Method for the automated polishing or grinding of a workpiece surface

Abstract

The invention relates to a method for the automated polishing or grinding of a workpiece surface, wherein the polishing or grinding is carried out by means of a machining device which comprises a robot device, an end effector mounted on the robot device, and a controller. The end effector comprises a passive tool, in particular a grindstone, for machining the workpiece surface. The method comprises identifying, on the basis of a geometry model of the workpiece surface, areal first sub-regions of the workpiece surface, the smoothed curvature of which at every point of the first sub-region is less than a predefined curvature limit value. The method also comprises determining a groove direction for the identified first sub-regions. Furthermore, the method comprises defining machining trajectories for the polishing or grinding of the identified first sub-regions of the workpiece surface, wherein, for the identified first sub-regions, machining trajectories for a cross-hatch machining process are defined which allow each of said first sub-regions to be machined in a cross-hatch pattern relative to the determined groove direction of said first sub-region.

Description

[0001] Process for automated polishing or grinding of a workpiece surface

[0002] Description

[0003] Field of the invention

[0004] The invention relates to methods for the automated polishing or grinding of a workpiece surface. Furthermore, the invention relates to a processing device comprising a robot device, an end effector attached to the robot device, and a controller.

[0005] Problem underlying the invention

[0006] It is an object of the invention to provide methods and a processing system for polishing or grinding a surface or a surface area of ​​a workpiece, which enable automated processing of the surface or the surface area.

[0007] Inventive solution

[0008] The stated object is achieved by a method for the automated polishing or grinding of a workpiece surface, wherein the polishing or grinding is carried out by means of a processing device comprising a robot device, an end effector attached to the robot device, and a controller. The end effector comprises a passive tool, in particular a grindstone, for processing the workpiece surface. The method comprises identifying, based on a geometric model of the workpiece surface, flat first partial regions of the workpiece surface whose smoothed curvature is less than a predetermined curvature limit value at every point of the first partial region. The method also comprises determining a scoring direction for the identified first partial regions.Furthermore, the method comprises defining machining trajectories for polishing or grinding the identified first partial areas of the workpiece surface, wherein machining trajectories for a cross-grinding operation are defined for the identified first partial areas, which machine the respective first partial area in a cross-grinding operation relative to the determined groove direction of the respective first partial area.

[0009] A smoothed curvature is defined as a curvature in which no local curvature fluctuations, such as those caused by local unevenness and grooves, occur. When the curvature is determined using a geometric model, such smoothed curvature values ​​are obtained.

[0010] According to the method according to the embodiments of the invention, the flat areas of the workpiece surface are first identified. These flat areas are characterized by a curvature characteristic, according to which the curvature at every point of the partial area is less than a predetermined curvature limit. The machining of these flat areas can then be carried out, for example, using a strategy adapted to the curvature characteristics, whereby machining using the cross-grinding method is advantageous for flat areas. For areas with different curvature characteristics, for example for transition, edge, and bead areas, a different machining strategy could be useful, as described below. The use of a cross-grinding method for machining the flat, extended areas has the particular advantage of uniform machining of the workpiece surface.Due to the uniform material removal achieved by the cross-grinding process, grinding of the workpiece is avoided. In particular, the cross-grinding process can be used to evenly remove existing grooves in the workpiece surface. For example, the cross-grinding process can achieve a high surface quality.

[0011] The invention also relates to a method for the automated polishing or grinding of a workpiece surface, wherein the polishing or grinding is carried out by means of a processing device comprising a robot device, an end effector attached to the robot device, and a controller. The end effector comprises a passive tool, in particular a grindstone, for processing the workpiece surface. The method comprises identifying, based on a geometric model of the workpiece surface, second subregions of the workpiece surface which, at each point of the second subregion, viewed in at least one direction of the workpiece surface, have a concave curvature that exceeds a predetermined curvature limit.The method also includes determining, for each identified second partial region of the workpiece surface, a directional path of the second partial region along the workpiece surface. Furthermore, the method includes guiding the tool along at least one dressing trajectory that has a dominant directional component in a direction transverse to the directional path of the respective second partial region.

[0012] According to embodiments of the present invention, for machining transition, edge, and bead regions that have a comparatively strong concave curvature, the tool is dressed prior to the actual polishing or grinding operation. For this purpose, for example, a number of dressing trajectories can be run through that have a dominant directional component in a direction transverse to the direction of travel of the respective second partial region. This can, for example, achieve an adaptation of the geometry of the tool to the curvature of the second partial region to be machined. In particular, the edges of the tool are rounded by dressing, so that the tool can increasingly adapt to the workpiece surface to be machined as a result of dressing. For example, this can prevent damage to the workpiece surface caused by the initially sharp edges of the tool.

[0013] The invention further relates to a machining device comprising a robot device, an end effector mounted on the robot device, and a controller. The end effector comprises a passive tool, in particular a grindstone, for machining the workpiece surface. The controller is designed to carry out one of the methods described above.

[0014] The invention relates to a method for the automated polishing or grinding of a workpiece surface, wherein the polishing or grinding is carried out by means of a processing device comprising a robot device, an end effector attached to the robot device, and a controller. The end effector comprises a passive tool, in particular a grindstone, for processing the workpiece surface. Quality control of the polishing or grinding process is carried out by means of a visual inspection system. The method comprises defining an area of ​​the workpiece surface to be processed and processing the surface area using processing trajectories adapted to the geometry of the workpiece surface in one or more grinding passes.The method also includes visually inspecting the surface using the visual inspection system and determining at least one quality parameter for the processing state of the surface area. Furthermore, the method includes comparing the at least one quality parameter with at least one predetermined target parameter and, if the at least one target parameter has not yet been achieved, continuing the polishing or grinding of the surface area.

