"Method and device to discriminate two ends of an article from each other"

DESCRIPTION Technical Field

The present invention relates to improvements to methods and to devices for handling articles having a longitudinal extension and two ends that differ from each other, for example and in particular tubular knitted articles such as stockings or socks.

More specifically, the present invention relates to a new method and a new device to detect, i.e. distinguish or discriminate from each other two ends of the same article that differ in shape from each other, such as, in particular although not exclusively, a sock or stocking or other tubular knitted article. State of the Art

In the manufacture of stockings and socks it is the trend to increasingly use automation of production processes, starting from knitting of the article up to sewing of the toe.

In this trend towards automation, one of the most critical aspects is represented by the difficulty in automatically detecting which of the two ends of a tubular article, such as a sock delivered from a circular knitting machine with the toe open and placed randomly in a container, is the band end and which is the toe, to allow subsequent automated handling of the article in the sewing machines.

Frequently, these operations are performed by hand: an operator picks up the individual semi-finished articles from a container, such as a basket, in which the semi-finished articles coming from knitting machines are placed randomly with the toe end still open, i.e. not yet sewn or linked. The operator then inserts the semi-finished article onto a conveying tube oriented in the correct direction, i.e. in the position required by the machinery downstream, which will automatically perform sewing or linking of the toe.

Attempts have been made to solve the problem of automating detection, i.e. discrimination, of the elastic band end and toe of semi-finished knitted articles. For example, US-A-5, 040,475 describes a complex machine that picks up individual tubular articles from a container in which these are placed randomly. By making the article follow a specific processing path, detection means first detect the orientation of the tubular article along the feed path,

identifying whether the article is oriented with the toe or with the elastic band end facing forward along said path. After this has been detected, the tubular article is disposed in an intermediate station, from which it is delivered in one direction or in the opposite direction as a function of the orientation with which it entered said station.

JP-A-7468502 and JP-A-1272801 describe other apparatus for handling tubular articles such as stockings or socks in order to orient them appropriately.

US-A-6719577 describes a device that longitudinally orients individual tubular articles coming from a container, in which they are placed randomly.

EP-A-1221502 describes a device in which the individual socks or other tubular articles are picked up from a container in which they are placed randomly and are then oriented so that one of the ends thereof is always the leading end by means of a particular pneumatic path and using detection systems capable of distinguishing the elastic band end from the toe of the article through different stretch characteristics of the fabric.

US-A-5769286 describes a spreading device for longitudinally spreading socks or other tubular knitted articles.

EP-A-178143 describes a further system to detect the orientation of a tubular textile article by distinguishing the elastic band end from the toe.

US-A-5511501 describes a complex machine that picks up individual tubular articles from a container, in which they are placed randomly and separates them placing each individual tubular article in a respective container of smaller dimensions. Subsequently, each article is picked up from the respective container of smaller dimensions and inserted in a specific pneumatic path inside which the tubular article is oriented so that it is delivered from the pneumatic path always with the same orientation.

US-A-5884822 describes a further device and a method to pick up individual tubular articles from a container. US-A-5992712 describes yet another device to pick up individual tubular knitted articles and orient them appropriately.

Similar problems of orientation of tubular articles can be found in the feed of stockings or socks to setting machines. For example, in producing feminine stockings, there is the problem of inserting each single stocking, with

the toe already sewed, onto a setting board and for this purpose the individual stockings must be picked up from a container in which they are placed randomly to insert them over the board.

The methods and devices currently known to automatically distinguish from one another a first end and a second end of an article, such as a tubular knitted article, are not particularly reliable and are costly. Objects and summary of the invention

An object of an embodiment of the present invention is to provide a method and a device that allow detection, i.e. automatic distinguishing from each other of the two ends of an article, which has ends that differ in shape from each other. According to a particular aspect, the object of the invention is to provide a method and a device which allow the toe or the elastic band end of a sock or stocking to be distinguished and detected.

Although in the present description reference is frequently made to the need to detect the toe of the sock still to be sewn, it must be understood that the teachings of the present invention can also be applied when the orientation of an already sewn article shall be detected, for example to perform thereon further operations required in the production and/or packaging cycle, such as orienting the article correctly before it is placed on setting boards, feeding the packaging machine or the like.

