Robotic milking system

Abstract

A robotic milking system (1) has milking means, and a robotic arm (3) having a teat detection system (5). The teat detection system has a laser source (10) for a laser beam (11, 12), a movable laser mirror (20, 25), a camera (8) for images of the laser beam on an object such as a teat, and an image processing device (9) for determining teat positions from the images, as well as a control system (40) for moving the mirror, having a coil (21, 27) with coil poles, and a power supply device for an adjustable current li. The power supply device has a DC power supply source (43) having a positive pole and a negative pole, switching elements, a current-measuring device, and a control, for the steps a) at tstart, connecting the power supply device to the coil poles until lm has risen to li + b or fallen to li - b, where b is a current range, b) disconnecting the coil pole from the DC voltage supply source, and connecting the coil poles until at least Tstart + T, where T is a switching period, c) connecting the power supply device to the coil poles with reversed polarity until, at t2, lm has fallen to li - b, or risen to li + b, d) connecting the coil poles until at least Tstart +2T, and e) updating Tstart to Tstart = max[Tstart + 2T; t2 + T]. A flexible, fast, stable and reliable laser beam control is thus obtained. This enables fast and effective teat detection, so that teat cups can be efficiently connected.

Description

[0001] Robotic milking system

[0002] The present invention relates generally to a robotic milking system having a teat detection system.

[0003] More particularly, the robotic milking system has milking means such as teat cups, for drawing milk from teats of a dairy animal, as well as a robotic arm for applying the milking means to the teats, having a teat detection system attached thereto for determining teat positions of the teats and for controlling the robotic arm based on the determined teat positions. The teat detection system comprises a laser source for emitting a laser beam, a movable mirror for directing the laser beam, an optical sensor for repeatedly capturing an image of reflections of the laser beam on an object such as a teat, and an image processing device for processing the repeatedly captured images and for determining said teat positions from the processed captured images.

[0004] A teat detection system is essential for enabling a robotic milking system to function. In the present invention, the teat detection system comprises an optical sensor which captures and processes reflections from a laser beam on, for example, the teats, in order to determine the teat positions therewith.

[0005] Milking systems with such teat detection systems are known per se. EP1555544A2 thus reveals a milking robot having a robotic arm with a teat detection system thereon, wherein two laser sources each emit a laser beam which is scattered via a rotating mirror with different surfaces to form a number of fixed horizontal lines. These lines, and in particular their reflections on teats, are used to determine the teat positions.

[0006] Although the known milking system works well per se, it is not always able to find the teat positions both reliably and quickly, so that in those cases the capacity of the robotic milking system as a whole is inadequate. In particular, as dairy animals are living creatures which can and will move during the connection of the milking means, it is very important to have a fast and reliable teat detection system.

[0007] An object of the present invention is to make the known robotic milking system faster, or at least to improve it such that the teat detection system can determine the teat positions more quickly and / or more reliably.

[0008] The invention achieves this object with a robotic milking system according to Claim 1.

[0009] The invention is based on the insight that the fixed course of the laser, i.e. horizontal laser lines through the rapidly rotating mirror(s) on a relatively large solid of revolution, form a limitation of the prior art. After all, not only will the robotic arm have to be moved in order to move the lines over, for example, a detected teat. This rotation is also inefficient, because the laser line usually points to a part of the field of view that is not relevant to the object detection. It is already very difficult to guide the laser line to a desired position and hold it there. In particular, in order to find a teat tip, which indicates most accurately where a milking means should be positioned, it is important to be able to accurately determine reflections of that teat from the radiation area. It therefore makes little sense to scan the further vicinity of the area in which a teat is seen or suspected with the laser beam that continues to sweep, but it is better to be able to guide the laser beam faster and in a more targeted manner. The invention provides for this by making the laser beam orientable by means of a galvanometer-like arrangement, with a coil having an adjustable magnetic field that can interact with a magnet and thereby generate or undergo a force or torque. This force or torque can be used to move the mirror. There are several possibilities here, which will be explained in more detail.