[0015] Preferred embodiments of the invention

[0016] Advantageous training and further developments, which can be used individually or in combination with one another, are the subject of the dependent claims and the following description.

[0017] According to a preferred embodiment, a curvature limit value is specified which corresponds to a curvature radius limit value which lies in the range between 20 cm and 2 m, more preferably in the range between 30 cm and 1 m.

[0018] Preferably, the machining trajectories for the cross-grinding machining comprise first cross-grinding trajectories which run on the workpiece surface at a first angle relative to the scoring direction, wherein the first angle lies in the range between +20° and +70°, more preferably between +35° and +55°, more preferably between +40° and +50° relative to the scoring direction.

[0019] Further preferably, the machining trajectories for the cross-grinding process comprise second cross-grinding trajectories that extend on the workpiece surface at a second angle relative to the scoring direction, wherein the second angle lies in the range between -20° and -70°, more preferably between -35° and -55°, more preferably between -40° and -50° relative to the scoring direction. It is advantageous if the first angle selected for the first cross-grinding trajectories corresponds in magnitude to the second angle selected for the second cross-grinding trajectories, but has the opposite sign.

[0020] According to an advantageous embodiment, the polishing or grinding process comprises a predetermined sequence of first and second cross-grinding trajectories.

[0021] Preferably, the geometry model of the workpiece surface comprises CAD and / or CAM data.

[0022] According to an advantageous embodiment, the geometric model of the workpiece surface is created by capturing the workpiece surface using at least one camera.

[0023] Preferably, determining the groove direction for the identified first sub-areas comprises:

[0024] - capturing at least one image of the workpiece surface using at least one camera;

[0025] - Evaluating at least one captured image using image processing.

[0026] It is advantageous if the at least one camera comprises a mobile camera attached to the robot device.

[0027] Preferably, determining the groove direction for the identified first sub-areas comprises:

[0028] - Deriving the groove direction from the geometric model, especially from CAM data.

[0029] According to an advantageous embodiment, the method comprises identifying, based on the geometric model of the workpiece surface, second partial regions of the workpiece surface which, at each point of the second partial region, viewed in at least one direction of the workpiece surface, have a concave curvature which exceeds a predetermined curvature limit value.

[0030] Preferably, the second subregions comprise transition, edge, and bead regions. According to an advantageous embodiment, the method comprises determining, for the identified second subregions of the workpiece surface, a directional profile of the second subregion along the workpiece surface, and establishing machining trajectories for the identified second subregions depending on the identified directional profile.

[0031] Preferably, the processing trajectories comprise at least one of the following:

[0032] - machining trajectories which comprise machining with a dominant directional component in a direction transverse to the directional course of the respective second sub-area;

[0033] - Machining trajectories that include machining with a dominant directional component following the directional course of the respective second sub-area.

[0034] Preferably, the second sub-regions comprise transition, edge and bead regions.

[0035] According to an advantageous embodiment, the method comprises performing a predetermined number of grinding passes along at least one of the dressing trajectories.

[0036] Preferably, after a predetermined number of grinding passes along at least one of the dressing trajectories, a visual inspection of the tool is carried out using at least one camera.

[0037] According to an advantageous embodiment, performing a visual inspection of the tool comprises:

[0038] - capturing at least one image of the tool using at least one camera,

[0039] - Determine, based on the at least one image, whether a curvature of the edges of the tool is smaller than a specified curvature limit.

[0040] Preferably, in the event that the visual inspection shows that the tool is not yet sufficiently dressed, the grinding passes are continued along at least one of the dressing trajectories.

[0041] According to an advantageous embodiment, as soon as sufficient dressing of the tool is achieved, the method comprises guiding the tool along machining trajectories which are selected depending on the directional course of the second partial region.

[0042] Preferably, the processing trajectories comprise at least one of the following:

[0043] - machining trajectories which comprise machining with a dominant directional component in a direction transverse to the directional course of the respective second sub-area;

[0044] - Machining trajectories that include machining with a dominant directional component following the directional course of the respective second sub-area.

[0045] It is advantageous if the contact pressure of the tool on the workpiece surface in dressing trajectories is higher than the contact pressure of the tool on the workpiece surface in machining trajectories.

[0046] It is also advantageous if machining trajectories provide for a higher feed rate of the tool relative to the workpiece surface than dressing trajectories.

[0047] Preferably, the visual inspection system comprises at least one stationary and / or at least one mobile camera attached to the robot device.

[0048] According to an advantageous embodiment, the visual inspection system comprises image processing.

[0049] Preferably, a diagnostic model is used to evaluate the images and assess surface quality. Furthermore, the diagnostic model preferably includes a machine learning model.

[0050] Preferably, a forecasting model is used to determine the processing steps required to achieve the at least one specified target parameter. The forecasting model preferably comprises a machine learning model.

[0051] It is advantageous if the achieved processing quality of the surface areas is highlighted in color using a projector or augmented reality glasses. Brief description of the drawings

[0052] Further advantageous embodiments are described in more detail below with reference to several exemplary embodiments shown in the drawings, to which the invention is not limited, however.

