METHOD FOR THE AUTOMATED MACHINING OF WORKPIECE SURFACES BY POLISHING, GRINDING OR PAINTING

Description

[0001] The invention relates to a method for the automated processing of workpiece surfaces, e.g., by polishing, grinding, or painting. The method can be carried out fully or at least partially automatically. The invention can be used in particular for the processing of small workpiece quantities. Processing parameters can also be individually adapted for changing surface types, particularly when processing different surface patterns, such as wood grain.

[0002] When painting or sanding wood, the path, force and other parameters usually have to be adjusted by hand until the desired result is achieved. This is particularly important with natural materials, as the processing parameters must be changed even within a workpiece in order to achieve the desired overall result. In many cases, however, the need to adjust processing parameters only becomes apparent during the process - due to the workpiece's reaction to the processing. However, defining the desired result in itself is a major challenge, as it is primarily based on empirical knowledge. Furthermore, the result cannot usually be objectively determined using a specific value, but is rather based on the overall impression based on experience.For this purpose, different sensory abilities of the human being are used to make an experience-based assessment, which is then usually translated into hand movements.

[0003] A similar approach is now being sought for the robot-based automation of sensitive components, even for very small batch sizes. (For higher batch sizes, the process makes sense when natural materials such as wood are involved, since the workpiece is macroscopically – geometrically, so to speak – identical (= series production), but differs in detail – for example, grain direction – (= single-piece production)).

[0004] Until now, sensitive components have been manufactured manually in small batches (especially single-batch quantities) or at least actively monitored by human operators. For example, a machine operator, especially with new or unknown components, has the option of adjusting the milling speed using a prominent rotary knob so that no chatter is audible.

[0005] Previous solutions for intelligent tool spindles focus on reducing vibrations and thus chatter marks through adapted speed control during milling processes. The workpiece is only monitored indirectly, and no definitive conclusions about the actual process results can be drawn, especially with sensitive materials and small batch sizes.

[0006] When deburring large quantities of metal or plastic, intelligent evaluation of integrated sensors can also be helpful. For example, integrated deflection and speed sensors can not only provide maintenance-relevant parameters, but also draw conclusions about fluctuations in component properties (especially size) during series production. Such fluctuations occur constantly in natural materials such as wood. Here, too, only indirect component monitoring is performed.

[0007] Force-controlled end effectors, which, for example, incorporate gravity in grinding operations, have also been developed to intelligently increase accuracy. However, they only control the force in an open loop with respect to the workpiece.

[0008] However, there are mainly solutions in the area of ​​predictive maintenance that rely on recording large amounts of data and machine learning. However, the primary goal here is to increase service life and reduce connection complexity by generating data directly in a spindle.

[0009] For larger batch sizes, processing parameters can now also be optimized automatically, for example, based on downstream quality assessments. However, this process is not effective for small batch sizes or even for batch sizes as small as 1.

[0010] Artificial intelligence can also be used to optimize the planning of the feed movement during machining, but this is currently not possible for small quantities due to the lack of sensory detection of the component in real time.

[0011] For example, DE 43 08 246 A1 discloses a method and a device for increasing the efficiency of processing machines.

[0012] WO 2020 / 128905 A1 relates to an automated surface treatment system.

[0013] DE 10 2019 200 482 A1 describes a method and a device for the automated machining of a workpiece with a machine tool.

[0014] The object of the invention is to provide possibilities for automated surface processing with small numbers of workpieces and / or changing surfaces, the processing results of which are at least close to or even superior to those of manual processing.

[0015] According to the invention, this object is achieved by a method having the features of claim 1. Advantageous embodiments and further developments of the invention can be realized with features defined in the dependent claims.

[0016] One component of the invention is an end effector, e.g. for polishing, grinding and painting tasks, especially of wood, which combines several disciplines - sensors, actuators, data and measurement signal processing - in order to achieve a uniform processing pattern on the surface to be processed compared to the respective neighboring regions.

[0017] In the method according to the invention, an end effector with at least one tool designed for the respective surface processing is moved relative to a respective surface to be processed. In this case, before, during, and after processing, in a processing area before, next to, and in the feed direction downstream of a processed surface area of ​​the respective processing, at least one sensor detects spatially resolved, preferably spatially and temporally resolved, measurement signals representing the currently achieved processing result and feeds them to an electronic evaluation and control unit located on the respective end effector.