[0010] Controlling the mirror by simply selecting a control current for the coil is very inefficient, unstable and unreliable. This explains the decision to opt for a control with alternating current, which, on average, has the desired value. Pulse width modulation, for example, can be used for this purpose, but the invention prefers a hysteretic control. The invention also makes use of the insight that the mirror can be moved even faster if a low self-inductance L of the coil can be chosen. As a result, when the supply voltage is applied, the current will increase very quickly, to values that exceed the maximum load capacity, also resulting in substantial losses. However, this can be counteracted by very quickly reversing the supply voltage to change from a rising current to a falling current. This, in turn, requires very high switching frequencies, which limits the choice of switching elements, but also causes higher switching losses and more EMI (electromagnetic interference). To counteract this, in turn, the invention incorporates the feature that every branch, both rising and falling, has a phase in which the coil is, as it were, short-circuited, so that the supply voltage cannot cause the current to rise further, but conversely the energy stored in the coil ensures that the current decreases, preferably only slowly. The values for self-inductance L, and resistance (and possibly capacitance) in the short- circuited coil circuit, can easily be chosen by a person skilled in the art within wide margins in order to achieve a sufficiently low switching frequency on one hand, and to ensure a sufficiently slow current drop on the other hand, thus preventing the circuit and control as a whole from becoming unstable. It is important to note, however, that the milking system having a teat detection system according to the present invention has a control of the laser beam that is stable and fast, without substantial EMI switching losses. This will be explained in detail below.

[0011] It should be noted here that, in the present application, the term "operably couplable or coupled" means that the coil is electrically connectible, or connected, to other elements. The switching elements will establish such connections, or break them.

[0012] Particular embodiments are described in the dependent claims, and also in the part of the description that now follows.

[0013] In embodiments, the robotic milking system also comprises a second mirror, wherein the first mirror is pivotable around a first axis of rotation and the second mirror is pivotable around a second axis of rotation, said first and second axes of rotation forming a non-zero angle between them, wherein the laser beam is directed at the first axis of rotation, and wherein the laser beam reflected by the first mirror is directed at the second axis of rotation. The laser beam can cover a solid angle here, and thus very effectively obtain a more complete image of objects in that solid angle. The advantage of the invention of faster, more reliable control of the laser beam is even more apparent here, since the independent and fast control of two mirrors allows the laser beam to be directed in two directions to a desired location. In principle, this could also be done in the prior art by moving the robot arm with the laser source as a whole vertically, but it will be clear that this will be done much more slowly than by pivoting one or more mirrors.

[0014] The first axis of rotation is advantageously substantially horizontal and the second axis of rotation is in a perpendicular plane to the first axis of rotation. More advantageously, the second axis of rotation is substantially vertical. Thus, when the mirrors are pivoted, the reflected laser beam will move in the respective perpendicular direction, i.e. vertically or horizontally, thereby simplifying the determination of the spatial position of an object on which the laser beam is reflected.

[0015] It should be noted here that "substantially" in any case includes "within the manufacturing tolerances" of the milking system, but also includes "within 5 degrees", preferably "within 2 degrees" in relation to the absolute vertical or horizontal. It should further be noted that reflecting / reflection also refers here to scattering / scatter, as long as the light from the laser beam striking the object returns to the sensor. On the other hand, it is certainly possible to maintain other angles, and then perform a coordinate transformation.

[0016] In embodiments, the first mirror is rigidly connected to a permanent magnet, and the laser illumination system also comprises a housing around said mirror, or mirrors, and the magnet comprises a housing containing said coil. If a second mirror is provided, it is rigidly connected in this embodiment to a respective second permanent magnet, and the laser illumination system also comprises a second housing around said second mirror and the second magnet containing a respective second coil, or said housing comprises both mirrors with a respective magnet and coil. This embodiment is comparable to a moving magnet galvanometer. One advantage is that current-carrying parts do not have to move, and furthermore that it is easy to cool the coil, for which purpose a cooling circuit, heat sink or similar is preferably provided for the coil. By being cooled, the coil can handle a greater maximum amperage, which in turn makes it possible to give the coil a greater (maximum) acceleration, thus making the system even faster.

[0017] Alternatively or additionally, the first mirror, and / or the second one if provided, is rigidly connected to said coil, to a second coil, respectively, and the laser illumination system further comprises a magnet provided around the coil or coils. This embodiment is comparable to a moving coil galvanometer. These embodiments can sometimes be made slightly more easily with a lower mass and self-inductance, so they can be faster than the moving magnet variant above. The magnet can, for example, extend to both sides of the coil in a horseshoe shape, or can consist of two or more partial magnets.

