Sorting system for managing made-to-order productions

Abstract

This invention concerns a new system for sorting components of complex products e.g. machinery which require customization, and hence a manufacturing model oriented to manage made-to-order production. The sorting system is feed with a set of indistinct parts which is sent through a recognition area via a conveyor belt, where a set of devices identifies distinctive features of the parts, such as the geometric properties. If the part is recognized, it is inserted by an anthropomorphic arm into the pallet dedicated to a specific active order. Otherwise, the unrecognized part is sent to the manual recognition area. The system is characterized by being fully automated and adaptable to various industries.

Description

[0001] TITLE: SORTING SYSTEM FOR MANAGING MADE-TO-ORDER PRODUCTIONS

[0002] TECHNICAL FIELD

[0003] The invention concerns technologies for production management, in particular for made-to- order production.

[0004] BACKGROUND ART

[0005] As is well known, made-to-order production differs from mass production in the level of customization, allowing specific customer needs to be meet. Such custom izations are generally achieved through outsourcing works to specialized companies.

[0006] This approach is followed not only by small and medium-sized enterprises, which cannot afford to insource specialized processing (due to the inability to effectively deploy the investment, a lack of specialized personnel, or simply a lack of adequate space), but also by large companies, which thus ensure both economies of scale and high production flexibility by outsourcing product customization to specialized subcontractors and centralizing only the final product assembly.

[0007] In the modern globalized economy, “mass-customization” i.e. the need to balance product standardization (to achieve the benefits of economies of scale) with product variety (to satisfy the diverse and changing needs of customers) is an issue that does not concern only the B2C sector but it is also frequent in the B2B sector, specifically in the production of industrial machinery.

[0008] In companies that operate according to a made-to-order model, the warehouse is a crucial department that requires adequate structure and management, as all incoming and outgoing materials must be managed quickly and precisely.

[0009] Specifically, incoming materials can be divided into two main categories: commercially available materials and custom-made materials. Management of commercially available materials is simple, as they are generally supplied in packages sorted by work order. Conversely, managing incoming materials that require outsourcing processing to be made is much more complex and time-consuming. The level of complexity clearly depends on the specific of the product to be made and on the manual vs. semi-automatic management system. For example, suppose a piece cut with plasma technology arrives from a supplier.

[0010] In the case of manual management of the work orders, the warehouse worker must perform the following operations: first, identify the order from the shipping document; then, print the drawing(s) of the items and perform a dimensional check; apply an identification tag (e.g. with the job number, drawing number, and item code); finally, place the piece on a “job pallet / box” ready for subsequent processing.

[0011] This sequence of tasks must be accomplished to all separate piece parts and repeated for all intermediate processes that are required for manufacturing the piece or its constituent parts, e.g., when the piece(or parts) is sent to an external supplier for welding and subsequent milling. Finally, when the welded and milled pieces are sent to an external supplier for final processing, such as painting, upon their return, the warehouse worker must: locate the order using the shipping document; print the drawing(s); apply an identification tag and finally, place the finished component (or an assembly) on a pallet / order box to be taken to the assembly area.

[0012] Machinery used in various sectors is made up of hundreds, if not thousands, parts and is typically made-to-order rather than mass-produced. This is the case, for example of winding machines that are used in the distribution and power transformer industry.

[0013] It is therefore clear that the management of complex production, such as industrial machinery, which by their very nature require specialized outsourcing, significantly requires specialized human resources for tedious and repetitive operations that nevertheless require great precision and accuracy. Indeed, if a warehouse worker makes an error, parts may undergo unplanned intermediate and final processing, or a semi-finished or finished component may be placed on a pallet or box that does not correspond to the intended order. Coding systems for parts and processes using barcodes, QR codes, or similar systems can improve a manual management system. However, the improvement is limited to process documentation, particularly the generation of the transport document, and the automation of part loading / unloading operations via the warehouse PC. However, the warehouse worker need to accomplish the same tasks described above as with the manual system.

[0014] Furthermore, similarly to the manual management system, the adoption of barcode-based systems does not significantly reduce recognition, management, and handling times, nor does it eliminate the possibility of errors.

[0015] Based on the inventor's extensive experience in manufacturing complex, custom-built machinery, there are currently no state-of-the-art technologies capable of solving or even mitigating the issues described above.

[0016] Therefore, it is still possible to innovate systems for managing made-to-order production, particularly those aimed at complex, customized products such as industrial machinery, characterized by outsourced manufacturing processes.

[0017] DISCLOSURE OF INVENTION

[0018] Object of the invention

[0019] The present invention intends to overcome the limitations and drawbacks of the solutions known in the art, by providing a system for managing made-to-order production which is characterized by a configuration and combination of its constituent parts suitable to impart a remarkable effectiveness and ease of use when applied to the management of complex productions susceptible to customization such as certain industrial machinery.