[0018] The electronic evaluation and control unit carries out a target-actual value comparison to determine whether a specified work result has been achieved or not. The result of the target-actual value comparison is used to end the processing of the respective surface or to adjust the parameters with which the respective tool is operated accordingly and to repeat the respective surface processing in the respective surface area with changed parameters or to adjust these parameters during a further feed movement.

[0019] Particularly during polishing or grinding, it is advantageous to acquire the measurement signal in a surface area surrounding the surface area currently being machined. This surface area can correspond in its geometric shape to the surface area currently being machined and can be, for example, circular or rectangular.

[0020] During painting, it is advantageous to record measurement signals directly in the processing area, either alone or in addition, so that it can be determined whether and how paint is currently being applied to the surface. Surface areas located outside of this area should also be monitored additionally, in particular to take into account the saturation of the surface substrate with the respective paint.

[0021] Sensors can be used that are selected from a force sensor, a torque sensor, a sensor for determining the instantaneous drive power of a tool, a sensor for determining the rotational speed of a tool spindle, a surface roughness sensor, an optical sensor, preferably an imaging optical sensor system, a flow rate sensor, a temperature sensor, a distance sensor, a gloss sensor, and a color sensor. Preferably, at least two different sensors should be used.

[0022] Depending on the measured signals recorded, a force, the rotational speed, the feed rate of the end effector, the distance to the surface to be machined, the volume flow of paint and / or an air flow with which a veil is to be formed around the currently painted processing area to protect sensors and to dry the paint can be controlled. Control cannot necessarily take place in the interpolation cycle of a robot. It can also take place more slowly and thus only become effective over several interpolation cycles. This can be useful in order to avoid sudden changes in processing parameters and, for example, to give the sensors in painting processes more time to observe the applied and slowly curing paint.

[0023] The measurement signals detected by the respective sensor(s) should be recorded with spatial resolution, with the recorded measurement signals being assigned to the respective position coordinates and taken into account during evaluation and control. It may be sufficient to consider the position coordinates two-dimensionally. For curved surfaces to be machined, three-dimensional recording and consideration may also be useful.

[0024] Particularly for surfaces with a pattern, such as grain, wavelength-resolved detection with an optical sensor can be performed alone or in addition. This can involve spectral analysis of reflected and / or scattered electromagnetic radiation. Image processing can also be used to detect and evaluate such surfaces.

[0025] Recorded measurement signals representing specified surface and processing types can be advantageously grouped and stored electronically. These groups can be used to perform a target-actual value comparison, which can then be used to control the end effector.

[0026] It is also possible to perform the target-actual value comparison using data determined by machining experts, which is preferably also grouped together. Manual machining can first be performed on an analog interface with the control of an end effector, and the corresponding measurement signals can be assigned and saved to the respective machining positions and operating parameters of the respective end effector.

[0027] However, an expert's evaluation of a machining result can also be used to assess the respective machining state achieved with known operating parameters of the end effector. For this purpose, at least one data set can be created and used for a target-actual value comparison. Multiple data sets can be assigned to at least one group.

[0028] Measurement signals acquired by at least one sensor can be fed into a learning process, creating a knowledge base that can be used for subsequent target-actual value comparisons and for controlling the respective end effector. Manually obtained measurement signals and associated control parameters of the end effector, as well as measurement signals with associated control parameters obtained during automated processing, can also be used for this purpose in order to achieve a self-learning system or to gradually improve a system.

[0029] An end effector can be fixedly positioned in space or by a manipulator, and a robot, as an example of a manipulator, can move a respective workpiece, whereby the robot also transmits the position coordinates of the respective machined surface position to the electronic evaluation and control unit, or at least communication can take place between the end effector and the electronic evaluation and control unit, which makes a decision as to whether the machining is completed at a specific position on the workpiece surface or whether further machining is required in the corresponding surface area.

[0030] Through different data structuring and analyses, underlying relationships between processing configurations and results can be captured and functionally implemented so that processing parameters can be extrapolated for unknown sensor measurement signal constellations.

[0031] This can be done by recording a new data set with spatially resolved measurement signals, particularly before the start of processing or before the start of further processing, at a position or in a specific surface area, which represent the processing state specified at that time or achieved up to that point or an end result achieved after processing (e.g. after drying during painting), and subjecting the results determined in this way to an evaluation, e.g. an expert evaluation. This can be used to determine whether the processing result achieved at this position or in this surface area has achieved a desired result or not. Once the processing result has been achieved, processing at this position or in this surface area can be ended.