In substance, in an embodiment of the invention there is provided a method for distinguishing from each other, i.e. for discriminating one from the other a first end and a second end of an elongated article, comprising the steps of: generating a curve that approximates the profile of one of said ends; and processing said curve to detect whether it approximates the profile of said first or of said second end.

In substance, according to this aspect of the invention, there is provided the generation by points of a curve that approximates the real profile of one of the two ends of the article, i.e. the one that is presented to the reading sensors. On the basis of the conformation of this approximation curve the device is able to detect whether this end is the first or the second of two ends characterized by profiles which are substantially different from each other, such as typically the toe and the elastic band end of a sock.

According to an aspect of an embodiment of the invention, the points of

the approximation curve are each characterized by two coordinates in a reference system, which can advantageously be a Cartesian reference system, although it would also be possible to use other reference systems, preferably two-dimensional, as the article will normally have a flat conformation.

In practice, in a possible embodiment, the method provides for moving with respect to each other an article and an array of photocells, for example aligned along a straight line, to gradually intercept the beams of the photocells with the end of the article. On the basis of the relative positions taken by the article and the photocells and of the electrical signals generated by gradual interception of the optical beams of the photocells, a plurality of points are determined lying on a profile that approximates the profile of the end of the article. As the profiles of the two ends differ from each other, the trends of the profiles reconstructed by point approximation will also differ from each other. By processing the curves that approximate these profiles the two ends can be discriminated from each other.

According to a possible embodiment, the method provides the steps of: arranging a plurality of photocells according to a predetermined arrangement; moving said arrangement of photocells and said article with respect to each other along a direction of relative movement, so that one end of said article intercepts the beams of said photocells during the relative movement; determining the coordinates of a plurality of points belonging to the profile of said end of the article that intercepts said beams, on the basis of: the relative positions taken by said article with respect to said photocells during the reciprocal movement, and of the arrangement of the photocells.

Preferably, in an embodiment of the invention, each time a photocell is intercepted by said article during said relative movement, the relative position between article and said photocells is detected, the position of the intercepted photocell and the reciprocal position between photocells and article along said direction of relative movement defining the coordinates of a point belonging to said profile.

The curve that approximates the profile of the end of the article that intercepts the photocells can be generated by interpolating the points belonging to the profile of said end, for example with a linear interpolation.

More complex processing would also be possible to obtain high degree interpolating polynomials of, although this is not absolutely necessary. An approximation through segments of straight line that join together the various points identified on the profile of the end of the article is usually sufficiently accurate. Therefore, in the present description and in the appended claims, curve is intended in general also as a broken line, obtained for example from the sequence of the portions of straight lines that join the points identified on the profile of the end of the article.

In a possible embodiment of the method according to the invention, the curve that approximates the profile of the end of the article read by the photocells is processed to determine the trend of the slope thereof, the first end and the second end of said article being detected and distinguished as a function of this trend of the slope. This can, for example, be performed by calculating the derivative of the curve obtained through the points identified on the end profile. The derivative can also be calculated by points, and therefore be constituted by a broken line that approximates, even roughly, the trend of the derivative.

In a particularly simple embodiment, although sufficiently accurate in many cases, the maximum and minimum values of the derivative are determined and the difference is calculated. The two ends of the article are detected, i.e. discriminated from each other, on the basis of a comparison between the maximum and minimum value of the derivative and a threshold value. This is possible thanks to the fact that, for example in the case of socks, the toe has a trend with a highly variable curvature, and therefore with a derivative that takes values differing greatly from one another along the extension of the profile representing the end, while the greater part of the band end is instead normally more or less rectilinear, and therefore its derivative has a flat trend.

According to another aspect, the invention relates to a device to distinguish from each other, i.e. to discriminate from one another a first end and a second end of an article, comprising an arrangement of sensors to identify a plurality of points belonging to the profile of an end of said article, and a control and processing unit which, as a function of the coordinates of said points, detects whether they belong to said first end or to said second

end.