[0020] Particularly, the main purpose of the invention is to provide a sorting system which, in association with a warehouse management software stored in the company server, automatically recognizes parts and displaces them to the pallet or box dedicated to a predetermined ongoing order, or, if recognition is not achieved, displaces them to a dedicated area where an operator can perform manual recognition.

[0021] In addition, a second purpose of the invention is to provide a sorting system which, once part recognition has been obtained, allows quality control of incoming components to be performed, i.e. by identifying components whose dimensions do not conform to those indicated in the Bill of Materials (BOM).

[0022] Finally, a final purpose of the invention is to provide a sorting system adaptable to various types of made-to-order production, particularly those involving several outsourced processes and therefore requiring significant management effort.

[0023] General description of the invention

[0024] These and still other purposes, which shall appear more clearly hereinafter, are achieved by a novel sorting system for managing made-to-order productions. The general features of said sorting system are defined by the enclosed claim 1 , while the features of advantageous embodiments are set forth in the corresponding dependent claims.

[0025] The present invention also discloses an apparatus for part recognition in the management of made-to-order production, which comprises two or more of said sorting systems as defined in the enclosed claim 12. The present invention further provides a method based on said system, the general characteristics of which are defined in the enclosed claim 13.

[0026] Finally, the present invention also provides a computer program that implements said method, which is defined in the enclosed claim 15.

[0027] The aforementioned claims, to which reference is made for the sake of brevity of exposition, are specifically and concretely defined below and are intended to be an integral part of this description.

[0028] In summary, the inventive concept underlying this invention involves comprises three interconnected elements: first, a plurality of recognition devices for identifying features of a part used in the production of a made-to-order product; the second element consists of association data that relates the Bill of Materials (BOM) used to produce said product to its distinguishing features; finally, the third element is a selection algorithm which, based on the features detected by the recognition devices, identifies the component which is then associated to an active order.

[0029] In this way, the system according to the invention acts as an automatic sorter which sorts the various parts required to fulfill a number of active orders, wherein said parts are delivered undifferentiated to a warehouse area of a production site and they can be either commercially available components or materials and components to be subjected to internal or external processing. The adoption of the system according to the invention allows for a remarkable reduction in the times related to the administrative management of the work-order, to the quality controls of the ingoing parts as well as the parts handling. At the same time, the system relieves warehouse workers from repetitive, boring and error-prone activities.

[0030] Brief description of drawings

[0031] The features and advantages of the present invention will be more fully understood by reference to the following drawings:

[0032] Figure 1 schematically shows the sorting system according to the present invention;

[0033] Figure 2 schematically shows a functional diagram of the sorting system of Figure 1 ;

[0034] Figure 3 illustrates the part handler which enables the acquisition of images from different perspectives;

[0035] Figure 4 is a pictorial representation of feature-based part recognition guided by a selection algorithm according to a parallel recognition process;

[0036] Figure 5 is a pictorial representation of feature-based part recognition guided by a selection algorithm according to a sequential recognition process.

[0037] These drawings illustrate and demonstrate various features and embodiments of the present invention but are not to be construed as limiting the invention. DETAILED DESCRIPTION

[0038] The subject of this invention is a system for sorting parts of a product (P) to be used in managing made-to-order production. This system, which with reference to the enclosed Figure 1 is indicated with the reference number (1), is characterized by the fact of comprising: a transport system (10) for displacing a plurality of pieces (80) to be recognized from a loading area (11) to an unloading area (12); an automatic recognition area (20) of said pieces (80) included between the loading and unloading areas (11 ,12), said recognition area (20) comprising one or more recognition devices (30) selected from: a device (31) for image acquisition in the visible, IR or UV, e.g. a camera or a 3d scanner; a device for measuring optical properties (32), e.g. a colorimeter; a device for measuring non-optical properties (33) e.g. a weighing device or a roughness meter; a combination of the previous devices; a processing unit (40) configured to exchange data with said recognition devices (30) and to access association data between a list of parts (81) of a product (P), i.e. the bill of materials, and a set of active orders (CA), said processing unit (40) including at least one memory unit (41) having a recognition software (SW) stored therein, which, when executed, processes the data received from said recognition devices (30) to identify one or more piece distinctive features, which enable each of said pieces (80) to be uniquely associated with an active order (CA); one or more pallets or boxes (50) placed in the unloading area (12), each configured to cooperate with the transport system (10) and each assigned to a work order belonging to said set of active orders (CA); a manual recognition area (50NOK) configured to cooperates with the transport system (10); a sorting system (70) for said pieces (80), included between the automatic recognition area (20) and the unloading area (12), said sorting system (70) configured to exchange data with the processing unit (40), said sorting system (70) being configured to convey the recognized pieces (80OK) towards one or more loading units (50OK), and the unrecognized pieces (80NOK) towards said manual recognition area (50NOK).