[0032] If the assessment is negative, processing may continue with modified processing parameters, preferably adapted to the achieved processing result. For this purpose, mathematical processes and tools, e.g. neural networks, can be used to take into account existing, previously determined data sets from a database that can be continually expanded over time with newly acquired data sets. For this purpose, processing parameters can be taken into account taking into account measurement signals recorded after further or previously carried out similar processing steps, which represent the achieved processing results. Measured values ​​can be, for example, the achieved surface feel, a two- or three-dimensional surface structure, an overall optical appearance (e.g. a wood pattern) and / or a gloss.

[0033] In particular, data sets that have been recorded on similar but differently geometrically designed surfaces and stored in a database can be taken into account.

[0034] For example, data sets can be classified to form clusters (groups) that can be used to reduce data and / or identify commonalities, for example, when obtaining machining results while taking into account the machining parameters adhered to. This allows for the use of experience gained in the past from identical or similar machining processes with correspondingly adhered sensor values ​​and final machining results.

[0035] The sensors can take into account not only the end effector (e.g. flow rates, speeds, forces, vibrations), as is already partially possible in the state of the art, but also the workpiece itself. The combined use of different sensors can be considered. For example, in addition to optical sensors, tactile sensors suitable for surface analysis can also be used. These sensors evaluate the surface properties of surface areas that have already been machined at least once in order to be able to draw conclusions about a machining result that has already been achieved or is still to be achieved. In this way, meaningful data can be obtained through sensor data fusion. The control loop around the workpiece can be expanded so that, thanks to integrated data processing, the actuators of the respective end effector can react directly to the current machining result on the workpiece and influence further machining.This can be a repetition of the machining of a surface area that has already been machined, as well as a subsequent machining of surface areas that can be machined as a result of a feed movement with the end effector.

[0036] It is important to explicitly record measurement signals in the immediate vicinity of a position on the workpiece surface to be machined and to respond to these conditions by adjusting the operating parameters of the tool or end effector accordingly. Extended measurement signal recording can also be performed at a later time, for example. This can occur during a previous processing step, such as polishing, which follows grinding or multiple grinding or polishing steps.

[0037] When sanding wood, this allows a careful approximation of the desired final geometry and / or surface roughness. The force required for sanding, with which the tool acts on the surface to be machined, can be gradually increased based on the workpiece's response to the action of the end effector.

[0038] When varnishing wood, for example, the absorbency of the wood can be taken into account and further layers of varnish can only be applied if necessary in certain surface areas.

[0039] This approach allows for the imitation of human working methods in a comparatively simple way. An experienced machine operator would also feel locally, but would also repeatedly scan the component as a whole to gain an impression of the overall appearance. To achieve a uniform appearance, the measurement signals recorded in surface areas adjacent to a surface area currently being machined, for example, in front of, next to, or in a radius around a surface area of ​​a workpiece that is currently being machined, can be included in the calculation of the subsequent machining parameters for the respective surface machining tool. Different algorithms can be considered for this, depending on the application and machining process.This can range from the relatively simple "nearest-neighbor interpolation" to complex solutions that allow for automated pattern continuation. However, a simple interpolation of the measurement signals acquired adjacent to the current machining surface can primarily ensure a uniform appearance of the respective machined surface. Such a simple version can be realized by recording the measurement signals and corresponding machining positions in a matrix in a local memory.

[0040] The invention should rely on local sensors instead of a camera located further away for two reasons. Firstly, this makes it much easier to set up a machining cell with sensors, as the end user only needs to attach a new end effector to a robot or machining center and doesn't have to worry about anything else. Secondly, lighting is particularly important for optical sensors. A controlled environment can be guaranteed much more easily locally than from a distance, where, for example, there may be many more scattered light sources, etc. The invention deliberately differs from the approach taken by experienced employees who, when machining particularly complex objects, sometimes take a step back to view the result from a distance.

[0041] The respective machining position can be provided by a robot or machine control system. However, it can also be calculated from the measurement signals of integrated acceleration sensors and gyroscopes, which allow the respective relative position of the tool to be determined in a simple and, above all, independent manner compared to the previous measurement positions.

[0042] On the one hand, the integrated logic can control the machining parameters managed directly by the tool spindle. This can affect, for example, flow rates, rotational speeds, or even integrated actuators for force control. However, information can also be transmitted to the respective robot or machine control system in order to adapt the respective feed direction of the tool or to re-traverse previously machined surface areas in order to achieve a desired machining result. For the end user, however, the entire process should be implementable with minimal programming effort. In most cases, this should be possible without any additional programming other than the initial path planning of the feed movement, which can also be created automatically based on a CAD model or a component scan. The technical design of the necessary control system can be adopted from the state of the art.