[0053] It shows schematically:

[0054] Figure 1 shows a representation of a processing system for grinding or polishing;

[0055] Figure 2 shows the end effector with the tool held in it;

[0056] Figure 3 shows the optical highlighting of a surface area of ​​the workpiece by means of a projection device;

[0057] Figure 4 shows a workpiece to be machined;

[0058] Figure 5 shows different surface areas of the workpiece, which have different curvature characteristics;

[0059] Figure 6 shows the different surface areas of the workpiece together with the groove direction of the workpiece;

[0060] Figure 7A shows the course of the first trajectories of a cross-grinding operation in an oblique view of the workpiece;

[0061] Figure 7B shows the course of the first trajectories of a cross-grinding operation in a plan view of the workpiece;

[0062] Figure 8A shows the course of the second trajectories of a cross-grinding operation in an oblique view of the workpiece; Figure 8B shows the course of the second trajectories of a cross-grinding operation in a plan view of the workpiece;

[0063] Figure 9 shows a grindstone before passing through a number of dressing trajectories;

[0064] Figure 10 shows the grindstone of Figure 9 after passing through the dressing trajectories;

[0065] Figure 11 shows the use of the dressed grindstone for grinding a concave transition, edge and bead area.

[0066] Detailed description of embodiments of the invention

[0067] In the following description of preferred embodiments of the present invention, like reference numerals designate like or comparable components.

[0068] Fig. 1 shows a machining system 2 according to embodiments of the present invention, which is designed to perform a final machining operation on a workpiece 4, in particular a grinding or polishing operation of the workpiece 4. The machining of the workpiece 4 is carried out partially or fully automated. The workpiece 4 to be machined can be, for example, a press tool, a die-casting tool, or an injection-molding tool. Such workpieces are subject to high demands regarding the quality of the surface finish.

[0069] To carry out the machining, the workpiece 4 can be mounted, for example, on a frame 6. The machining of the workpiece 4 is carried out by means of a robot device 8. In order to be able to machine even large workpieces 4, for example pressing tools, the base 10 of the robot device 8 is slidably mounted on a linear guide 12. The linear guide 12 is mounted on a carrier unit 14. Solutions are conceivable in which the carrier unit 14 is provided with rollers in order to be able to move the carrier unit 14, together with the robot device 8 mounted thereon, within a production environment to the workpiece 4 to be machined. The robot device 8 shown as an example in Fig. 1 comprises five links 16a to 16e, which are connected to one another via joints, but the robot device 8 can also comprise more links, for example six or seven links.Attached to the last link 16e is a robot flange 18, to which an end effector 20 can be releasably attached, for example, by means of one or more screw connections. A tool 22 is held in the end effector 20, which is designed in particular for grinding or polishing the workpiece 4. The tool 22 can be, for example, a passive tool, such as a grindstone or whetstone, which is guided over the surface of the workpiece 4 in predeterminable trajectories by the robot device 8. Alternatively, the tool 22 can be an active tool, for example, a grinding or polishing device driven by a motor.

[0070] Fig. 2 shows an enlarged view of the last member 16e of the robot device 8, the robot flange 18, the end effector 20, and the tool 22. The end effector 20 comprises a tool holder 24 in which the respective tool 22, for example a grindstone, is held. The tool holder 24 can be designed, for example, in the form of a clamping device in which the tool 22 can be clamped. The tool holder 24 can be pivoted relative to the robot flange 18 about two axes of rotation 26, 28 shown in Fig. 2. The two axes of rotation 26, 28 are preferably arranged at right angles to one another, viewed in the normal direction to the workpiece surface. Since the tool holder is pivotably mounted, the angular position of the tool 22 is automatically adjusted when the tool 22 is pressed against the workpiece 4.When the tool 22 is guided over the tool 4 during grinding, the angular position of the tool 22 automatically adapts to the topology of the workpiece surface so that the tool 22 rests optimally on the workpiece 4. The end effector 20 can also comprise a translational adjustment unit. The translational adjustment unit can be designed, in particular, to move the tool 22 relative to the robot flange 18 in the normal direction to the workpiece surface. The translational adjustment unit can, for example, be designed to press the tool 22 against the workpiece surface with a predetermined contact pressure or predetermined contact force. The components of the end effector 20 are protected from grinding dust by a bellows 30.

[0071] A mobile camera 32 can also be attached to the robot flange 18 via a camera mount 31, which can be moved into different positions relative to the surface of the workpiece 4 by means of the robot device 8 in order to record images of the workpiece surface. With the help of the mobile camera 32, for example, an optical inspection of the surface of the workpiece 4 can be carried out in order to record the surface roughness of the workpiece surface or to determine, for example, a groove direction on the workpiece 4, which is predetermined by grooves that were created by a previous milling operation. The processing system 2 can comprise one or more light sources for illuminating the surface of the workpiece 4. Fig. 1 also shows a projection device 34, for example a projector, which can also be used to illuminate the surface.For optical inspection of the workpiece surface, it is particularly advantageous to illuminate the workpiece surface with grazing incidence and to capture the resulting light-dark pattern using the mobile camera 32. By evaluating the light-dark pattern, the surface roughness and the course of the grooves, for example, can then be determined. The mobile camera 32 is preferably designed as a stereo camera. Alternatively or in addition to the mobile camera 32, a lidar scanner can be attached to the camera mount 31, which is designed to determine a three-dimensional contour of the workpiece surface.

[0072] For optically capturing the workpiece surface, the processing system 2 can further comprise one or more stationary cameras 36. Preferably, a plurality of cameras 36 are provided, for example, to be able to generate an image of the workpiece surface free of obscurations from different camera images. The processing system 2 can further comprise a display unit 38, wherein, for example, a representation 40 of the workpiece surface can be shown on the display. The display unit 38 is preferably designed in the form of a touch display to facilitate user interaction with the processing system 2. For example, the user can select an area of ​​the workpiece surface to be processed by tapping the touch display.

