herd management system, for example. Alternatively, it can be stored in a separate unit. Reference data stored in the herd management system or the separate unit can be used to identify the locations and / or the livestock. An evaluation unit, which can be part of the herd management system or the separate unit, is used to carry out the process, particularly the step of analyzing the images or image sequences for identification.

[0013] In an advantageous embodiment of the method, a marker is arranged at the at least one milking or feeding station to identify it. The marker can comprise an optically readable line or area pattern, such as a barcode or a QR code. Such a marker is robust and cost-effective and can be easily manufactured in a suitable size. Alternatively or additionally, the marker can comprise at least one human-readable character. The marker can then be used both by users and within the framework of the method, by evaluating the character(s) using OCR (Optical Character Recognition) methods.

[0014] Alternatively or additionally to using a marker, the position of at least one milking stall or feeding station within the milking or feeding system can also be used to identify it. Particularly with fixed milking stalls, e.g., in group or row milking parlors, this may eliminate the need for a marker or allow identification even if an existing marker is obscured.

[0015] In a further advantageous embodiment of the method, pattern recognition based on a characteristic coat pattern or body shape is used to identify the farm animal. The pattern recognition can advantageously be AI-supported, in particular using a self-learning and adaptive algorithm. This allows for a high recognition rate, even in the event of changes, such as in the animal's body shape. To train the AI-supported method, an additional identification of the farm animal can be performed temporarily, for example, by reading an RFID chip with which the animal is tagged. The RFID chip is preferably read within the detection range of the at least one camera to ensure unambiguous identification during the training process.

[0016] In a further advantageous embodiment of the method, the at least one camera captures the farm animal before it enters the at least one milking or feeding area, and the animal is identified before entering the at least one milking or feeding area. Its movement to the milking or feeding area is then tracked. This increases the time available for identification, thus improving the reliability of identification. Furthermore, the animal is recorded from different perspectives during its movement, increasing the total area of ​​the animal's coat captured. This can also contribute to a further increase in the reliability of identification.

[0017] A milking or feeding system with at least two milking stalls or feeding stations and at least one camera, within whose detection range the at least two milking stalls or feeding stations are located, and an evaluation unit, is characterized according to the invention in that the milking or feeding system is configured to carry out such a method. Preferably, each of the at least two milking stalls or feeding stations is assigned a unique, optically readable marker. The advantages mentioned in connection with the method result from this.

[0018] The invention is explained in more detail below with reference to exemplary embodiments and the accompanying figures. The figures show:

[0019] Figure 1 shows a schematic representation of a milking carousel;

[0020] Figure 2 shows a schematic representation of a group milking parlor; and

[0021] Figure 3 shows a flowchart of a method according to the invention.

[0022] Figures 1 and 2 show schematic top-view views of milking systems suitable and configured for carrying out a method according to the invention. In Figure 1, the milking system is a rotary milking parlor 1, and in Figure 2, it is a group milking parlor 11. In both figures, identical reference numerals denote identical or equivalent elements. For clarity, not every element in the figures is provided with a reference numeral. By way of example, the rotary milking parlor 1 of Figure 1 and the group milking parlor 11 of Figure 2 are used by cows as livestock 10, hereinafter also referred to as animals 10. In other embodiments of the milking systems, they can also be configured for the automated milking of other milk-producing livestock, for example, buffalo, sheep, or goats.

[0023] The milking carousel 1 of Figure 1 has a rotating platform 2 that rotates around a pivot point 3, with the rotation in the illustrated case being counterclockwise. A plurality of milking stalls 4 are arranged along the outer circumference of the milking carousel 1. In the illustrated example, the rotating platform 2 is annular in shape, with the inner part of the milking carousel 1 being stationary. Alternatively, the rotating platform can also be circular with a rotating inner section.

[0024] In the illustrated example, milking carousel 1 is a medium-sized milking carousel with 40 milking stalls. This number is purely exemplary. A method according to the invention can also be implemented with a milking carousel with a smaller or larger number of milking stalls. The method according to the invention can also be implemented with a different arrangement of milking stalls than the circular arrangement in milking carousel 1. For example, the method according to the invention can also be implemented in a row milking parlor in which a number of milking stalls are positioned next to or behind each other in a row arrangement.