[0018] For the sake of clarity, in the case of the above-mentioned two embodiments, if two mirrors are provided, both mirrors are connected in the same way to a corresponding permanent magnet or coil. In addition, for example, a housing is then provided around the first mirror and around the second mirror, but advantageously around both mirrors.

[0019] In advantageous embodiments, the control comprises image recognition software which is configured to process the repeatedly captured images, to recognize at least said object in said processed images and to determine a position of said object. The control is advantageously configured to control at least one of the mirrors by means of the control system on the basis of the processed images, in particular on the basis of the position of the recognized object. It is known per se in the prior art for the teat detection system to be controlled on the basis of the captured images. Nevertheless, as already explained above, the present invention offers advantages in terms of speed of adjustment of the setting of the mirror(s) and thus of the position scanned by the laser beam, by changing the / , for the coil associated with that mirror.

[0020] The camera can in this case be a 2D camera, i.e. a conventional video camera. This allows to determine the direction of the laser beam onto the object. Furthermore, if the distance between the laser source and the camera is known, i.e. the base, the teat detection system can also calculate the 3D coordinates of the reflection point on the object such as the teat, and it is advantageously configured for this purpose. Alternatively, if the laser source also comprises a distance sensor, as is known per se in laser distance meters, the teat detection system can be configured to determine a distance from the object to the laser source, which is then advantageously configured accordingly. However, the camera / sensor can also be a 3D camera, such as a stereo camera. The position of the object, such as the teat or a leg or tail of the animal, can therefore similarly be easily determined in three dimensions. The camera could also be a time-of-flight camera (ToF camera) which is configured to measure the transit time of the laser beam from the laser source to the ToF camera, and from that the distance between the camera and the reflection point of the laser beam. The 3D coordinates of the reflection point, and therefore the object, can then be determined based on the 2D coordinates, which can also be determined with the ToF camera.

[0021] The values to be used in the robotic milking system for various quantities are not particularly limited per se. The switching period T is thus freely selectable within broad limits. In embodiments, however, T is between 0.3 and 10 ps. This allows sufficiently fast control of the mirror(s), with high reliability and stability, without the need for costly and / or complex components suitable for higher frequencies, with only limited switching losses and no significant problems due to EMI. In addition, the value of the (maximum) current h is to a large extent freely selectable, and is, for example, a few amperes. On this basis, a value can be chosen for b which ensures sufficient stability, such as, for example, a maximum of 10% of / ,, or an absolute maximum of, for example, 200 mA. This value can also be chosen within a wide range per se. The values for e.g. the self-inductance L of the coil and the resistance R of the short-circuited coil circuit can further be determined on the basis of the above. L- 50 pH, for example, has been chosen. Such a value is common, and is suitable for quickly changing the current in the coil without imposing excessive demands on the electrical supply. However, other values, such as smaller values in particular, are quite possible. A person skilled in the art can further determine suitable values for the (net) resistance R of the short-circuited coil circuit in order to ensure that the current in the phase with the short-circuited coil decreases sufficiently slowly. Here, R comprises the resistance of the coil itself, and possibly an additional resistance. It is guaranteed, for example, that the current decreases by a maximum of 10% of / ,, or a maximum of 200 mA, in a whole period T. Equally, the value for the range b can be chosen within wide margins, e.g. similarly 200 mA, or smaller or greater, as long as the desired stability is ensured.

[0022] In embodiments, the measuring device comprises at least one current- measuring device, provided in each case between one of said positive pole and said negative pole on one hand and one of the controllable switching elements on the other hand, or between one of said controllable switching elements on one hand and one of said first or second coil poles on the other hand. The location of the current measurement is not particularly limited per se, and any location in the circuit through the coil is possible. It can be advantageous to measure the current in more than one location in order to obtain a redundant and more reliable measurement. Measuring the current directly in series with the coil, i.e. constantly connected to one of the coil poles, offers the advantage that the coil current can be measured continuously, but has the disadvantage that the losses are slightly greater, and also that the overall circuit is slightly more complex. Alternatively, the current can be measured in series with one of the poles of the power supply source. Although the current cannot be measured during one of the steps (b) or (d), it is not necessary in principle to do so, since the course of the current itself will be known from the properties of the remaining circuit, in particular the (self-)inductance L of the coil, and the resistance in the coil and the connections in the "short circuit". The circuit as a whole is therefore also somewhat simpler.

