Computer-implemented method for an optical metrology system with optical tool measuring devices, and optical metrology system

Abstract

The invention relates to a computer-implemented method using an optical metrology system, which comprises an optical tool setting and measuring device, and using a computer system, comprising at least the following method steps: receiving and / or retrieving a digital data sheet of a tool, which includes a technical drawing of the tool and a dimensional specification of the tool, by the computer system, reading out the technical drawing and the dimensional specification from the digital data sheet at least by means of an analysis and evaluation module in order to extract at least some of the data and / or information which is contained in the digital data sheet and describes the tool, said module being part of the computer system and comprising at least one trained algorithm of machine learning, creating at least one measurement program for the optical tool setting and measuring device on the basis of measurement features, which were determined on the basis of the readout, by at least one trained algorithm of machine learning of the analysis and evaluation module and / or selecting a measurement program for the optical tool setting and measuring device from a list of measurement programs stored in the computer system, on the basis of the determined measurement features, by means of at least one trained algorithm of machine learning of the analysis and evaluation module.

Description

[0001] October 2, 2025

[0002] Computer-implemented method for an optical metrology system with optical tool measuring devices and optical metrology system

[0003] State of the art

[0004] The invention relates to a method according to claim 1, a computer system according to claim 16 and an optical measurement system according to claim 17.

[0005] Tools come in countless different shapes and sizes. Measurement systems for measuring and / or adjusting these tools, for example, before use in a machine tool, therefore also offer a very large number of measurement options and programs for the various tools. In some cases, several different measurement options / programs exist for the same tool, depending on its subsequent use. Selecting the optimal and / or appropriate measurement procedure for each tool's subsequent use is therefore a complex task requiring advanced knowledge of the respective tools, their specific applications, and the individual measurement system itself. This knowledge is currently imparted to operators through training courses or individual consultations, for example, by the manufacturer or supplier of the measurement systems.Due to the sheer number of possibilities, training conducted within a reasonable timeframe can only ever cover a portion of everything. Furthermore, consultations are always tailored to a specific tool, tool group, or measurement system, meaning that any change to one of these parameters (ZO 17881 WO) necessitates a new consultation. Simple programming along the lines of "use measurement sequence Y for tool X" is rigid and quickly reaches its limits given the rapidly evolving landscape of tools and measuring instruments.

[0006] The object of the invention is, in particular, to provide a generic method with advantageous properties with regard to user comfort and / or efficiency of use. This object is achieved according to the invention by the features of the independent and dependent claims, while advantageous embodiments and further developments of the invention can be found in the dependent claims.

[0007] Advantages of the invention

[0008] A computer-implemented method, in particular a computer-implemented tool measurement method, is proposed, comprising an optical measurement system which includes one or more optical tool measuring devices and / or optical tool component and measuring devices, and a computer system arranged internally and / or externally to the optical measurement system, preferably connected to the optical measurement system at least by a data communication link, comprising at least the following method steps: a) Receiving and / or retrieving a digital data sheet of a tool, for example a shank tool or complete tool, which includes one or more technical drawings of the tool and one or more dimensional specifications of the tool, for example a scale, dimension, or dimension of the tool, by the computer system; b) Reading the technical drawing or several of the technical drawings and the dimensional specification ora) extracting at least some of the data and / or information contained in the digital datasheet and describing the tool, using at least one machine learning algorithm trained specifically for this task, from several of the dimensional specifications from the digital datasheet; c) optionally determining possible and / or required measurement characteristics for the tool described in the digital datasheet by analyzing the data and / or information obtained from the readout using at least one machine learning algorithm trained specifically for this task; d) creating at least one measurement program for at least one of the optical tool measuring devices and / or the optical tool part and measuring devices of the optical metrology system based on measurement characteristics determined from the readout by at least one machine learning algorithm trained specifically for this task.a machine learning algorithm (specifically trained for this task) of the analysis and evaluation module and / or e) selection of a measurement program for at least one of the optical tool measuring devices and / or the optical tool part and measuring devices of the optical metrology system from a list of measurement programs stored in the computer system based on the determined measurement characteristics by at least one machine learning algorithm (specifically trained for this task) of the analysis and evaluation module. This can advantageously simplify and / or improve the finding and setting of suitable measurement programs. Advantageously, the user-friendliness of the metrology systems can be improved. Advantageously, the efficiency of the metrology systems can be improved. In particular, the effort required for tool measurement, especially of new or previously unknown tools, can be advantageously reduced. In particular, aPersonnel costs and / or workload for personnel can be reduced. In particular, downtime of measurement systems and / or connected machine tools can be advantageously reduced. Furthermore, measurement result quality can be advantageously improved, especially by ensuring that the optimal measurement program can always be run, regardless of the operator's experience. The degree of automation of the measurement system can also be advantageously increased. ZO 17881 WO

[0009] The optical metrology system can comprise a single optical tool measuring device or multiple such devices. If several such devices are present, they can be configured differently from one another, e.g., for different measuring tasks, different measuring ranges, with different measuring methods, for different accuracies, etc. When creating or selecting the measurement programs, the optimal device(s) from among several devices of the optical metrology system is preferably chosen. A "tool measuring device" is understood to mean, in particular, a device that is designed to at least partially detect and / or adjust at least one length, at least one angle, at least one contour, and / or at least one external shape of a tool.The optical metrology system may include known tool measuring devices and / or optical tool part-measuring and measuring devices, such as those distributed by E. ZOLLER GmbH & Co. KG Setting and Measuring Devices (Pleidelsheim, Germany, www.zoller.info) at the time of preparation and submission of this document. The optical metrology system is preferably based on camera-based and / or image analysis-based optical measurement methods. At least one of the optical tool measuring devices and / or the optical tool part-measuring and measuring devices is intended for transmitted light measurement, in particular transmitted light contour measurement, of tools. At least one of the optical tool measuring devices and / or the optical tool part-measuring and measuring devices may alternatively or additionally be intended for reflected light measurement, in particular reflected light surface measurement, of tools.At least one of the optical tool measuring devices and / or the optical tool part and measuring devices may be provided for in addition to tactile measurement of tools. "Provided for" is to be understood in particular as being specifically programmed, designed, and / or equipped. The fact that an object is intended for a specific function is to be understood in particular as meaning that the object ZO 17881 WO fulfills and / or performs this specific function in at least one application and / or operating state.