In practice, the control and processing unit can be programmed so as to generate a curve that approximates the profile of said end as a function of said points identified by said sensors. According to a preferred embodiment of the invention, the sensors comprise an arrangement of photocells, and there are also provided handling members to move with respect to each other, along a direction of relative movement, said article and said arrangement of photocells, with a system to detect the reciprocal position between said article and said photocells. The control and processing unit can be programmed to determine the plurality of points associating the positions of the photocells that are intercepted by said article, during the relative movement between the article and the photocells, with the position taken each time by the article with respect to the photocells along said direction of relative movement. The arrangement of photocells can be of various types. Preferably, they will be disposed according to a linear string of photocells. This string of photocells is preferably disposed according to an alignment orthogonal to said direction of relative movement. In this way, a series of points in a system of Cartesian coordinates is obtained simply by translating the article and the array of photocells with respect to each other.

Advantageously, the control and processing unit can be programmed so that each time a photocell is intercepted by said article during said relative movement, the relative position between article and photocells is detected, the position of the intercepted photocell and the reciprocal position between article and photocells along said direction of relative movement, forming the coordinates of a point belonging to said profile.

Further advantageous features and embodiments of the method and of the device according to the invention are indicated in the appended claims and will be described in greater detail hereunder with reference to non-limiting possible embodiments of the invention. Brief description of the drawings

The invention will be better understood by following the description and accompanying drawing, which shows practical non-limiting embodiments of the invention. In the drawing:

Figure 1 shows a diagram of the device in a front view;

Figure 2 shows a section according to H-Il in Figure 1 ;

Figures 3A-3D schematically show a sequence of movement of an article like a stocking or sock, which is positioned with the toe facing the optical reading system;

Figure 4 shows a diagram of the profile of the toe of the article and the derivative thereof;

Figures 5A-5B schematically show a sequence of movement of an article which is positioned with the band end facing the optical reading system; Figure 6 shows a diagram of the profile of the band end and the derivative thereof;

Figure 7 schematically shows a different method of processing to discriminate the first and the second end of an article from each other, on the basis of the trend of a curve that approximates the profile of the end of the article viewed by the sensors of the device;

Figures 8 and 9 show a further embodiment of the method according to the present invention; and

Figures 1OA and 10B show a modified embodiment of the invention. Detailed description of a preferred embodiment of the invention Figures 1 and 2 very schematically show the device according to the invention. The numeral 1 indicates a supporting surface of an article M, which in the example illustrated is a sock with a toe P to be sewn and an elastic band end B.

Above the supporting surface 1 , which can be constituted by a conveyor or the like, there is disposed a movable element 3, such as a presser provided with a lifting and lowering movement fv and a translational movement according to the double arrow fo. The presser 3 is moved according to the double arrow fv by a piston-cylinder actuator 5, in turn carried by a slide 7 sliding in a guide (not shown). The movement according to fo of the slide 7 along the guide is controlled by a belt 9 driven around two pulleys 11, 13 the second of which is motorized by means of a motor 15. The numeral 17 indicates an encoder or other angular transducer that allows detection of the movement of the motor 15 and consequently of the slide 7 according to the double arrow fo. Instead of an encoder associated with the motor 15, any

other means could be used to detect the movements of the slide 7 or even of the presser 3 directly. This detection means (the encoder 17 in the example) is connected to a programmable control unit, indicated with 19, interfaced, for example, with a monitor 20 or other user interface. The control unit 19 can be provided with a keypad and other peripherals and units typical of a programmable control device usable to control the device.

The numeral 21 indicates a system of photocells comprising a linear array of photoemitters and a linear array of photoreceivers, the two arrays being parallel to and opposite each other. In the drawing the numeral 23 indicates a first of said arrays, for example the array of photoemitters, while the numeral 25 indicates the opposite array, for example the array of photoreceivers. As will be apparent from the description below of two operating cycles, the arrangement is such that the article M is made to translate with one or other of the ends thereof facing towards the detecting or reading device 21, to intercept with said end the beams of the photocells to detect whether the end that intercepts the photocells is the band end B or the toe P.