[0023] In embodiments, the controllable switching elements are or comprise fieldeffect transistors (FETs). These are ideally suited to achieving a fast and low-loss control system, and likewise working method. All types of FET can be used here depending on further desired specifications. Nevertheless, other switching elements are not excluded.

[0024] In advantageous embodiments, at least one current-measuring device is provided in one of said field-effect transistors. This provides a lossless, or at least low- loss, current measurement facility at a favourable price, albeit with often slightly poorer accuracy.

[0025] The invention will be explained below with reference to the drawing, which shows some non-limiting exemplary embodiments, wherein is shown in:

[0026] - Figure 1 a schematic, perspective view of a robotic milking system according to the invention;

[0027] Figure 1 shows a schematic, perspective view of a robotic milking system 1 according to the invention for milking a dairy animal 100. The robotic milking system 1 comprises a milking robot 2 having a robotic arm 3 and teat cup 4, as well as a teat detection system 5 with a working area 6, all in a milking parlour 7. The dairy animal 100 shown has an udder 101 with teats 102, and hind legs 103.

[0028] In principle, the parts shown are present on every robotic milking system, so there is no need to provide many associated details here. It should be noted, however, that although the teat detection system 5 is shown for the sake of clarity on the milking robot 2, or at least a housing thereof, it is in practice almost always provided on the robotic arm 3, if only to ensure an optimal view of the teats 102 when the teat cup 4 is connected. Only one of these teat cups is shown here, as in the case of systems which pick up a teat cup 4 from a store and connect it separately to the teat 102. Other systems carry all teat cups 4, e.g. four teat cups for cows, on a carrier, wherein the robotic arm remains under the dairy animal 100 during milking, as in the Lely Astronaut® system. In addition, the milking parlour 7 can be provided as an individual, ground-based milking parlour, or a parlour on a rotating milking device or carousel.

[0029] Figure 2 shows a slightly more detailed, schematic side view of a part of the robotic milking system 1 having the teat detection system 5, but now on the robotic arm 3. The teat detection system 5 comprises a camera 8 with an image processing device 9, as well as a laser source 10 which emits two laser beams 11 and 12, within a solid angle a, said beams 11 and 12 being received as reflected beams 1 T and 12' respectively.

[0030] The laser source 10 is shown here as a source which emits two laser beams 11 , 12. These beams 11 , 12 do not necessarily have to be emitted at the same time, but merely represent some possible directions that can be provided within the solid angle a. This solid angle can therefore be determined as the set of directions that can be given to the laser beam 11 , 12, but could also be regarded as the set of directions from which reflected light 11', 12' can be seen by the camera 8. Ideally, these definitions should be substantially the same, or at least the field of view as seen by the camera 8 will be at least as large as the solid angle for emitting laser beams. It should also be noted that the reflected radiation 1 T, 12' is shown here as a single beam, but in practice it will be a diffusion cone that is depicted by the internal optics of the camera 8 as coming from one direction.

[0031] The laser beam 11 is shown here as striking a teat 102, while the laser beam 12 strikes a hind leg 103. This can be determined by the teat detection device 5 using the camera 8, which can, for example, capture an image in which the laser beam assumes these positions 11 and 12, obviously along with many other directions, in order to create an image in the solid angle in front of the camera 8. The laser beam does not necessarily have to be switched on continuously, but can be switched on if required at desired times only, in order to illuminate individual points. A more continuous illumination can obviously also be provided, so that lines are created in desired directions. The objects irradiated by the laser beam are thus illuminated in the image from the camera 8. The image processing device 9 is also configured to detect and recognize objects in the one or more images from the camera 8, according to image processing methods known per se. Thus, for example, the hind leg 103 can be recognized, so that it can be avoided when the robotic arm 3 is controlled as the milk cup 4 is connected to the teat 102. The robotic arm 3 can be controlled here by the image processing device 9 or by a computer operably connected thereto.

