WO2010023122A2 - Arrangement and method for determining positions of the teats of a milking animal - Google Patents

Arrangement and method for determining positions of the teats of a milking animal Download PDF

Info

Publication number
WO2010023122A2
WO2010023122A2 PCT/EP2009/060583 EP2009060583W WO2010023122A2 WO 2010023122 A2 WO2010023122 A2 WO 2010023122A2 EP 2009060583 W EP2009060583 W EP 2009060583W WO 2010023122 A2 WO2010023122 A2 WO 2010023122A2
Authority
WO
WIPO (PCT)
Prior art keywords
milking
animal
milking animal
images
arrangement
Prior art date
Application number
PCT/EP2009/060583
Other languages
French (fr)
Other versions
WO2010023122A3 (en
Inventor
Bohao Liao
Uzi Birk
Emil HÄLLSTIG
Original Assignee
Delaval Holding Ab
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Delaval Holding Ab filed Critical Delaval Holding Ab
Publication of WO2010023122A2 publication Critical patent/WO2010023122A2/en
Publication of WO2010023122A3 publication Critical patent/WO2010023122A3/en

Links

Classifications

    • AHUMAN NECESSITIES
    • A01AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01JMANUFACTURE OF DAIRY PRODUCTS
    • A01J5/00Milking machines or devices
    • A01J5/017Automatic attaching or detaching of clusters
    • A01J5/0175Attaching of clusters
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/50Depth or shape recovery
    • G06T7/55Depth or shape recovery from multiple images
    • G06T7/593Depth or shape recovery from multiple images from stereo images