As in general the two ends P, B of the article M have profiles which differ from each other, by feeding the article M gradually with the end thereof under the linear array of photocells of the reading system 21 it is possible to read and detect the form of the profile of said end, using the following data: the position in space of the photocells; the signal of each photocell; the position of the presser 3 along the direction of movement fo and consequently the position of the article along the direction of feed according to the arrow F. Figures 3A-3D and 4 show detection of a toe P of the article M. These figures show four relative positions of the toe P with respect to the array of photocells 21 , indicated in these diagrams with X1-X8. In the examples eight emitter-receiver pairs are shown, but it must be understood that a different number of photocells can also be used. The number is dictated by requirements of cost ' and sufficient detection accuracy. In the position in Figure 3A none of the photocells is intercepted by the toe P of the article. In Figure 3B the photocell X4 is intercepted, while in the position in Figure 3C the position photocells X3-X6 are intercepted. In Figure 3D the position photocells X2-X7 are intercepted.

The positions of the photocells X1-X8 represent a corresponding number of values on the axis of the abscissas of a Cartesian diagram reproduced in Figure 4, while the positions of the article M, determined by the encoder 17 associated with the motor 15, represent the ordinates. By reproducing on a diagram X ₁ Y (Figure 4) in the abscissas the coordinates corresponding to the photocells that for each discrete value of Y (position of the article) are intercepted, a set of points P1-P6 is obtained which, interpolated, approximately provide the profile PF of the end of the article. More specifically, the point P4 is defined by the coordinate on the axis of the abscissas (x) defined by the position of the photocell X4 (the first which is intercepted) (Figure 3B) and by the coordinate YO. It is not necessary to know the absolute value of the coordinate YO on the axis of the ordinates, as what matters is the position of the coordinates on said axis of the subsequent points of the profile. As the article M advances, the position photocells X3, X5 are intercepted at the value of the ordinate Y1 , followed by the photocell X6 (Figure 3C) at the ordinate Y2 and finally the photocell X7 at the ordinate Y3 (Figure 3D). In this way the points of the profile PF are obtained. The value of the ordinate YO can be taken as the zero value and the subsequent ordinates are determined on the basis of the relative movement performed by the presser 3 and therefore by the article M.

The same occurs in the case in which the end that is read by the array of photocells is the band end. This case is represented schematically in Figures 5A and 5B, and Figure 6 shows the curve PF which is obtained by appropriately interpolating the points identified by the coordinates X ₁ Y thus determined. The points on the diagram are again indicated with P1-P6.

It can be seen that the diagrams PF in Figures 4 and 6 have a highly different shape in the two cases. In particular, it can be seen that the interpolating curve in Figure 4 has a much more variable slope with respect to the curve in Figure 6. Having generated the interpolating curve, with any suitable criterion, the derivative can be calculated. The trend of the derivative (again reproduced for the two cases in the diagrams in Figures 4 and 6 and indicated with D) represents the variation of the slope of the curve and is therefore indicative of the greater or lesser curvature of the profile PF detected by the photocells.

The derivative can be used, for example through a simple criterion of comparison with a threshold value, to detect i.e to distinguish the toe from the band end. In fact, it is for example possible simply to determine the maximum and minimum values of the derivative and calculate the difference. The difference in value can be compared with an experimentally determined threshold. If the difference in value exceeds the threshold value, this is indicative of a curved profile and consequently identifies a toe, while if the difference in value does not exceed the threshold, this is indicative of a flat profile and consequently identifies a band end. Experimentally, it is easy to identify a threshold value according to which the difference between maximum and minimum value of the derivative is significantly below and above this threshold for band end B and toe P profiles of the article, respectively.

Figure 7 schematically shows a different embodiment of the invention. In substance, in this case the device remains substantially as described with reference to Figures 1 and 2. The methods with which the points PI-Pn belonging to the profile of the end of the article that is read by the sensors, for example constituted by the photocells 23, 25, also remain substantially the same. Conversely, the methods with which the data representing the points belonging to the approximation curve are processed change.

Figure 7A represents a Cartesian reference system, the abscissas and ordinates of which are represented respectively with X and Y. P1-P5 indicate points belonging to the profile of the end of the article M. In this example five points are shown, but it must be understood that the number of points can vary (in this as in the previous embodiment) on the basis of the conformation of the system and of the shape of the article.