[0010] The computer system can be configured as a single control unit, e.g., one of the optical tool measuring devices and / or the optical tool part and measuring devices of the optical metrology system; as a plurality of networked control units; as a computer assigned to one or more of these devices and colocated with them; as a computer that is external and not colocated with the device(s); as a distributed computing system, such as a cloud computing system; or as a combination of two or more of the aforementioned systems. In cases where at least part of the computer system is not colocated with the optical tool measuring devices and / or the optical tool part and measuring devices of the optical metrology system, the data communication link can be wired or wireless.The digital data sheet is preferably an electronic document or file containing essential technical and / or functional information about a specific tool in a structured format. The digital data sheet does not need to be specifically created for the described process. It can be the same as the data sheet typically prepared for any tool, for example, by its manufacturer.

[0011] The digital datasheet can be a manufacturer's datasheet. It can be in various electronic formats (e.g., PDF, HTML, CSV, etc.). The digital datasheet can be provided, for example, via the manufacturer's website for the tool or a manufacturer's digital catalog.

[0012] Receiving the digital data sheet can be achieved, among other methods, by reading a storage medium (e.g., provided or pre-installed by an operator of the optical measurement system), by automatically or manually retrieving an online database, or by another digital data transmission method. It is also conceivable that the digital data sheet could be received by scanning and / or ZO 17881 WO

[0013] A photograph of a physical data sheet can be taken. Accessing the digital data sheet can involve accessing a digital file containing the data sheet stored on a local or external storage medium. For example, it is conceivable that the computer system obtains digital data sheets directly from databases of tool manufacturers or from manufacturers of optical tool measuring devices and / or optical tool part and measuring devices. The tools are designed in particular as shank tools, preferably as rotary shank tools, for example, drills, milling cutters, profile tools, and / or reamers, wherein preferably a shank of the shank tools is provided for mounting in a tool holder. A complete tool comprises in particular a specially matched and / or related combination of tool chuck and tool, which is in particular demountable.A technical drawing is, in particular, a scaled representation of an object or system that includes one or more views of the object or system. The technical drawing may contain information about tolerances that can be taken into account in the described procedure during measurement program creation and / or selection. The technical drawing may be standardized. However, the described procedure is also intended for any non-standardized technical drawing.

[0014] The trained machine learning algorithm(s) used for data extraction is / are preferably specifically trained to extract technical drawings and / or dimensional data from digital tool datasheets. The trained machine learning algorithm(s) receive the digital datasheet(s) as input and deliver as output the tool-descriptive data and / or information contained in the digital datasheet in the form of drawings, text, and / or symbols. The analysis and evaluation module of the computer system is preferably a specialized software component and / or hardware functional unit of the computer system that extracts, processes, analyzes, and / or interprets the data to provide the measurement program selection or to create the measurement program.The analysis and evaluation module of the computer system can, for example, be configured as a storage unit or storage partition of the computer system that interacts with a processor unit of the computer system. The measurement program can include a selection of sensors from optical tool measuring devices and / or optical tool part-measuring and measuring devices. The measurement program can include a selection and / or arrangement of individual process steps of a measurement program for optical tool measuring devices and / or optical tool part-measuring and measuring devices. The measurement program can include a measurement setting and / or alignment setting for the sensors, e.g., for a reflected light camera, a transmitted light camera, illumination of the reflected light or transmitted light camera, etc., of the optical tool measuring devices and / or optical tool part-measuring and measuring devices.The measurement program can include a selection of measurement positions / ranges / tool ​​holding positions for the optical tool measuring devices and / or optical tool presetting and measuring devices. The measurement program can include a selection of measurement steps for the optical tool measuring devices and / or optical tool presetting and measuring devices. The measurement program can include a time sequence of measurement steps for the optical tool measuring devices and / or optical tool presetting and measuring devices. The measurement program can include a precision setting of components of the optical tool measuring devices and / or optical tool presetting and measuring devices. The measurement program can include a selection of an output format for the measurement results from the optical tool measuring devices and / or optical tool presetting and measuring devices.The measurement program may include a specification of a tool clamping type and / or clamping position in the optical tool measuring devices and / or optical tool part and measuring devices. The measurement program may further include parameters of the optical metrology system that are required or advantageous for tool measurement and / or tool setting. ZO 17881 WO.

[0015] The machine learning algorithm(s) used to determine the possible and / or required measurement characteristics (specifically trained for this purpose) is / are preferably specifically trained to determine measurement characteristics based on data and / or information describing a tool obtained from digital datasheets. The trained machine learning algorithm(s) used for this determination receive the data and / or information describing the tool as input and deliver the possible and / or required measurement characteristics for the tool described in the digital datasheet as output. The tasks of "reading" and "determining" can be performed by separately trained and executed machine learning algorithms. However, it is also conceivable that at least the tasks of "reading" and "determining" are combined and performed by a single, appropriately trained machine learning algorithm.

[0016] The trained machine learning algorithm used to create the measurement programs is preferably specifically trained for creating measurement programs for optical metrology systems. This algorithm receives as input the data and / or information describing the tool and / or the determined possible and / or required measurement characteristics, and outputs as the measurement program created specifically for the tool for implementation in the optical metrology system or for controlling the optical metrology system. The task of "creating the measurement program" can be performed by a trained machine learning algorithm that is trained and / or executed separately from the machine learning algorithms responsible for the tasks of "reading" and "determining."However, it is also conceivable that at least the task of "determining," or even the tasks of "reading" and "determining" together with the task of "creating the measurement program," could be combined and performed by a single, appropriately trained machine learning algorithm. ZO 17881 WO.