Definitions

  • the present invention generally relates to dairy farm robot milking and to automatic attachment of teat cups related thereto.
  • EP 1170987 discloses a method for recognizing and determining the position of at least one teat of a milking animal, which includes the steps: moving a scanning head including a light source to a region containing the teat or teats, capturing at least one image formed by said light, evaluating the image or images so as to determine if each image describes said teat or teats.
  • the scanning head (1) is moved to a fixed initial position (A) in the room which is under the animal and clearly in front of an udder and thereby the teats of all known relevant animals, thereafter the scanning head is moved in determined steps (A - J) under the animal: upwards towards the animal, downwards, backwards towards the udder, upwards, downwards etc., while carrying out the scanning procedure.
  • the invention also concerns an apparatus carrying out the process and a milking robot including such an apparatus .
  • a drawback of such an approach is that the scanning head including the light source has to be moved extensively. Such repeated scanning is time consuming, which renders the method slow. Further, the scanning head is mounted on a robot arm, which has to be moved up and down in order to find the teats of the animal. Such movement of the entire robot arm just for detecting the teats of the animal seems to be inflexible and slows down the process yet further.
  • WO 03/055297 discloses a method for determining the position of at least one teat of an udder of an animal for milking, comprising illuminating points on the udder with at least one directed light source in a predetermined pattern; detecting points illuminated with the light source in the pattern with recording means; determining position data of illuminated points from a detection of the recording means on the basis of triangulation; forming a three-dimensional representation of the udder from the position data and the pattern; recognizing the teat in the three-dimensional representation; and determining the location of the teat on the basis of the recognition and with the position data.
  • a drawback of such a method is that it seems to be complex and the determination of position data of illuminated points from a detection of the recording means on the basis of triangulation takes time, which will cause the vision system to be slow.
  • various activities depending on such determination such as the attachment of teat cups to the teats of a milking animal, will take considerable time.
  • scanning over the entire udder is needed for each teat cup attachment since the animal has time to move between the attachments.
  • the method may be very slow due to the extensive scanning.
  • p. 1162 is disclosed the use of an arrangement comprising a number of cameras fixedly mounted in the milking stall instead of using a video camera mounted on a movable robot arm.
  • two or three video cameras can be mounted at each side of the milking animal to be milked, preferably on the walls of the milking stall or just outside thereof.
  • the video cameras are directed diagonally upwards towards the region where the teats are when the milking animal has been positioned in the milking stall.
  • a computer is provided for e.g. selecting two cameras, which together create a stereoscopic image, which by means of image processing enables a substantially exact determination of the position of a teat.
  • Two further cameras may be used to confirm the three-dimensional position of a teat.
  • a robot arm is then moved to the teat based on the determined position.
  • a milking environment is an environment, in which it is very difficult to perform stereo vision measurements.
  • the environment is non-clean and dirt may settle on camera lenses. Further the milking animals are moving, and teats may not be visible to the cameras due to self-occlusion.
  • it is an object of the present invention to provide an arrangement and a method for determining positions of the teats of a milking animal in a milking system comprising a milking stall for housing the milking animal during milking, a movable robot arm for automatically attaching teat cups to the teats of the milking animal in the milking stall, and a control device for controlling the movement of the robot arm based on determined positions of the teats of the milking animal, which arrangement and method solve at least some of the problems of the prior art as set forward above.
  • an arrangement for determining positions of the teats of a milking animal comprises an active wavefront sampling camera and image processing means.
  • the active wavefront sampling camera is provided for being directed towards the udder of the milking animal in the milking stall and for repeatedly recording images of the udder of the milking animal.
  • the image processing means is provided for repeatedly forming three-dimensional images from the images repeatedly recorded by the active wavefront sampling camera and for detecting the teats of the milking animal and determining their positions in all three spatial dimensions based on the repeatedly formed three-dimensional images.
  • the control device may control the robot arm very fast and accurately.
  • the image recordings by such camera and the subsequent processing of the measurement data provide high quality information extremely fast and as a result the milking system is capable of operating at higher speed. The milking time will be shorter and the throughput of animals is increased.
  • the inventive arrangement is implemented in a rotary milking system, wherein the active wavefront sampling camera is preferably located at a fix position with respect to a floor, on which the rotary milking system is installed and with respect to which a rotary platform of the rotary milking system rotates during milking.
  • the active wavefront sampling camera is preferably located at a fix position with respect to a floor, on which the rotary milking system is installed and with respect to which a rotary platform of the rotary milking system rotates during milking.
  • the position of the teat cups when being stored in e.g. a magazine, i.e. when not being used, may not be known.
  • the active wavefront sampling camera of the inventive arrangement may be directed towards teat cups located in the magazine and provided to repeatedly record images of the teat cups.
  • the image processing means is capable of forming three-dimensional images from the images recorded by the active wavefront sampling camera and detecting the teat cups and determining their positions in all three spatial dimensions in the magazine based on the formed three- dimensional images.
  • the active wavefront sampling camera is mounted on the movable robot arm and the image processing means determines the positions of the teats of the milking animal in all three spatial dimensions relative the movable robot arm. Additionally positions for other objects such as legs, udders, bodies, magazines, and teats can be determined by an active wavefront sampling camera mounted on the movable robot arm.
  • the active wavefront sampling camera is mounted in a fixed position with respect to the milking stall and the image processing means determines the positions of the teats of the milking animal in all three spatial dimensions relative the milking stall.
  • the active wavefront sampling camera can be located at a side of the milking animal and be directed essentially sideways, i.e., horizontally, towards the udder of the milking animal.
  • the active wavefront sampling camera can be located below the milking animal and be directed upwards towards the udder of the milking animal.
  • the above positions/orientations of the camera seem to be the most favorable ones in order to fast locate the udder and the teats of the milking animals.
  • the camera positions/orientations seem to provide the best contrast between the udder/teats and other objects in the view field of the camera. Other orientations may be conceivable.
  • the camera may be movable between several positions/orientations depending on the application or status of the camera.
  • a method for determining positions of the teats of a milking animal is provided.
  • an active wavefront sampling camera is directed towards the teats of the milking animal in the milking stall. Images of the teats of the milking animal are repeatedly recorded by said active wavefront sampling camera and three-dimensional images of the teats of the milking animal are repeatedly formed from the images repeatedly recorded by the active wavefront sampling camera. Finally, the teats of the milking animal are repeatedly detected and their positions in all three spatial dimensions are repeatedly determined based on the repeatedly formed three-dimensional images .
  • An advantage of the present invention is that the determination of the positions of the teats of an animal in all three spatial coordinates is made very fast.
  • the invention provides in particular for the recording of images by the active wavefront sampling camera, wherein all visible teats can be detected, and the positions of all the teats can be determined, simultaneously. Hence, teat cup attachment can be made very fast .
  • Figs. 1-3 display each a schematically outlined milking system including an arrangement for determining positions of the teats of a milking animal according to a respective embodiment of the present invention.
  • Figs. 1 and 3 are top views while Fig. 2 is a perspective view.
  • Fig. 4 displays schematically, in a top view, an arrangement for determining the identity of a milking animal according to an embodiment of the present invention.
  • Fig. 5 displays schematically, in a side elevation view, an arrangement for determining the weight of a milking animal according to an embodiment of the present invention.
  • Fig. 1 a milking system, in which an arrangement for determining positions of the teats of milking animals according to an embodiment of the invention is implemented.
  • the rotary milking system 3 comprises a plurality of milking stalls 5, which milking animals 8 enter in a sequential order.
  • Each of the milking stalls 5 comprises milking equipment including teat cups that are attached to the teats of the milking animal present in the milking stall prior to milking.
  • teat cups 11 are illustrated only for one of the milking stalls 5.
  • the rotary milking system 3 may be of parallel, tandem, or herringbone configuration.
  • the longitudinal directions of the milking stalls and of the milking animals therein extend radially (the milking animals stand side by side)
  • the longitudinal directions of the milking stalls and of the milking animals therein extend circumferentially (the milking animals stand one after the other)
  • the longitudinal directions x of the milking stalls and of the milking animals therein extends partly radially, partly circumferentially .
  • a robot 14 provided with a robot arm 15 is provided for automatically attaching teat cups 11 to the teats of the milking animals 8 present in the milking stalls 5 under the control of a control device 19, which is operatively connected to the milking robot 14.
  • the milking robot 14 is preferably stationary with respect to a rotatable carousel or rotating platform 20 of the rotary milking system 3, which forms the support for the milking stalls 5.
  • the milking robot 14 is movable back and forth in e.g. a circumferential direction .
  • the rotating platform 20 may, for each of the milking animals, be kept still while the milking robot 14 automatically attaches teat cups 11 to the teats of the milking animal 8, and is rotated there in between. Alternatively, the rotating platform is rotated continuously during the attachment of the teat cups and the milking of the milking animals 8.
  • an active wavefront sampling camera 21 is provided.
  • AWS Active wavefront sampling
  • AWS is a method of imaging in three dimensions from a single lens based on imaging the wavefront of a lens in time.
  • AWS is little more than an off-axis aperture placed in an optical path that rotates such that images are recorded from different sections along a circular path of a lens.
  • Fig. 1 Moving objects can be imaged in three dimensions using AWS as long as its motion is steady (either velocity or acceleration) relative to the framing rate of the camera.
  • the inventors of the present invention have concluded that AWS lends itself well to applications for teat detection in automatic milking systems .
  • the AWS camera 21 is, for each of the milking animals in the milking stalls, directed towards the udder of the milking animal, wherein the AWS camera repeatedly records images of the udder of the milking animal in real time.
  • the AWS camera comprises an aperture arrangeable in at least three different positions during the repeated recording of images of the udder of the milking animal.
  • Image processing means 22 is provided, for each of the milking animals, for repeatedly forming three-dimensional images from the images recorded by the AWS camera 21; and detecting the teats of the milking animal and determining their positions in all three spatial dimensions by a calculation method based on the repeatedly formed three-dimensional images.
  • the AWS camera 21 together with the image processing means 22 thus determine directly the position in all three spatial coordinates for each object point that is found in the pixels of the image recorded.
  • the image processing means 22 is preferably implemented as a software module in the control device 19, which thus is operatively connected to the AWS camera 21, or in any other device operatively connected to the AWS camera 21 and the control device 19. Alternatively, the image processing means 22 is integrated in the AWS camera 21.
  • the calculation method used can be any calculation method suitable for use in the field of digital image processing.
  • the AWS camera 21 can be mounted in fixed positions with respect to the different parts of the milking system 3 depending on the application in question.
  • the camera 21 can be located at a fix position with respect to a floor, on which the rotary milking system 3 is installed and with respect to which the rotary platform 20 of the rotary milking system 3 rotates during milking.
  • the camera 21 may be mounted in a ceiling of a building in which the milking system is installed.
  • the camera 21 is mounted on the rotating platform 20 or on the robot arm 15.
  • the camera 21 is movable along some path.
  • the AWS camera 21 can be movable between a first position, in which the camera 21 is located during the detections of the images, and a second position, in which the camera 21 is located there in between, that is, when being idle.
  • the AWS camera 21 can further be located in a number of different positions and orientations depending on the application in question.
  • the camera 21 can, during each of the recordings of the images of the udder of each of the milking animals, be located at a side of the milking animal and be directed essentially sideways in the horizontal plane and perpendicular to a longitudinal direction of the milking animal.
  • the camera 21 can, during each of the recordings of the images of the udder of each of the milking animals, be located below the milking animal and be directed upwards.
  • two or more AWS cameras can be provided operatively connected to the control device 19 in order to provide further detailed three-dimensional information of the positions and orientations of the teats of the milking animals.
