Method and apparatus for handling parts of moving article by means of robot

Abstract

The invention relates to a method (100) for treating a part of a moving article, in particular a sole, by means of a robot, comprising: (a) creating (110) a first path (240) with respect to the part to be treated, comprising a list of points or a predetermined trajectory; (b) determining (120) a 3D scan (610) of the component to be processed by the 3D scanner; (c) calculating (130) a global offset between the first path (240) and the 3D scan (610) of the component to be processed; (d) calculating (150) a plurality of local offsets between the first path (240) and the 3D scan (610) of the component to be processed; (e) creating (160) a second path (810) based on the first path (240), the calculated global offset, and the calculated local offset; and (f) conveying (170) the second path (810) to a robot that performs processing of the component based on the second path (810).

Description

Technical Field

[0001] The present invention relates to a method and an apparatus for processing components of a sports article by a robot, based on a 3D scan of a component to be processed, in particular a sole and / or an upper of a shoe. Background Art

[0002] The production of sports articles - in particular soles and / or uppers of shoes - includes a plurality of processing steps in which components of the sports article are processed. These processing steps may include deburring, cutting, decorating, grinding, gluing or cementing of these components. Due to the material instability of the components - which in particular leads to shrinkage, distortion, bending or warping of the components - such processing steps are difficult to automate and are usually carried out manually. Compared to automated methods, manual processing or post-processing results in higher production costs and may lead to components not being processed in an ideal manner, which results in defective components and thus an increase in costs.

[0003] Previous automated methods for processing components of sports articles by a robot were based on the so-called "best fit" principle. First, a computer-aided design (CAD) model of an ideal component is created. In addition, an ideal processing path is created along which the processing of the component is carried out. This ideal processing path can be taught to the system by learning techniques or can be explicitly predetermined, for example, by a sequence program. Then the predetermined ideal path is applied to the CAD model. In both cases, the path usually contains a list of points at which the robot is to start processing. In addition, a camera system is used, which detects the position and orientation of the entire real component to be processed, for example, by scanning. The data and information thus obtained are processed by the system and used to calculate a global offset by matching the scan with the CAD model. The global offset can include both translation and rotation and allows the system to accurately determine the position and orientation of the component to be processed. The offset thus obtained is valid for the entire processing path of the robot and the processing path is corrected according to the global offset, i.e., the processing path is rotated and / or shifted. This allows the system to correct the incorrect position of the component as a whole.

[0004] However, deviations between the real part and the CAD model due to material instability of the part cannot be corrected, and such deviations especially lead to shrinkage, distortion, bending or kinking of the part. Due to material instability, the real part to be processed no longer corresponds to the CAD model. Even after global offset correction between the real part and the CAD model, the robot still continues to follow the ideal processing path related to the CAD model, so the actually executed robot motion does not correspond to the optimal real processing path. This results in inaccurately machined parts, which reduces the quality of the final product. Depending on the magnitude of the deviation, the part may even be unusable after processing and must be excluded from further processing or must be post-processed manually. This results in increased manufacturing costs per part, higher working hours, and may lead to poorer quality of the part or the final product. Summary of the Invention

[0005] Accordingly, an object of the present invention is to provide a method and a device that enable the automated processing of parts of a moving article using a robot.

[0006] This object is achieved by a method for processing parts of a moving article using a robot, the method comprising the following steps: a) creating a first path for the part to be processed, the first path comprising a list of points or a predefined trajectory; b) determining a 3D scan of the part to be processed by means of a 3D scanner; c) calculating a global offset between the first path and the 3D scan of the part to be processed; d) calculating a plurality of local offsets between the first path and the 3D scan of the part to be processed; e) creating a second path based on the first path, the calculated global offset and the calculated local offsets; and f) transmitting the second path to the robot, which performs the processing of the part based on the second path.

[0007] The method according to the invention minimizes the deviation between the real part to be processed of a moving article and an ideal computer-generated model, maximizes the individual adaptability of the path, and thus optimizes the quality of the automatically processed part. Within the scope of the present invention, a part of a moving article is understood to be a spatially delimited part of a moving article. The method can be used to process a plurality of such parts of a moving article. In addition, a part according to the invention can also be understood as the entire moving article. This especially applies to cases where the moving article cannot be subdivided into spatially definable parts. The moving article can be a sports shoe, and the part can be the sole of the sports shoe or a part of the sole. In addition, the part can also be the upper.