[0017] The trained machine learning algorithm used to select measurement programs is preferably specifically trained to select measurement programs for optical metrology systems from a predefined, known preselection. This algorithm receives as input the data and / or information describing the tool and / or the determined possible and / or required measurement characteristics. It outputs the most suitable measurement program from the stored list for implementation in the optical metrology system / for controlling the optical metrology system. The "measurement program selection" task can be performed by a trained machine learning algorithm that is trained and / or executed separately from the machine learning algorithms responsible for the "reading" and "determining" tasks.However, it is also conceivable that at least the "determine" task, or even the "read" and "determine" tasks together with the "select measurement program" task, could be combined and performed by a single, appropriately trained machine learning algorithm. It is also conceivable that the "select measurement program" and "determine measurement program" tasks could be performed by the same specially trained machine learning algorithm. For example, the trained machine learning algorithm could individually define some of the measurement steps of a measurement program and select others from one or more predefined lists. For instance, the machine learning algorithm could first attempt to select from a list, and if none of the measurement programs in the list are suitable, it could create a custom measurement program.The list of measurement programs is preferably stored on a storage unit of the computer system or on a storage unit accessible to a program of the computer system, in particular the trained machine learning algorithm. Distributed storage of the list is also conceivable. ZO 17881 WO.

[0018] Furthermore, it is proposed that a created and / or selected measurement program be output to the optical metrology system and automatically preset in at least one of the optical tool measuring devices and / or the optical tool part and measuring devices of the optical metrology system. This allows for high user-friendliness and / or high efficiency. In addition, it can advantageously achieve high operational reliability, particularly by preventing operator errors. Subsequently, the tool is entered into the optical metrology system and / or measured by the optical metrology system, at least using the measurement program. Alternatively or additionally, the measurement program can also be output to a user of the optical metrology system, e.g., via an operating unit / computer screen, who can then preferably set it manually.Alternatively or additionally, it is also conceivable that the user is offered a selection of presets from various selected and / or created suitable measurement programs for acceptance, rejection or selection.

[0019] It is proposed that several measurement programs be created for measuring the tool described in the digital datasheet and / or selected from the list of measurement programs stored in the computer system. This can advantageously achieve a high degree of flexibility and / or operational reliability. In particular, it is conceivable that a number of measurement programs are useful for a tool contained in a digital datasheet, depending on the planned future use, the desired precision, etc. The multiple created and / or selected measurement programs then form the basis for the selection option described above by the user.

[0020] Furthermore, it is proposed that, based on the determined measurement characteristics, at least one trained machine learning algorithm of the analysis and evaluation module, particularly before the creation and / or selection of the measurement program(s), should be used to select the appropriate device from a plurality of optical tool measuring devices and / or optical tool part and measuring devices included in the optical metrology system. This can achieve high user-friendliness and / or high efficiency. In addition, it can advantageously achieve high operational reliability, particularly by preventing operator errors, especially in optical metrology systems with many individual devices.The trained machine learning algorithm used for device selection is preferably specifically trained to select optical tool measuring devices and / or optical tool part and measuring devices from a known optical metrology system. The trained machine learning algorithm used for selecting the measurement programs receives as input the data and / or information describing the tool and / or the determined possible and / or required measurement characteristics, and provides as output the required device(s) of the optical metrology system. The "device selection" task can be performed by a trained machine learning algorithm that is trained and / or executed separately from the machine learning algorithms performing the "read" and "determine" tasks and / or from the "measurement program selection" and "measurement program determination" tasks.However, it is also conceivable that at least the "determine" task, or even the "read out" and "determine" tasks together with the "device selection" task, could be combined and performed by a single, appropriately trained machine learning algorithm. It is also conceivable that the "measurement program selection" and "device selection" tasks, and / or the "measurement program determination" and "device selection" tasks, could be performed by the same specially trained machine learning algorithm. Furthermore, it is conceivable that the trained machine learning algorithm responsible for device selection could consider the current availability / planning / capacity of the individual devices in the optical metrology system as an additional input, thus influencing the measurement program creation and / or selection. ZO 17881 WO.

[0021] Furthermore, if at least one trained machine learning algorithm of the analysis and evaluation module, based on the determined measurement characteristics, selects a device function from a plurality of functions of the optical tool measuring device and / or optical tool part and measuring device selected in the device selection, particularly before the creation and / or selection of the measurement program(s), a particularly high level of efficiency and / or precision can be achieved. The aforementioned trained machine learning algorithm intended for the task of "device selection" can also be used for the task of "device function selection." Alternatively, a separate machine learning algorithm can be specifically trained for this purpose, which, for example, receives the device selection as an additional input and outputs the device function selection.

[0022] If the available device functions include at least various measuring devices of the selected optical tool measuring device and / or optical tool part-measuring device, such as a reflected light measuring device, a transmitted light measuring device and / or a tactile measuring device, and / or at least various tool holders, clamping types and / or adapters for presenting the tool in front of the measuring device during a measuring process by the selected optical tool measuring device and / or optical tool part-measuring device, a particularly optimized measuring program can be advantageously run. Furthermore, a particularly high degree of automation can be advantageously achieved.

[0023] It is further proposed that the measurement programs of the majority of measurement programs, the device selection, and / or the device function selection be output to a user of the optical measurement system for selection and / or activation. This can advantageously result in particularly good, efficient, and / or user-friendly operation of the optical ZO 17881 WO.

[0024] The output can be achieved via the measurement system. The output is preferably provided via the control unit / computer screen of the optical measurement system.

[0025] Furthermore, it is proposed that the analysis and evaluation module include at least one hybrid AI system / a multimodal AI system with two or more interacting machine learning algorithms. This can advantageously achieve particularly efficient and / or precise creation and / or selection of measurement programs. It can also advantageously lead to greater robustness of the results and / or greater efficiency of the processes. A hybrid AI system is, in particular, an artificial intelligence that combines two or more different AI approaches. The goal is to combine the strengths of at least two different methods in order to solve more complex problems.