[0025] The milking stations 4 of milking carousel 1 can be configured as manual, semi-automatic, or fully automatic milking stations. In manual and semi-automatic milking stations, a milking unit (not shown separately here) is manually attached to the teats of the animals 10. The subsequent milking process is then automated in a semi-automatic milking station and is generally also automatically monitored. This monitoring includes, for example, tracking the milk flow to determine when the milking process is complete. In a fully automatic milking system, all of these steps are performed automatically. Milking is preceded or followed by a cleaning and / or care process for the teats of the animals 10 and usually also by a final grooming treatment, which is carried out manually or automatically.In addition, a cleaning process for the milking equipment is usually provided, which is also carried out manually or automatically.

[0026] Outside the rotating platform 2, there is an entrance area 6 where the animals 10 are guided individually towards the rotating platform 2 by means of appropriate barriers, for example, gates. The depicted milking carousel 1 is a so-called "external milking carousel" in which the animals 10 stand on the milking stalls 4 with their heads oriented towards the pivot point 3. The milking stall 4 is appropriately limited in the head area so that the animals 10 are stopped as soon as they have reached the desired milking position on the milking stall 4.

[0027] In entrance area 6, the milking stations 4 are initially open to the outside, so that the animals 10 can enter the next milking station 4 from entrance area 6.

[0028] Each milking station 4 is assigned a unique marker 5. This unique marker 5 can, for example, consist of a line or area pattern that is clearly visible and arranged at the respective milking station 4. In particular, a barcode can be used as a line pattern or a QR (Quick Response) code as an area pattern. However, it is also conceivable that the marker 5 alternatively or additionally displays one or more (human-readable) characters, such as a sequence of numbers or letters, which can be automatically evaluated using OCR (Optical Character Recognition) methods. The marker 5 can be black and white or display one or more colors. A colored marker 5 can contribute to greater recognition reliability due to its contrast with the coat colors of the animals 10. Furthermore, the use of color allows for the encoding of additional information or increases the information content of the marker 5 while maintaining the same size.

[0029] Contrary to the direction of rotation of the milking carousel 1, an exit area 8 is located adjacent to the entrance area 6, where the animals 10 can leave the rotating platform 2. The milking stalls 4 in this exit area 8 are also open to the outside. The animals 10 back out of the corresponding milking stall 4, which takes more time than entering the milking stall 4 in the entrance area 6. For this reason, the exit area 8 is wider and, in this case, spans the width of three milking stalls 4. To prevent the animals 10 from falling backward out of one of the milking stalls 4 while riding on the rotating platform 2, the milking carousel 1, with the exception of the entrance area 6 and the exit area 8, is surrounded by a railing 9.

[0030] In or next to the entrance area 6, at least one camera 7, shown here schematically enlarged, is positioned as an optical detection device. Camera 7 can, for example, be a two-dimensional imaging video camera. For the purposes of this application, a camera can also be understood to be a laser scanner, preferably operating in two dimensions.

[0031] In the example shown, camera 7 is positioned to capture both the animal 10, which is currently entering or has entered milking station 4 accessible via entrance 6, and the marker 5 assigned to this milking station 4. The size of the captured area can be selected so that adjacent milking stations 4 are also recorded with regard to each entering animal 10 and the marker 5. Furthermore, the capture area can be selected so that animals 10 are recorded shortly before entering the rotating platform 2, and their movement can be tracked.

[0032] Camera 7 is connected to an evaluation unit that enables the recognition of a recorded animal 10 and the marker 5. Animal 10 can be identified using one or more images or a sequence of images, for example, by examining a characteristic spot pattern in its fur or other characteristic physical features. For this purpose, the image of animal 10 is compared with stored images in a database, which can be maintained, for example, within a herd management system.

[0033] If not only the immediate area of ​​milking stall 4 on the rotating platform 2 is recorded, but also at least part of the entrance area 6, an animal 10 approaching milking stall 4 can be detected and tracked even before entering one of the milking stalls 4. This increases the time available for identification, thus improving identification reliability. Furthermore, the animal 10 is captured from different perspectives as it moves, increasing the total area of ​​the animal 10's coat captured. This can also contribute to a further increase in identification reliability. The evaluation unit, which analyzes images or image sequences from camera 7, can be a standalone unit that may access the aforementioned database of the herd management system. Alternatively, the analysis of the images or image sequences from camera 7 can also take place within the herd management system itself. Identification is carried out, for example,through pattern recognition, in particular by means of an AI-supported method.

[0034] Furthermore, an additional detection device (not shown here) can be positioned in entrance area 6. This device reads, for example, an identification number assigned to animal 10, which is stored, for instance, on a chip attached to the animal's ear. The chip could be an RFID chip, and the detection device could accordingly include an RFID reader. Such an additional detection device could also be installed in a different position, for example, hanging close to the heads of animals 10 passing by on the milking carousel 1. The read identification number is sent from the additional detection device to the evaluation unit, possibly involving the herd management system. The additional detection device can be used, in particular, temporarily to train the evaluation unit for pattern recognition.

