TECHNICAL FIELD

The present invention relates generally to the initializing of an automatic milking equipment before operation. Especially, the invention relates to a system for determining the tool positions in an automatic milking arrangement and a method implemented in such a system. The invention also relates to rotary milking plat form, a computer program and a non-volatile data carrier.

BACKGROUND

Today’s automatic milking arrangements are highly complex in stallations. This is particularly true for rotary milking platforms, where a relatively large number of milking stations are served by a milking robot, or similar automatic equipment. Inter alia, this means that the milking robot attaches teatcups and other tools, e.g. cleaning cups, to the animals in a fully automatic manner. Therefore, the milking robot must be capable of automatically retrieving relevant tools from a storage place and possibly retur ning them thereto after completing each stage of the milking procedure. This, in turn, requires that the milking robot has very accurate knowledge about the respective position for each tool.

In the prior-art solutions, an operator teaches the milking robot relevant tool pick-up positions by controlling, for example via a joystick, a grip device of the milking robot to the space coordi nate where the grip device shall be positioned when picking up a particular tool. When the grip device has a desired pick-up po sition relative to the tool, the operator programs this position in to the control unit for the milking robot by entering a confirming command. Such semi manual programming of the tool positions is not ideal, for instance due to various accuracies in the user operated control link for the milking robot, and because - for safety reasons - the operator may need to be located at a place from which it is difficult, or even impossible, to see if the grip device is actually located at the desired pick-up position.

A rotary milking platform typically has a rather large number of milking stalls, say up to 80, and sometimes even more. Here, each milking stall has its own set of tools in the form of teat- cups, and perhaps one or more cleaning cups.

Of course, programming the individual positions for all these tools into the milking robot by said semi manual manner is a very tedious as well as error-prone task. Therefore, in practice, the operator normally teaches the milking robot the respective tool positions for one stall as a reference. Then, the space coor dinates for these reference positions are translated into corres ponding space coordinates representing the positions for the tools in all the remaining stalls. In the light of the error sources mentioned, it is obvious that this strategy may result in consider able errors in the individual tool positions, especially for the lar ger types of rotary milking platforms.

SUMMARY

The object of the present invention is therefore to offer a conve nient solution for determining the respective positions of the tools in an automatic milking arrangement in an efficient and ac curate manner.

According to one aspect of the invention, the object is achieved by a system for determining the tools positions in an automatic milking arrangement. The system contains a camera and a cont rol unit. The camera is configured to, from an origin location, re gister three-dimensional image data of at least four tools whose respective positions are to be determined. The control unit is configured to cause the camera to obtain three-dimensional ima ge data representing the at least four tools. The control unit is configured to identify tool candidates in the three-dimensional image data using an algorithm involving matching the image data against reference data. The control unit is configured to calcula te a respective position for the at least four tools based on the origin location and data expressing respective distances from the origin location to each of the identified tool candidates. The at least four tools are presumed to be stored in a dedicated space and be arranged according to a spatially even distribution rela tive to one another. Therefore, the control unit is further configu red to disregard any tool candidate, which is detected at such a position that the position for the candidate deviates from the spatially even distribution.

This system is advantageous because it avoids a semi manual involvement of a human operator. Furthermore, each respective position of all the tools in all milking stations of an automatic milking arrangement can be specifically registered with high ac- curacy and stored for later use in a fully automatic manner, ir respective of whether the tools are organized along a line, along an arc or in a square.

Preferably, the control unit is configured to apply a linear regres sion classification algorithm on the three-dimensional image data to determine the spatially even distribution of the at least four tools.

Although, of course, each tool occupies a volume extending in all three space dimensions, it is advantageous to define the position for a particular tool to be a particular point on a depicted object in the three-dimensional image data. For example, the particular point may be a well-defined point on an identified tool candidate.

According to one embodiment of this aspect of the invention, the control unit is configured to store the respective positions for the at least four tools in a memory unit. The stored respective posi- tions are retrievable from memory unit by a tool-pickup system in connection with picking up at least one of the at least four tools for attachment to an animal. Hence, the automatic milking arran gement can be put into operation directly. Preferably, the at least four tools are placed in a tool rack and the system includes a grip device which is arranged on a robotic arm. After that the respective positions have been stored in the memory unit, the control unit is further configured to retrieve the stored respective positions from the memory unit, and control the robotic arm and the grip device to pick up at least one of the at least four tools from the tool rack. Further preferably, the respec tive position for each tool is expressed in terms of space co ordinates for a particular point on an object depicted in the three-dimensional image data.