[0039] The sorting system (1) according to the invention recognizes the pieces (80) within an automatic recognition area (20) having one or more devices (30) that are able to identify features of the pieces (80) and, through the association data, establish probabilistically whether or not they are associated with a given active order (CA).

[0040] The recognition devices (30) are chosen from: a device for acquiring images (31) in the visible, IR or UV range; a device for measuring optical properties (32); a device for measuring non- optical properties (33).

[0041] By way of example and not limitation, recognition devices (30) may be: a camera, a video camera, a 3d scanner, a colorimeter, detectors of visible (VIS), IR or UV radiation reflected by the surface directly or diffusely, meters of VIS, IR or UV radiation absorbed by the surface, a weighing device, a roughness meter or CMM (coordinate-measuring machine) probes.

[0042] Imaging devices (31) identify a geometric feature that includes: the length of an edge, the area of a surface, the curvature of a surface, the shape, or others.

[0043] Devices based on the measurement of optical and non-optical properties (32,33) identify a geometric feature such as: the color coordinates; the specular reflectance VIS, IR or UV; the diffuse reflectance VIS, IR or UV; the specular absorbance VIS, IR or UV; the diffuse absorbance VIS, IR or UV; the roughness; the electrical resistance.

[0044] Combinations of such geometric and physical parameters are possible.

[0045] Within the recognition area (20), the devices (30) can be arranged within the same volume, e.g. a load cell (33) and two cameras (31 , 3T) arranged within a single recognition tunnel.

[0046] Alternatively, the recognition devices (30) are arranged in a number of separate but contiguous volumes (20j) (i is an integer, with i>1). For example, a colorimeter (32) and a load cell (33) are arranged in a first volume (20i) separated from a second volume (2O2) containing two cameras (31 , 3T).

[0047] Regardless of the configuration of the automatic recognition area (20), two recognition modes are possible, which are schematically illustrated in Figure 4 and Figure 5.

[0048] The first involves a parallel or incremental recognition process i.e. the distinctive features, as they are detected by the recognition devices (30), are used as selection parameters of the association data to speed up the recognition process.

[0049] The second recognition method differs from the previous one in that the recognition devices (30) first identify the distinctive features and then filter the association data to identify the active order (CA) to which the piece (80) must be associated.

[0050] In order to speed up the recognition process, the recognition software (SW) stored in the processing unit (40) is assisted by artificial intelligence algorithms selected from the group consisting of: recurrent neural networks, convolutional neural networks, decision trees, random forests algorithms, support vector machines, k-nearest neighbor algorithms, or a combination thereof.

[0051] In any case, recognition by the sorting system (1) is a binary process that can lead to a positive outcome (defined as "OK") or a negative outcome ("NOK") in the case of non-recognition. The probability that a given piece (80) is recognized is given by the formula: wherein NOK and NNOK indicate, respectively, the number of positive and negative recognitions. In this regard, for the sake of clarity the reference numbers (80), (80OK) and (80NOK) shall be used to designate in this specification, respectively, the pieces to be recognized by the sorting system (1), the pieces recognized by the system (1), and the pieces not recognized by the system (1). Both pieces (80OK, 80NOK) represent a part of a product, for example a machine, whose production is managed through the made-to-order production system according to the invention. This product part, designated with the reference (81), is a finished component (or assembly) to be used as is or a rough component (or assembly) to be subjected to internal or external processing before assembly.

[0052] Clearly, the list of parts (81) constitutes the bill of materials stored in tabular format in the memory unit (41) of a company server which is accessible by the processing unit (40) of the sorting system (1) according to the invention.

[0053] For the purposes of implementing the present invention, each of the parts (81) within the bill of materials (or component register) is identified by a unique code, by views of CAD drawings, and by some distinctive features that characterize a given part (81). Examples of features include: weight, maximum dimensions (width, length, height), color, surface reflectance / absorbance, surface roughness, the presence of a detail having a particular shape. As illustrated schematically in Figure 2, a subset of the list of parts (81) is associated with an active work order (CA): in other words, there are association data that correlate the active work order (CA) with the parts (81) required for their fulfillment.

[0054] The set of active work order (CA) also includes sub-job orders (SCA) for the management of those components or assemblies that require internal or external processing, e.g. welding or painting processes.

[0055] The bill of materials (81) and the association data are typically stored in tabular format stored in an ERP (Enterprise Resource Planning) or a WSM (Warehouse System Management) software and are accessible from the processing unit (40).

[0056] Once a piece (80) has been positively recognized in the automatic recognition area (20), the association data allows such piece (80) to be associated with an active order and accordingly to be delivered by the sorting system (70) to the pallet (50OK) corresponding to that order.