[0008] In order to individually adapt the path for processing a sports article part, a first path regarding the part is first created. This path relates to the ideal part and can be given by a list of points or by a curve along which the robot is intended to move. The list of points or the curve is defined relative to the robot coordinate system and serves as the starting point for the processing of the part, regardless of its individual characteristics.

[0009] In the next step, a 3D scanner is used to scan or detect the part to be processed. This enables the recording of the individual characteristics and features of the part to be processed, and the individual adaptation of the second path is based on these characteristics and features. These individual characteristics and features can include deviations from the ideal part and can include shrinkage, distortion, bending, kinking, etc.

[0010] In the next step, the global offset between the first path and the 3D scan of the part to be processed is calculated. This global offset can include a translation or rotation of the first path and serves as the first adaptation of the path defined relative to the ideal part and the real part.

[0011] In the next step, a plurality of local offsets between the first path and the 3D scan of the part to be processed are calculated. In addition to including the global characteristics of the real part in the previous method step, this also allows for the inclusion of individual local characteristics. Based on these local offsets, a second path is then calculated starting from the first path and the determined global offset, which takes into account the local individual characteristics and features of the real part. Thus, this second path minimizes the deviation between the real part and the first path generated relative to the ideal part.

[0012] In the final step, the second path is transmitted to the robot or the robot system. Based on this second path, the robot performs the processing of the part previously detected by 3D scanning. The processing can include, for example, cutting or trimming the part, such as a sole or a part of a sole. In addition, the processing can include polishing the part or applying an adhesive to the part. Furthermore, within the scope of the processing, decorative elements can be applied to the part, such as a colored layer.

[0013] The first path can be determined based on the CAD (Computer-Aided Design) model of the part to be processed. The advantage offered by CAD is that the part to be processed can be developed cost-effectively and quickly on a computer. If CAD data for the part to be processed exists, these data can be directly incorporated into the method according to the invention. In this way, the CAD data is used not only for the design of the part but also for its production or processing.

[0014] The detection or scanning of the part to be processed can be performed by an optical 3D scanner. This enables rapid imaging and high detailing.

[0015] The calculation of the global offset can be performed based on the features of the part to be processed, where the features can be detected by a 3D scanner. The features are part-dependent and can be, for example, depressions, protrusions, or bevels. The selection of features can depend on the degree of detailing of the 3D scanner used.

[0016] The part to be processed can be divided into a plurality of spatial segments using the CAD model on which the first path is based, and the calculation of the local offset can be performed with respect to the plurality of spatial segments of the part to be processed. Thus, the greater the number of spatial segments, the greater the number of local offsets resulting, and thus the more precise or individualized the creation of the second robot path.

[0017] The plurality of spatial segments created with respect to the CAD model on which the first path is based can be cuboids. This allows for a simple and unambiguous division of the CAD model and thus a simple and unambiguous division of the first path. Cuboids are mathematically easy to handle and facilitate fast digital processing.

[0018] The first path and the second path can have a plurality of spatial points. With a greater number of spatial points, in particular the curvature of the part to be processed can be taken into account.

[0019] Between the points of the first path and the points of the second path, there can be a bijective correspondence. This means that there is a one-to-one assignment between the points of the first and second paths. Through the bijective correspondence between the points of the first path and the second path, it is ensured that the curvature of the part to be processed can also be taken into account in the second path. In addition, the bijective correspondence limits the geometric complexity of the second path and thus the running time of the robot with respect to the second path. Thus, before the creation of the second path, i.e., through the creation of the first path, the running time of the robot has already been affected.

[0020] Another aspect of the invention relates to a device configured to perform the method described herein. It goes without saying that what has been applied herein regarding the method of processing a moving article part by a robot also applies to the device, in particular regarding the functions, features, embodiments, and advantages of the device and the method. Therefore, these are not mentioned separately here. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Exemplary embodiments of the invention are described below with reference to the drawings. The drawings show: Figure 1 : A flowchart of a possible embodiment of the method according to the invention; Figure 2 : CAD models of a sole, a gripper, and a flange, and a CAD model of a first robot path created based on the CAD model of the sole; Figure 3: The CAD model of the calibration adapter, where the calibration adapter consists of a calibration flange, a calibration plate, and three calibration balls, and the three calibration balls are used to define the coordinate system; Figure 4 : Constructing a coordinate system using three calibration balls; Figure 5 : The CAD model of the calibration adapter and the 3D scan of the manufactured calibration adapter, and the transformation of the 3D scan data onto the calibration adapter; Figure 6 : The area of the distinct features of the sole, which is defined relative to the CAD model; Figure 7 : Subdividing the sole CAD model into 3D grid regions; Figure 8 : The first robot path created relative to the CAD model and the second robot path defined based on local offsets; and Figure 9 : Illustrating the treatment of the sole by a trimming tool, where the sole is attached to the robot arm by a gripper. Detailed Description of the Invention