  • AWS cameras arranged stationary on the floor or on the rotating platform could be combined with one AWS camera mounted on the robot arm, or a single AWS camera mounted on the robot arm could be used in the invention.
  • the speed and accuracy of the arrangement for determining teat positions are of outermost importance in order to provide a milking system with accurate and fast automatic attachments of teat cups to teats of milking animals. This is of particular importance in a rotary milking system where not only the robot arm for attaching the teat cups is movable, but also the entire robot may be movable. Further, the rotating platform and the milking stalls are moving (rotating) .
  • the AWS camera 21 of the inventive arrangement may thus be provided to record images of the teat cups 11 in real time, particularly when the teat cups 11 are located in the magazine, and the image processing means 22 forms three-dimensional images from the recorded images, and detects the teat cups and determines their positions in all three spatial dimensions by the above calculation method based on the recorded images of the teat cups when being located in the magazine.
  • the AWS camera 21 can be used to establish a position of each of the milking animals in at least one spatial dimension, e.g. a longitudinal direction x or in a direction perpendicular thereto, with respect to the milking stalls in which they are housed.
  • the AWS camera 21 is, for each of the milking animals, directed towards a leg or an outer contour of the milking animal in the milking stall, wherein the AWS camera 21 records images of the leg or the outer contour of the milking animal.
  • the image processing means 22 forms a three-dimensional image from the recorded images, and detects the milking animal and determines its position in the spatial dimension by the calculation method based on the recorded images of the leg or the outer contour of the milking animal.
  • Such embodiment may be advantageous in a milking system where a detection device is used to detect the teats of a milking animal in a milking stall based on the prior knowledge of where in the milking stall the milking animal is located and possibly on the physical dimensions of the milking animal itself.
  • control device 19 may, for each of the milking animals, hold information in, or receive information from, a database 23 regarding the position of the udder of the milking animal relative the milking animal itself. This may be a single approximate figure valid for all the animals. Further, the control device 19 directs the AWS camera 21 and controls the robot arm 15 of the milking robot 14 to move towards the udder of the milking animal based on the information of the position of the at least one teat of the milking animal relative the milking animal itself, and on the detected position of the milking animal in the spatial dimension relative the milking stall.
  • the information of the position of the at least one teat of the milking animal relative the milking animal itself can be deduced from the recording of visual detections of the milking animal in connection with an earlier milking of the milking animal .
  • control device 19 may, for each of the milking animals, hold information in, or receive information from, the database 23 regarding the position of each of the teats of the milking animal relative the other teats of the milking animal. Further, the control device 19 controls the robot arm 15 of the milking robot 14 to move a teat cup to an obscured teat of the milking animal based on the information of the position of each of the teats of the milking animal relative the other teats of the milking animal and on a determined position of a repeatedly detected teat of the milking animal, which is not obscured.
  • the above algorithm is applied in an arrangement where the AWS camera 21 is, during each of the recordings of the images, located behind the milking animal and is directed forward towards the back of the milking animal.
  • the spatial dimension determined is then preferably in a direction perpendicular to a longitudinal direction of the milking stall.
  • the leg or outer contour of the milking animal may include one or both back legs of the milking animal, preferably the inner contours of the back legs.
  • the determination of the positions of the teats of the milking animal can be made in the following manner. First the milking animal is detected and the position thereof is determined. Based on this information the contour of the back legs and the body therein between is detected and the positions thereof are determined. Based on this information an area of interest (i.e. where the teats most probably are found) is located, and in this area the udder of the milking animal is searched for. When the udder has been detected and located the teats are searched for and located. Finally, when the positions of the teats have been determined the teat cup attachment can be initiated.
  • an area of interest i.e. where the teats most probably are found
  • Such algorithm is characterized by fast object recognition and can be applied in situations where a new milking animal is detected or where a milking animal has a position of the legs which renders the determination of the teat positions more difficult.
  • the arrangement of the present invention may be used as a multifunctional detection system for detecting any of the following: (i) presence of a milking animal in a milking stall, (ii) the behavior of a milking animal, (iii) the activity of a milking animal, (iv) the body shape of a milking animal, (v) an incorrect teat cup attachment, (vi) a teat cup kick-off, (vii) presence of an obstacle in the working area of the milking robot, (viii) a parameter related to the operation of the milking robot, and (ix) a parameter related to the operation of the camera.
  • the embodiment of Fig. 1 may comprise a light source such as a coherent laser source 12 and a speckle pattern generating arrangement 13, both being preferably operatively connectable to the control device 19 and/or the image processing means 22.
  • the light source 12 and the speckle pattern generating arrangement 13 are provided to illuminate the object to be detected by the AWS camera 21 by light having a speckle pattern.
  • the light source 12 and the speckle pattern generating arrangement 13 creates an illuminated speckle pattern on the object, which can be used for correlating pixels from different images recorded by the AWS camera 21 to one another. This may be particularly advantageous when detecting a teat since color and texture of the teat surface are similar to that of the udder surface and thus it may be difficult to interpret the recorded images. The situation is even more complicated by the fact that the size, shape, color structure, morphological structure and texture may vary quite much from animal to animal.
  • the light source 12 and the speckle pattern generating arrangement 13 thus provide means for creating patterns on the udder and teats of the milking animal which can be used to facilitate or speed up the detection.
  • a voluntary milking system or station 3 comprising an enclosure having an inlet gate 4a and an outlet gate 4b, which are both capable of being opened automatically.
  • the front end of the milking station 3 is denoted by 3a, the back end is denoted by 3b, the sides are denoted by 3c and 3d.
  • the milking station 3 comprises further an automatic milking machine (not explicitly illustrated) including teat cups 11 connected to an end unit by means of milk lines (only the portions attached to the teat cups 11 are shown in Fig. 2) .
  • the milking station further includes a milking robot 14 having a movable robot arm 15 provided with a gripper.
  • the milking robot 14 is arranged to automatically apply the teat cups 11 of the milking machine to the teats of a milking animal 8 present in the milking station 3 prior to milking.
  • three of the teat cups 11 are arranged in a teat cup rack or magazine 16, whereas the fourth one is held by the gripper of the robot arm 15.
  • a teat cleaning device including e.g. a teat cleaning cup 24 or brushes 25 may be provided for cleaning the teats of the milking animal 8 prior to milking.
  • the milking station 3 comprises an identification device
  • control device 19 which is responsible for controlling of the milking system, which inter alia includes the initiation of various activities in connection with the milking such as e.g. opening and closing of the gates 4a and 4b, and control of the milking machine and its handling device 14.
  • the arrangement for determining teat positions comprises an AWS camera 21 for repeatedly recording images of the udder of the milking animal.
  • Image processing means 22 of e.g. the control device 19 forms three-dimensional images and detects repeatedly the teats of the milking animal and determines their positions in all three spatial dimensions based on the repeatedly recorded images of the udder of the milking animal 8.
  • the AWS camera 21 is mounted on the movable robot arm 15 of the milking robot 14 and the control device 19 is thus provided for determining the positions of the teats of the milking animal in all three spatial dimensions relative the movable robot arm 15.
  • the inventive arrangement may further be arranged for determination of the positions and orientations of the teat cups 11, the teat cleaning cup 24, and the brushes 25.
  • Fig. 3 illustrates a further embodiment of the invention wherein the arrangement for determining teat positions is implemented in a voluntary milking system of the above kind.
  • two AWS cameras 21 are mounted in fixed positions with respect to the milking station 3.
  • One AWS camera 21 is arranged to record images of the udder of a milking animal from behind and one AWS camera 21 is arranged to record images of the udder of the milking animal from a side.
  • the image processing means of the control device 19, to which the AWS cameras 21 are operatively connected, is provided for forming three-dimensional images from, and determining the positions of the teats of the milking animal in all three spatial dimensions relative the milking station 3 based on, images recorded by the two AWS cameras 21.
  • inventive arrangement for determining teat positions of the present invention may be implemented in virtually any kind of milking system where the teat positions need to be found automatically.
  • Fig. 4 shows a novel arrangement for determining the identity of a milking animal based on the above AWS concept.
  • An AWS camera 41 is directed towards the milking animal 42, or a part thereof, wherein the AWS camera 41 is arranged to record images of the milking animal, or the part thereof.
  • An image processing device 43 is provided for forming a three-dimensional image from the images recorded by the AWS camera 41; for digitally processing the formed three-dimensional image; and for determining the identity of the milking animal among a group of milking animals based on comparisons between the digitally processed three-dimensional image and (i) physical parameters characteristic for each milking animal of the group of milking animals or (ii) previously recorded digitally processed three- dimensional images of each milking animal of the group of milking animals.
  • the AWS camera 41 is provided to record the images from a position below the udder of the milking animal 42.
  • the AWS camera 41 is positioned elsewhere and/or oriented differently.
  • the physical parameters characteristic for the milking animals can be found by detailed analysis of digital images of the milking animals, e.g. as recorded by the AWS camera 41.
  • the above arrangement is less complicated than prior art systems.
  • the milking animals do not have to be provided with transducers or tags.
  • An AWS camera which may already be provided for other purposes such as e.g. the above determination of teat positions, can be used for the identification.
  • less equipment is needed for the achievement of a certain number of functions of the milking system.
  • the AWS imaging technology provides for rapid and direct identification of milking animals.
  • the arrangement may be provided for detecting damaged or swollen teats or any other abnormal teat condition and to alert a dairy farmer of such condition.
  • the above arrangement may be implemented in a milking system or a milking stall, but alternatively it may be implemented in a feeding station, at a gate arrangement, or elsewhere at a dairy farm.
  • Fig. 5 shows a novel arrangement for determining the weight of a milking animal 52 based on the above AWS concept.
  • An AWS camera 51 is directed towards the milking animal 52, wherein the AWS camera 51 is provided to record images of the milking animal.
  • An image processing device 53 is provided for forming a three-dimensional image from the recorded images; for digitally processing the formed three-dimensional image; and for determining the weight of the milking animal based on the digitally processed three-dimensional image and a table which correlates dimensions or volumes with weights for the species or breed, to which the milking animal belongs.
  • the AWS camera 51 is located above the milking animal 52 and is directed downwards towards the milking animal
  • the camera may be located at an angle CC with respect to a vertical plane parallel with a longitudinal direction of the milking animal 52.
  • the arrangement comprises a second AWS camera (not illustrated) directed towards the milking animal, wherein the AWS cameras are provided to record images of the milking animal at different view angles, wherein the image processing device 53 is provided for forming three-dimensional images from the recorded images; for digitally processing the recorded three-dimensional images; and for determining the weight of the milking animal based on the digitally processed three- dimensional images and the table which correlates dimensions with weights for the breed, to which the milking animal belongs .
  • the images of the milking animal at different view angles may be recorded by a single AWS camera, which is movable between at least two different positions.
  • the AWS camera may be movable between two positions located at each side of a vertical plane parallel with a longitudinal axis of the animal.
  • the above arrangement is less bulky than a conventional weighing device and the arrangement may be used for other purposes as well.
  • the arrangement may be implemented at any location of a dairy farm. It shall be noted that the arrangement of Fig. 5 may be arranged for determining or calculating a body score index (BSI) , which can be used as a parameter related to the health of the milking animal. While any of the above described camera locations may be used for body score index determinations, it seems like that locating the AWS camera 51 above the milking animal 52 at an angle CC with respect to a vertical plane parallel with a longitudinal direction of the milking animal 52 and directing the camera diagonally downwards towards the milking animal 52 is a preferred option.
  • BAI body score index
  • a calculation of the volume can be made. From a reference value for each milking animal stored in a database, a current body score index can be determined.
  • a multifunctional arrangement may be provided for determining teat positions, for identifying animals, and for weighing animals.