[0057] If multiple orders (CA) are active, the recognized piece (80OK) can be assigned to the most urgent active order, or the active order closest to be fulfilled, or the most profitable active order one based on the settings parameters of the recognition software (SW) set by the user. When the order is closed, it means that all the parts (81) necessary for the production of the product are available.

[0058] The unrecognized pieces (80NOK) are instead sent by the sorting system (70) to the pallet (50NOK) for manual recognition. These pieces (80NOK) can refer to the following categories: the first includes those pieces included in the bill of materials (81) for which the recognition devices (30) were unable to identify distinctive features; the second includes those pieces included in the bill of materials (81) for which the recognition devices (30) identified distinctive features in an ambiguous manner, i.e. referring to multiple products; the third category refers to those new pieces not included in the bill of materials (81) since the distinctive features detected by the recognition devices (30) are absolutely not attributable to the existing components (81); the fourth category includes non-compliant pieces (80NOK) which include those pieces (80NOK) not included in the bill of materials (81), but “similar” to some already present in the master data, or those pieces (80NOK) included in the bill of materials (81) but not attributable to any active order (CA).

[0059] By statistically analyzing the results of a number / V of recognition tests, the processing unit (40) or the operator in charge of manual sorting process are able to distinguish to which category the unrecognized piece (80NOK) belongs. In this way, it is possible to implement corrective actions to reduce the number of pieces to be manually identified, e.g. by optimizing the image acquisition methods of the cameras (31 , 3T) or by inserting additional distinctive features to make the recognition unambiguous.

[0060] Based on the description provided, it will be clear to those skilled in the art that the sorting system (1) according to the invention not only allows for the recognition of pieces (80) and their attribution to a specific active order (CA) but is also able to assist quality control operations. In fact, based on the geometric properties or other distinctive features such as color or mass, it is able to identify products belonging to the fourth category, i.e. non-compliant products.

[0061] In this regard, similar components (80) can be distinguished in the list of parts (81) by indicating components similar to a given component or by introducing an appropriate similarity parameter that quantifies how much a given component is “similar” to others.

[0062] In this way, families of similar components can be recognized (e.g. a plate present in the bill of materials (81) with lengths of 100, 125, and 150 mm) and errors in the generation of a work order or non-conformities related to a sub-job order can be managed (e.g. a 125 mm plate mistakenly included or machined instead of a 100 mm can be recognized).

[0063] Further details of the present invention shall become apparent by the detailed description of two preferred but non-exclusive embodiments which follow.

[0064] First preferred embodiment

[0065] In the first preferred embodiment of the invention, herein described by way of example and not limitation, the sorting system (1) is configured to implement a recognition mode defined as “single piece mode”, i.e. by passing a single piece (80) at a time through the automatic recognition area (20).

[0066] For this purpose, the loading area (11) includes means for conveying the pieces (80) along a recognition direction (RD), e.g. one that facilitates the recognition of the piece (80) within the recognition area (20).

[0067] In this embodiment, a first anthropomorphic robot (13) picks up the pieces to be recognized (80) from a loading bay or a container and places them on a transport system (10) of the type comprising a conveyor roller which connects the loading area (11) with the unloading area (12). The type of gripping hand of the robot (13) depends on the specific production to be managed and in particular depends on the size and average mass of the pieces to be recognized.

[0068] In the case of mechanical parts, made of metal or plastic, the gripping hand is preferably an electrically actuated gripper. The first anthropomorphic robot (13) places the piece (80) to be recognized in correspondence with the recognition direction (RD), which is printed on the axis of the conveyor roller to provide a reference for the first robot (13).

[0069] In the first embodiment, the conveyor roller (10) is of the segmented type, i.e. it is made up of three distinct and contiguous conveyor belts (101 , 102, 103): the first (101) extends from the area where the robot (13) unloads the pieces up to the entrance of the recognition area (20); the second roller (102) moves the pieces within the automatic recognition area (20); finally, the third conveyor belt (103) directs the recognized pieces to a pallet (50OK) and the unrecognized ones towards the manual recognition area (50NOK).

[0070] Each of the three conveyor belts is driven by its own motor at a speed that can be set via the processing unit (40) by an operator or by a sensor signal, such as a load cell, or a camera, in order to prevent pieces overload. In particular, the speed of the second conveyor belt (102) is governed by the type and characteristics of the recognition devices (e.g. the acquisition time of an image) as well as by the outcome of the recognition process. If the piece (80) is not recognized, the motor of the second conveyor belt (102) can reverse its motion to repeat the recognition step.

[0071] Depending on the characteristics of the made-to-order productions to be managed, for greater construction simplicity, the conveyor roller (10) could be single and extend without interruptions from the loading area (11) to the unloading area (12).