[0026] If the hybrid AI system includes an OCR (Optical Character Recognition) algorithm, particularly one based on deep learning or neural networks, which is designed to extract the measurement(s) from the digital datasheet, then a particularly reliable and accurate extraction of all textual information (including numbers and symbols) from the digital datasheet can be achieved. In particular, measurement data in digital datasheets can be captured very reliably. Alternatively, an ICR (Intelligent Character Recognition) algorithm could be used. The OCR / ICR algorithm is preferably a machine learning algorithm specialized in recognizing text in image files.Image text recognition is one of the flagship disciplines of machine learning, so the training and / or application of corresponding machine learning OCR algorithms falls within the expertise of a person skilled in the art (see, among others, https: / / de.wikipedia.org / wiki / Texterkennung, as of June 4, 2024, 6:04 PM UTC). AI-based OCR tools, or their machine learning algorithms, that could be used are known to those skilled in the art and are also partially available as open-source tools (e.g., newer versions of Tesseract, Calamari OCR, Kraken, or EasyOCR). In particular, the trained machine learning OCR algorithm and / or one or more of the other machine learning algorithms mentioned herein are executed by a processor unit of the analysis and evaluation module or to which the analysis and evaluation module has access.In particular, the OCR algorithm of machine learning and / or one or more of the other machine learning algorithms mentioned herein are stored on the storage unit of the analysis and evaluation module or on which the analysis and evaluation module has access. The analysis and evaluation module includes, in particular, an input for receiving the digital datasheets, an information processing unit for analyzing the digital datasheets, and an output unit for at least the output of measurement programs based on the results of the evaluation to other components of the optical measurement system. Alternatively, however, it is also conceivable that the described method uses a non-AI-based OCR system to extract textual content from the digital datasheet.

[0027] Alternatively or additionally, if the hybrid AI system includes a computer vision machine learning algorithm, particularly one based on deep learning or neural networks, such as a CNN (Convolutional Neural Network) or a Capsule Networks machine learning algorithm, which is at least designed to read the technical drawing(s) from the digital datasheet, then a particularly reliable and accurate extraction of all pictorial / graphical information from the digital datasheet can be achieved. In particular, technical drawings in digital datasheets can thus be captured very reliably. The machine learning algorithm is preferably a trained machine learning algorithm specialized in the recognition of objects / geometries in image files.Object recognition / geometry recognition from images is also one of the flagship disciplines of machine learning (ZO 17881 WO), so that the training and / or application of corresponding machine learning algorithms falls within the expertise of a specialist (see, among others, https: / / en.wikipedia.org / wiki / Outline_of_object_recognition, as of October 30, 2023, 12:14 UTC). For example, one of the well-known CNN machine learning algorithms described in the following publications can be used in the procedure: a) AlexNet: Alex Krizhevsky, Imagenet classification with deep convolutional neural networks, Communications of the ACM 60.6, pp. 84-90 (2017); b) MobileNet: Andrew G. Howard, MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications, CoRR, abs / 1704.04861, (2017). c) Xception : Francois Chollet, Xception: Deep Learning with Depthwise Separable Convolutions, CoRR, abs / 1610.02357, (2016); d) LeCun Y, Bengio Y, Hinton G (2015) Deep learning; Nature 521 :436{444, DOI 10.1038 / naturel4539; e) Lin H, Li B, Wang X, Shu Y, Niu S (2019); Automated defect inspection of LED chip using deep convolutional neural network; J Intell Manuf; 30:2525{2534, DOI 10.1007 / sl0845-018-1415-x; f) Fu G, Sun P, Zhu W, Yang J, Cao Y, Yang MY, Cao Y (2019); A deep-learning-based approach for fast and robust steel surface defects classification; Opt Laser Eng 121 :397{405, DOI 10.1016 / j.optlaseng.2019.05.005; g) Lee KB, Cheon S, Kim CO (2017) A Convolutional Neural Network for Fault Classification and Diagnosis in Semiconductor Manufacturing Processes; IEEE T Semiconduct M 30: 135{142, DOI 10.1109 / TSM.2017.2676245; h) Goncalves DA, Stemmer MR, Pereira M (2020) A convolutional neural network approach on bead geometry estimation for a laser cladding system; Int J Adv Manuf Tech 106:1811 {1821 , DOI.

[0028] 10.1007 / s00170-019-04669-z; i) Karatas A, Kölsch D, Schmidt S, Eier M, Seewig J (2019) Development of a convolutional autoencoder using deep neuronal networks for defect detection and generating ideal references for cutting edges; Munich, Germany, DOI 10.1117 / 12.2525882; j) Stahl J, Jauch C (2019) Quick roughness evaluation of cut edges using a convolutional neural network; In: Proceedings SPIE 11172, Munich, Germany, DOI 10.1117 / 12.2519440; oder k) ein CNN des unter dem Namen “TensorFlow” bekannten Open Source ZO 17881 WO

[0029] Frameworks. Alternative CNN algorithms known to experts, such as Region Proposals (R-CNN, Fast R-CNN, Faster R-CNN), Detectron, Single Shot MultiBox Detector (SSD), or You Only Look Once (YOLO, for example, in version 9, which was licensable at the time of application), etc., are of course also conceivable. Simple open-source solutions are available for applying many of these machine learning algorithms (see the Wikipedia article referenced above). For example, the open-source frameworks "TensorFlow CapsNet," "PyTorch CapsNet," and Keras "KapsNet" exist for the Capsulate Networks machine learning algorithms. Alternatively, however, it is also conceivable that the described method uses a non-AI-based computer vision system to extract pictorial / graphical content from the digital datasheet.