[0035] The system then identifies the assigned milking stall 4, which an identified animal enters or has entered, using a unique marker 5. This unique marker 5 could, for example, be a barcode or pattern code that is clearly visible to camera 7 and positioned at the respective milking stall 4. Alternatively, the marker 5 could be a (human-readable) number assigned to milking stall 4, which is recognized in the images or image sequences from camera 7 using pattern recognition. By identifying both animal 10 and milking stall 4 simultaneously, ideally using a single image or image sequence, the evaluation of the information from camera 7 allows for a unique assignment between animal 10 and milking stall 4.

[0036] Figure 1 shows, by way of example and schematic representation, the camera 7 positioned next to the entrance area 6. In further embodiments, alternatively or additionally, another camera 7 can be positioned above the entrance area or above the rotating platform 2, adjacent to the entrance area 6, in order to identify the milking station 4 and the animal 10 that has entered this milking station 4. Preferably, the camera 7 is arranged at least near the entrance area 6 so that information about the association between animal 10 and milking station 4 is available as early as possible, in particular before the start of an automated care and / or cleaning process for the teats of animal 10 and in any case before the start of the actual milking process, after the animal has entered the rotating platform 2.

[0037] Figure 2 shows a group milking parlor 11 as a further example of a milking system in which the application procedure is implemented, in a schematic top view.

[0038] The group milking parlor 11 has two groups 12, each with a plurality, specifically 9, of milking stalls 4. Within each group 12, the milking stalls 4 are arranged side by side in a row. Therefore, with regard to the arrangement of the milking stalls 4, the group milking parlor 11 shown here is a row milking parlor with two rows of milking stalls 4. As in the example of Figure 1, each milking stall 4 is equipped with a marker 5.

[0039] Each group 12 is assigned an entrance area 6 where animals 10 can gather. Each entrance area 6 narrows towards an entry gate 13, with the aid of railings 9, forming a passage that can only be entered by one animal 10 at a time. The entry gate 13 separates the entrance area 6 from the respective group 12 of milking stalls 4.

[0040] Between the two groups 12 is a milking passage 14, in which milkers stand in a manual or semi-automatic milking system to attach milking equipment to the animals 10 in the milking stalls 4. The milking system of Figure 2 can alternatively be fully automatic, in which case milking robots are installed at the milking stalls.

[0041] In the group 12 shown on the right in Figure 2, all milking stalls 4 are already occupied by animals 10 and the entrance gate 13 is closed. In the group 12 shown on the left in Figure 2, the milking stalls 4 have just been vacated by animals that have already been milked, and the entrance gate 13 has reopened to admit animals 10 ready for milking.

[0042] To carry out a method according to the invention, four cameras 7 are positioned in the milking system of Figure 2, by way of example, such that their detection range covers at least the existing milking stalls 4. Using the cameras 7, as in the first embodiment, an animal 10 occupying each milking stall 4 and preferably also the marker 5 assigned to it can be detected for each milking stall 4. From an evaluation of images or image sequences from the cameras 7, the animal 10 and, if applicable, the marker 5 can be identified, and an assignment between animal 10 and milking stall 5 can be made. Optionally, more than the two cameras 7 shown can be present per group 12, so that preferably animal 10 and marker 5 are detected at each milking stall 4.

[0043] Since the group milking parlor 11 is stationary, unlike the milking carousel 1 of Figure 1, the milking places 4 can alternatively or additionally be identified by their position within the group milking parlor 11, instead of detecting the markers 5.

[0044] The cameras 7 are advantageously arranged for each group 12 such that their respective detection areas are at least seamlessly connected and preferably partially overlap. Furthermore, it can be provided that, in addition to the milking stations 4, at least parts of the milking passage 14 and / or the entrance areas 6 are also captured by one or more of the cameras 7. This allows an animal 10 to be identified as soon as it approaches one of the milking stations 4, thereby increasing the time available for identification and thus the reliability of the identification. Moreover, the animal 10 is recorded from different perspectives while moving, which increases the total area of ​​the animal 10's coat captured. This also contributes to increased identification reliability.

[0045] Figure 3 shows an embodiment of a method according to the invention in the form of a flowchart. The method is explained below by way of example with reference to the two milking systems shown in Figures 1 and 2 and using the reference numerals used therein.