[0022] Only some possible embodiments of the present invention will be described in detail below. It goes without saying that these exemplary embodiments can be modified and combined with each other in various ways as long as they are compatible, and certain features can be omitted as long as they are unnecessary.

[0023] It goes without saying that not all features of the described aspects / embodiments must be present in order to achieve the technical advantages of the present disclosure, which is defined by the subject matter of the claims. The disclosed aspects / embodiments can be modified by combining certain features of one aspect / embodiment with one or more features of another aspect / embodiment. In particular, those skilled in the art will understand that the features and / or functional elements of one aspect / embodiment can be combined with the technically compatible features and / or functional elements of any other aspect / embodiment of the present disclosure as long as the resulting combination falls within the definition of the present disclosure.

[0024] In the drawings and the description of the present invention, the same reference numerals denote the same elements. For the sake of clarity and conciseness, certain aspects of the components or steps of certain embodiments are presented without excessive detail - if such details are implicit to those skilled in the art based on the teachings contained herein, and / or if such details would obscure the understanding of the more relevant aspects of the embodiments.

[0025] Within the meaning of a person skilled in the art and / or to avoid redundancy, reference is also made to the explanations in the foregoing sections, which are also applicable to the following detailed description. Furthermore, for the sake of brevity and clarity, not all features, parts, elements, aspects, components and / or steps are explicitly identified by reference numerals. This applies in particular if a person skilled in the art recognizes that these features, parts, elements, aspects, components and / or steps are present in plurality.

[0026] As already explained, as will be explained below with reference to some embodiments, the method according to the invention or the device according to the invention can be particularly used for trimming or cutting soles. However, the invention is not limited to the treatment of soles or their components.

[0027] As Figure 1 shown, a possible exemplary embodiment of the method 100 for processing a moving product part by a robot or a robot system is to trim one or more expulsions of a sole using an optical 3D scanner by a computer-controlled robot arm. Such expulsions on the part can be caused by material leaving through an opening or gap that is created between two or more molding bodies. The opening can be provided to vent the mold during molding or injection molding.

[0028] Soles are mass-produced articles and must thus be produced to a large extent. This particularly includes producing a specific sole model in a specific size multiple times. In a first method step 110, a CAD model 200 of a sole 210 is created, which is used as a reference model in the following steps. The model reproduces the shape of an ideally formed sole. As Figure 2 shown, in the exemplary embodiment shown here, the created CAD model 200 also includes a gripper 220, which can be mounted on a robot arm (not shown) via a flange 230. The gripper is used to fix or hold the sole and thus forms the connection between the sole and the robot. In another embodiment, the created CAD model can also consist only of a CAD model of the sole.

[0029] Regarding the CAD model of the sole 210, an ideal robot path 240 is now defined, which is either generated by means of learning techniques or directly predetermined. The data associated with the ideal robot path 240 is a list of points defined relative to the robot system, and these points define the trajectory that the robot is intended to follow. The predetermined trajectory in the CAD model 200 of the sole 210 including the gripper 220 exactly corresponds to the real gripper model. If this is not the case, deviations will necessarily occur in subsequent method steps, especially when calculating the corrected trajectory. The number of points in this list can be automatically calculated as a function of the path curvature. The actual robot path is then a polygonal sequence of the points contained in the list. Alternatively, the robot path can also emerge from the predetermined list of points by means of polynomial interpolation, spline interpolation or other interpolation techniques.