[0032] In addition to a laser generator (not separately shown), the laser source 10 comprises one or more mirrors for directing the laser beam 11 , 12, said one or more mirrors being moved by means of coils which are controlled by means of a control system. This will be explained in more detail with reference to Figures 3-6.

[0033] Figure 3 shows a highly schematic top view of a part of the teat detection system.

[0034] The figure shows a first mirror 20 which is pivotable in the direction of the double arrow A around a first axis 21 and which is operably connected to a first coil 21 , and also a second mirror 25 which is pivotable in the direction of the double arrow B around a second axis 26 and which is operably connected to a second coil 27. A laser beam 11 -1 , a once-reflected laser beam 11 -2 and a twice-reflected laser beam 11 -3 are further shown.

[0035] The laser beam 11-1 as emitted by the laser source (not shown here) is directed onto the first axis 21 , which preferably coincides with the surface of the first mirror 20. The laser beam 11 -1 will thus be reflected as beam 11-2 in a fan shape from the tangent point 23 of the laser beam 11 -1 on the first mirror. The once-reflected laser beam 11 -2 is then directed here onto the second axis 26, so that the twice-reflected laser beam 11 -3 appears to originate from the reflection of the tangent point 23 in the second mirror 25.

[0036] By varying the setting of the first mirror 20 and / or the second mirror 25, the twice-reflected laser beam 11-3 can be given a direction within a solid angle (cf. a in Figure 2) in order to be able to scan and portray an image of that area, that field of view. The setting of the mirrors 20, 25 can be changed by means of the coils 22 and 27.

[0037] Figure 4 shows a detail of a possible arrangement for a movable mirror with coil control.

[0038] The figure shows a mirror 20' which is rotatable around the axis 2T in the direction of the double arrow C, and which is connected by a rod to the coil 22'. A laser generator is denoted by 28, a control device by 30, and a north pole and south pole of a permanent magnet by 31 and 32 respectively.

[0039] The coil 22', which is drawn here highly schematically, is connected to the rod 29 and therefore to the mirror 20', so that every (rotational) movement of the coil 22' is copied by the mirror 20'.

[0040] In order to rotate the coil 22' in the directions C, the coil is placed in a permanent magnetic field generated by the magnet having north pole 31 and south pole 32. The current, i.e. , for example, an adjustment current A, which flows through the coil 22', generates a magnetic field therein which is influenced by the external, permanent magnetic field. In other words, the external magnetic field of the magnet 31 , 32 exerts a moment of force on the coil, under the influence of the current flowing through it. The magnitude of the moment of force is dependent on the current. The rotation or the final position will therefore also be dependent on the current. The zero position of the coil 22’, and therefore of the mirror 20', can be determined here by means of a mechanical spring or similar (not shown here). This mechanism is known perse in galvanometers. However, it is not necessary, as galvanometer arrangements are also known wherein the mirror, even if the coil is de-energized, can be stationary over the entire range, such as over +30° to -30°.

[0041] Alternative embodiments, compared to the moving coil galvanometer arrangement described above, are possible, such as a moving magnet galvanometer arrangement. The mirror is rigidly connected here to a permanent magnet, and this is completely surrounded by the coil to be controlled, or possibly simultaneously controlled coils on both sides of the mirror with the magnet. The (or any) coil can be accommodated in a housing around the mirror, said housing usually being made of metal. In such an arrangement, cooling can be provided for the coil(s), which can then handle higher currents. This, in turn, has the advantage that the acceleration to be provided to the mirror can be greater, so that the system as a whole becomes faster.

[0042] Figure 5 shows schematically a simple control system 40, which comprises a coil 22" having a first coil pole 22"-1 and a second coil pole 22"-2, as well as a DC electrical source 43 with a positive pole 44-1 and a negative pole 44-2. Four switching elements are indicated as FETs 45-1 to 45-4. Ammeters are denoted by 46-1 , 46-2 and 46-3. The circuit comprises two nodes 47-1 and 47-2, and parts of the circuit are denoted by 48-1 to 48-5. A control is denoted by 50.

[0043] For the sake of clarity, a number of elements are not shown here, such as an object that is moved with the coil, and the active connections of the ammeters 46 and (most of) the switching elements 45 with the control 50. This will be explained in more detail below.