According to another embodiment of this aspect of the invention, the at least four tools are arranged relative to one another in a predefined pattern. Here, the control unit is configured to use in formation about the predefined pattern to confirm and/or disre- gard at least one of the tool candidates.

For example, the at least four tools may be arranged along a li ne. In such a case, the control unit is configured to disregard any tool candidate that is detected at an outlier distance exceeding a second threshold distance from an estimated line interconnecting at least two other tool candidates in a set of tool candidates for said tools.

If the at least four tools are arranged with an equal distance bet ween each neighboring tool of the tools in the line, the control unit is configured to disregard a tool candidate that is detected at such a position that the tool candidate results in that a diffe rence between first and second inter-distances exceeds a third threshold distance; where the first inter-distance is an interspace between a primary pair of neighboring tool candidates including the tool candidate and a first tool candidate and the second in- ter-distance is an interspace between a secondary pair of neigh boring tool candidates including the tool candidate and a second tool candidate. Hence, so-called false positive tool candidates can be excluded from the process. According to yet another embodiment of this aspect of the in vention, if a processing time after having obtained the three-di mensional image data in the control unit, less than a predefined number of tool candidates have been identified, say four, the control unit is configured to reposition the camera to a new origin location. The control unit is configured to, from the new origin lo cation, cause the camera to obtain updated three-dimensional image data representing the at least four tools. The three-dimen sional image data may be dynamic. This means that the data may be represented by a video sequence and/or be built up from multiple still images registered from one or more origin locations in one or more angles. Further, the control unit is configured to calculate a respective position for the at least four tools based on the new origin location, and data expressing respective distances from the new origin location to each of the identified tool candidates. As a result, high data quality can be obtained also if, for some reason, there is a temporary interference obstructing the camera’s view of the tools.

According to another embodiment of this aspect of the invention, the camera is arranged on the robotic arm. Naturally, this is ad vantageous because it allows convenient movement and repo sitioning of the camera. Moreover, the camera that is used du ring normal operation of the milking arrangement can also be used for programming the tool positions into the system. According to another aspect of the invention, the object is achie ved by a rotary milking platform having a plurality of milking stalls. Each milking stall, in turn, includes at least four tools, and the rotary milking platform contains the above-described system for determining the positions of tools. In addition to the above advantages, this rotary milking platform is advantageous becau se regardless of the platform size, no coordinate translations are required to register the tool positions.

According to another aspect of the invention, the object is achie ved by a method for determining the positions of tools in an au- tomatic milking arrangement. The method involves registering, via a camera at an origin location, three-dimensional image data representing at least four tools whose positions are to be deter mined. The method further involves identifying tool candidates in the three-dimensional image data using an algorithm involving matching the image data against reference data. A respective position for the at least four tools is then calculated based on the origin location and data expressing respective distances from the origin location to each of the identified tool candidates The at least four tools are arranged according to a spatially even distribution relative to one another. The method further involves disregarding any tool candidate detected at such a position that the position for the candidate deviates from the spatially even distribution. The advantages of this method, as well as the pre- ferred embodiments thereof, are apparent from the discussion above with reference to the system.

According to a further aspect of the invention, the object is achieved by a computer program loadable into a non-volatile da ta carrier communicatively connected to a processing unit. The computer program includes software for executing the above method when the program is run on the processing unit.

According to another aspect of the invention, the object is achie ved by a non-volatile data carrier containing the above computer program. Further advantages, beneficial features and applications of the present invention will be apparent from the following description and the dependent claims.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention is now to be explained more closely by means of preferred embodiments, which are disclosed as examples, and with reference to the attached drawings.

Figure 1 shows a system for determining the positions of tools according to one embodiment the invention;

Figures 2-4 illustrate, schematically, how identified tool candi dates can be discarded according to embodiments the invention;

Figure 5 shows a rotary milking platform according to one embodiment the invention; and

Figure 6 illustrates, by means of a flow diagram, the gene ral method according to the invention.