[0030] Additionally, it is proposed that the hybrid AI system include a machine learning algorithm specifically designed, preferably specially trained, for determining measurement characteristics from the data and / or information extracted from the digital datasheet. This can advantageously achieve high efficiency and / or accuracy in finding and setting suitable measurement programs. It can also advantageously improve the user-friendliness and efficiency of the measurement systems.In particular, the machine learning algorithm specifically designed and preferably specially trained for determining measurement characteristics from the data and / or information extracted from the digital datasheet analyzes at least the graphical data extracted from the digital datasheet by the computer vision machine learning algorithm to determine the measurement characteristics. Furthermore, the textual data obtained using the OCR machine learning algorithm can also be considered in this determination of measurement characteristics. Measurement characteristics can include specific measurement points, surfaces, edges, angles, undercuts, and more. The specially trained machine learning algorithm is ZO 17881 WO.

[0031] Learning to determine measurement characteristics can, for example, be done in a training method known to experts, using numerous example cases for tools and associated measurement characteristics determined by experts.

[0032] Furthermore, it is proposed that the hybrid AI system include a machine learning algorithm specifically trained for creating at least one measurement program and / or for selecting at least one measurement program from the list of measurement programs stored in the computer system, particularly based on the measurement characteristics. This can advantageously achieve high efficiency and / or accuracy in finding and setting suitable measurement programs. It can also advantageously improve the user-friendliness and efficiency of the measurement systems.In particular, the machine learning algorithm specifically designed for creating and / or selecting measurement programs (from the measurement features), preferably a specially trained algorithm, analyzes at least the measurement features extracted from the digital datasheet by combining the computer vision machine learning algorithm with the OCR machine learning algorithm to create the measurement program. The specially trained machine learning algorithm for selecting and / or creating measurement programs can, for example, be trained in a training manner known to those skilled in the art using numerous example cases for tools and associated expert measurement program selections, especially for different optical metrology systems.

[0033] Furthermore, it is proposed that the machine learning algorithm responsible for creating and / or selecting the measurement program be further trainable, in particular through supervised learning, reinforcement learning, or a human-in-the-loop approach, with training feedback for further training of the machine learning algorithm provided by a (user) measurement program change to a measurement program created or suggested for selection by the analysis and evaluation module (ZO 17881 WO).

[0034] The measurement program is formed by a user selection from one of several measurement programs available for measuring the tool described in the digital datasheet and suggested by the analysis and evaluation module. This allows for a high degree of user-friendliness. User-specific requirements, wishes, or habits can be advantageously incorporated into the creation and / or selection of measurement programs. The aforementioned training approaches for further training machine learning algorithms—supervised learning, reinforcement learning, or the human-in-the-loop approach—are known to those skilled in the art, so no further explanation regarding their feasibility appears necessary here.The relevant expertise is summarized in particular in the English-language articles of the online encyclopedia "Wikipedia" entitled "Reinforcement learning", "Reinforcement learning from human feedback" and "Supervised learning" (see https: / / en.wikipedia.org / wiki / Reinforcement_learning, accessed: October 9, 2024, 03:22 UTC, https: / / en.wikipedia.org / wiki / ReinforcementJearningJrom_humanJeedback, accessed: October 9, 2024, 07:11 UTC and https: / / en.wikipedia.org / wiki / Supervised_learning, accessed: August 11, 2024, 13:16 UTC).

[0035] Furthermore, it is proposed that the machine learning algorithm responsible for creating the measurement program and / or selecting the measurement program from the list of measurement programs stored in the computer system take into account the usage history of at least one, preferably all, of the optical tool measuring devices and / or the optical tool part and measuring devices of the optical metrology system. This advantageously allows for a high level of user-friendliness. User-specific requirements, wishes, or habits can also be advantageously incorporated into the measurement program creation and / or selection retrospectively. Preferably, when adding an optical tool ZO 17881 WO

[0036] The usage history of the optical tool measuring device(s) and / or optical tool-piece measuring device(s) is read out and used for further training of the trained machine learning algorithm and / or as a parameter in the trained machine learning algorithm. This is done either for the optical measuring system subjected to the procedure or, in the case of a new setup of the procedure for an optical measuring system, for the optical tool-piece measuring system subjected to the procedure. The AI ​​is thus preferably trained to include the usage history in the analysis.The trained machine learning algorithms used to create the measurement programs can therefore receive, in addition to the aforementioned inputs (text, images and / or measurement features), a previous usage history of the respective optical tool measuring device(s) and / or optical tool component and measuring device(s) of the optical measurement system.

[0037] Furthermore, a localized or distributed computer system is proposed, comprising at least one processor, in particular the processor unit, and at least one data storage device, in particular the storage unit, on which a computer program, in particular comprising the hybrid AI system, is stored. This computer program, when executed by the processor, is designed to perform the aforementioned procedure. This advantageously simplifies and / or improves the finding and setting of suitable measurement programs. It also advantageously improves the user-friendliness and efficiency of the measurement systems.

[0038] Furthermore, the optical metrology system is equipped with one or more optical tool measuring devices and / or with one or more optical tool part and measuring devices, wherein the optical metrology system includes the localized computer system, by means of a data communication link with an externally arranged localized computer system or with the distributed ZO 17881 WO

[0039] It is suggested that the system be connected to a computer. This can advantageously simplify and / or improve the finding and setting of suitable measurement programs. It can also advantageously improve the user-friendliness and efficiency of the measurement systems.

[0040] The inventive method, the inventive computer system, and the inventive optical measurement system are not limited to the application and embodiment described above. In particular, the inventive method, the inventive computer system, and the inventive optical measurement system may, to achieve a functionality described herein, comprise a different number of individual elements, components, and units than the number specified herein.

[0041] Drawings

[0042] Further advantages will become apparent from the following description of the drawings. The drawings illustrate an embodiment of the invention. The drawings, the description, and the claims contain numerous features in combination. A person skilled in the art will expediently consider the features individually and combine them into meaningful further combinations.