[0044] The control 50 is configured to allow a desired current h to flow through the coil 22", so that the coil can undergo a current-dependent moment in an (external) magnetic field (not shown here). To do this, the control 50 can connect the coil 22" to the poles 44-1 and 44-2 of the electrical power supply source 3 through targeted switching of the switching elements 45-1 to 45-4. The control 50 uses a measurement of the current through the coil 22" here to determine whether the current should further rise or fall, or to determine that the current is within a predetermined range b around the desired adjustment current / ,. If the direction of the current through the coil 22" is also important, the term 'rise' is not to be understood as an increase in the absolute value, but as an increase in the value of the current on the real number axis, wherein 'positive' refers to a current flowing from the first coil pole 22"-1 to the second coil pole 22"-2, and vice versa. Other definitions are clearly also possible.

[0045] The current is, for example, zero, and the current should be +1 A. The control 50 then switches the first FET 45-1 and the fourth FET 45-4 to 'conductive', and the second FET 45-2 and the third FET 45-3 to 'non-conductive1. The parts 48-2, between the positive pole 44-1 and the branch point 47-2, and 48-4 between the branch point 47- 1 and the negative pole 44-2 are thus blocked, and a rising current flows from the positive pole 44-1 via the parts 48-1 , 48-3 and 48-5. The current will increase here due to the inductive effect of the coil 22". The current through the coil is measured with one or more of the ammeters 46-1 , 46-2 and 46-3. It is expressly stated here that it is not necessary to provide all three of those ammeters, but that it is sufficient to provide only one of them, and that several ammeters supplied serve, in principle, only to obtain a redundant current measurement. It is further possible not to provide the current measurement at the locations shown, but on one or more other parts of the circuit, and / or integrated into a FET, e.g. combined with an Infineon IR25750L. With a combination such as the latter, no further current-measuring resistors or transformers, with their associated losses, are required.

[0046] If the desired adjustment current h is attained, the coil will either undergo a corresponding acceleration or assume a corresponding position if a spring or other restoring actuator is provided. In the first case, the acceleration leads to a change in position, wherein the acceleration must be adjusted until the desired position is attained. Delay of the coil using an opposite current / voltage will then be an inherent part of the control. In the other case, the counteracting actuator, such as the spring, ensures deceleration until the position corresponding to the adjustment current is attained. This position, and therefore the corresponding current, must be held until the next change in the adjustment current, and is therefore net zero in the first case, and the adjustment current in the second case.

[0047] The second case, with a desired adjustment current that must be maintained, is discussed below. The hysteresis circuit according to the invention is used to provide a stable situation. For this purpose, the control 50 initially allows the current to change further to a range b beyond the desired adjustment current A. For example, b is chosen as a few percent of the maximum adjustment current. If the measured current attains the value h + b, the control 50 switches the switching elements such that the current through the coil 22" regenerates, i.e. the coil poles 22"-1 and 22"-2 are interconnected, and the circuit with the electrical supply source 43 is disconnected. To do this, for example, either the FETs 45-1 and 45-2 are switched to 'non-conductive', while the FETs 45-3 and 45-4 are switched to 'conductive', or vice versa. Other switching elements that can switch quickly enough are obviously also possible, such as other types of transistors.

[0048] The current through the coil 22" will now drop toward zero, as there is no longer an external power supply source. On one hand, the range b and, on the other hand, the electrical quantities L of the coil and the resistance R of the shorted circuit through the coil 22" are advantageously chosen such that the current falls more slowly than the current rises when the power supply source 43 is connected to the coil. It is then possible to allow the current to fall in the switchover time T, which is moreover also chosen, to a limited extend only, for example less than 2b, before the next step in the cycle of the method according to the invention is initiated. Alternatively, if the supply voltage across the coil 22" were only reversed, the current would fall just as quickly as it rose. That would mean that the current would again have to be reversed much faster in order to maintain a stable average value. A polarity reversal of this type, which is must faster than required according to the present invention, would need much faster switching components, with all the disadvantages thereof.

[0049] An example of a method to be used in the robotic milking system according to the invention, and in particular the control system thereof, will now be explained in more detail with reference to Figures 5 and 6, which show schematically a time-current diagram. Here, 1 A is chosen as the first adjustment current , and 60 mA is chosen for the range b. In practice, a wider range of, for example, 0.3 A is furthermore not unusual.