[0030] In a second alternative method step 115, a calibration phase is first performed. This calibration phase is used to align the coordinate system of the 3D scanner with the robot coordinate system. For this purpose, the calibration adapter 300 is constructed as a CAD model and manufactured accordingly. As Figure 3 shown, the calibration adapter 300 consists of a clamping flange 310, a calibration plate 320 and three optical calibration spheres 330a - 330c. These three optical calibration spheres 330a - 330c (which are attached to the calibration plate 320) define a coordinate system 400 that can be traced back to the robot flange, i.e., the connecting element of the robot arm and the calibration adapter 300 or the gripper 220, and thus to the origin of the robot coordinate system. As Figure 4 shown, the coordinate system 400 is defined by two vectors X 410 and Y 420, which are defined by the centers of the spheres 330a and 330c or 330a and 330b. These two vectors have the same origin 440, i.e., the center of the sphere 330a, and thus define a plane in three-dimensional space. The third coordinate axis Z 430 is then a vector perpendicular to this plane, and the coordinate axis Z also has the center of the sphere 330a as its base point.

[0031] The calibration step 115 is optional and does not have to be performed every time a component is processed. For example, it is sufficient to perform the calibration only once during system setup or at specific time intervals. Therefore, the optional calibration step 115 is only explained here for the sake of completeness.

[0032] For calibration, the calibration adapter 300 is now optically detected using a 3D scanner. In this case, the calibration adapter 300 is presented to the scanner at the same spatial position and orientation as the sole to be processed, which will later be mounted on the gripper. By this optical detection of the calibration adapter, the center of the ball and the calibration plate can be identified. Using the CAD model of the calibration adapter 300 and based on the nominal position of the center of the ball relative to the robot flange, a transformation matrix can then be calculated by the computer system. This transformation matrix now allows the position and orientation of the robot flange relative to the 3D scanner to be calculated. The obtained transformation matrix is stored on the computer system for later individual evaluation.

[0033] Figure 5 An optical 3D scan of the calibration adapter 510 as seen from the perspective of the 3D scanner, and the CAD model of the calibration adapter 300 are shown. The transformation matrix can be calculated based on the identification of the center of the ball and the calibration plate 320, such that the optically detected calibration adapter 510 can be traced back to the CAD model of the calibration adapter 300. In another method step performed after the calibration, generation, and storage of the transformation matrix, the calibration adapter is replaced by a gripper that can be attached to the robot flange. The sole to be processed is now mounted or placed on this gripper. In this case, the robot arm or the robot flange is already in the initial position, i.e., the position where the calibration scan has been performed, or is moved to this position after the sole has been mounted or placed on the gripper. Thereafter, in the next method step, the determination 120 of the 3D scan 610 of the actual sole is performed by the optical 3D scanner.

[0034] Due to the flash areas 640 that may occur during the sole production process, as Figure 6 shown, and due to the instability of the material, the actual sole does not exactly correspond to the ideal sole model. In particular, the 3D scan 610 of the actual sole can deviate from the sole in the stored CAD model 210.

[0035] In the next method step 130, a global pre-orientation of the CAD model 210 of the sole including the gripper 220, and the obtained optical 3D scan 610 of the actual sole mounted on the gripper takes place. This global pre-orientation is used for the registration of the scan data of the CAD model 210. Depending on the sole to be processed, a distinct feature 620 can be assigned to it, which can be detected by the 3D scanner and can be associated with an area 630. In the embodiment shown here, there is a depression 620 in the area 630 around the center of the sole.

[0036] In a first pre - orientation step 130, a rough orientation occurs between the 3D scan 610 of the real sole and the CAD model 210 based on a selected region 630 containing one of the distinct features 620. In a specific embodiment, the data of the 3D scan 610 and the CAD model 210 are given as point clouds. Now a transformation is calculated that globally minimizes the distance between the two point clouds. This can include minimizing the distance metric between the point clouds with respect to the set of all transformation matrices. Thus, the result of this first global pre - orientation step is a transformation matrix.

[0037] In a second pre - orientation step 130, the so - called global fine - orientation or "ICP registration" (iterative closest point registration algorithm), the transformation determined in the first pre - orientation step 130 is further optimized. This new orientation or registration further improves the global orientation of the 3D scan 610 of the real sole with respect to the CAD model 210.

[0038] After the global orientation of the 3D scan 610 of the real sole and the CAD model 210, and thus in particular after the global orientation of the ideal robot path with respect to the 3D scan 610 of the real sole, now a local correction occurs in another method step 140. This local correction allows for a precise orientation between the 3D scan 610 and the CAD model 210 that goes beyond just a translation or rotation of the model as a whole.