[0043] They show:

[0044] Fig. 1 shows a schematic representation of an optical measurement system.

[0045] Fig. 2a is an example data sheet,

[0046] Fig. 2b shows another example data sheet and

[0047] Fig. 3 shows a schematic flowchart of a process using the optical measurement system.

[0048] Description of the Exemplary Implementation Figure 1 shows a schematic representation of an optical metrology system 10, which includes, by way of example, an optical tool presetting and measuring device 14 and an optical tool measuring device 12. The exemplary optical tool presetting and measuring device 14 is shown in more detail, while the exemplary optical tool measuring device 12 is shown only in a highly simplified form as a box. Both devices 12 and 14 can also be completely different types and / or varieties of optical tool presetting and measuring devices 14 or optical tool measuring devices 12. Furthermore, the optical metrology system 10 can also include other or identical optical tool presetting and measuring devices or other or identical optical tool measuring devices, as indicated by three dots in Figure 1.The optical tool part and measuring device 14 and the optical tool measuring device 12 are each intended for optical measurement of tools 20.

[0049] The optical tool-single-piece measuring device 14, shown in more detail as an example, has various measuring devices 30. The measuring devices 30 each serve to measure the tools 20. One of the measuring devices 30 is a transmitted light measuring device 34 with an illumination 44 and a transmitted light camera 46, which are arranged on opposite sides of a measuring area 48 of the optical tool-single-piece measuring device 14. The transmitted light measuring device 34 is designed to capture shadow outlines in a known manner. One of the measuring devices 30 is a reflected light measuring device 32. The reflected light measuring device 32 has a reflected light camera 52 and an illumination 44. The illumination 44 of the reflected light measuring device 32 is identical to the illumination 44 of the transmitted light measuring device 34 in the embodiment shown in Fig. 1. Alternatively, separate illuminations 44 are of course also conceivable.The incident light measuring device 32 is designed for surface measurement in a known manner. One of the measuring devices 30 is a tactile measuring device 36. The tactile measuring device 36 comprises a probe 50. The probe 50 is designed for the tactile measurement of the surfaces of tools 20. It is conceivable that a tool adapter 38 is necessary for measuring some tools 20, by means of which the tool 20 or a tool holder (not shown) or a tool chuck with the tool 20 can be inserted into the respective optical tool assemblies and measuring device 14 or tool measuring device 12. The tool adapter 38 serves to present the tool 20 in front of the measuring device(s) 30 during a measurement process by the optical tool assemblies and measuring device 14 or the optical tool measuring device 12.

[0050] The optical measurement system 10 includes a computer system 16. In Fig. 1, the computer system 16 is shown by way of example as a localized computer system 16, which forms part of the measurement system 10. Alternatively or additionally, however, an externally arranged localized computer system and / or an externally distributed computer system, such as a cloud computing computer system, connected to the optical measurement system 10 via a data communication link, would also be conceivable. The computer system 16 includes a processor unit with a processor 40. The computer system 16 includes a memory unit with a data storage device 42. The computer system 16 comprises an analysis and evaluation module 26. The analysis and evaluation module 26 includes the data storage device 42. A computer program is stored on the data storage device 42. The computer program is intended to be called and executed by the processor 40.The computer program is designed to execute a computer-implemented method, as described in particular in this context. The computer program comprises one or more trained machine learning algorithms. The analysis and evaluation module 26, in particular the computer program of the analysis and evaluation module 26, preferably comprises a hybrid AI system with two or more than two specifically interacting trained machine learning algorithms. A list 28 of measurement programs for the optical measurement system 10 is also stored on the data storage device 42.

[0051] Figures 2a and 2b show exemplary digital data sheets 18 for exemplary tools 20. The digital data sheets 18 may originate from physical data sheets that were converted into digital data sheets 18 by scanning or photographing, or they may already exist as digital data sheets 18. Each digital data sheet 18 contains data and / or information describing the tool 20. In both cases, a shank tool is shown as an example in the digital data sheet 18. However, the tool 20 could also be a different tool, e.g., a complete tool assembly. Each digital data sheet 18 includes one or more technical drawings 22 of the respective tool 20. Each digital data sheet 18 includes one or more dimensional specifications 24 of the respective tool 20, for example, a scale, dimension, or measurement of the tool 20.The two digital datasheets 18 shown in Figures 2a and 2b, which are structured very differently in form, are intended to serve as examples to show that any types and forms of digital datasheets 18, regardless of their formatting, are intended and suitable for use in the described procedure.

[0052] Figure 3 shows a schematic flowchart of a process using the optical measurement system 10. Most or all of the process steps can be computer-implemented. Therefore, the process can be a computer-implemented method, preferably a computer-implemented tool measurement method. The process is preceded by several training steps, not detailed here, to prepare the combined trained machine learning algorithm(s). The process can be subdivided into process steps.

[0053] In at least one optional process step 90, a physical data sheet is digitized and the digital data sheet 18 is created. In at least one ZO 17881 WO

[0054] In process step 100, the digital data sheet 18 of the tool 20 is received and / or accessed by the computer system 16. In at least one process step 110, the dimensions 24 are read from the digital data sheet 18. The reading of the dimensions 24 from the digital data sheet 18 is performed by the analysis and evaluation module 26 of the computer system 16 using one of the trained machine learning algorithms, in particular the hybrid AI system. The hybrid AI system includes an OCR (Optical Character Recognition) machine learning algorithm. The OCR machine learning algorithm is preferably based on deep learning or a neural network. The OCR algorithm is specifically designed to read at least the dimensions 24 from the digital data sheet 18. In process step 110, the technical drawings 22 are read from the digital data sheet 18.The technical drawings 22 are read from the digital datasheet 18 using one of the trained machine learning algorithms, in particular the hybrid AI system, by the analysis and evaluation module 26 of the computer system 16. The hybrid AI system comprises a computer vision machine learning algorithm, e.g., a CNN (Convolutional Neural Network) machine learning algorithm or a Capsulate Networks machine learning algorithm. The computer vision machine learning algorithm is preferably based on deep learning or a neural network. The computer vision machine learning algorithm is specifically designed to read the technical drawings 22 from the digital datasheet 18. During the reading of the digital datasheet 18, at least some or all of the data and / or information contained in the digital datasheet 18 and describing the tool 20 are extracted.