[0072] In any case, the pieces emerging from the automatic recognition area (20) can be picked up from the third conveyor belt (103) by a second robot (73) which places the unrecognized pieces in a box or pallet (50NOK) for manual recognition, and the recognized pieces in the boxes, or pallets (50OK) corresponding to the active order number (CA) which is assigned by the recognition software (SW) as a result of the recognition process.

[0073] The use of a second robot (73) is preferable when there is a need to manage a large number of orders simultaneously or pieces of a certain size. Alternatively, a conveyor comprising 2 or more bars hinged to the structure of the unloading area (12) can be used. The bars are controlled by the processing unit (40) and orient synchronously so as to direct the pieces transported by the third conveyor belt (103) towards the box or pallet (50OK) corresponding to the active order (CA) or to manual recognition area (50NOK).

[0074] In the first preferred embodiment of the invention, herein described by way of example and not limitation, the automatic recognition area (20) is a single recognition volume consisting of a tunnel having the following recognition devices (30) therein: a weighing device (33) consisting of a load cell located underneath the second conveyor belt (102); two cameras (31 , 3T) fixed to the tunnel structure, e.g. on opposite sides and oriented so as to obtain images of the pieces (80) in transit from different angles.

[0075] The cameras (31 ,3T) equipped with appropriate frame grabbers detect a plurality of images which are sent to the processing unit (40) together with the weight detected by the load cell (31). In the first embodiment, the cameras (31 , 3T) are fixed, however they could be mounted on motorized turrets that can be oriented via the processing unit (40).

[0076] To identify the piece (80) with a high degree of confidence, it is essential to take images from different angles in order to precisely calculate a geometric characteristic or to maximize the likelihood that the distinctive feature of the piece (80) is identified and recognized. For this purpose, the recognition tunnel (20) includes handling means (21) that allow the piece (80) to be rotated during recognition so that images from different perspectives can be taken by the cameras (31 , 3T).

[0077] If the piece (80) is small in size, a suitable handling means (21) is a blower which, once appropriately connected to a pneumatic circuit (not shown), blows a jet of compressed air onto the piece (80) through a nozzle.

[0078] If, however, the piece (80) is larger, greater than approximately 100 mm, a handling means (21) consisting of a manipulator represents a better solution.

[0079] In the first preferred embodiment, the handling device (21) is a small anthropomorphic arm which is governed by a control logic to guide the gripping hand towards the piece (80). The gripping hand can be of an electric or pneumatic type. Advantageously, the control logic is functionally connected via the processing unit (40) to one or both cameras (31 , 3T) so that the images acquired can be exploited in the recognition process.

[0080] Alternatively, a handling device (21) such as the manipulator illustrated in Figure 3 is useful for rotating pieces (80) of even considerable dimensions. This is a pneumatic manipulator (21) with two degrees of freedom based on a gripping hand (212), e.g. a pneumatic or electric gripper, which can be moved easily along an axis (211) fixed to the top or to the walls of the recognition tunnel (20).

[0081] To facilitate the gripping of the piece (80), the axis (211) is aligned with the direction of entry of the pieces (80) into the tunnel, which has been defined as the “recognition direction” and is indicated in the enclosed figures with the reference (RD).

[0082] Optionally, a sensor can also be inserted to identify the position of the piece (80) inside the tunnel (20).

[0083] The gripper hand (212) can be moved along the axis (211) by means of a first motor (213), of a linear or rotary type, and a corresponding first transmission system (214), e.g. a toothed belt (214) mechanically connected to the shaft of a rotary electric motor (212). Clearly, the motor is conveniently fixed inside the tunnel.

[0084] To rotate the gripper hand (212), two solutions can be chosen. The first involves constraining the gripper hand (212) to the axis (211) not directly but via a second rotary motor (213'): in this configuration, the gripper hand (212) is rotatably constrained to the axis (211). In the second solution, the gripper hand (212) is fixed to the axis (211), which is rotatably constrained to the tunnel (20) by means of bearings. If the shaft ends are toothed, they can be connected to a second rotary motor (213') via two toothed belts.

[0085] Advantageously, if the gripping points of the gripper (212) are made conductive and are connected to a voltage generator of approximately ten volts, it is possible to rotate the piece (80) during recognition and at the same time detect its electrical properties, for example, recognize plastic parts (81) from metal ones.

[0086] Other equivalent solutions will appear obvious to those skilled in the art based on this disclosure, for example combinations of the handling means (21) described above.

[0087] In any case, the choice of the manipulator (21) depends on the type of parts (80) to be handled, the characteristics of which (e.g., the average weight or average dimensions) are known a priori.

[0088] In the first preferred embodiment of the invention, herein described by way of example and not limitation, the sorting system (1) includes a processing unit (40), for example an on-board computer, which exchanges data with the load cell (33) and with the two cameras (31, 3T) in the recognition tunnel (20).

[0089] The recognition software (SW) stored in the computer (40) recognizes the piece (80) based on its weight and the distinctive geometric features extracted from the image processing.