In this embodiment, the curve on which the points P1-P5 lie is sloping with respect to the coordinates X-Y. This means that the article has reached the photocells 23, 25 in an inclined and not in a straight position. This embodiment includes a preliminary step to process the data detected to correct this error. In substance, the curve on which the points P1-P5 lie is reproduced in a system of reference X'-Y ¹ rotated by an angle α, the straight line X' of the abscissas of which is more or less orthogonal to the axis of the article, consequently passing through the end points PI ₁ P5 of the curve.

lndicating with (a) and (b) the differences between the abscissa and the ordinate of the point P5 and the abscissa and the ordinate of the point P1 , therefore α = arctan(b/a) the equations that transform the coordinates of the points P1-P5 from the system X-Y to the system X'-Y' are: X' = Xcos(-α) - Ysin(-α) Y= Xsin(-α) + Ycos(-α)

Figure 7B represents the curve obtained by interpolation of the points P1-P5 in the reference system X'-Y ¹ , where the coordinates have been placed in horizontal and vertical position respectively. It is understood that this preliminary operation to correct the slope of the approximation curve of the profile of the end of the article (and consequently the interpolation curve of the points belonging to the profile) can be performed as a preliminary step also in the embodiment described with reference to Figures 3 to 6.

Once the curve has been reproduced in the reference system X'-Y ¹ , the difference between the maximum value and the minimum value of the ordinates of the points P1-P5, indicated in Figure 7B with δY', can be determined. As the two ends of the article M (toe P and band end B) are characterized by profiles that differ by degree of curvature, the value δY' of the difference of ordinates will differ significantly for the two ends. Figure 7B reproduces with a broken line the curve approximating the band end B of the article and δC indicates a comparison value determined experimentally on the basis of analysis of the shape of the ends of an adequate number of articles. The device can be programmed so that all the ends of articles in which δY'>δC are classified as "toes" and all the ends of articles in which δY'<δC are classified as "band ends".

Figures 8A and 8B describe an improved procedure or method to discriminate between toe and band end, which prevents discrimination errors due, for example, to interception of the photocells by the side of the article rather than by the end profile (band end or toe) if the article is positioned with a particularly high inclination with respect to the direction of movement.

In Figures 8A and 8B, B indicates the band end of the article M, FO, F1... FN indicate the photocells in number equal to N+1 and X, Y or X', Y' indicate

the Cartesian reference axes with respect to which the coordinates of the points belonging to the profile of the end (toe or band end) of the article M that intercepts the photocells are identified (the system X ₁ Y is constituted by axes parallel respectively to the alignment of the photocells and to the relative movement between photocells and article, while the system XT is the one rotated by an angle corresponding to the angle of inclination, with respect to the axis X, of the line joining the first and the last point of the profile considered for discrimination, according to the criterion of rotation described with reference to the previous embodiment). Figures 8A and 8B represent an article M (such as a sock) from the side of the band end while it passes in front of the photocells slightly inclined with respect to the direction of relative motion, represented by the arrow F. The amount of movement (H), i.e. the relative movement of the article with respect to the photocells, is established in advance and the article moves forward by said amount irrespective of the number of photocells effectively intercepted at the end of this movement. The amount H is preferably just sufficient to allow the profile of the band end B of an inclined article (angle β in the figure) to pass, to prevent a high number of photocells from intercepting the side of the article in the event of an inclined article. The amount (H) is determined experimentally on the basis of the type of article.

As can be easily understood by observing Figure 8A, in order not to distort the analysis, not all the points of the profile that have been intercepted by the photocells must be considered, but it is advisable to discard the two end points, as the point that intercepts the photocell marked with F8 would cause a distorted signal.

To discard both the first and the last point of the profile an algorithm described below can be used. To understand the following, it must be mentioned that before starting a detection operation the variables involved in the calculations are put equal to zero and ordered from 0 to N (with N+1 equal to the number of photocells) in the sense that the first photocell is assigned the input 0 of a PLC or other programmable control unit and the corresponding captured ordinate is YO; the second photocell is assigned the input 1 and the ordinate Y1 and so forth up to the last with input N and ordinate YN. Moreover, the second point of the profile intercepted by the photocells is

- - indicated with n1 and the second last point with n2 respectively.