[0043] Low technical integration hurdles can be achieved by using connections that are as standardized as possible (e.g., a power connection and a ProfiNet data connection). This keeps integration costs into an existing machining cell low, eliminating the need for significant modifications to the machining cell.

[0044] An end effector can be mounted either on the robot, on a machine tool, or fixed in the room. With a fixed installation in the room, for example, a robot would handle the component in front of the end effector (and transmit the respective position on the workpiece to the logic unit via data). Otherwise, the functionality would be identical to the other two possible arrangements.

[0045] Four variants are conceivable, each with an increased level of difficulty: 1.) Active tool spindle (easily integrated into almost all robot systems or processing machines or centers): The tool spindle operates almost independently of the robot or system control system. Only the "on / off" command should come from the system control system; the respective speed and / or force can be controlled within the tool spindle. Such a tool spindle could be used, for example, for sanding wood or for polishing or deburring. For such applications, a programmed robot path should traverse the entire workpiece surface to be machined at least once with the appropriate feed motion in order to machine every position on the respective workpiece surface. Wood, as a natural product, does not respond evenly to the tool across a surface to be machined.In the simpler version, the active tool spindle can take over complete speed control, while in the more complex version, integrated force control can also be enabled by having the tool spindle with an additional axis and force sensors. While the surface of the workpiece to be machined is traversed by the tool spindle and the respective tool, the integrated sensors (e.g. a camera system) can record the local condition before and / or after machining of a surface area of ​​a workpiece. Based on, for example, learned empirical knowledge, specified machining guidelines and data collected in advance during the use of the active tool spindle can independently decide at which speed and force the workpiece should be machined in the respective area. 2.) Active painting unit (easily integrated into almost all robot systems): This application is the one described under 1.) explained it in a very similar way, with the difference that the end effector can actively control the amount of paint to be applied and other parameters of a spray unit to influence the volume flow of paint to be applied to the respective surface area and the distribution of the individual paint droplets. The robot path along which the appropriately designed end effector is to be moved should also travel over the entire surface of the workpiece to be treated at least once. The spray unit can control the painting parameters locally and autonomously based on measurement signals and empirical knowledge (or other predetermined values). This way, for example, a uniform primer can be achieved on wood with locally varying absorption properties. 3.) Inclusion of the robot controller without path changes (requires integration into the robot controller). With this variant, in addition to the capabilities described above, the tool spindle can also execute pre-programmed elements from the robot control program. This means, for example, that the feed rate of the respective tool stored as a parameter in the robot program can be adapted to the requirements of the workpiece, or actions already provided for in the robot program, such as multiple or selected traversal of paths, can be executed. For this purpose, the tool spindle can control the corresponding signals on the robot side for moving the robot with the end effector in relation to the surface to be machined, or the surface to be machined in relation to the end effector. 4.) Incorporation of the robot control system, including path generation (requires very deep integration into the robot control system). In the most complex version, the tool spindle of a tool defines the robot actions independently and in near real time. For example, in this version, a completely new robot path, adapted to the requirements of the workpiece, could be suggested and executed on the spindle side. This would require very deep intervention in the robot or machine control system.

[0046] The following explanations will be given for influencing an end effector during grinding or polishing. In a force-controlled grinding process perpendicular to the surface (using the example of a simply rotating spindle), the transverse forces and torque, which can be determined using appropriately designed sensors, can be used to determine the current machining status directly beneath the tool. This can take advantage of the fact that a rough surface causes greater tangential forces on the grinding tool than a smooth surface. In addition, the roughness around the actual tool can be measured directly on the surface using at least one appropriately designed sensor. Optical sensors can also be used for this purpose. The sensor data can be stored together in an electronic memory, preferably sorted into predefined groups, and compared with the decisions of experienced employees (expert knowledge) who use their usual instruments or systems to make the decision.Measuring devices can be used (usually just hands for feeling and eyes). From this comparison, a decision can be derived based on a variety of information. A simple shear force sensor, such as the one from , can be used as a sensor outside the spindle. https: / / connect.nissha.com / filmdevice / en / shear-force-sensor / is known, which is mounted in a ring around the actual grinding tool and can rotate around the spindle axis to continuously collect information in front of, behind, and beside the tool in the area around the currently machined surface area. Alternatively, the surface roughness could also be measured directly, as is the case, for example, with https: / / www.sciencedirect.com / science / article / pii / S09244247163119 92 is known.This allows for effective comparison of measurement signals before and after machining in a machined surface area. Imaging sensors can also be used to monitor the area surrounding the surface area currently being machined. Using suitable software, the recorded measurement signals can be evaluated, for example using image processing, and taken into account for controlling operating parameters. Imaging sensors can be arranged in a ring around the actual machining spindle. For polishing tasks, at least one sensor for gloss measurement can be present. This sensor uses optical sensors to detect electromagnetic radiation reflected and / or scattered by the machined surface, preferably with spatial and / or wavelength resolution, and the respective recorded intensity can be taken into account. Painting