[0090] In this preferred embodiment, the part recognition process (80) occurs in parallel with the acquisition of the data acquired by the devices (30) and their processing.

[0091] According to this method, the following operations are executed in order: a) the processing unit (40) acquires the value of the mass of the piece (80) from the load cell (33) placed along the recognition direction (RD) below the second conveyor belt (102); b) the recognition software (SW) filters the association data between the list of parts (81) and the set of active orders (CA) based on the value read by the load cell (33), selecting only those parts (81) having a mass compatible with that value; c) the processing unit (40) acquires the images from the two cameras (31 , 3T) fixed inside the recognition tunnel (20); d) the recognition software (SW) processes the images using known algorithms, attempting to extract distinctive geometric features, such as length, width, area, or calculating a similarity index obtained by comparing the images from the cameras (31 , 3T) with the CAD images stored among the association data; e) the recognition software (SW) filters the dataset obtained in step b) using the results obtained in the previous step d) and evaluates whether the recognition process allows the piece (80) to be uniquely identified as belonging to one of the open orders (CA); f) if in step e) more than one compatible part (81) is identified, to define a positive recognition the processing unit (40) activates the handling devices (21) to rotate the piece (80) to allow further images from different perspectives to be acquired by the cameras. These images are used to start a new recognition cycle by repeating steps c) - e).

[0092] Once the piece (80) has been recognized, the processing unit (40) using the association data determines the identifier of the active order (CA) to which the piece (80) is associated. The processing unit (40) sends a command signal to the second robot (73) which picks up the piece (80OK) from the third conveyor belt (103) and places it on the pallet (50OK). If multiple jobs are active, the recognized piece (80OK) is assigned to the most urgent job (50OK).

[0093] If the recognition process fails, the second robot (73) picks up the piece (80NOK) and places it on the pallet (50NOK) for manual recognition.

[0094] It shall be evident to those skilled in the art that the procedure described here allows for speeding up the recognition process of the pieces (80) provided that the list of parts (81) can be divided into distinct clusters based on an easily identifiable distinctive features. For instance, if the feature is the mass of the piece (80), a Cluster “A” is made up of parts (81) with a mass lower than 0.1 kg, a Cluster “B” of parts (81) with a mass between 0.1 and 2 kg, a Cluster “C” parts (81) with a mass higher than 2 kg.

[0095] Alternative embodiments still based on “incremental” recognition can be obtained by simply replacing the weighing of the piece with the detection of another physical characteristic, such as color or diffuse reflectance in the visible range. In this case, it is advisable to install the colorimeter on a first recognition volume (20i) and the two cameras (31 , 3T) on a second recognition volume (2O2).

[0096] Second preferred embodiment

[0097] In the second preferred embodiment of the invention, herein described by way of example and not limitation, the sorting system (1) is configured to implement a recognition mode defined as “multiple pieces mode”, i.e. by passing several pieces (80) through the automatic recognition area (20) simultaneously.

[0098] In this preferred embodiment, the automatic recognition area (20) is a single recognition volume consisting of a tunnel having two cameras (31 , 3T) therein which are fixed to the tunnel structure, for example on opposite sides and oriented so as to obtain images of the pieces (80) in transit from different angles.

[0099] With this configuration, the recognition process of the piece (80) is performed in series with respect to the acquisition of the data acquired by the devices (30) and their processing.

[0100] According to this method, the following operations are executed in order: a) the processing unit (40) acquires the images from the two cameras (31 , 3T) fixed inside the recognition tunnel (20); b) the recognition software (SW) processes the images using known algorithms, attempting to extract distinctive geometric features, such as length, width, area, or calculating a similarity index obtained by comparing the images from the cameras (31 , 3T) with the CAD images stored among the association data; c) the recognition software (SW) filters the dataset using the results obtained in the previous step d) and evaluates whether the recognition process allows the piece (80) to be uniquely identified as belonging to one of the open orders (CA); d) if in step c) more than one compatible part (81) is identified, to define a positive recognition the processing unit (40) activates the handling devices (21) to rotate the piece (80) to allow further images from different perspectives to be acquired by the cameras. These images are used to start a new recognition cycle by repeating steps a) - d).

[0101] Once the piece (80) has been recognized, the processing unit (40) determines the identifier of the active work order (CA) to which the piece (80) is associated by using the association data. Further details for implementing this embodiment may be derived from the previous embodiment or from general knowledge well known to those skilled in the art.

[0102] CONCLUSIONS

[0103] To conclude, it has been shown that the invention described hereinabove fully achieves the intended aim and objects.

[0104] Furthermore, it shall be apparent to those skilled in the art that the sorting system disclosed herein is novel, and its development required significant inventive effort. Furthermore, the sorting system is particularly useful when used for managing made-to-order productions.