Landscapes

  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Animal Husbandry (AREA)
  • Environmental Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Length Measuring Devices By Optical Means (AREA)

Abstract

An arrangement is provided for determining positions of the teats of a milking animal (8) in a milking system (3) comprising a milking stall (5) for housing the milking animal during milking, a movable robot arm (15) for automatically attaching teat cups (11) to the teats of the milking animal in the milking stall, and a control device (19) for controlling the movement of the robot arm based on determined positions of the teats of the milking animal. The arrangement comprises an active wavefront sampling camera (21) directed towards the udder of the milking animal in the milking stall, wherein the active wavefront sampling camera is provided to repeatedly record images of the udder of the milking animal; and image processing means (22) provided for repeatedly forming three-dimensional images from the images recorded by said active wavefront sampling camera; and for detecting the teats of the milking animal and determining their positions in all three spatial dimensions based on the repeatedly formed three-dimensional images.

Description

ARRANGEMENT AND METHOD FOR DETERMINING POSITIONS OF THE TEATS OF A MILKING ANIMAL
TECHNICAL FIELD OF THE INVENTION
The present invention generally relates to dairy farm robot milking and to automatic attachment of teat cups related thereto.
DESCRIPTION OF RELATED ART AND BACKGROUND OF THE INVENTION
EP 1170987 discloses a method for recognizing and determining the position of at least one teat of a milking animal, which includes the steps: moving a scanning head including a light source to a region containing the teat or teats, capturing at least one image formed by said light, evaluating the image or images so as to determine if each image describes said teat or teats. The scanning head (1) is moved to a fixed initial position (A) in the room which is under the animal and clearly in front of an udder and thereby the teats of all known relevant animals, thereafter the scanning head is moved in determined steps (A - J) under the animal: upwards towards the animal, downwards, backwards towards the udder, upwards, downwards etc., while carrying out the scanning procedure. The invention also concerns an apparatus carrying out the process and a milking robot including such an apparatus .
A drawback of such an approach is that the scanning head including the light source has to be moved extensively. Such repeated scanning is time consuming, which renders the method slow. Further, the scanning head is mounted on a robot arm, which has to be moved up and down in order to find the teats of the animal. Such movement of the entire robot arm just for detecting the teats of the animal seems to be inflexible and slows down the process yet further.
WO 03/055297 discloses a method for determining the position of at least one teat of an udder of an animal for milking, comprising illuminating points on the udder with at least one directed light source in a predetermined pattern; detecting points illuminated with the light source in the pattern with recording means; determining position data of illuminated points from a detection of the recording means on the basis of triangulation; forming a three-dimensional representation of the udder from the position data and the pattern; recognizing the teat in the three-dimensional representation; and determining the location of the teat on the basis of the recognition and with the position data.
A drawback of such a method is that it seems to be complex and the determination of position data of illuminated points from a detection of the recording means on the basis of triangulation takes time, which will cause the vision system to be slow. As a consequence various activities depending on such determination, such as the attachment of teat cups to the teats of a milking animal, will take considerable time. Probably, scanning over the entire udder is needed for each teat cup attachment since the animal has time to move between the attachments. Further, there is a risk that the animal has time to move during the scanning or between scanning and teat cup attachment. Thus, the method may be very slow due to the extensive scanning.
In Research Disclosure, July 2002, p. 1162 is disclosed the use of an arrangement comprising a number of cameras fixedly mounted in the milking stall instead of using a video camera mounted on a movable robot arm. For instance, two or three video cameras can be mounted at each side of the milking animal to be milked, preferably on the walls of the milking stall or just outside thereof. Advantageously, the video cameras are directed diagonally upwards towards the region where the teats are when the milking animal has been positioned in the milking stall. A computer is provided for e.g. selecting two cameras, which together create a stereoscopic image, which by means of image processing enables a substantially exact determination of the position of a teat. Two further cameras may be used to confirm the three-dimensional position of a teat. A robot arm is then moved to the teat based on the determined position.
While such an arrangement has several advantages such as a faster determination of the teats of the milking animal, a smaller and lighter robot arm, possibilities to better protect the cameras from dirt and from being damaged by kicks from the milking animal, and capabilities to monitor the complete milking operation for e.g. detecting a teat cup falling off the teat, it may be difficult to obtain a sufficiently robust, accurate, precise, and fast implementation, which is capable of controlling the robot arm to obtain a sufficiently high number of correct teat cup attachments.
A milking environment is an environment, in which it is very difficult to perform stereo vision measurements. The environment is non-clean and dirt may settle on camera lenses. Further the milking animals are moving, and teats may not be visible to the cameras due to self-occlusion.
Another problem arises since each milking animal's physiology differs; the udders of the milking animals may be located at quite different locations, which puts limitations on the positions of the cameras. For instance, the area where the teats most likely are has been found by empirical studies to be at least 500 x 600 x 480 mm3 large.
Still another problem arises since both color and texture of the teats are similar to the surface of the udder, which means that teat detection will be an arduous task: the contrast is low and color filters are of no use. The situation is even more complicated by the fact that the size, shape, color structure, morphological structure and texture may vary quite much from animal to animal . SUMMARY OF THE INVENTION
Accordingly, it is an object of the present invention to provide an arrangement and a method for determining positions of the teats of a milking animal in a milking system comprising a milking stall for housing the milking animal during milking, a movable robot arm for automatically attaching teat cups to the teats of the milking animal in the milking stall, and a control device for controlling the movement of the robot arm based on determined positions of the teats of the milking animal, which arrangement and method solve at least some of the problems of the prior art as set forward above.
It is a further object of the invention to provide such an arrangement and such a method, which are robust, effective, fast, precise, accurate, reliable, safe, easy to use, and of low cost.
It is still a further object of the invention to provide such an arrangement and such a method, which are capable of obtaining a very high number of correct teat cup attachments.
These objects among others are, according to the present invention, attained by arrangements and methods as claimed in the appended patent claims.
According to one aspect of the invention an arrangement for determining positions of the teats of a milking animal comprises an active wavefront sampling camera and image processing means. The active wavefront sampling camera is provided for being directed towards the udder of the milking animal in the milking stall and for repeatedly recording images of the udder of the milking animal. The image processing means is provided for repeatedly forming three-dimensional images from the images repeatedly recorded by the active wavefront sampling camera and for detecting the teats of the milking animal and determining their positions in all three spatial dimensions based on the repeatedly formed three-dimensional images. By such arrangement the control device may control the robot arm very fast and accurately. The image recordings by such camera and the subsequent processing of the measurement data provide high quality information extremely fast and as a result the milking system is capable of operating at higher speed. The milking time will be shorter and the throughput of animals is increased.
In one embodiment the inventive arrangement is implemented in a rotary milking system, wherein the active wavefront sampling camera is preferably located at a fix position with respect to a floor, on which the rotary milking system is installed and with respect to which a rotary platform of the rotary milking system rotates during milking. By such provisions a single active wavefront sampling camera can serve several or all of the milking stalls of the milking system. The fast operation of the inventive arrangement is required since the milking stalls are moving (rotating) with respect to the active wavefront sampling camera camera.
In such a milking system the position of the teat cups when being stored in e.g. a magazine, i.e. when not being used, may not be known. However, the active wavefront sampling camera of the inventive arrangement may be directed towards teat cups located in the magazine and provided to repeatedly record images of the teat cups. Hereby the image processing means is capable of forming three-dimensional images from the images recorded by the active wavefront sampling camera and detecting the teat cups and determining their positions in all three spatial dimensions in the magazine based on the formed three- dimensional images.
In another embodiment of the invention the active wavefront sampling camera is mounted on the movable robot arm and the image processing means determines the positions of the teats of the milking animal in all three spatial dimensions relative the movable robot arm. Additionally positions for other objects such as legs, udders, bodies, magazines, and teats can be determined by an active wavefront sampling camera mounted on the movable robot arm.
In still another embodiment of the invention the active wavefront sampling camera is mounted in a fixed position with respect to the milking stall and the image processing means determines the positions of the teats of the milking animal in all three spatial dimensions relative the milking stall.
The active wavefront sampling camera can be located at a side of the milking animal and be directed essentially sideways, i.e., horizontally, towards the udder of the milking animal. Alternatively, the active wavefront sampling camera can be located below the milking animal and be directed upwards towards the udder of the milking animal.
The above positions/orientations of the camera seem to be the most favorable ones in order to fast locate the udder and the teats of the milking animals. The camera positions/orientations seem to provide the best contrast between the udder/teats and other objects in the view field of the camera. Other orientations may be conceivable.
Further, the camera may be movable between several positions/orientations depending on the application or status of the camera.
Further embodiments of the inventive arrangement are set out in the dependent claims.
According to a second aspect of the invention a method for determining positions of the teats of a milking animal is provided. According to the method an active wavefront sampling camera is directed towards the teats of the milking animal in the milking stall. Images of the teats of the milking animal are repeatedly recorded by said active wavefront sampling camera and three-dimensional images of the teats of the milking animal are repeatedly formed from the images repeatedly recorded by the active wavefront sampling camera. Finally, the teats of the milking animal are repeatedly detected and their positions in all three spatial dimensions are repeatedly determined based on the repeatedly formed three-dimensional images .
An advantage of the present invention is that the determination of the positions of the teats of an animal in all three spatial coordinates is made very fast. The invention provides in particular for the recording of images by the active wavefront sampling camera, wherein all visible teats can be detected, and the positions of all the teats can be determined, simultaneously. Hence, teat cup attachment can be made very fast .