[0047] Sensors for color and gloss can be used in the painting process. These sensors, at least in a similar way to the one previously explained, register reflected and / or scattered electromagnetic radiation in the work area and can be used to evaluate and control operating parameters. Measurements are preferably performed while the paint is wet, although the actual evaluation should only take place when the applied paint is dry. This approach allows for the determination and exploitation of a correlation between the measured values ​​recorded by sensors while wet and the quality achieved while dry. Here, too, the sensors should be mounted directly on the spray end effector. Ideally, the sensors should be shielded from paint particles, for example, by means of a sealing air or a protective film that can be placed in front of at least one sensor.

[0048] Sensor measurement signals can be recorded before, during, and after machining a surface area (hereinafter referred to as sensor measurement signals). Measurement signals can be recorded during machining, provided the accessibility of the sensors allows it. Typically, only forces can be measured directly beneath the machining spindle. In the case of painting, accessibility is not compromised, but increased overspray occurs, from which an air curtain can protect the sensors.

[0049] Additionally, the respective processing parameters can be recorded. The sensor measurement signals can be recorded and considered together with their relative position coordinates, preferably at least two-dimensionally.

[0050] The objective should be to establish a connection between machining parameters, sensor measurement signals, and machining results, so that in the future, the parameters for influencing the respective tool during the currently performed surface machining can be adjusted based on sensor measurement signals. The results can be assessed by experienced employees during a training phase. Sensor measurement signals should be saved with a position and time stamp (spatially and temporally resolved), including information on the currently used machining parameters (e.g., rotation speed). This way, the respective tool or an end effector can know what faster means in numbers when deciding to "rotate faster" in a subsequent machining step, thus enabling the machining result to be achieved more quickly and effectively.

[0051] Approaches for the algorithmic evaluation of sensor data: Data structuring and analysis: In this phase, data sets are to be broken down into smaller, individual surface areas depending on the specific three-dimensional geometry of the component's surface to be machined. These areas can then be transferred to other workpieces with the same or similar geometry and geodesy. This includes, among other things, particularly conspicuous geometries (e.g., a distinction between edges, straight surfaces, concave and convex surface areas) that can be detected by sensors in order to separately include them in the further calculation of the tool control variables (machining parameters). It can be taken into account that different parameters must be used on edges than on flat surfaces, for example, to avoid inadvertently grinding away an edge. 1. For implementation, a classification of data sets of expert assessments with regard to the type of geometry, the quality of the results and / or post-processing steps, differentiated according to different parameter types, can establish basic structures. Using clustering, groups of processing parameters and sensor measurement signals can be extracted from the data, whereby the following objectives are decisive: For implementation, a classification of data sets of expert assessments with regard to the type of geometry, the quality of the results and / or post-processing steps, differentiated according to different parameter types, basic structures can be established; Using clustering, groups of processing parameters and sensor measurement signals can be extracted from the data, whereby the following objectives are decisive: Identification of similar data sets: ∘ Identification of similar data sets in both areas;∘ Identifying factors that characterize similar data sets; ∘ Overlaying groups (clusters), for example to check for a group of sensor measurement signals that have common properties for an associated processing (cluster analysis on this subset); ∘ and from this extraction of processing parameters and sensor measurement signal combinations that can be related. In addition, data compression methods such as principal component analysis may be used to focus on relevant data sets. 2. Establishing the functional relationship between processing parameters and sensor measurement signals. Classifying new data sets of measurement signals assigned to positions into the determined groups;Checking whether the groups can also be used sensibly for new data sets Testing different options to extrapolate the assignment of sensor measurement signals-processing parameter pairs sensibly, e.g.: ∘ Regression ∘ Neural networks ∘ Reinforcement learning ∘ ... Using the groups for extrapolation, ie training individual configurations / instances of an artificial intelligence (AI) for the different groups, thereby exploiting the lower variability;

[0052] The invention enables the autonomous processing of sensitive workpiece surfaces, such as wood, in very small batches while maintaining a desired overall impression of the processing result. The greatest advantage is the automatability of previously handcrafted unique pieces, where the primary goal is a uniform, "attractive" overall appearance rather than precise dimensions. For this purpose, the intelligent end effector used in the invention can be coupled to a variety of existing robots or machines without requiring costly modifications or complex programming. Furthermore, no additional external sensors are required to use the end effector.