[0105] In particular, the system described above minimizes the time required for administrative management of the order (e.g., managing warehouse loading / unloading delivery notes), quality control of the parts of the order (the recognition system is certainly faster and more precise), and also for handling (using an automatic system, the warehouse worker is only responsible for checking). Furthermore, the use of an automatic system eliminates, or at least minimizes, errors throughout the entire order management process.

[0106] It is worth noting that although the description and examples provided contain many details, these should not be construed as limiting the scope of the invention but simply as illustrations of some embodiments of the present invention. Therefore, the present disclosure is not limited to the exemplary embodiments shown and described herein.

[0107] Hence, any modification of the present invention which falls within the scope of the following claims is considered to be part of the present invention.

[0108] Where the characteristics and techniques mentioned in any claim are followed by reference signs, these reference marks have been applied solely for the purpose of increasing the intelligibility of the claims and consequently these reference marks have no limiting effect on the interpretation of each element identified by way of example from these reference signs.

Claims

CLAIMSWhat is claimed:1) Sorting system (1) for parts of a product (P) to be used in managing made-to-order productions, said system characterized in that it comprises: a transport system (10) for displacing a plurality of pieces (80) to be recognized from a loading area (11) to an unloading area (12); an automatic recognition area (20) of said pieces (80) included between the loading and unloading areas (11 ,12), said recognition area (20) comprising one or more recognition devices (30) selected from: a device (31) for image acquisition in the visible, IR or UV, e.g. a camera or a 3D scanner; a device for measuring optical properties (32), e.g. a colorimeter; a device for measuring non-optical properties (33) e.g. a weighing device or a roughness meter; a combination of the previous devices; a processing unit (40) configured to exchange data with said recognition devices (30) and to access association data between a list of parts (81) of a product (P) and a set of active orders (CA), said processing unit (40) including at least one memory unit (41) having a recognition software (SW) stored therein, which, when executed, processes the data received from said recognition devices (30) to identify one or more piece distinctive features, which enable each of said pieces (80) to be uniquely associated with an active order (CA); one or more pallets or boxes (50) placed in the unloading area (12), each configured to cooperate with the transport system (10) and each assigned to a work order belonging to said set of active orders (CA); a manual recognition area (50NOK) configured to cooperates with the transport system (10);a sorting system (70) for said pieces (80), included between the automatic recognition area (20) and the unloading area (12), said sorting system (70) configured to exchange data with the processing unit (40), said sorting system (70) being configured to convey the recognized pieces (80OK) towards one or more loading units (50OK), and the unrecognized pieces (80NOK) towards said manual recognition area (50NOK).2) Sorting system (1) according to claim 1 wherein said distinctive feature is selected from: a geometric feature selected from: the length of an edge, the area of a surface, the curvature of a surface, the shape, a combination thereof; or a physical feature selected from: weight, color; visible, IR or UV specular reflectance; visible, IR or UV diffuse reflectance; visible, IR or UV specular absorbance; visible, IR or UV diffuse absorbance; roughness; electrical resistance; a combination thereof.3) Sorting system (1) according to claim 1 or 2 wherein the automatic recognition area (20) is divided into two or more recognition volumes (20j), wherein i is an integer, i>1 , said volumes being contiguous but separated, said volumes including a single type of said one or more recognition devices.4) Sorting system (1) according to claim 1 or 2 or 3 wherein: the transport system (10) is a conveyor roller connecting the loading area (11) with the unloading area (12), said transport system (10) that guides the pieces (80), one at a time, into the automatic recognition area (20), and then to a pallet or box (50) or to the manual recognition area (50NOK) by means of the sorting system (70); the loading area (11) includes means for conveying the pieces (80), along a recognition direction (RD); the automatic recognition area (20) is a tunnel comprising recognition devices (30)which recognize said pieces (80), one at a time, said devices (30) comprising a weighing device (33) and at least two cameras (31 ,31’) configured to obtain images from different angles of said pieces (80); the recognition software (SW) which, when executed, recognizes the piece (80) according to a distinctive feature and determines the identifier of the active order (CA) to which said piece (80,81) is associated.5) Sorting system (1) according to claim 1 or 2 or 3 wherein: the transport system (10) is a conveyor roller connecting the loading area (11) with the unloading area (12), said transport system (10) that guides the pieces (80), more than one at a time, into the automatic recognition area (20), and then to a pallet or box (50) or to the manual recognition area (50NOK) by means of the sorting system (70); the automatic recognition area (20) is a tunnel comprising recognition devices (30) which recognize said pieces (80), more than one at a time, said devices (30) comprising at least two cameras (31 ,31’) configured to obtain images from different angles of said pieces (80); the recognition software (SW) which, when executed, recognizes the piece (80) according