[0046] In a first step S1, an image or image sequence is created by at least one of the cameras 7, which is aimed at one or more of the milking stalls 4. An image of one of the animals 10, as well as a marker 5 at the milking stall 4, is recorded. The camera 7 can be a single camera located in the entrance area 6 of the milking system or in an area above or next to the milking stalls 4. Instead of a single camera 7, it is also possible to use several cameras, particularly if, for perspective reasons, it is difficult to capture both the animal 10 and the marker 5 clearly from a single viewpoint. Even when multiple cameras 7 are used, a joint and coordinated evaluation is performed so that the association between animal 10 and milking stall 4 can be reliably determined.Marker 5 can, for example, include a barcode, a QR code, or any other optically easily decodable character.

[0047] In a next step S2, at least one milking station 4 and, if necessary, several milking stations 4 are identified from the image or image sequence of the at least one camera 7 by means of the marker 5 or by means of their position on the image or in the image sequence.

[0048] In a subsequent step, S3 again uses the image or image sequence from at least one camera 7 to identify one or more animals 10 and assign them to one of the identified milking stations 4. For example, a characteristic pattern of spots on the animal's coat can be used for identification. Alternatively and / or additionally, other characteristic physical features can be used. As mentioned previously, the recognition is preferably carried out using a self-learning, AI-supported algorithm, which can be trained, for example, using RFID chips worn at least temporarily by the animals 10.

[0049] In a subsequent step S4, a found assignment or possibly several found assignments are saved, for example in a database that the milking system can access in the further course of the milking process.

[0050] The procedure then branches back to step S1 and is executed again accordingly. Reference numeral

[0051] 1 milking carousel

[0052] 2. Rotating platform 3. Pivot point

[0053] 4 milking stalls

[0054] 5 markers

[0055] 6 Entrance area

[0056] 7 Camera 8 Exit area

[0057] 9 railings

[0058] 10 (farm) animals

[0059] 11 group milking parlor

[0060] 12 groups (of milking stalls) 13 entrance gates

[0061] 14 milking

[0062] S1-S4 Process step

Claims

Claims 1. Method for determining the occupancy of a milking stall of a milking system by a farm animal (10) or for determining the occupancy of a feeding stall of a feeding system by a farm animal (10) comprising the following steps: - Capturing at least one milking place (4) or feeding place by at least one camera (7) in one image or image sequence; - Evaluating the image or image sequence to identify the at least one milking place (4) or feeding place; - Evaluating the image or image sequence to identify a farm animal (10) located in the at least one milking place (4) or feeding place; - Create and save an association between the at least one identified milking place (4) or feeding place and the identified farm animal (10).

2. Method according to claim 1, wherein a marker (5) arranged at the at least one milking station (4) or feeding station is used to identify the at least one milking station (4) or feeding station.

3. Method according to claim 2, wherein the marker (5) comprises an optically evaluable line or area pattern.

4. The method of claim 3, wherein the optically evaluable line or area pattern is a barcode or a QR code.

5. Method according to any one of claims 2 to 4, wherein the marker (5) comprises a human-readable character which is evaluated using OCR methods.

6. Method according to one of claims 1 to 5, wherein a position of the at least one milking stall (4) or feeding stall within the milking system or feeding system is used to identify the at least one milking stall (4) or feeding stall.

7. Method according to any one of claims 1 to 6, wherein for the identification of the farm animal (10) a pattern recognition based on a characteristic coat pattern of the farm animal (10) or based on a body shape.

8. Method according to claim 7, wherein the pattern recognition is performed using AI.

9. Method according to claim 8, wherein an additional identification of the farm animal (10) is temporarily carried out, on the basis of which the AI-supported pattern recognition is trained.

10. Method according to claim 9, wherein the additional identification of the farm animal (10) is based on reading an RFID chip with which the animal (10) is provided.

11. Method according to claim 10, wherein the reading takes place within a detection range of the at least one camera (7).

12. Method according to claim 10 or 11, wherein the reading takes place within a detection range of an additional camera.

13. Method according to one of claims 1 to 12, wherein the at least one camera (7) detects the farm animal before it enters the at least one milking place (4) or feeding place and the farm animal (10) is identified before it enters the at least one milking place (4) or feeding place and its movement to the milking place (4) or feeding place is tracked.

14. Milking or feeding system with at least two milking stalls (4) or feeding stalls and at least one camera (7) in whose detection range the at least two milking stalls (4) or feeding stalls are located, and an evaluation device, characterized in that the milking or feeding system is equipped to carry out a method according to one of claims 1 to 13.

15. Milking or feeding system according to claim 14, wherein each of the at least two milking stations (4) or feeding stations is assigned a unique, optically readable marker (5).

Images