This being stated, the detection algorithm is developed in the following steps:

Assignment of the variables 1) at the start of the procedure all the variables Yj and all the pointers j and z are reset, where j is the pointer or counter relative to the photocells (from 0 to N) and z is a flag required for correct execution of the algorithm, and which can take two alternative values: 0 or 1 ;

2) after resetting the values of the variables the relative motion between photocells and article M is started. It is necessary to take into account that the distance between the article and the photocells is not known in advance and therefore the movement continues until the profile of the article intercepts the first photocell;

3) at the first signal (interception of the first photocell by the article) the position Yj is captured (as according to the aforesaid j=0). This value is considered as the "zero" value of the relative movement, in the sense that the final value of the movement is set at this point. As H is the maximum travel between the instant in which interception starts and stopping of the relative movement between article and photocell, consequently 4) the position at which the motor that feeds the article stops is determined equal to the sum Yj + H

5) during the movement from YO to Y0+H the interception positions of the profile (ordinates Yi) are captured and memorized. I.e.: during the relative movement equal to the travel H the values of the relative movement corresponding to each single interception signal of a photocell are collected;

6) the motion is stopped at the position Y+H. Definition of the polygonal that approximates the profile:

The polygonal is determined considering the coordinate points Yj excluding the first and the last point intercepted. With reference to Figure 8A, the end points of the polygonal which must be considered are the points n1 and n2 (second and second last). The procedure to discard the point preceding n1 and the point subsequent to n2 is as follows:

7) reading of the ordinate Yj indicated by the pointer j

if Yj = 0 and z = 0 , skip to step 8 if Yj > 0 and z = 0 , skip to step 9 if Yj = 0 and z = 1 , skip to step 9 if Yj > 0 and z = 1 , skip to step 8 8) j = j + 1 and return to step 7

9) if z = 0, n1 = j + 1 , z = z + 1 , skip to step 8 if z = 1 , n2 = j - 1 , skip to step 10

At the end of this iteration the points n1 and n2 have been identified and consequently the polygonal that approximates the profile intercepted by the photocells is given by the totality of the points n1, n2 and intermediate points. In the subsequent step analysis of the polygonal is performed. The coordinates in the system X, Y of the points n1 and n2 respectively are indicated with Xn1 , Yn1 , and with Xn2, Yn2.

10) A = arctan ((Yn2-Yn1)/((Xn2-Xn1)) 11) rotation of the coordinates (the equations already described with reference to the preceding example are applied)

12) search for the minimum and maximum values of the ordinates Y

13) comparison with the value set to assign the type of profile: if the difference between maximum and minimum value of the ordinates Y ¹ is greater than the predetermined value, the profile intercepted belongs to a toe, otherwise it belongs to a band end.

It is understood from the algorithm described and from Figure 8 that by discarding the first and the last point identified by passage of the profile of the article in front of the photocells, the end identified is prevented, for example in the case in Figure 8, from being interpreted as a toe, while it is actually a band end. Due to the inclination between axis of the article and direction of motion, this error could derive from interception of the point n3 belonging to the side of the article by the photocell F8.

In the embodiment previously described it was assumed that the article M will be positioned with a toe P always substantially flattened frontally, i.e. with an arched profile as shown in the figures. However, this is not always the case and in particular according to how the article is handled and flattened, it can be positioned lying on one side, so that the toe P does not have the arched profile, shown for example in Figure 2, but a stepped profile, as shown

in Figure 9, with an approximately rectilinear area.

This could cause difficulty in interpretation, i.e. a risk of incorrect detection, as both the toe and the band end would in this case have rectilinear profile portions, although of different lengths, if the amount H of reciprocal movement between article M and photocells is below the length L. On the other hand, in the case of inclined articles (angle β Figure 8), to prevent an excessive number of photocells from being intercepted by the sides of the article rather than by the end profile, the amount H must be limited.

To avoid an error in discriminating between toe and band end in the case in which the toe is positioned as shown in Figure 9, rather than using the criterion of discrimination described above, based on the difference between maximum and minimum height with respect to the system of reference X', Y', of the points of the profile, a toe is discriminated from a band end by counting the number of photocells that have intercepted the profile, as in the toe these are undoubtedly fewer than those involved in a band end. According to an improved embodiment of the method of the present invention, therefore, after performing the movement of a degree H between article and photocells, in the first place the total number of photocells intercepted by the profile of the article is determined. If this is below a given minimum number, then the end that has passed before the photocells is a toe. If the number of photocells intercepted is higher than said given minimum, then the discrimination procedure described is performed on the basis of the difference of coordinates.