[0073] In addition to the display unit 38, the projection device 34 can also be used for interaction with the user. By means of the projection device 34, information can be projected onto the surface of the workpiece 4, for example. For example, certain areas of the workpiece surface can be highlighted by projecting light, in particular colored light. Fig. 3 shows an example of how a specific surface area 42 of the workpiece 4 can be marked with the aid of the projection device 34 by projecting light, in particular colored light. The color of the projected light can, for example, indicate which processing quality has already been achieved in the respective surface area 42. Different levels of processing quality can, for example, be marked with different colors.Alternatively, the projected light can be used to show the user, for example, those areas of the workpiece surface that still need to be machined. Fig. 3 also shows how the optically highlighted surface area 42 of the workpiece surface can also be highlighted in parallel in the representation 40 of the workpiece surface shown on the display unit 38 by a suitable marking 44.

[0074] The machining system 2 also includes a magazine 46 for tool changes. The magazine 46 can, for example, contain a plurality of differently sized grinding stones, preferably with different grits. To perform a tool change, the end effector 20 is moved toward the magazine 46. There, the tool holder 24 deposits the previous tool and picks up a new tool from the magazine 46.

[0075] The machining system 2 shown in Fig. 1 further comprises a controller, for example, a programmable logic controller (PLC), implemented on a computer 48. A model of the workpiece 4, for example, a CAD (Computer-Aided Design) model or a CAM (Computer-Aided Manufacturing) model of the workpiece 4, which specifies the geometry of the workpiece 4, serves as the basis for controlling the machining system 2. Alternatively, the surface topology of the workpiece 4 can also be captured using optical methods, for example, using the mobile camera 32 and the stationary cameras 36, or using a lidar scanner.

[0076] In addition, a model of the tool 22 is available on the controller side, which specifies the geometry of the tool 22 currently in use. For example, the current degree of wear of the tool 22 can also be taken into account. For example, the end effector 20 with the tool 22 can be moved by the robot device 8 at regular or irregular intervals toward one of the stationary cameras 36 in order to record an image of the tool 22 and to determine the degree of wear of the tool through image analysis. This information can then be taken into account in the tool model provided by the controller.

[0077] The controller is designed to generate trajectories for grinding or polishing the workpiece 4 based on the model of the workpiece 4 and the tool model. The tool 22, in particular the grindstone, is then moved by the robot device 8 according to these trajectories with a suitable feed rate and suitable contact force or suitable contact pressure over the workpiece surface. The controller is also designed to generate, based on these trajectories, suitable control signals for controlling the robot device 8 and, if necessary, also control signals for controlling the linear guide 12 in order to guide the tool 22 in a suitable manner over the surface of the workpiece 4. Furthermore, the controller can, for example, specify the number of re-grinding operations required to achieve a desired surface quality.In addition, the control system initiates, for example, the necessary tool changes and generates control signals for controlling the robot device 8 and, if necessary, also control signals for the magazine 46.

[0078] The controller can further be designed to evaluate the images of the workpiece surface captured by the mobile camera 32 and / or the stationary cameras 36 by means of image processing, for example in order to determine the roughness, the waviness and the surface quality of the surface of the workpiece 4.

[0079] To evaluate the images provided by the cameras 32, 36, the controller can, for example, comprise a diagnostic model that is designed or trained to derive, for example, at least one of the following from the images captured by the cameras: a measure of roughness, a measure of waviness, a quality measure. The diagnostic model can preferably be a machine learning model that is designed or trained to derive at least one parameter for roughness, waviness, and quality from the captured images of the material surface. This machine learning model can, for example, be trained in advance using a training data set that includes a plurality of images of workpiece surfaces and associated parameters for characterizing at least one of roughness, waviness, and quality. The machine learning model can, in particular, comprise a convolutional neural network (CNN).

[0080] In addition, the control system can comprise a forecast model that can be configured or trained, for example, to create a forecast based on the current processing status of a surface area, as to which further processing is still required to achieve a predeterminable processing quality of this surface area. The current processing status can be specified to the forecast model, for example, by specifying key values ​​and parameters. The forecast model can, in particular, be configured or trained to estimate the number of over-grinding passes still required to achieve a specific surface quality. For example, the forecast model can be configured to create the forecast regarding the number of over-grinding passes still required based on at least one of the following:

[0081] - the last recorded surface condition of the surface area, which can be provided, for example, by the diagnostic model,

[0082] - the previous number of roundings,

[0083] - the feed of the tool,

[0084] - the contact force of the tool,

[0085] - the contact pressure of the tool,

[0086] - the tool data, in the case of a grindstone, for example, data on grain size, binder and bond strength, and

[0087] - Information on the surface category of the surface area, for example data on area, curvature, radii of curvature, convex or concave characteristics of the surface area.

[0088] Based on one or more of these key figures, the prediction model then creates a forecast of the future machining process, in particular the number of over-grinding operations still required to achieve a predefined machining quality. The diagnostic model can preferably be designed in the form of a machine learning model that is designed or trained to create a forecast of the future machining process based on the current machining state of the surface area and, in particular, to determine the number of over-grinding operations still required to achieve a predefined machining quality. The diagnostic model can be trained, for example, using training data sets, with each training data set relating a machining process to the change in roughness, waviness, or surface quality of the machined workpiece surface determined during machining.