DETAILED DESCRIPTION

In Figure 1 , we see a system according to one embodiment of the invention. Here, the system is arranged to determine the respective positions of four different tools, for example a set of teatcups 141 , 142, 143 and 144.

The system includes a camera 130 and a control unit 120. The camera 130 is configured to register three-dimensional image data Di _mg 3 _D of the tools 141 , 142, 143 and 144 whose respective positions are to be determined. Preferably, the camera 130 is a time-of-flight camera (ToF camera), i.e. a range imaging camera system that resolves distance based on the known speed of light. According to the invention, however, the camera 130 may be any alternative imaging system capable of determining the respective distances to the objects being imaged, for example a 2D camera emitting structured light or a combined light detection and ranging, LIDAR, camera system. Moreover, the three-dimen- sional image data Di _mg 3 _D may be dynamic. This means that the three-dimensional image data Di _mg 3 _D can be represented by a video sequence and/or be built up from multiple still images re gistered from one or more origin locations Pc in one or more an gles. Thus, the camera 130 is positioned at an origin location Pc, and may either be arranged on a robotic arm 1 10, as illustrated in Figure 1 , or be placed on another suitable structure, e.g. a tri- pod. In any case, the space coordinates (xc , yc , zc) of the origin location Pc are known with high accuracy. As a result, the three- dimensional image data Di _mg 3 _D registered by the camera 130 forms a basis for determining the position of any object within a view field VF of the camera 130. Namely, the space coordinates for the particular point can be calculated based on the origin lo cation Pc(xc , yc , zc) and data di(xi , yi , zi), e.g. a space vector, expressing a distance in three dimensions from the origin loca tion Pc(xc , yc , zc) to the particular point on a depicted object. For example, the particular point can be a well-defined point on a first tool candidate Tci , such as an intersection between a symmetry center of the teatcup body and the liner’s edge to the teatcup body.

The control unit 120 is configured to cause the camera 130 to obtain three-dimensional image data Di _mg 3 _D representing the tools 141 , 142, 143 and 144, and forward the three-dimensional image data Di _mg 3 _D to the control unit 120. The control unit 120 is then configured to identify tool candidates Tci , Tc2, Tc3 and Tc4 in the three-dimensional image data Di _mg 3 _D using an algorithm involving matching the image data Di _mg 3 _D against reference data. For example, the reference data may comprise one or more cha racteristic patterns of the tool and/or a typical tool outline. The control unit 120 is configured to calculate a respective position PTI , PT2 , P T3 and P T ₄ for each of the tools 141 , 142, 143 and 144 based on the origin location Pc(xc , yc , zc) and data expressing respective distances from the origin location Pc(xc , yc , zc) to each of the identified tool candidates Tci , Tc2, Tc3 and Tc4.

To improve the efficiency of the positioning process, the follo wing strategy is applied according to the invention. Since it is sa- fe to assume that each of the tools 141 , 142, 143 and 144 is sto red in a dedicated space, such as at a given position in a rack 150, it can be presumed that the tools 141 , 142, 143 and 144 ha ve predefined locations relative to one another. More precisely, the tools 141 , 142, 143 and 144 are presumed to be arranged according to a spatially even distribution relative to one another. In Figure 1 , this inter-tool relationship is illustrated by means of an equal distance Ad separating each neighboring tool from one another in the rack 150.

The control unit 120 is configured to disregard any tool candida- te Tc2, which is detected at such a position that the position for the candidate Tc2 deviates from the spatially even distribution. To determine whether or not a tool candidate deviates from the spatially even distribution, the control unit 120 may be configu red to formulate the pattern recognition problem in the image da- ta Dimg3 _D in terms of linear regression.