[0039] For the local orientation or correction 140, the sole is subdivided into spatial sections 710 along the sole using the CAD model, as Figure 7 shown. Additionally, these sections are divided transversely to the sole. This results in the sole being subdivided into cuboid spatial sections 710, which are hereinafter referred to as 3D cells. The spatial extent of each individual 3D cell 710 can be adjustable separately or the same for all 3D cells 710. Now the data of the 3D scan 610 of the real sole are assigned to these spatial sections, and the scan data are extracted from the point clouds regarding these sections.

[0040] Now in another method step 150, within the regions defined with respect to the sections, the scan data itself defined with respect to the sections are again registered with respect to the CAD model 210 of the sole. For this purpose, each region within the 3D cell 710 is projected onto the CAD model 210 of the sole by means of area mapping. For this purpose, within the 3D cell 710, local compensation planes are determined in both the CAD model 210 and the 3D scan 610 of the sole. These two compensation planes, determined separately for each 3D cell 710, are subsequently projected onto each other via their centroids. This projection results in a local transformation matrix for each 3D cell.

[0041] This local correction of the robot path allows it to react locally to deviations in the segment and locally correct the robot path. This results in a more precise adaptation of the 3D scan 610 of the sole and the CAD model 210, and thus in a more precise adaptation between the ideal robot path 240 defined with respect to the CAD model 210 and the individual characteristics of the real sole. This in turn leads to a more precise trimming of the flash area 640, thereby improving the quality of the trimmed sole. This allows for the automatic trimming of the flash area 640 in the sole, resulting in a reduction in working costs and an improvement in product quality.

[0042] After determining the local transformation, the path points of the ideal robot path 240 of the CAD model 210 are transformed back 160 onto the 3D scan 610 of the real sole. For this inverse transformation, the 3D grid area 710 is used again, such that the affected path points can be determined for each subsection. For each 3D grid 710, the path points are then transformed back to the calibration data of the 3D scanner. This provides a list of corrected path points 810 as Figure 8 shown, and thus a corrected robot path.

[0043] The transformed path points including the newly oriented calculations are stored. Then, this data can be sent 170 to the robot or the robot system. Then, based on the obtained data, the sole 930 held by the gripper can be guided 170 around the trimming tool 900 that is immovable relative to the robot, as Figure 9 shown. In this case, the sole is guided along the stop 940. This stop 940 forms a reaction piece against which the oscillating knife 920 runs, separating the flash. In this case, the pressing device 910 brings the flash into the correct position for this case, causing it to protrude upwards.

[0044] In another embodiment, the sole 930 can already be attached to a position fixed relative to the robot arm, and the robot arm guides the trimming tool 900 to process it around the sole 930.

Claims

1. A method (100) for processing parts of sports articles, in particular soles, by means of a robot, the method comprising: (a) creating (110) a first path (240) about a part to be processed, the first path comprising a list of points or a predetermined trajectory; (b) determining (120) a 3D scan (610) of the part to be processed by means of a 3D scanner; (c) calculating (130) a global offset between the first path (240) and the 3D scan (610) of the part to be processed; (d) calculating (150) a plurality of local offsets between the first path (240) and the 3D scan (610) of the part to be processed; (e) creating (160) a second path (810) based on the first path (240), the calculated global offset and the calculated local offset; as well as (f) transmitting (170) the second path (810) to the robot, and the robot performing processing on the component based on the second path (810).

2. The method according to claim 1, wherein: The first path (240) is determined based on a CAD model (210) of the component to be processed.

3. The method according to claim 1, wherein: The determination of the 3D scan is performed by an optical 3D scanner ( 120 ).

4. The method according to claim 1, wherein: The calculation (130) of the global offset is performed based on the features (620) of the part to be processed that can be detected by the 3D scanner.

5. The method according to claim 2, wherein: Using the CAD model (210) on which the first path (240) is based, the component to be processed is divided into a plurality of space segments (710), and calculation of local offsets is performed for the plurality of space segments of the component to be processed (150).

6. The method according to claim 5, wherein: The spatial segment (710) is defined by a cuboid.

7. The method according to claim 1, wherein: The first path (240) and the second path (810) have a plurality of spatial points.

8. The method according to claim 7, wherein: There is a bijective correspondence between points of the first path and points of the second path.

9. An apparatus configured to perform the method according to one of claims 1-8.

10. A computer program product comprising a computer program, the computer program comprising instructions which, when the computer program is executed on a computer, cause the computer to perform the method according to one of claims 1 to 8.

Images