Further characteristics of the invention, and advantages thereof, will be evident from the following detailed description of preferred embodiments of the present invention given hereinafter and the accompanying Figs. 1-5, which are given by way of illustration only, and are thus not limitative of the present invention .
BRIEF DESCRIPTION OF THE DRAWINGS
Figs. 1-3 display each a schematically outlined milking system including an arrangement for determining positions of the teats of a milking animal according to a respective embodiment of the present invention. Figs. 1 and 3 are top views while Fig. 2 is a perspective view.
Fig. 4 displays schematically, in a top view, an arrangement for determining the identity of a milking animal according to an embodiment of the present invention. Fig. 5 displays schematically, in a side elevation view, an arrangement for determining the weight of a milking animal according to an embodiment of the present invention.
DETAILED DESCRIPTION OF EMBODIMENTS
In Fig. 1 is shown a milking system, in which an arrangement for determining positions of the teats of milking animals according to an embodiment of the invention is implemented. The rotary milking system 3 comprises a plurality of milking stalls 5, which milking animals 8 enter in a sequential order. Each of the milking stalls 5 comprises milking equipment including teat cups that are attached to the teats of the milking animal present in the milking stall prior to milking. For sake of simplicity teat cups 11 are illustrated only for one of the milking stalls 5. The rotary milking system 3 may be of parallel, tandem, or herringbone configuration. In the parallel configuration the longitudinal directions of the milking stalls and of the milking animals therein extend radially (the milking animals stand side by side) , in the tandem configuration the longitudinal directions of the milking stalls and of the milking animals therein extend circumferentially (the milking animals stand one after the other) , and in the herringbone configuration, which is illustrated in Fig. 1, the longitudinal directions x of the milking stalls and of the milking animals therein extends partly radially, partly circumferentially .
A robot 14 provided with a robot arm 15 is provided for automatically attaching teat cups 11 to the teats of the milking animals 8 present in the milking stalls 5 under the control of a control device 19, which is operatively connected to the milking robot 14. The milking robot 14 is preferably stationary with respect to a rotatable carousel or rotating platform 20 of the rotary milking system 3, which forms the support for the milking stalls 5. Alternatively, the milking robot 14 is movable back and forth in e.g. a circumferential direction .
The rotating platform 20 may, for each of the milking animals, be kept still while the milking robot 14 automatically attaches teat cups 11 to the teats of the milking animal 8, and is rotated there in between. Alternatively, the rotating platform is rotated continuously during the attachment of the teat cups and the milking of the milking animals 8.
In order to determine positions of the teats of the milking animals 8 present in the milking stalls 5, and thus be capable of moving the teat cups 11 to the teats of the milking animal, an active wavefront sampling camera 21 is provided.
Active wavefront sampling (AWS) is a method of imaging in three dimensions from a single lens based on imaging the wavefront of a lens in time. In its simplest form, AWS is little more than an off-axis aperture placed in an optical path that rotates such that images are recorded from different sections along a circular path of a lens. Despite a relatively small baseline, very high accuracy can be achieved relative to stereo based 3D because alignment error is virtually eliminated and multiple perspectives provide significant processing redundancy, Fig. 1. Moving objects can be imaged in three dimensions using AWS as long as its motion is steady (either velocity or acceleration) relative to the framing rate of the camera. The inventors of the present invention have concluded that AWS lends itself well to applications for teat detection in automatic milking systems .
For further details regarding the AWS technique and the AWS camera reference is made to the following publications, the contents of which being hereby incorporated by reference: Rohaly J. and Hart D. P. (2001), Monocular 3-D Active micro-PTV, 4th International Symposium on Particle Image Velocimetry, Gottingen, Germany, September 17-18, 2001;
Frigerio F. and Hart D. P. (2006), Calibrationless Aberration Correction through Active Wavefront Sampling (AWS) and Multi- Camera Imaging, 13th Int Symp on Applications of Laser Techniques to Fluid Mechanics, Lisbon, Portugal, 26-29 June, 2006;
Prakash H. (2007), The Active Wave-front Sampling based 3D endoscope, Thesis (S.M.), Massachusetts Institute of Technology, Dept . of Mechanical Engineering; and, (S.M.), Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2007;
Frigerio F. (2006), 3-dimensional surface imaging using Active Wavefront Sampling, Thesis (Ph. D.), Massachusetts Institute of Technology, Dept. of Mechanical Engineering, 2006; and
WO 2007/095307 Al.
The AWS camera 21 is, for each of the milking animals in the milking stalls, directed towards the udder of the milking animal, wherein the AWS camera repeatedly records images of the udder of the milking animal in real time.
Preferably, the AWS camera comprises an aperture arrangeable in at least three different positions during the repeated recording of images of the udder of the milking animal.
Image processing means 22 is provided, for each of the milking animals, for repeatedly forming three-dimensional images from the images recorded by the AWS camera 21; and detecting the teats of the milking animal and determining their positions in all three spatial dimensions by a calculation method based on the repeatedly formed three-dimensional images. The AWS camera 21 together with the image processing means 22 thus determine directly the position in all three spatial coordinates for each object point that is found in the pixels of the image recorded.
The image processing means 22 is preferably implemented as a software module in the control device 19, which thus is operatively connected to the AWS camera 21, or in any other device operatively connected to the AWS camera 21 and the control device 19. Alternatively, the image processing means 22 is integrated in the AWS camera 21.
The calculation method used can be any calculation method suitable for use in the field of digital image processing.
The AWS camera 21 can be mounted in fixed positions with respect to the different parts of the milking system 3 depending on the application in question. The camera 21 can be located at a fix position with respect to a floor, on which the rotary milking system 3 is installed and with respect to which the rotary platform 20 of the rotary milking system 3 rotates during milking. In such instance, the camera 21 may be mounted in a ceiling of a building in which the milking system is installed. Alternatively, the camera 21 is mounted on the rotating platform 20 or on the robot arm 15. Still alternatively, the camera 21 is movable along some path. In one version, for instance, the AWS camera 21 can be movable between a first position, in which the camera 21 is located during the detections of the images, and a second position, in which the camera 21 is located there in between, that is, when being idle.
The AWS camera 21 can further be located in a number of different positions and orientations depending on the application in question. The camera 21 can, during each of the recordings of the images of the udder of each of the milking animals, be located at a side of the milking animal and be directed essentially sideways in the horizontal plane and perpendicular to a longitudinal direction of the milking animal. Alternatively, the camera 21 can, during each of the recordings of the images of the udder of each of the milking animals, be located below the milking animal and be directed upwards.
Still further, two or more AWS cameras can be provided operatively connected to the control device 19 in order to provide further detailed three-dimensional information of the positions and orientations of the teats of the milking animals.
For instance, the use of several of the AWS cameras seems to be advantageous if they are not mounted on the robot arm since at least one teat would always be obscured. Further, one or several
AWS cameras arranged stationary on the floor or on the rotating platform could be combined with one AWS camera mounted on the robot arm, or a single AWS camera mounted on the robot arm could be used in the invention.
The speed and accuracy of the arrangement for determining teat positions are of outermost importance in order to provide a milking system with accurate and fast automatic attachments of teat cups to teats of milking animals. This is of particular importance in a rotary milking system where not only the robot arm for attaching the teat cups is movable, but also the entire robot may be movable. Further, the rotating platform and the milking stalls are moving (rotating) .
Since also the milking stalls and the milking equipments belonging to the milking stalls are moving it is typically not known - at least not in a fixed coordinate system - where the teat cups are located, particularly when they are stored in a magazine (not illustrated) in each of the milking stalls.
In a rotary milking system it may be particularly difficult to find the magazines since they may appear in different positions in different milking stalls, and the position of a magazine in a milking stall can be stored only if the position of the milking stall is monitored.
The AWS camera 21 of the inventive arrangement may thus be provided to record images of the teat cups 11 in real time, particularly when the teat cups 11 are located in the magazine, and the image processing means 22 forms three-dimensional images from the recorded images, and detects the teat cups and determines their positions in all three spatial dimensions by the above calculation method based on the recorded images of the teat cups when being located in the magazine.
In another embodiment of the present invention the AWS camera 21 can be used to establish a position of each of the milking animals in at least one spatial dimension, e.g. a longitudinal direction x or in a direction perpendicular thereto, with respect to the milking stalls in which they are housed. Thus, the AWS camera 21 is, for each of the milking animals, directed towards a leg or an outer contour of the milking animal in the milking stall, wherein the AWS camera 21 records images of the leg or the outer contour of the milking animal. The image processing means 22 forms a three-dimensional image from the recorded images, and detects the milking animal and determines its position in the spatial dimension by the calculation method based on the recorded images of the leg or the outer contour of the milking animal.
Such embodiment may be advantageous in a milking system where a detection device is used to detect the teats of a milking animal in a milking stall based on the prior knowledge of where in the milking stall the milking animal is located and possibly on the physical dimensions of the milking animal itself.
Thus, the control device 19 may, for each of the milking animals, hold information in, or receive information from, a database 23 regarding the position of the udder of the milking animal relative the milking animal itself. This may be a single approximate figure valid for all the animals. Further, the control device 19 directs the AWS camera 21 and controls the robot arm 15 of the milking robot 14 to move towards the udder of the milking animal based on the information of the position of the at least one teat of the milking animal relative the milking animal itself, and on the detected position of the milking animal in the spatial dimension relative the milking stall.
The information of the position of the at least one teat of the milking animal relative the milking animal itself can be deduced from the recording of visual detections of the milking animal in connection with an earlier milking of the milking animal .
This arrangement is particularly advantageous in situations where one or more teats are obscured and is/are not clearly detectable by the AWS camera 21. Thus, the control device 19 may, for each of the milking animals, hold information in, or receive information from, the database 23 regarding the position of each of the teats of the milking animal relative the other teats of the milking animal. Further, the control device 19 controls the robot arm 15 of the milking robot 14 to move a teat cup to an obscured teat of the milking animal based on the information of the position of each of the teats of the milking animal relative the other teats of the milking animal and on a determined position of a repeatedly detected teat of the milking animal, which is not obscured.