[0050] At time to, the control 50 switches the FETs 45-1 to 45-4 such that the current, which is zero at that time, will rise to the new value of 1 A plus the range of 0.06 A, i.e. 1.06 A. This value is attained at ti. Here, ti is greater than to + T, where T is a predetermined time period, such as 1 ps. In this case, the control 50 will not short-circuit the coil first, but will immediately proceed to the next step in the cycle, in this case step c), i.e. operably connecting the supply voltage to reduce the current to h - b, i.e. 0.94 A. To do this, the FETs 45-1 ... 45-4 are suitably switched. Figure 6 shows that the current drops quickly from / ,- + b to h- b at time t2. At that time, the coil 22" is short-circuited by appropriately switching the FETs 45-1 ... 45-4 once more. The current will now only drop slowly and until the changeover time ti + T. At that time, the supply voltage is re-applied across the coil 22", but reversed. The current will now rapidly increase once more to h + b, at time ts, whereafter the coil is again short-circuited until time ti + 2T, wherein the current has again slowly dropped. The reverse-polarity connection is then re-established until time t4, whereafter the coil is again short-circuited until time ti + 37, whereafter the reverse-polarity connection is re-established until time ts, the coil is short-circuited until time ti + AT, and the reverse-polarity connection is established until time ts. It will of course be clear that this can be repeated as long as the desired adjustment current h remains the same, i.e. in this case 1 A.

[0051] The remainder of the diagram shows what can happen in embodiments if the adjustment current h changes. The coil 22" is again short-circuited at time ts. However, there is now no need to wait until time ti + 5T, since the desired adjustment current is reduced to 0.5 A before that time, i.e. at time t?. In this embodiment, the control 50 will not wait until the end of the switching period T, but will immediately reduce the adjustment current and also the FETs 45-1 ... 45-4 such that this current is also attained as quickly as possible. In this case, through suitable switching of the FETs 45-1 ...45-4 by the control 50, the current will fall at time t? to the new value of A + b, i.e. 0.5 - 0.06 = 0.44 A, said value being attained at ts. Given that ts occurs more than period T after fz, the current will be immediately reversed once more, so that it quickly rises to h + b - 0.56 A, at time tg. At that time, the coil 22" will again be short-circuited, and the current will slowly fall until time ts + T, whereafter the cycle continues once more with the new values.

[0052] It should be noted here that many variants are possible within the scope of the invention. Thus, for example, if the absolute value of the new adjustment current A becomes less, it is not necessary for the current to be modified to a range b beyond the new adjustment current h, but alternatively the current changes less, for example to a value which is still b away from the new adjustment current. In the example shown in Figure 6, this works as follows. The new adjustment current of 0.5 A is set at time frand the current drops to a range beyond that value, i.e. to 0.44 A. Alternatively, it is possible to stop just at a range before the new adjustment current is attained with the coil kept connected to the supply voltage, i.e., in this case when the measured current attains the value of 0.56 A. After all, the coil 22" will then be short-circuited, whereafter the current can continue to fall slowly. Conversely, if the value of the adjustment current increases, the coil 22"can be short-circuited when the measured current has risen to the new value h - b. Although the current will fall again slowly after short-circuiting the coil 22", the stabilizing effect of the alternating polarity reversal and short circuiting will then be achieved slightly earlier.

[0053] When the adjustment current is changed in the very first step, i.e. wherein current is changed, it is also generally possible to already short-circuit the coil 22" at a different measured current value that lies within the range from h - b to h + b, but thereafter, in the following steps a) and c), the coil is short-circuited only when these limits are reached. It is certainly also possible not to skip step b) or d) if changing the adjustment current / , lasts longer than T, or to shorten these steps when the adjustment current is changed but still attains the new value within period T. In other words, there is always a short-circuit step (b) or (d) in such embodiments. Such embodiments are a little slower, but are easier to implement. It is also possible, for example, in the event that the adjustment current is changed, not to restart the new switching periods from the time when the coil 22" is short-circuited for the first time in a new step (b) or (d), but to count through the periods T. In other words, in this case, wherein the time when the measured current attains the new limit value, and therefore the time of the next short circuit, will occur at a time tnew modulo T, the next short circuit will last until the next multiple of the period 7" is attained.