[0089] The following discusses methods for defining machining trajectories for the workpiece and for performing grinding or polishing operations on the workpiece. The objective of these methods is to translate the human approach to finishing a workpiece, which demonstrably leads to very high surface quality, into automated machining of the workpiece surface using a tool. Both active tools, i.e., motor-driven tools, and passive tools, particularly grinding or honing stones, can be used. However, the focus is on the use of passive tools, particularly grinding stones.To perform the grinding operation, the passive tool is guided over the workpiece surface along machining trajectories under controlled contact pressure. The tool, held in the tool holder, continuously adapts to the potentially complex surface topology of the workpiece. The determination of suitable machining trajectories for this purpose will be discussed in more detail below.

[0090] As a starting point, Fig. 4 shows a workpiece 50 whose workpiece surface 52 is identified by grid lines. Starting from a geometric model of the surface topology, the workpiece surface 52 is divided into different surface areas with different curvature characteristics. The geometric model of the workpiece surface 52 can be provided, for example, in the form of CAD (computer-aided design) data or CAM (computer-aided manufacturing) data of the workpiece. Alternatively, it would be conceivable to generate a geometric model of the workpiece surface 52 using other methods, for example, using optical methods or using a lidar scanner.

[0091] First, the workpiece surface 52 is divided into surface areas with different curvature characteristics. For this purpose, for example, flat areas can be identified where the curvature at every point of the flat area is comparatively low and lies below a predetermined curvature limit. Furthermore, transition, edge, and bead areas can be identified on the workpiece surface 52, which have a curvature at every point of their surface that lies above the predetermined curvature limit in at least one direction.

[0092] Such transition, edge and bead regions generally extend along the workpiece surface 52 following a directional course. Such transition, edge and bead regions can further be subdivided according to whether the strong curvature detected is a convex curvature or a concave curvature.

[0093] Fig. 5 shows, by way of example, which surface regions with different curvature characteristics can be identified on the workpiece surface 52. The workpiece surface 52 comprises, in particular, a first flat region 54 with a slight curvature. Adjacent to the first flat region 54 is a first convex transition, edge, and bead region 56, and adjacent thereto is a first concave transition, edge, and bead region 58. Adjacent to the first concave transition, edge, and bead region 58 is a second flat region 60 with a slight curvature, which is adjoined by a second convex transition, edge, and bead region 62.

[0094] The two planar areas 54 and 60 are particularly characterized by the fact that at each point of the planar areas the smoothed curvature in each direction is smaller than a predetermined curvature limit value K G is.

[0095] A smoothed curvature is understood to mean a curvature value smoothed over a local surrounding area. The local surrounding area over which the smoothing is performed is selected to be large enough that local unevenness and, in particular, grooves caused, for example, by previous machining, disappear when determining the smoothed curvature value as a result of the smoothing and do not, or only minimally, influence the smoothed curvature value. On the other hand, the curvature profile of the workpiece surface 52 should be traceable based on the smoothed curvature values. If one starts from a CAD or CAM data set of the workpiece 50 when determining the curvature values, one automatically obtains such smoothed curvature values.

[0096] The curvature limit KG can, for example, be specified as a curvature radius limit RG, where KG = 1 / RG. For example, RG=400 mm could be selected as the curvature radius limit. If RG=400 mm is selected as the curvature radius limit, the smoothed curvature at every point in the flat areas is less than (1 / 400 mm).

[0097] In contrast, the convex transition, edge and bead regions, for example the convex transition, edge and bead regions 56 and 62, have at each point at least in one direction a smoothed curvature that is greater than the curvature limit value KG, this curvature being a convex curvature.

[0098] Also in the concave transition, edge and bead areas, for example in the concave transition, edge and bead area 58, the smoothed curvature is at least in one direction greater than the predetermined curvature limit value KG at each point of the concave transition, edge and bead area, this curvature being a concave curvature.

[0099] The next step involves defining suitable machining trajectories for grinding or polishing for the identified surface areas, depending on the determined curvature characteristics of the respective surface area. The goal is to define the machining trajectories so that the machining process largely corresponds to a grinding process performed by a human on the workpiece surface.

[0100] According to the exemplary embodiments of the present invention, machining using the cross-grinding method is provided for the flat regions of the workpiece surface 52. During cross-grinding, grinding or polishing is initially carried out with a grinding direction of +a to a preferred direction. The angle +a is preferably between 20° and 70° relative to the preferred direction, more preferably between 35° and 55°, more preferably between 40° and 50°, and more preferably approximately 45° relative to the preferred direction. After the first grinding pass, in which the machining direction is rotated by an angle +a to the preferred direction, a second machining process follows, in which the machining direction is rotated by an angle -a relative to the preferred direction.

[0101] Preferably, the scoring direction of the tool is used as the preferred direction. The scoring may, for example, originate from a previous machining operation. The cross-grinding process initially comprises a grinding pass with first trajectories oriented at an angle of +α relative to the scoring direction, followed by a grinding pass with second trajectories oriented at an angle of -α relative to the scoring direction. Using the cross-grinding process, the scoring can be gradually ground away in several grinding passes.

[0102] In contrast, for concave transition, edge, and bead areas, it is advisable to use trajectories that are determined depending on the directional course of the respective concave transition, edge, and bead area. The trajectories used can, for example, include trajectories that follow the directional course of the concave transition, edge, and bead area, and also, for example, include trajectories that run perpendicular to the directional course of the concave transition, edge, and bead area. For trajectories that follow the directional course of the concave transition, edge, and bead area, particular care should be taken to avoid grinding out the concave transition, edge, and bead area along the trajectories.For this purpose, it may be useful, for example, to additionally impose a component oriented transversely to the directional course on the trajectory following the directional course in order to obtain, for example, a zigzag course of the trajectory following the directional course.