In one version the above algorithm is applied in an arrangement where the AWS camera 21 is, during each of the recordings of the images, located behind the milking animal and is directed forward towards the back of the milking animal. The spatial dimension determined is then preferably in a direction perpendicular to a longitudinal direction of the milking stall. The leg or outer contour of the milking animal may include one or both back legs of the milking animal, preferably the inner contours of the back legs.
The determination of the positions of the teats of the milking animal can be made in the following manner. First the milking animal is detected and the position thereof is determined. Based on this information the contour of the back legs and the body therein between is detected and the positions thereof are determined. Based on this information an area of interest (i.e. where the teats most probably are found) is located, and in this area the udder of the milking animal is searched for. When the udder has been detected and located the teats are searched for and located. Finally, when the positions of the teats have been determined the teat cup attachment can be initiated.
Such algorithm is characterized by fast object recognition and can be applied in situations where a new milking animal is detected or where a milking animal has a position of the legs which renders the determination of the teat positions more difficult.
It shall further be appreciated that the arrangement of the present invention may be used as a multifunctional detection system for detecting any of the following: (i) presence of a milking animal in a milking stall, (ii) the behavior of a milking animal, (iii) the activity of a milking animal, (iv) the body shape of a milking animal, (v) an incorrect teat cup attachment, (vi) a teat cup kick-off, (vii) presence of an obstacle in the working area of the milking robot, (viii) a parameter related to the operation of the milking robot, and (ix) a parameter related to the operation of the camera.
Yet further, the embodiment of Fig. 1 may comprise a light source such as a coherent laser source 12 and a speckle pattern generating arrangement 13, both being preferably operatively connectable to the control device 19 and/or the image processing means 22. The light source 12 and the speckle pattern generating arrangement 13 are provided to illuminate the object to be detected by the AWS camera 21 by light having a speckle pattern. The light source 12 and the speckle pattern generating arrangement 13 creates an illuminated speckle pattern on the object, which can be used for correlating pixels from different images recorded by the AWS camera 21 to one another. This may be particularly advantageous when detecting a teat since color and texture of the teat surface are similar to that of the udder surface and thus it may be difficult to interpret the recorded images. The situation is even more complicated by the fact that the size, shape, color structure, morphological structure and texture may vary quite much from animal to animal.
The light source 12 and the speckle pattern generating arrangement 13 thus provide means for creating patterns on the udder and teats of the milking animal which can be used to facilitate or speed up the detection.
With reference next to Fig. 2 a further embodiment of the present invention will be described. The arrangement for determining teat positions is here implemented in a voluntary milking system or station 3 comprising an enclosure having an inlet gate 4a and an outlet gate 4b, which are both capable of being opened automatically. The front end of the milking station 3 is denoted by 3a, the back end is denoted by 3b, the sides are denoted by 3c and 3d.
The milking station 3 comprises further an automatic milking machine (not explicitly illustrated) including teat cups 11 connected to an end unit by means of milk lines (only the portions attached to the teat cups 11 are shown in Fig. 2) . The milking station further includes a milking robot 14 having a movable robot arm 15 provided with a gripper. The milking robot 14 is arranged to automatically apply the teat cups 11 of the milking machine to the teats of a milking animal 8 present in the milking station 3 prior to milking. In Fig. 2 three of the teat cups 11 are arranged in a teat cup rack or magazine 16, whereas the fourth one is held by the gripper of the robot arm 15. Typically, a teat cleaning device including e.g. a teat cleaning cup 24 or brushes 25 may be provided for cleaning the teats of the milking animal 8 prior to milking.
Further, the milking station 3 comprises an identification device
(not illustrated) provided to identify a milking animal approaching the milking station 3, and a control device 19, which is responsible for controlling of the milking system, which inter alia includes the initiation of various activities in connection with the milking such as e.g. opening and closing of the gates 4a and 4b, and control of the milking machine and its handling device 14.
The arrangement for determining teat positions comprises an AWS camera 21 for repeatedly recording images of the udder of the milking animal. Image processing means 22 of e.g. the control device 19 forms three-dimensional images and detects repeatedly the teats of the milking animal and determines their positions in all three spatial dimensions based on the repeatedly recorded images of the udder of the milking animal 8.
The AWS camera 21 is mounted on the movable robot arm 15 of the milking robot 14 and the control device 19 is thus provided for determining the positions of the teats of the milking animal in all three spatial dimensions relative the movable robot arm 15.
The inventive arrangement may further be arranged for determination of the positions and orientations of the teat cups 11, the teat cleaning cup 24, and the brushes 25.
Fig. 3 illustrates a further embodiment of the invention wherein the arrangement for determining teat positions is implemented in a voluntary milking system of the above kind.
Here, two AWS cameras 21 are mounted in fixed positions with respect to the milking station 3. One AWS camera 21 is arranged to record images of the udder of a milking animal from behind and one AWS camera 21 is arranged to record images of the udder of the milking animal from a side.
The image processing means of the control device 19, to which the AWS cameras 21 are operatively connected, is provided for forming three-dimensional images from, and determining the positions of the teats of the milking animal in all three spatial dimensions relative the milking station 3 based on, images recorded by the two AWS cameras 21.
It shall be appreciated that the inventive arrangement for determining teat positions of the present invention may be implemented in virtually any kind of milking system where the teat positions need to be found automatically.
There are many prior art techniques for determining the identities of milking animals, however, many of them being complex or requiring the providing of each of the milking animals to be identified with a transducer or tag.
Fig. 4 shows a novel arrangement for determining the identity of a milking animal based on the above AWS concept. An AWS camera 41 is directed towards the milking animal 42, or a part thereof, wherein the AWS camera 41 is arranged to record images of the milking animal, or the part thereof. An image processing device 43 is provided for forming a three-dimensional image from the images recorded by the AWS camera 41; for digitally processing the formed three-dimensional image; and for determining the identity of the milking animal among a group of milking animals based on comparisons between the digitally processed three-dimensional image and (i) physical parameters characteristic for each milking animal of the group of milking animals or (ii) previously recorded digitally processed three- dimensional images of each milking animal of the group of milking animals. Preferably, the AWS camera 41 is provided to record the images from a position below the udder of the milking animal 42. Alternatively, the AWS camera 41 is positioned elsewhere and/or oriented differently.
The physical parameters characteristic for the milking animals can be found by detailed analysis of digital images of the milking animals, e.g. as recorded by the AWS camera 41.
The above arrangement is less complicated than prior art systems. The milking animals do not have to be provided with transducers or tags. An AWS camera, which may already be provided for other purposes such as e.g. the above determination of teat positions, can be used for the identification. Thus, less equipment is needed for the achievement of a certain number of functions of the milking system.
Further, the AWS imaging technology provides for rapid and direct identification of milking animals.
Still further, the arrangement may be provided for detecting damaged or swollen teats or any other abnormal teat condition and to alert a dairy farmer of such condition.
It shall be appreciated that the above arrangement may be implemented in a milking system or a milking stall, but alternatively it may be implemented in a feeding station, at a gate arrangement, or elsewhere at a dairy farm.
In the prior art the weights of milking animals are typically determined by weighing devices, onto which the milking animals are led or guided. Such devices, however, are bulky and costly.
Fig. 5 shows a novel arrangement for determining the weight of a milking animal 52 based on the above AWS concept. An AWS camera 51 is directed towards the milking animal 52, wherein the AWS camera 51 is provided to record images of the milking animal. An image processing device 53 is provided for forming a three-dimensional image from the recorded images; for digitally processing the formed three-dimensional image; and for determining the weight of the milking animal based on the digitally processed three-dimensional image and a table which correlates dimensions or volumes with weights for the species or breed, to which the milking animal belongs.
Preferably, the AWS camera 51 is located above the milking animal 52 and is directed downwards towards the milking animal
52. The camera may be located at an angle CC with respect to a vertical plane parallel with a longitudinal direction of the milking animal 52.
In one version, the arrangement comprises a second AWS camera (not illustrated) directed towards the milking animal, wherein the AWS cameras are provided to record images of the milking animal at different view angles, wherein the image processing device 53 is provided for forming three-dimensional images from the recorded images; for digitally processing the recorded three-dimensional images; and for determining the weight of the milking animal based on the digitally processed three- dimensional images and the table which correlates dimensions with weights for the breed, to which the milking animal belongs .
Alternatively, the images of the milking animal at different view angles may be recorded by a single AWS camera, which is movable between at least two different positions. For instance, the AWS camera may be movable between two positions located at each side of a vertical plane parallel with a longitudinal axis of the animal.
The above arrangement is less bulky than a conventional weighing device and the arrangement may be used for other purposes as well. The arrangement may be implemented at any location of a dairy farm. It shall be noted that the arrangement of Fig. 5 may be arranged for determining or calculating a body score index (BSI) , which can be used as a parameter related to the health of the milking animal. While any of the above described camera locations may be used for body score index determinations, it seems like that locating the AWS camera 51 above the milking animal 52 at an angle CC with respect to a vertical plane parallel with a longitudinal direction of the milking animal 52 and directing the camera diagonally downwards towards the milking animal 52 is a preferred option.
By placing the camera so that the important body parts for body score index can be exposed and then making a computer "cut out" of the model, a calculation of the volume can be made. From a reference value for each milking animal stored in a database, a current body score index can be determined.
It shall further be noted that the arrangements of Figs. 4 and 5 may be applied to other animals than milking animals.
It shall be appreciated by a person skilled in the art that various features of the above embodiments can be combined to form yet further embodiments of the present invention. Particularly, a multifunctional arrangement may be provided for determining teat positions, for identifying animals, and for weighing animals.