[0055] In at least one optional process step 120, possible and / or required measurement characteristics for the tool 20 described in the digital data sheet 18 are determined by analyzing the data and / or information obtained through the readout. The determination of the measurement characteristics is carried out by ZO 17881 WO, a further machine learning algorithm of the analysis and evaluation module 26, which is specifically trained for this task and can also be part of the hybrid AI system.

[0056] In at least one process step 130, a device selection is made from a plurality of optical tool measuring devices 12 and / or optical tool part and measuring devices 14 encompassed by the optical metrology system 10, based on the determined measurement characteristics, by at least one trained machine learning algorithm of the analysis and evaluation module 26, in particular the hybrid AI system. In at least one process step 131, the device selection is output to a user of the optical metrology system 10 for selection and / or activation.

[0057] In at least one process step 140, a device function selection is made from a plurality of device functions of the optical tool measuring device 12 and / or optical tool part-measuring device 14 selected in the device selection, based on the determined measurement characteristics, by at least one trained machine learning algorithm of the analysis and evaluation module 26, in particular the hybrid AI system. The device functions available for selection include the various measuring devices 30 of the selected optical tool part-measuring device 14. The device functions available for selection include various tool holders, various tool clamping types, and / or the various tool adapters 38 for presenting the tool 20 in front of the measuring device 30 during a measurement process by the optical tool part-measuring device 14.In at least one process step 141, the device function selection is output to a user of the optical metrology system 10 for selection and / or activation. Alternatively, the device selection and / or the device function selection can also result from the process steps described below with reference numbers 150a and / or 150b and / or 151b, e.g., if a measurement program and / or a device function is permanently assigned to a device, in particular a ZO 17881 WO optical tool measuring device 12 and / or an optical tool presetting and measuring device 14.Alternatively, the operator could be offered a selection of possible devices, in particular optical tool measuring devices 12 and / or optical tool parting and measuring devices 14 of the optical metrology system 10, or a corresponding selection from the selection of possible devices, in particular optical tool measuring devices 12 and / or optical tool parting and measuring devices 14 of the optical metrology system 10, could be automatically selected based on further parameters, e.g. based on the utilization of the devices, in particular the optical tool measuring devices 12 and / or the optical tool parting and measuring devices 14 of the optical metrology system 10.

[0058] In at least one further process step 150a, an optimal measurement program for at least one of the optical tool measuring devices 12 and / or the optical tool part and measuring devices 14 of the optical metrology system 10 is created based on the measurement characteristics determined by the readout. Instead of just a single measurement program, several possible measurement programs for measuring the tool 20 described in the digital data sheet 18 can also be created. The measurement program or programs are created by a specially trained machine learning algorithm of the analysis and evaluation module 26. The hybrid AI system includes the machine learning algorithm specially trained for creating the at least one measurement program.The machine learning algorithm responsible for creating the measurement program takes into account a usage history of the optical tool measuring devices 12 and / or the optical tool part and measuring devices 14 of the optical measurement system 10.

[0059] In at least one alternative or additional process step 150b to process step 150a, an optimal measurement program for at least one of the optical tool measuring devices 12 and / or the optical tool part and measuring devices 14 of the optical metrology system 10 ZO 17881 WO is selected from the list 28 of (pre-configured) measurement programs stored in the computer system 16, based on the measurement characteristics determined by the readout. Instead of just a single measurement program, several possible measurement programs for measuring the tool 20 described in the digital data sheet 18 can also be selected from the list 28. The measurement program or programs are selected by a specially trained machine learning algorithm of the analysis and evaluation module 26.The hybrid AI system comprises a machine learning algorithm specifically trained for selecting at least one measurement program from the list 28 measurement programs stored in computer system 16. The machine learning algorithm responsible for selecting the measurement program from the list 28 measurement programs stored in computer system 16 takes into account the usage history of the optical tool measuring devices 12 and / or the optical tool part and measuring devices 14 of the optical metrology system 10. In at least one process step 151b, when multiple measurement programs are selected, the multiple measurement programs of the plurality of measurement programs are output to the user of the optical metrology system 10 for selection and / or activation.

[0060] In at least one process step 160, the machine learning algorithm responsible for creating and / or selecting the measurement program is further trained, in particular through supervised learning, reinforcement learning, or a human-in-the-loop approach. For this purpose, user training feedback is recorded in process step 160. The training feedback for the further training of the machine learning algorithm is generated by a (user) modification of a measurement program created or suggested for selection by the analysis and evaluation module 26, and / or by a user selection of one of several measurement programs from the majority of those possible for measuring the tool 20 described in the digital data sheet 18 and suggested by the analysis and evaluation module 26.

[0061] In at least one process step 170, the created and / or selected measurement program is output to the optical measurement system 10. In at least one process step 180, the output measurement program is automatically preset in at least one of the optical tool measuring devices 12 and / or the optical tool presetting and measuring devices 14 of the optical measurement system 10.