[0103] Different machining strategies can be selected for grinding the convex transition, edge, and bead areas, for example, the convex transition, edge, and bead areas 56 and 62. A first machining strategy provides for grinding these convex transition, edge, and bead areas together with adjacent flat areas, whereby the cross-grinding process selected for the flat area can extend beyond the boundaries of the flat area, either completely or partially into the convex transition, edge, and bead area. In this way, flat areas and the adjacent convex transition, edge, and bead area can be ground together, completely or partially.

[0104] Alternatively, according to a second machining strategy, it may also be provided to grind the convex transition, edge, and bead regions separately. The trajectories for grinding the convex transition, edge, and bead regions are preferably determined depending on the directional course of the respective convex transition, edge, and bead region. The trajectories for grinding can, for example, comprise trajectories running transversely to the directional course or obliquely to the directional course of the respective convex transition, edge, and bead region. Furthermore, the trajectories for grinding can, for example, also comprise trajectories that follow the directional course of the respective convex transition, edge, and bead region.

[0105] The following discusses in more detail the cross-grinding of the flat regions 54 and 60 of the workpiece surface 52. As shown in Fig. 6, the workpiece 50 may, for example, have grooves extending along a groove direction 64 as a result of a previous machining operation, in particular a milling operation. It is advantageous if this groove direction 64 is used as the preferred direction in the cross-grinding process.

[0106] The scoring direction 64 can be determined, for example, by optical inspection of the workpiece surface 52, for example with the aid of the mobile camera 32 attached to the robot flange 18 of the robot device 8. When determining the scoring direction 64, it is advantageous to illuminate the workpiece surface 52 with light at grazing incidence. By means of image analysis, the scoring direction 64 can then be determined based on the images of the workpiece surface 52 recorded by the mobile camera 32. Alternatively, the scoring direction 64 can also be derived from the geometric model of the workpiece 50, for example from CAM (Computer-Aided Manufacturing) data. The scoring direction 64 can be derived from this CAM data. For example, the CAM data sets can also specify, among other things, the tool paths of the preceding machining operation, so that the scoring direction 64 can be derived from the CAM data.

[0107] The controller can then determine the trajectories for the cross-grinding of the flat areas 54 and 60. For the cross-grinding, the controller defines two different trajectories, namely first trajectories 66, which run at an angle of +α to the groove direction 64, and second trajectories 68, which run at an angle of -α relative to the groove direction 64.

[0108] The first trajectories 66 are shown in Fig. 7A and Fig. 7B. Fig. 7A shows the workpiece 50 in an oblique view, while Fig. 7B shows a plan view of the surface of the workpiece 50. The starting point for determining the first trajectories 66 is the geometric model of the workpiece surface 52, for example the CAD or CAM data of the workpiece surface 52. The first trajectories 66 are selected such that the tool 22, as shown in Fig. 7A and Fig. 7B, moves over the first flat region 54 and the second flat region 60 at an angle of +α to the groove direction 64. The generated trajectory consists of a sequence of surface points that are combined in spline movements. The trajectories are selected such that the robot flange 18 of the robot device 8 is always aligned as parallel as possible to the contact surface between the tool 22 and the workpiece 50.Any misalignment angles occurring in the longitudinal and transverse directions of the tool 22 are mechanically compensated by the two rotational degrees of freedom of the tool holder 24.

[0109] Fig. 8A and Fig. 8B show how the flat regions 54 and 60 are traversed by the tool 22 along the second trajectories 68, which are oriented at an angle of -α relative to the scoring direction 64. Fig. 8A shows the workpiece 50 in an oblique view, while Fig. 8B shows a plan view of the workpiece 50. In the context of a cross-grinding operation, for example, a first number of first trajectories 66 can be traversed first, followed by a second number of second trajectories 68.

[0110] In the embodiment shown in Figs. 7A, 7B, 8A, 8B, the first trajectories 66 and the second trajectories 68 are limited to the flat regions 54 and 60. Alternatively, however, it can also be provided to machine the convex transition, edge, and bead regions 56 and 62 adjacent to the flat regions 54 and 60, together with the flat regions 54, 60, in whole or in part in a common grinding pass. In this case, the first trajectories 66 and the second trajectories 68 would be positioned such that they extend beyond the flat regions 54, 60, completely or partially into the adjacent convex transition, edge, and bead regions 56, 62.

[0111] During the grinding process, grindstones progressively wear down. As a result of this wear, the shape of the grindstone increasingly adapts to the contour of the surface being machined. Because of this increasing adaptation of the grindstone's geometry to the surface being machined, in many cases, a qualitatively improved machining result can be achieved precisely through the wear of the grindstone. This is especially true when machining surface areas with a pronounced curvature, for example, when machining concave transitions, edges, and bead areas.

[0112] With the solution according to the invention, the wear of the grindstone can be systematically utilized to dress the grindstone before the actual grinding process. This is particularly advantageous when machining concave transition, edge, and bead areas. Here, there is a risk of the sharp edges of the grindstone damaging the highly curved surface.

[0113] As an example, Fig. 9 shows the grinding of the first concave transition, edge, and bead region 58 of the workpiece surface 52 of the workpiece 50. Also shown is the grindstone 70 used for the machining, which is clamped in the tool holder 24 of the end effector. It can be seen that the edges of the grindstone 70 are not yet worn and are therefore comparatively sharp-edged. If the grindstone 70 were to be used for grinding in the direction of the concave transition, edge, and bead region 58, there would be a risk that the workpiece surface 52 would be damaged by the sharp edges of the grindstone 70. Therefore, according to one embodiment of the invention, the grindstone 70 is specifically dressed before the actual grinding of the workpiece 50 is carried out in order to better adapt the contour of the grindstone 70 to the contour of the surface region to be machined.