Claims

1. An arrangement for determining positions of the teats of a milking animal (8) in a milking system (3) comprising a milking stall (5) for housing the milking animal during milking, a movable robot arm (15) for automatically attaching teat cups
(11) to the teats of the milking animal in the milking stall, and a control device (19) for controlling the movement of said robot arm based on determined positions of the teats of the mi l king animal , c h a r a c t e r i z e d i n that said arrangement comprises:
- an active wavefront sampling camera (21) provided for being directed towards the udder of the milking animal in the milking stall, wherein the active wavefront sampling camera is provided to repeatedly record images of the udder of the milking animal; and
- image processing means (22) provided for repeatedly forming three-dimensional images from said images repeatedly recorded by said active wavefront sampling camera; and for detecting the teats of the milking animal and determining their positions in all three spatial dimensions based on said repeatedly formed three-dimensional images.
2. The arrangement of claim 1 wherein said active wavefront sampling camera comprises an aperture arrangeable in at least three different positions during the repeated recording of images of the udder of the milking animal.
3. The arrangement of claims 1 or 2 wherein
- said active wavefront sampling camera is directed towards teat cups located in a magazine (16) , wherein the active wavefront sampling camera is provided to repeatedly record images of the teat cups; and - the image processing means is provided for repeatedly forming three-dimensional images of the teat cups from said images of the teat cups, and for repeatedly detecting the teat cups and determining their positions in all three spatial dimensions based on said repeatedly formed three-dimensional images of the teat cups .
4. The arrangement of any of claims 1-3 wherein said arrangement is provided for determining positions of the teats of the milking animal in all three spatial dimensions in a rotary milking system.
5. The arrangement of claim 4 wherein said active wavefront sampling camera is, during the recordings of said three- dimensional images, located at a fix position with respect to a floor, on which said rotary milking system is installed and with respect to which a rotary platform of said rotary milking system rotates during milking.
6. The arrangement of any of claims 1-4 wherein
said active wavefront sampling camera is mounted on said movable robot arm; and
- said image processing means is provided for determining the positions of the teats of the milking animal in all three spatial dimensions relative said movable robot arm.
7. The arrangement of any of claims 1-4 wherein
- said active wavefront sampling camera is mounted in a fixed position with respect to said milking stall; and
- said image processing means is provided for determining the positions of the teats of the milking animal in all three spatial dimensions relative said milking stall.
8. The arrangement of any of claims 1-7 wherein said active wavefront sampling camera is movable between a first position, in which said active wavefront sampling camera is located during the recording of said images, and a second position, in which said active wavefront sampling camera is located there in between.
9. The arrangement of any of claims 1-8 wherein
- said active wavefront sampling camera is directed towards a leg or an outer contour of the milking animal in the milking stall, wherein the active wavefront sampling camera is provided to record an image of the leg or the outer contour of the milking animal; and
- the image processing means is provided for forming a three- dimensional image of the leg or the outer contour of the milking animal from said image of the leg or the outer contour of the milking animal, and for detecting the milking animal and determining its position in at least one spatial dimension based on said formed three-dimensional image of the leg or the outer contour of the milking animal.
10. The arrangement of claim 9 wherein said leg or outer contour of the milking animal includes a back leg of the milking animal .
11. The arrangement of claim 9 or 10 wherein
- said image processing means is provided to first detect and determine the position of the milking animal, and to thereafter detect and determine the position of the udder of the milking animal based on the position of the milking animal.
12. The arrangement of any of claims 1-11 wherein
- said active wavefront sampling camera is, during each of the recordings of said images, located behind the milking animal and is directed forward towards the back of the milking animal; and
said at least one spatial dimension is in a direction perpendicular to a longitudinal direction of the respective milking stall.
13. The arrangement of any of claims 1-11 wherein said active wavefront sampling camera is, during each of the recordings of said images, located at a side of the milking animal and is directed essentially sideways.
14. The arrangement of any of claims 1-11 wherein said active wavefront sampling camera is, during each of the recordings of said three-dimensional images, located below the milking animal and is directed upwards.
15. The arrangement of any of claims 1-14 comprising
- a control device (19, 22) for holding or receiving information of the position of each of the teats of the milking animal relative the other teats of the milking animal and for controlling said robot arm to move to an obscured teat of the milking animal based on said information of the position of each of the teats of the milking animal relative the other teats of the milking animal and on the determined position of a repeatedly detected teat of the milking animal, which is not obscured.
16. The arrangement of any of claims 1-15 wherein said arrangement is a multifunctional detection system further provided for detecting any of the following: (i) presence of a milking animal in a milking stall, (ii) the behavior of a milking animal, (iii) the activity of a milking animal, (iv) the body shape of a milking animal, (v) an incorrect teat cup attachment, (vi) a teat cup kick-off, (vii) presence of an obstacle in the working area of the robot arm, (viii) a parameter related to the operation of the robot arm, and (ix) a parameter related to the operation of the three-dimensional camera.
17. The arrangement of any of claims 1-16 wherein
- said active wavefront sampling camera is directed towards the milking animal, or a part thereof, wherein the active wavefront sampling camera is provided to record further images of the milking animal, or the part thereof; and
- the image processing means is provided for forming a three- dimensional image of the milking animal, or the part thereof, from the further images recorded by the active wavefront sampling camera; for digitally processing the further three- dimensional image; and for determining the identity of the milking animal based on comparisons between the digitally processed further three-dimensional image and (i) physical parameters characteristic for each of a group of milking animals or (ii) previously recorded digitally processed three- dimensional images of each of a group of milking animals.
18. The arrangement of claim 17 wherein said active wavefront sampling camera is provided for recording the images of the milking animal, or the part thereof, as images of the udder of the milking animal, preferably from a position below the udder of the milking animal.
19. The arrangement of any of claims 1-18 wherein
- said active wavefront sampling camera is directed towards the milking animal, wherein the active wavefront sampling camera is provided to record yet further images of the milking animal; and
- the image processing means is provided forming a yet further three-dimensional image from the yet further images recorded by the active wavefront sampling camera; for digitally processing the yet further three-dimensional image; and for determining the weight of the milking animal based on the digitally processed yet further three-dimensional image and a table which correlates dimensions or volumes with weights for the species or breed, to which the milking animal belongs.
20. The arrangement of claim 19 wherein said active wavefront sampling camera is located above the milking animal and is directed downwards towards the milking animal during the recording of the further image of the milking animal.
21. The arrangement of claim 19 wherein
- said arrangement comprises a second active wavefront sampling camera (21) directed towards the milking animal, wherein the active wavefront sampling cameras are provided to record images of the milking animal at different view angles; and
- the image processing means is provided for forming three- dimensional images from the images of the milking animal recorded at different view angles by the active wavefront sampling cameras; for digitally processing the three- dimensional images formed from the images recorded at different view angles; and for determining the weight of the milking animal based on the digitally processed three-dimensional images formed from the images of the milking animal recorded at different view angles, and the table which correlates dimensions with weights for the breed, to which the milking animal belongs.
22. The arrangement of any of claims 1-21 further comprising a light source (12) and a speckle pattern generating arrangement
(13) provided to illuminate the udder of the milking animal, during the recordings of the images, by light having a speckle pattern.
23. A milking system (3) comprising a milking stall (5) for housing the milking animal during milking; a movable robot arm (15) for automatically attaching teat cups to the teats of a milking animal in the milking stall; a control device (19) for controlling the movement of said robot arm based on determined positions of the teats of the milking animal; and the arrangement of any of claims 1-22.
24. A method for determining positions of the teats of a milking animal (8) in a milking system (3) comprising a milking stall (5) for housing the milking animal during milking, a movable robot arm (15) for automatically attaching teat cups to the teats of the milking animal in the milking stall, and a control device (19) for controlling the movement of said robot arm based on determined positions of the teats of the milking animal, said method being cha r a c t e r i z e d b y the steps of:
- directing an active wavefront sampling camera (21) towards the teats of the milking animal in the milking stall;
repeatedly recording images of the teats of the milking animal by said active wavefront sampling camera;
- repeatedly forming three-dimensional images of the teats of the milking animal from the images repeatedly recorded by said active wavefront sampling camera; and
repeatedly detecting the teats of the milking animal and determining their positions in all three spatial dimensions based on said repeatedly formed three-dimensional images.
25. The method of claim 24 wherein said active wavefront sampling camera directed towards the teats of the milking animal in the milking stall comprises an aperture arrangeable in at least three different positions during the repeated recording of images of the udder of the milking animal.
26. The method of claim 24 or 25 comprising illuminating the udder of the milking animal, during the recordings of the images, by light having a speckle pattern by means of a light source (12) and a speckle pattern generating arrangement (13) .
27. An arrangement for determining positions of the teats of a milking animal (8) in a milking system (3) comprising a milking stall (5) for housing the milking animal during milking, a movable robot arm (15) for automatically attaching teat cups to the teats of the milking animal in the milking stall, and a control device (19) for controlling the movement of said robot arm based on determined positions of the teats of the milking animal, cha r a c t e r i z e d i n that said arrangement comprises :
- an active wavefront sampling camera (21) provided for being directed towards the udder of the milking animal in the milking stall, wherein the active wavefront sampling camera is provided to repeatedly record images of the udder of the milking animal;
- image processing means (22) provided for repeatedly forming three-dimensional images of the udder of the milking animal from the images repeatedly recorded by said active wavefront sampling camera; and for detecting a teat of the milking animal, which is not obscured, and determining its position in all three spatial dimensions based on said repeatedly formed three- dimensional images; and
- means for holding or receiving information of the position of each of the teats of the milking animal relative the other teats of the milking animal and for controlling said robot arm to move to an obscured teat of the milking animal based on said information of the position of each of the teats of the milking animal relative the other teats of the milking animal and on the determined position of the repeatedly detected teat of the milking animal, which is not obscured.
28. An arrangement for determining the identity of an animal (42) , cha r a c t e r i z e d i n that said arrangement comprises :
- an active wavefront sampling camera (41) provided for being directed towards the animal, or a part thereof, wherein the active wavefront sampling camera is provided to record images of the animal, or the part thereof; and
image processing means (43) provided for forming a three- dimensional image of the animal, or the part thereof, from said images recorded by said active wavefront sampling camera; for digitally processing the recorded three-dimensional image; and for determining the identity of the animal among a group of animals based on comparisons between the digitally processed three-dimensional image and (i) physical parameters characteristic for each animal of the group of animals or (ii) previously recorded digitally processed three-dimensional images of each animal of the group of animals.
29. An arrangement for determining the weight of an animal (52) , cha r a c t e r i z e d i n that said arrangement comprises:
- an active wavefront sampling camera (51) provided for being directed towards the animal, wherein the three-dimensional camera is provided to record images of the animal; and
- image processing means (53) provided for forming a three- dimensional image of the animal from the images recorded by said active wavefront sampling camera; for digitally processing the recorded three-dimensional image; and for determining the weight of the animal based on the digitally processed three- dimensional image and a table which correlates dimensions or volumes with weights for the species or breed, to which the animal belongs.
30. An arrangement for determining the body score index of an animal (52) , cha r a c t e r i z e d i n that said arrangement comprises :
- an active wavefront sampling camera (51) provided for being directed towards the animal, wherein the active wavefront sampling camera is provided to record images of the animal; and
image processing means (53) provided for forming a three- dimensional image of the animal from the images recorded by said active wavefront sampling camera; for digitally processing the recorded three-dimensional image; and for determining the body score index of the animal based on the digitally processed three-dimensional image.
PCT/EP2009/060583 2008-08-29 2009-08-14 Arrangement and method for determining positions of the teats of a milking animal WO2010023122A2 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
SE0801861-6 2008-08-29
SE0801861 2008-08-29