[0053] The invention relates to an integrated solution that can be used in a variety of scenarios without great effort. For this purpose, it can primarily be assumed that the at least one sensor, the actuators for performing the respective processing, including the respective tool, and an electronic evaluation and control unit can be integrated into an end effector used in the invention or implemented as a complete system, so that only the usual connections (e.g., power and data) are required. Improved robot control and external sensor technology can be dispensed with (e.g., by integrating specific optical sensors including evaluation logic into a robot or similar).

[0054] Potential applications include all current or future robot-assisted processes, particularly those requiring a combination of sensitivity and minimal programming effort while maintaining the overall appearance of the machined workpiece surface. These include manual processes, such as sanding, polishing, or painting wood or other sensitive materials, especially in small batches. Developed primarily for robots, the solution for sensitive and intelligent tool spindles can also be used in machine tools or fixed in space to move a workpiece surface to be machined in front of the tool spindle, for example, using a robot.

Claims

1. A method for the automated processing of workpiece surfaces by, for example, polishing, grinding or painting, in which at least one tool with an end effector, which is designed for the respective surface processing, is moved relative to a respective surface to be processed, wherein before, during and after processing in a processing region before, next to and after carrying out the respective processing in the direction of feed movement of the end effector, spatially resolved measurement signals, which represent the processing result currently achieved, are recorded and fed to an electronic evaluation and control unit using at least one sensor, which electronic evaluation and control unit is present on the respective end effector, wherein the electronic evaluation and control unit carries out a target / actual value comparison to determine whether a predetermined work result has been achieved or not, wherein the result of the target / actual value comparison causes the processing of the respective surface to terminate or parameters with which the respective tool is operated to be adjusted accordingly and the respective surface processing is repeated in the respective surface region with changed parameters or these parameters are adjusted during a further feed movement.

2. The method according to claim 1, characterized in measurement signals which represent the currently achieved processing result are recorded in a spatially and time-resolved manner using at least one sensor and are fed to the electronic evaluation and control unit which is present on the respective end effector.

3. The method according to one of the preceding claims, characterized in that at least one sensor is used which is selected from a force sensor, a torque sensor, a sensor for determining the instantaneous drive power of a tool, a sensor for determining the rotational speed of a tool spindle, a surface roughness sensor, an optical sensor, preferably an imaging optical sensor system, a flow rate sensor, a temperature sensor, a distance sensor and a gloss level sensor.

4. The method according to one of the preceding claims, characterized in that the measurement signals detected with the respective sensor(s) are recorded with spatial and / or wavelength resolution, the recorded measurement signals being assigned to respective position coordinates and taken into account in the evaluation and control.

5. The method according to one of the preceding claims, characterized in that detected measurement signals representing predetermined surface and processing types are combined in groups and fed to an electronic memory, with which the target / actual value comparison is carried out.

6. The method according to one of the preceding claims, characterized in that the target / actual value comparison is carried out with data which have been determined by means of processing experts and which are preferably also combined in groups.

7. The method according to one of the preceding claims, characterized in that measurement signals recorded with the at least one sensor are fed to a teach-in process, with which a knowledge base is created, which is used for subsequent target / actual value comparisons and the control of the respective end effector.

8. The method according to one of the preceding claims, characterized in that at least two different sensors are used simultaneously.

9. The method according to one of the preceding claims, characterized in that the end effector is fixedly positioned in space and a robot moves a respective workpiece, wherein the robot also transmits the respective position coordinates of the respective processed surface position to the electronic evaluation and control unit, or at least communication takes place between the end effector and the electronic evaluation and control unit takes place, which makes a decision as to whether processing is to be completed at a specific position on the workpiece surface or further processing is to take place in the respective surface region.

10. The method according to one of the preceding claims, characterized in that background correlations of the processing configurations and results are detected by various types of data structuring and analyses and these are functionally implemented so that processing parameters can be extrapolated for unknown sensor data constellations.