to a distinctive feature and determines the identifier of the active orders (CA) to which said pieces (80,81) are associated.6) Sorting system (1) according to one or more of the claims 1 to 5 characterized in that the automatic recognition area (20) further includes a blower (21) which, once appropriately connected to a pneumatic circuit, blows a jet of compressed air onto the piece (80) or around it.7) Sorting system (1) according to one or more of the claims 1 to 6 characterized in that the automatic recognition area (20) further includes a manipulator (21) having twodegrees of freedom and configured to rotate the piece to be recognized (80), said manipulator comprising: an axis (211) aligned with the recognition direction (RD) and fixed to the top or to the walls of said tunnel or of one of the recognition volumes (20j); a gripping hand (212), pneumatic or electric, slidably moveable along said axis (211) by means of a first motor (213) and a first transmission system (214), said first motor (213) being of a linear or rotary type, a device for rotating the gripping hand (212) selected from: a second rotary motor (213’) rotatably constrained directly to the gripping hand (212); or, a second rotary motor (213’) rotatably constrained to the axis (211) by means of a second transmission system (214’), preferably a pair of toothed belts; optionally, a sensor to identify the position of the piece (80) within the automatic recognition area (20).8) Sorting system (1) according to one or more of the 1 to 7 characterized in that the automatic recognition area (20) further includes a manipulator (21) consisting of an anthropomorphic arm.9) Sorting system (1) according to one or more of the preceding claims wherein said association data between the list of parts (81) of the product (P) and the set of active orders (CA), comprise at least one distinctive characteristic of said pieces (80).10) Sorting system (1) according to one or more of the preceding claims wherein the loading area (11) includes at least one anthropomorphic robot (13) which picks up the pieces to be recognized (80) and places said pieces on the transport system (10).11) Sorting system (1) according to one or more of the preceding claims wherein the sorting system (70) comprises: at least one anthropomorphic robot (73) which picks up and places the recognizedpieces (80OK) on a pallet or box (50OK), and picks up and places the unrecognized pieces (80NOK) on the manual recognition area (50NOK); or conveyor bars (71) suitably hinged to a fixed element of the unloading area (12), said bars (71) which direct the recognized pieces (80OK) towards a pallet or box (50OK) or direct the unrecognized pieces (80NOK) towards the manual recognition area (50NOK), wherein said pallet or box (50) corresponds to the active order (CA) identified by the recognition software (SW).12) An apparatus (2) for parts of a product (P) recognition to be used in managing made- to-order productions, said apparatus (2) comprising two or more sorting systems (1) according to one or more of claims 1 to 11.13) Method for sorting parts of made-to-order machines comprising the following steps: a) storing on a processing unit (40) association data between a list of parts (81) and a set of active orders (CA), wherein said list of parts (81) includes distinctive features such as weight, shape or dimensions; b) defining the probability threshold Powhich corresponds to a positive recognition, according to said distinctive features, i.e. the piece to be recognized (80) is identical to one of the parts in the list of parts (81) of the product (P); c) obtaining a system (1) for managing made-to-order productions according to one or more of claims 1 to 12 or an apparatus according to claim 13; d) arranging a plurality of pieces to be recognized (80) on the loading area (11) of a transport system (10) served by said system (1); e) conveying said pieces to be recognized (80) to an automatic recognition area (20), downstream of said loading area (11), and comprising one or more recognition devices (30) selected from: a device for image acquisition (31) in the visible, IR orUV, such as a camera or a 3D scanner; a device for measuring optical properties (32), such as a colorimeter; a device for measuring non-optical properties (33) such as a weighing device or a roughness meter; a combination thereof; f) executing a recognition software (SW) on said processing unit (40), which receives as input the readings of said one or more recognition devices (30), queries said association data and provides as output: for each piece to be recognized (80), one or more piece distinctive features that allow each of said pieces (80) to be uniquely associated with an active order (GA); the identifier of said at least one active order (CA); g) conveying the recognized pieces (80OK) which are identified in a unique way towards an unloading area (12), downstream of said automatic recognition area (20), said unloading area (12) comprising one or more pallets or boxes (50) assigned to said at least one active order (CA); h) conveying the unrecognized pieces (80NOK), or the pieces which are identified in a non-unique way, towards a manual recognition area (50NOK); i) optionally, signal when the recognized pieces (80OK) loaded in a pallet or box (50) coincide with the list of parts (81) constituting said at least one active order (CA), j) optionally, print or apply or read a barcode or QRCode to apply or read additional information on the recognized piece (80OK) or unrecognized piece (80NOK).14) A method for sorting parts according to claim 13 wherein the recognition step f) is assisted by artificial intelligence algorithms selected from the group consisting of: recurrent neural networks, convolutional neural networks, decision trees, random forests algorithms, support vector machines, k-nearest neighbor algorithms, or a combination thereof.15) A computer program implementing the sorting method according to one or more of claims 9 to 14.

Images