[0054] A person skilled in the art will easily conceive still further variants.

Claims

CLAIMS1 . Robotic milking system for milking a dairy animal, comprising:- milking means such as teat cups, for drawing the milk from teats of the dairy animal, and- a robotic arm for applying the milking means to the teats, and- a teat detection system attached to the robotic arm for determining teat positions of the teats and for controlling the robotic arm based on the determined teat positions, wherein said teat detection system comprises a laser source for emitting a laser beam in a radiation area, a first movable mirror for directing the laser beam by means of reflection, a camera for repeatedly capturing an image of reflections from the radiation area of the laser beam on an object such as a teat, and an image processing device for processing the repeatedly captured images and for determining said teat positions from the processed captured images, wherein the teat detection system further comprises a control system for said movement of the first mirror with a coil which is operably couplable or coupled to said mirror, having a first coil pole and a second coil pole, and also having an electrical power supply device which is configured to provide the coil with an adjustable desired current A, wherein the electrical power supply device comprises:- a DC power supply source having a positive pole and a negative pole, configured to provide the adjustable current to the coil,- a group of controllable switching elements for selectably interconnecting, by means of switching, the positive pole, the negative pole, and the first and second coil poles,- a measuring device configured to measure an instantaneous value lmof a current through the coil or a quantity equivalent thereto,- a control which is operably connected to the group of switching elements and the measuring device, wherein the control is configured to have the switching elements repeatedly perform the following cycle of steps: a) at a time tstart, connecting the positive pole to one of the first coil pole and the second coil pole, and the negative pole to the other of the first coil pole and the second coil pole until, at a time ti, the measured value of lmhas risen to A + b if the current prior to said connecting was less than A + b, or the current through the coil has fallen to A - b if the current prior to said connecting was greater than A + b, where b is a predetermined current range,b) disconnecting the first coil pole from at least one of the positive pole and the negative pole, and connecting and keeping connected the first coil pole and the second coil pole, until at least a time tstart + T, in particular until a time ti + T, where T is a predetermined switching period, c) connecting the positive pole to said other of the first coil pole and second coil pole, and the negative pole to said one of the first coil pole and the second coil pole, until, at a time t2, the measured value of lmhas fallen to h - b if the current prior to said connecting was greater than A - b, or the current through the coil has risen to / ,- + b if the current prior to said connecting was less than / ,- + b, d) disconnecting the first coil pole from at least one of the positive pole and the negative pole, and connecting and keeping connected the first coil pole and the second coil pole, until at least a time tstart + 2T, in particular until a time t2 + T, and e) updating tsta t to tstart = max[fSfart + 2T;+ 7] by means of the control.

2. Robotic milking system according to Claim 1 , further comprising a second mirror, wherein the first mirror is pivotable around a first axis of rotation, and the second mirror is pivotable around a second axis of rotation, said first and second axes of rotation forming a non-zero angle between them, wherein the laser beam is directed at the first axis of rotation, and wherein the laser beam reflected by the first mirror is directed at the second axis of rotation.

3. Robotic milking system according to Claim 2, wherein the first axis of rotation is substantially horizontal and the second axis of rotation is perpendicular to the first axis of rotation.

4. Robotic milking system according to one of the preceding claims, wherein the first and / or the second mirror is / are rigidly connected to a permanent magnet, to a second permanent magnet, respectively, and wherein the laser illumination system further comprises a housing around said first and / or second mirror and the respective magnet with therein said coil, a second coil, respectively.

5. Robotic milking system according to one of Claims 2 to 4, wherein the first and / or the second mirror is / are rigidly connected to said coil, to a second coil, respectively, and wherein the laser illumination system further comprises a magnet provided around said coil or coils.

6. Robotic milking system according to one of the preceding claims, wherein the control comprises image recognition software configured to process the repeatedly captured images, to recognize at least said object in said processed images and todetermine a position of said object.

7. Robotic milking system according to Claim 6, wherein the control is configured to control the first and / or the second mirror by means of the control system on the basis of the processed images, in particular on the basis of the position of the recognized object.

8. Robotic milking system according to one of the preceding claims, wherein 7" is between 0.3 and 10 ps.

Images