Publications (2)

Publication Number Publication Date
WO2010023122A2 true WO2010023122A2 (en) 2010-03-04
WO2010023122A3 WO2010023122A3 (en) 2010-12-16

Family

ID=41722011

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2009/060583 WO2010023122A2 (en) 2008-08-29 2009-08-14 Arrangement and method for determining positions of the teats of a milking animal

Country Status (1)

Country Link
WO (1) WO2010023122A2 (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2013187821A1 (en) 2012-06-12 2013-12-19 Delaval Holding Ab Arrangement and method for a milking system
US8794181B2 (en) 2011-03-17 2014-08-05 Mirobot Ltd. System and method for three dimensional teat modeling for use with a milking system
US20150366156A1 (en) * 2013-02-06 2015-12-24 Delaval Holding Ab Teat treatment method and apparatus
EP3241431A1 (en) * 2016-05-03 2017-11-08 Lely Patent N.V. Method for performing a teat-related operation, and robot device therefor
US9848575B2 (en) 2011-03-17 2017-12-26 Mirobot Ltd. Human assisted milking robot and method
US10701905B2 (en) 2017-12-06 2020-07-07 International Business Machines Corporation Imaging and three dimensional reconstruction for weight estimation

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
SU1777728A1 (en) * 1990-07-12 1992-11-30 Inst Mash Im A A Blagonravova Automated milking plant
RU2004964C1 (en) * 1991-06-14 1993-12-30 Ленинградский сельскохоз йственный институт Animal identification method and device
RU2065573C1 (en) * 1993-04-16 1996-08-20 Александр Павлович Парфенов Tv camera for stereogrammetric videorecording of images of objects
WO2001067039A1 (en) * 2000-03-07 2001-09-13 Pheno Imaging, Inc. Three-dimensional measuring system for animals using structured light
US20070188601A1 (en) * 2006-02-13 2007-08-16 Janos Rohaly Three-channel camera systems with non-collinear apertures

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5412420A (en) * 1992-10-26 1995-05-02 Pheno Imaging, Inc. Three-dimensional phenotypic measuring system for animals
AU6862198A (en) * 1997-04-04 1998-10-30 Alfa Laval Agri Ab Method and apparatus for generating image information when carrying out animal related operations
US6974373B2 (en) * 2002-08-02 2005-12-13 Geissler Technologies, Llc Apparatus and methods for the volumetric and dimensional measurement of livestock
JP5519693B2 (en) * 2008-12-03 2014-06-11 デラヴァル ホルディング アーベー Apparatus and method for determining a score of an animal's physical condition

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
SU1777728A1 (en) * 1990-07-12 1992-11-30 Inst Mash Im A A Blagonravova Automated milking plant
RU2004964C1 (en) * 1991-06-14 1993-12-30 Ленинградский сельскохоз йственный институт Animal identification method and device
RU2065573C1 (en) * 1993-04-16 1996-08-20 Александр Павлович Парфенов Tv camera for stereogrammetric videorecording of images of objects
WO2001067039A1 (en) * 2000-03-07 2001-09-13 Pheno Imaging, Inc. Three-dimensional measuring system for animals using structured light
US20070188601A1 (en) * 2006-02-13 2007-08-16 Janos Rohaly Three-channel camera systems with non-collinear apertures

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Boguslaw Cyganek, J. Paul Siebert: "An Introduction to 3D Computer Vision Techniques and Algorithms" 2009, , XP002603902 Chapter 1, Section 1.2, in particular second paragraph *
HIDEO MINAGAWA: "Surface area, volume, and projected area of Japanese-Short-horn cattle measured by stereo photogrammetry using non-metric cameras" J. AGRIC. METEOROL., vol. 50, no. 1, 1994, pages 17-22, XP002603903 *
MALASSIOTIS S ET AL: "A face and gesture recognition system based on an active stereo sensor" PROC. 2001 INT. CONF. ON IMAGEE PROCESSING; [INTERNATIONAL CONFERENCE ON IMAGE PROCESSING], THESSALONIKI, GREECE LNKD- DOI:10.1109/ICIP.2001.958283, vol. 3, 7 October 2001 (2001-10-07), pages 955-958, XP010563510 ISBN: 978-0-7803-6725-8 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8794181B2 (en) 2011-03-17 2014-08-05 Mirobot Ltd. System and method for three dimensional teat modeling for use with a milking system
US9848575B2 (en) 2011-03-17 2017-12-26 Mirobot Ltd. Human assisted milking robot and method
WO2013187821A1 (en) 2012-06-12 2013-12-19 Delaval Holding Ab Arrangement and method for a milking system
US9510553B2 (en) 2012-06-12 2016-12-06 Delaval Holding Ab Arrangement and method for a milking system
US20150366156A1 (en) * 2013-02-06 2015-12-24 Delaval Holding Ab Teat treatment method and apparatus
US9826709B2 (en) * 2013-02-06 2017-11-28 Delaval Holding Ab Teat treatment method and apparatus
EP3241431A1 (en) * 2016-05-03 2017-11-08 Lely Patent N.V. Method for performing a teat-related operation, and robot device therefor
NL2016721B1 (en) * 2016-05-03 2017-11-10 Lely Patent Nv Method for performing a teat-related operation, and robot device therefor
US10701905B2 (en) 2017-12-06 2020-07-07 International Business Machines Corporation Imaging and three dimensional reconstruction for weight estimation

Also Published As

Publication number Publication date
WO2010023122A3 (en) 2010-12-16

Similar Documents

Publication Publication Date Title
EP2244556B1 (en) Arrangement and method for determining positions of the teats of a milking animal
CA2626286C (en) Arrangement and method for visual detection in a milking system
CA3034793C (en) Time of flight teat location system
US8794181B2 (en) System and method for three dimensional teat modeling for use with a milking system
EP2526762A1 (en) Arrangement and method for determining positions of the teats of a milking animal
WO2010023122A2 (en) Arrangement and method for determining positions of the teats of a milking animal
WO2003055297A1 (en) Method and apparatus for detection of teats
CA2775395C (en) Vision system for robotic attacher
WO2010020457A1 (en) Arrangement and method for controlling a movable robot arm
WO2010023121A2 (en) Arrangement and method for determining positions of the teats of a milking animal
EP3503715B1 (en) Arrangement and method for classifying teats with respect to size measures
CA2849212C (en) Vision system for robotic attacher
CA2898603C (en) Vision system for robotic attacher
O'Callaghan et al. Design of automatic milking system for use in pasture-based systems
WO2011031210A1 (en) Arrangement and method for determining positions of the teats of a milking animal
SE527840C2 (en) Teat position determining arrangement for use in milking system, includes image processor to determine position of teat of milking animal relative to robot arm, when fixing teat cup to teat

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 09781882

Country of ref document: EP

Kind code of ref document: A2

NENP Non-entry into the national phase in:

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 09781882

Country of ref document: EP

Kind code of ref document: A2