Using a fundamental concept that patterns from a single-object class lie on a linear subspace, a linear model can be developed, which represents the image data Di _mg 3 _D as a linear combination of class-specific galleries. The inverse problem may then be sol- ved using the least-squares method. To this aim, the control unit 120 is preferably configured to apply a linear regression classifi cation algorithm (LRC) on the three-dimensional image data Di _mg 3 _D to determine any deviation from the spatially even dist ribution of the 141 , 142, 143 and 144. The ability to distinguish the evenly distributed tools from other objects in the image data Di _mg 3 _D is beneficial, inter alia because the control unit 120 can thereby avoid regarding a pole, other stalling equipment or similar tool like object, in proximity to the tools as a tool candidate. Referring now to Figure 2, according to the invention, the control unit 120 is configured to disregard any second tool candidate Tc2 that is detected at a separation distance d _s from a first tool can didate Tci , if the separation distance d _s exceeds a first threshold distance d _t m in relation to the predefined relative locations Ad. For example, if the equal distance Ad is 10 centimeters, any se cond tool candidate Tc2 detected more than 15 centimeters from a first tool candidate Tci may be discarded. Of course, this stra tegy is actually applicable to any number of tools larger than two.

Provided that there are three or more tools, say four tools 141 , 142, 143 and 144, these tools may be arranged relative to one another in a predefined pattern, for instance in the corners of a square, along an arc or along a line L as shown in Figure 1 . In such a case, the control unit 120 is preferably configured to use information about the predefined pattern to confirm and/or disregard at least one of the tool candidates Tci , Tc2, Tc3 and T c4. Below, this will be exemplified by two embodiments of the invention.

Figure 3 illustrates a situation where a set of tools are stored in a rack 150 according to a linear arrangement. Here, the camera 130 has registered three-dimensional image data Di _mg 3 _D in which tool candidates Tci , Tc2, Tc3 and Tc4 have been identified. As can be seen, a third tool candidate Tc3 has been detected at an outlier distance d ₀ from an estimated line L _e interconnecting at least two of the other tool candidates, here Tci , Tc2, and Tc4 res pectively, in a set of tool candidates for the tools. Provided that the outlier distance d ₀ exceeds a second threshold distance dtn2, the control unit 120 is configured to disregard the detected tool candidate Tc3 according to one embodiment of the invention.

Moreover, if there is a plurality of tools 141 , 142, 143 and 144, the control unit 120 may draw further conclusions based on the predefined pattern in which the tools are arranged relative to one another. Figure 4 illustrates how an identified tool candidate Tc2 can be discarded according to a third embodiment the invention.

Here, four tools 141 , 142, 143 and 144 are arranged in a line L (see Figure 1 ) with an equal distance Ad between each neighbo ring tool in the line L. The control unit 120 is configured to use the information about the predefined tool pattern to disregard a tool candidate Tc2, which is detected at such a position that said tool candidate Tc2 results in that a difference between a first inter-distance Adi and a second inter-distance Ad2 exceeds a third threshold distance. The first inter-distance Adi is an inter space between a primary pair of neighboring tool candidates, say Tci and Tc2, that includes said tool candidate Tc2 and a first other tool candidate Tci . The second inter-distance is an inter- space between a secondary pair of neighboring tool candidates; say T c2 and Tc3, which includes said tool candidate Tc2 and an other second tool candidate Tc3.

Analogously, the predefined pattern in which the tools organized may also be used by the control unit 120 to confirm a tool can- didate. I.e. if, for example, a second tool candidate is found at the expected distance Ad from a first tool candidate, the position for the second tool candidate can be confirmed.

To facilitate making efficient use of the calculated tool positions [P] when operating the automatic milking arrangement, according to one embodiment of the invention, the system includes a memory unit 160, e.g. a storage medium in the form of a Flash memory or a Read Only Memory (ROM). The control unit 120 is further configured to store the respective positions [P] for the at least two tools 141 , 142, 143 and 144 in the memory unit 160. The stored respective positions [P] are retrievable from the me mory unit 160 by a tool-pickup system in connection with picking up at least one of the at least four tools 141 , 142, 143 and 144, typically for attachment to an animal.

According to one embodiment of the invention, the at least four tools 141 , 142, 143 and 144 are placed in a tool rack 150, as il lustrated in Figure 1 . The system for determining the positions of tools in the automatic milking arrangement further contains a grip device 1 15, which is arranged on a robotic arm 1 10. After that the respective positions [P] have been stored in the memory unit 160, the control unit 120 is configured to retrieve the stored respective positions [P] from the memory unit 160, and control the robotic arm 1 10 and the grip device 1 15 to pick up at least one of the at least four tools 141 , 142, 143 and 144 from the tool rack 150, so that for example this/these tool(s) can be attached to an animal. The grip device 1 15 may contain one or more elec tro-magnets configured to cooperate with one or more magnetic members on each of the at least four tools 141 , 142, 143 and 144. Alternatively, or in addition thereto, the grip device 1 15 may contain at least one mechanical gripping claw configured to grasp around the tool itself or a part thereof, e.g. a grip bar.