In conclusion, the more efficient system comprises both the criterion of rotation of the coordinates of the profile and therefore of comparison between Ymax and the set value, and the one that takes account of the number of photocells that intercept the profile.

In the above described embodiments, only one end of the article is caused to interact with the photocells or other sensors, in order to recognize whether said end is the first end or the second end of the article. Once said end is recognized as being the first end (for example the toe), the other end is assumed to be the second end (for example the band end). This way of performing the method is not however the only possible alternative. Indeed, e.g. in order to achieve a higher degree of accuracy and reproducibility of the method, or else in order to reduce the risk of misinterpretation of the detected

data, in an improved embodiment of the method according to the invention both ends of one and the same article are subject to detection by means of the same or of two different sensor arrangements.

In one possible embodiment, described here below with reference to Figs. 1OA and 10B, the device includes a single linear array of photocells 21. In an embodiment, said photocells include photoemitters 23 and photoreceivers 25 arranged on two sides of an intermediate position where the article of manufacture M is caused to pass, as described above. However, in a different embodiment (which can be applied in the above disclosed embodiments as well) the photocells can include photoemitters arranged on the same side of e.g. a reflective surface or conveyor on which the article M is placed.

Two handling members 5A and 5B are arranged, one on each side of the photocells arrangement 21. In an embodiment the members 5A, 5B are movable in a vertical direction according to double arrow fv and are suspended from a supporting structure schematically shown at 8. In an embodiment the two handling members 5A, 5B include respective endless belts 6A, 6B, entrained around rollers (three for each handling member in the example shown). At least one of each three-roller cluster is motorized while the others can be idle. The motorized roller causes the respective belt 6A, 6B to move according to double arrow fo.

An article of manufacture M having a first end (e.g. a toe end P) and a second end (e.g. a band end B) is moved under the handling members 5A, 5B. As in the previous examples the orientation of the article M is random, i.e. it is not known beforehand which is the toe end and which is the band end. In the example shown the band end B is on the left-hand side of the figure and the toe P is on the right-hand side.

Once the article M is placed underneath one or the other of the two members 5A, 5B, the members are lowered (arrow fv), such that at least one of said handling members touch the article M and presses it against the supporting surface 1 , for example a conveyor moving orthogonally to the figure. One or both belts 6A, 6B are put in motion in the same direction, for example such as to push the article towards the left-hand side. During this motion one of the ends of the article pass between the surface 1 and the

emitters 23, i.e. intercepts the emitter-receiver pairs 23, 25. In the example shown, in the position shown in Fig. 1OA the band end B has just intercepted the photocells. Once this happened, the device is able to determine, on the basis of one of the above disclosed modes, whether the end which just passed between the photoemitters 23 and photoreceivers 25 is the first or second end. The motion of the belts 6A, 6B can be reversed or continued (depending on the starting position of the article) such that also the second end passes through the photocells 21. The device determines whether the second end is a band end or a toe end, adopting one of the above disclosed methods.

If the two readings are consistent (i.e. if the first reading determines that the end first detected is e.g. the band end, as in the example shown, and the second is the toe end) then the orientation of the article M is properly determined. If, however, the readings are inconsistent (e.g. if both ends are read as being toe ends or both as being band ends), then the control unit considers the reading as aborted and discards the article. The article can in such case e.g. be resent to the basket or other container from which it is picked up again for a subsequent detection process.

Errors (double-toe detection or double-band end detection) can be caused e.g. if the article is not properly spread on the surface 1.

If both ends are read the system reduces the chance of errors to a really negligible value, even if a somewhat less reliable algorithm is used to discriminate between toe and band end portions respectively. Thus a reliable system can be implemented using a substantially simple algorithm. It is understood that the drawing only shows an example provided purely as a practical embodiment of the invention, which can vary in forms and arrangements without however departing from the scope of the concept underlying the invention. Any reference numerals in the appended claims are provided to facilitate reading of the claims with reference to the description and to the drawing, and do not limit the scope of protection represented by the claims.