[0114] Fig. 9 illustrates that the grindstone 70 passes through a predetermined number of dressing trajectories, which are illustrated in Fig. 9 by the double arrow 72. These dressing trajectories have a dominant movement component in the direction transverse to the extension direction of the concave transition, edge, and bead region 58. To accelerate the wear of the grindstone 70, the grindstone 70 is preferably pressed against the workpiece surface 52 with a comparatively high contact pressure, caused by a contact force 74, wherein the contact pressure is preferably greater in the dressing trajectories than in regular machining trajectories. In addition, the movement of the grindstone 70 along the dressing trajectories occurs at a reduced speed compared to regular machining trajectories. Fig. 10 shows the grindstone 70 after passing through a number of dressing trajectories.It can be seen that the edges of the grindstone 70 have significantly larger radii than at the beginning of the dressing process. In order to monitor the degree of wear and the edge radii of the grindstone 70, the end effector can be moved by means of the robot device 8, for example, in front of one of the stationary cameras 36, in order to capture an image of the grindstone 70 and evaluate it using image processing. For example, it can be checked whether the edge radii are already above a predetermined edge radius limit value due to wear. As an alternative to such a visual inspection of the grindstone, provided the wear behavior of the grindstone 70 is known, a number of dressing trajectories can be determined, even without a final visual inspection, which the grindstone 70 must follow before the actual grinding of the concave transition, edge, and bead area 58 begins.

[0115] As soon as the grindstone 70 is sufficiently adapted to the contour of the surface area to be machined, it can be used for regular grinding. After dressing, the grindstone 70 can be used for any machining trajectories, which can be oriented both along the directional course and transversely or obliquely to the directional course of the concave transition, edge, and bead area. As illustrated in Fig. 11 by the double arrow 76, the grindstone 70 can, for example, be moved along a machining trajectory following the directional course of the concave transition, edge, and bead area 58. During the movement along the machining trajectory, the grindstone 70 is preferably pressed against the workpiece surface 52 with less contact pressure, caused by the contact force 78, than during dressing.

[0116] When machining the workpiece surface 52, a predetermined number of grinding processes is first carried out. The quality of the surface is then assessed by visual inspection. For this purpose, one or more images of the workpiece surface 52 can be captured, for example, using the mobile camera 32 or at least one of the stationary cameras 36. Based on these images, a measure of the machining quality can be determined, for example using image processing. If this quality measure shows that the quality of the workpiece surface 52 is not yet sufficient, machining is continued. Once sufficient quality is achieved, machining is terminated. To evaluate the images of the workpiece surface 52 captured by the cameras 32, 36, the diagnostic model described above, which preferably comprises a machine learning model, can be used in particular.This diagnostic model is designed or trained to derive parameters from the captured images of the workpiece surface that characterize the surface quality, roughness, waviness, etc. The parameters provided by the diagnostic model can be used to decide whether sufficient machining quality has already been achieved or whether machining needs to be continued.

[0117] To determine the further processing of the workpiece surface 52 still required, the above-described prediction model, which preferably comprises a machine learning model, can also be used. The prediction model is designed or trained to create a prediction, based on the current processing status, of what further processing of the workpiece surface is still required to achieve a desired processing quality. In particular, the prediction model can, for example, determine the number of re-grinding steps still required based on the current processing status.

[0118] Information on the current machining status can preferably be provided to the prognostic model by the diagnostic model. The diagnostic model can, in particular, provide the prognostic model with parameters that characterize the surface condition, for example, parameters relating to the quality, waviness, and roughness of the workpiece surface. The prognostic model can then predict the further machining process based on this information on the status quo.

[0119] Through the close cooperation of two differently designed machine learning models, accurate planning of workpiece machining can be ensured. The forecast model, which is trained to estimate the future machining process based on a large number of already known machining processes, can also suggest optimized machining parameters for further machining, for example, adjusting the feed rate, initiating machining with a different grain size, varying the contact pressure, etc. This can reduce the machining effort and the required machining time. The features disclosed in the above description, claims, and drawings can be important for the implementation of the invention in its various forms, both individually and in any combination.