Preferably, if a robotic arm 1 10 is included in the system, the ca mera 130 is arranged on the robotic arm 1 10. Namely, this highly facilitates implementing the above above-described procedure. It is generally advantageous if the control unit 120 and the ca mera 130 are configured to effect the above-described proce dure in an automatic manner by executing a computer program 127. Therefore, the control unit 120 may include a memory unit 126, i.e. non-volatile data carrier, storing the computer program 127, which, in turn, contains software for making processing cir cuitry in the form of at least one processor in the central control unit 120 execute the above-described actions when the compu ter program 127 is run on the at least one processor.

Figure 5 shows a rotary milking platform 500 according to one embodiment the invention. The rotary milking platform 500 has a plurality of milking stalls 520,, here 24 altogether. Each of the milking stalls 520, contains at least four tools, symbolically indi cated by the reference numerals 141 , 142, 143, 144. The rotary milking platform 500 also includes a system for determining the respective tool positions, which is here symbolically indicated via the robotic arm 510. Said positioning system 510 is arranged to calculate a respective position Pn , PT2 , PT3 and PT ₄ for the at least four tools 141 , 142, 143 and 144 respectively in each of said milking stalls 520, according to the procedure described above. Preferably, the rotary milking platform 500 is stepwise ro tated pass the system 510 for calculating the respective tool po sitions for all the milking stalls 520, on the platform 500 in a con secutive manner. In order to sum up, and with reference to the flow diagram in Figure 6, we will now describe the general method according to the invention for determining the positions of tools in an auto matic milking arrangement. In a first step 610, three-dimensional image data are registered via a camera at a known origin location. The three-dimensional image data represent at least four tools of the milking arrange ment whose respective positions are to be determined.

In a subsequent step 620, tool candidates are identified in the three-dimensional image data using an algorithm that involves matching the image data against reference data.

Thereafter, in step 630, a respective position is calculated for the at least four tools. The position calculations are based on the known origin location of the camera and data expressing res- pective distances from the origin location to each of the identified tool candidates. The at least four tools have predefined locations relative to one another. More precisely, the at least four tools are arranged according to a spatially even distribution relative to one another. Any tool candidate disregarded, which is detected at such a position that the position for the candidate deviates from the spatially even distribution.

Subsequently, the procedure ends.

All of the process steps, as well as any sub-sequence of steps, described with reference to Figure 6 may be controlled by means of a programmed processor. Moreover, although the em bodiments of the invention described above with reference to the drawings comprise processor and processes performed in at least one processor, the invention thus also extends to computer programs, particularly computer programs on or in a carrier, ad- apted for putting the invention into practice. The program may be in the form of source code, object code, a code intermediate source and object code such as in partially compiled form, or in any other form suitable for use in the implementation of the pro- cess according to the invention. The program may either be a part of an operating system, or be a separate application. The carrier may be any entity or device capable of carrying the prog ram. For example, the carrier may comprise a storage medium, such as a Flash memory, a ROM (Read Only Memory), for ex ample a DVD (Digital Video/Versatile Disk), a CD (Compact Disc) or a semiconductor ROM, an EPROM (Erasable Program mable Read-Only Memory), an EEPROM (Electrically Erasable Programmable Read-Only Memory), or a magnetic recording medium, for example a floppy disc or hard disc. Further, the car rier may be a transmissible carrier such as an electrical or opti cal signal which may be conveyed via electrical or optical cable or by radio or by other means. When the program is embodied in a signal which may be conveyed directly by a cable or other de- vice or means, the carrier may be constituted by such cable or device or means. Alternatively, the carrier may be an integrated circuit in which the program is embedded, the integrated circuit being adapted for performing, or for use in the performance of, the relevant processes. The term“comprises/comprising” when used in this specification is taken to specify the presence of stated features, integers, steps or components. However, the term does not preclude the presence or addition of one or more additional features, inte gers, steps or components or groups thereof. The invention is not restricted to the described embodiments in the figures, but may be varied freely within the scope of the claims.