[0120] List of reference symbols

[0121] 2 processing system

[0122] 4 Workpiece

[0123] 6 frame

[0124] 8 Robot device

[0125] 10 feet of the robot device

[0126] 12 Linear guide

[0127] 14 carrier unit

[0128] 16a to 16e segments

[0129] 18 Robot flange

[0130] 20 End effector

[0131] 22 tools

[0132] 24 tool holders

[0133] 26, 28 rotation axes

[0134] 30 bellows

[0135] 31 Camera mount

[0136] 32 mobile cameras

[0137] 34 Projector device

[0138] 36 stationary cameras

[0139] 38 Display unit

[0140] 40 Representation

[0141] 42 Surface area

[0142] 44 Marking

[0143] 46 Magazine

[0144] 48 computers

[0145] 50 workpieces

[0146] 52 Workpiece surface

[0147] 54 first flat area

[0148] 56 first convex transition, edge or bead area

[0149] 58 first concave transition, edge or bead area

[0150] 60 second flat area

[0151] 62 second convex transition, edge or bead area

[0152] 64 Groove direction

[0153] 66 first trajectories

[0154] 68 second trajectories

[0155] 70 Whetstone

[0156] 72 Double arrow

[0157] 74 contact pressure

[0158] 76 Double arrow

[0159] 78 contact pressure

Claims

PATENT CLAIMS 1. A method for the automated polishing or grinding of a workpiece surface, wherein the polishing or grinding is carried out by means of a processing device comprising a robot device, an end effector attached to the robot device, and a controller, wherein the end effector comprises a passive tool, in particular a grindstone, for processing the workpiece surface, the method comprising the following steps: a) starting from a geometric model of the workpiece surface, identifying flat first partial regions of the workpiece surface whose smoothed curvature at each point of the first partial region is less than a predetermined curvature limit value, b) determining a groove direction for the identified first partial regions, c) defining processing trajectories for the polishing or grinding of the identified first partial regions of the workpiece surface,wherein machining trajectories for a cross-grinding operation are defined for the identified first sub-areas, which machine the respective first sub-area in cross-grinding relative to the determined groove direction of the respective first sub-area.

2. Method according to claim 1, characterized in that the machining trajectories for the cross-grinding machining comprise first cross-grinding trajectories which run on the workpiece surface at a first angle relative to the scoring direction, wherein the first angle is in the range between +20° and +70°, more preferably between +35° and +55°, more preferably between +40° and +50° relative to the scoring direction.

3. Method according to claim 1 or claim 2, characterized in that the machining trajectories for the cross-grinding machining comprise second cross-grinding trajectories which run on the workpiece surface at a second angle relative to the scoring direction, wherein the second angle is in the range between -20° and -70°, more preferably between -35° and -55°, more preferably between -40° and -50° relative to the scoring direction.

4. Method according to claim 3 with reference to claim 2, characterized in that the first angle selected for the first cross-cut trajectories corresponds to the first angle selected for the second The magnitude of the second angle selected for the cross-section trajectories corresponds to that but has the opposite sign.

5. Method according to one of claims 1 to 4, characterized in that determining the groove direction for the identified first partial areas comprises: - capturing at least one image of the workpiece surface using at least one camera; - Evaluating at least one captured image using image processing.

6. Method according to one of claims 1 to 4, characterized in that determining the groove direction for the identified first partial areas comprises: - Deriving the groove direction from the geometric model, especially from CAM data.

7. Method according to one of claims 1 to 6, characterized by the following step: - starting from the geometric model of the workpiece surface, identifying second partial regions of the workpiece surface which, at each point of the second partial region, viewed in at least one direction of the workpiece surface, have a concave curvature which exceeds a predetermined curvature limit value.

8. Method according to one of claims 1 to 7, characterized by the following step: - for the identified second partial areas of the workpiece surface, determining a directional course of the second partial area along the workpiece surface, and defining machining trajectories for the identified second partial areas depending on the identified directional course.

9. A method for the automated polishing or grinding of a workpiece surface, wherein the polishing or grinding is carried out by means of a processing device comprising a robot device, an end effector attached to the robot device and a controller, wherein the end effector comprises a passive tool, in particular a grindstone, for processing the workpiece surface, wherein the method comprises the following steps: a) starting from a geometric model of the workpiece surface, identifying second partial regions of the workpiece surface which, at each point of the second partial region, viewed in at least one direction of the workpiece surface, have a concave curvature that exceeds a predetermined curvature limit value, b) determining, for each identified second partial region of the workpiece surface, a directional profile of the second partial region along the workpiece surface, c) guiding the tool along at least one dressing trajectory that has a dominant directional component in the direction transverse to the directional profile of the respective second partial region.

10. Method according to claim 9, characterized by the following step: - Performing a predetermined number of grinding passes along at least one of the dressing trajectories.

11. Method according to claim 9 or claim 10, characterized in that after a predetermined number of grinding passes along at least one of the dressing trajectories, a visual inspection of the tool is carried out by means of at least one camera.

12. The method according to claim 11, characterized in that performing a visual inspection of the tool comprises: - capturing at least one image of the tool using at least one camera, - Determine, based on the at least one image, whether a curvature of the edges of the tool is smaller than a specified curvature limit.

13. Method according to one of claims 11 or 12, characterized in that, if the visual inspection shows that the tool is not yet sufficiently dressed, the grinding passes are continued along at least one of the dressing trajectories.

14. Method according to one of claims 9 to 13, characterized by the following step: - as soon as the tool has been sufficiently dressed, guiding the tool along machining trajectories which are selected depending on the direction of the second sub-area.

15. Process for the automated polishing or grinding of a workpiece surface, wherein the polishing or grinding processing is carried out by means of a processing device which comprises a robot device, an end effector attached to the robot device and a controller, wherein the end effector comprises a passive tool, in particular a grindstone, for processing the workpiece surface, wherein a quality control of the polishing or grinding processing is carried out by means of a visual inspection system, wherein the method comprises the following steps: a) defining an area of ​​the surface of the workpiece to be processed, b) processing the surface area by means of processing trajectories adapted to the geometry of the workpiece surface in one or more grinding passes, c) visually inspecting the surface by means of the visual inspection system and determining at least one quality parameter for the processing state of the surface area;d) comparing the at least one quality parameter with at least one predetermined target parameter and, if the at least one target parameter has not yet been achieved, continuing the polishing or grinding of the surface area; 16. A machining device comprising a robot device, an end effector attached to the robot device and a controller, wherein the end effector comprises a passive tool, in particular a grindstone, for machining the workpiece surface, wherein the controller is designed to carry out the method according to one of claims 1 to 8 or the method according to one of claims 9 to 14.

Images