[0062] Reference sign

[0063] 10 Optical measurement system

[0064] 12 Optical tool measuring device

[0065] 14 Optical tool parting and measuring device

[0066] 16 computer systems

[0067] 18 Digital Datasheet

[0068] 20 tools

[0069] 22 Technical drawing

[0070] 24 Measurement

[0071] 26 Analysis and Evaluation Module

[0072] List of 28

[0073] 30 Measuring device

[0074] 32 Incident light measuring device

[0075] 34 Transmittance measuring device

[0076] 36 Tactile measuring device

[0077] 38 tool adapters

[0078] 40 processor

[0079] 42 Data storage devices

[0080] 44 Lighting

[0081] 46 Transmissive camera

[0082] 48 measuring range

[0083] 50 buttons

[0084] 52 Reflection camera

[0085] 90th process step

[0086] 100th process step

[0087] 110 Procedure step

[0088] 120th process step

[0089] 130 Procedure step

[0090] 131st procedural step

[0091] 140 Process step Process step Process step Process step Process step Process step

Claims

October 2, 2025 Claims 1. Computer-implemented method, in particular a computer-implemented tool measuring method, comprising an optical measuring system (10) which includes one or more optical tool measuring devices (12) and / or optical tool part measuring devices (14), and a computer system (16) arranged internally and / or externally to the optical measuring system (10) and preferably connected to the optical measuring system (10) at least by a data communication link, comprising at least the method steps (100, 110, 150a, 150b): - Receiving and / or accessing a digital data sheet (18) of a tool (20), for example a shank tool or complete tool, which includes one or more technical drawings (22) of the tool (20) and one or more dimensional specifications (24) of the tool (20), for example a scale, dimension or dimension of the tool (20), by the computer system (16), - Reading the technical drawing (22) or several of the technical drawings (22) and the dimension (24) or several of the dimension specifications (24) from the digital data sheet (18) at least by means of an analysis and evaluation module (26) of the computer system (16) comprising at least one trained machine learning algorithm for the extraction of at least part of the data and / or information contained in the digital data sheet (18) and describing the tool (20), - Creating at least one measurement program for at least one of the ZO 17881 WO optical tool measuring instruments (12) and / or the optical tool part and measuring instruments (14) of the optical metrology system (10) based on measurement features determined by reading out data by at least one trained machine learning algorithm of the analysis and evaluation module (26) and / or - Selecting a measurement program for at least one of the optical tool measuring devices (12) and / or the optical tool presetting and measuring devices (14) of the optical metrology system (10) from a list (28) of measurement programs stored in the computer system (16) based on the determined measurement characteristics by at least one trained machine learning algorithm of the analysis and evaluation module (26).

2. Method according to claim 1, characterized in that a created and / or selected measurement program is output to the optical measurement system (10) and is automatically preset in at least one of the optical tool measuring devices (12) and / or the optical tool part and measuring devices (14) of the optical measurement system (10).

3. Method according to claim 1 or 2, characterized in that several measurement programs possible for measuring the tool (20) described in the digital data sheet (18) are created and / or selected from the list (28) of measurement programs stored in the computer system (16). ZO 17881 WO 4. Method according to one of the preceding claims, characterized in that, based on the determined measurement characteristics, a device selection is made from a plurality of optical tool measuring devices (12) and / or optical tool part measuring devices (14) encompassed by the optical measuring system (10) by means of at least one trained machine learning algorithm of the analysis and evaluation module (26).

5. Method according to claim 4, characterized in that, based on the determined measurement features, at least one trained machine learning algorithm of the analysis and evaluation module (26) selects a device function from a plurality of device functions of the optical tool measuring device (12) and / or optical tool part and measuring device (14) selected in the device selection.

6. Method according to claim 5, characterized in that the device functions available for selection comprise at least different measuring devices (30) of the selected optical tool measuring device (12) and / or optical tool part and measuring device (14), such as a reflected light measuring device (32), a transmitted light measuring device (34) and / or a tactile measuring device (36), and / or at least different tool holders, clamping types and / or adapters (38) for presenting the tool (20) in front of the measuring device (30) during a measuring process by the selected optical tool measuring device (12) and / or optical tool part and measuring device (14). ZO 17881 WO 7. Method according to one of claims 3 to 6, characterized in that the measurement programs of the plurality of measurement programs, the device selection and / or the device function selection are output to a user of the optical measurement system (10) for selection and / or activation.

8. Method according to one of the preceding claims, characterized in that the analysis and evaluation module (26) comprises at least one hybrid AI system with two or more cooperating machine learning algorithms.

9. Method according to claim 8, characterized in that the hybrid AI system comprises an OCR (Optical Character Recognition) algorithm of machine learning, in particular based on deep learning or neural networks, which is at least intended to read the measurement (24) or the measurements (24) from the digital data sheet (18).

10. Method according to claim 8 or 9, characterized in that the hybrid AI system comprises a computer vision algorithm of machine learning, in particular based on deep learning or neural networks, e.g. a CNN (Convolutional Neural Network) algorithm of machine learning or a Capsulate Networks algorithm of machine learning, which is at least designed to read the technical drawing (22) or technical drawings (22) from the digital data sheet (18).

11. Method according to one of claims 8 to 10, characterized in that the hybrid AI system comprises a machine learning algorithm specifically provided, preferably specifically trained, for determining the measurement characteristics from the data and / or information extracted by reading from the digital data sheet (18).

12. Method according to one of claims 8 to 11, characterized in that the hybrid AI system comprises a machine learning algorithm specifically trained for the creation of the at least one measurement program.

13. Method according to one of claims 8 to 12, characterized in that the hybrid AI system comprises a machine learning algorithm specifically trained for selecting the measurement program from the list (28) of measurement programs stored in the computer system (16). ZO 17881 WO 14. Method at least according to one of claims 1 or 3, characterized in that the machine learning algorithm responsible for creating the measurement program and / or for selecting the measurement program is further trainable, in particular by supervised learning, by reinforcement learning or by a human-in-the-loop approach, wherein training feedback for the further training of the machine learning algorithm is formed by a (user) measurement program change to a measurement program created or proposed for selection by the analysis and evaluation module (26) and / or by a user selection of one of several measurement programs from the plurality of measurement programs possible for measuring the tool (20) described in the digital data sheet (18) and proposed by the analysis and evaluation module (26).

15. Method according to one of the preceding claims, characterized in that the machine learning algorithm responsible for creating the measurement program and / or for selecting the measurement program from the list (28) of measurement programs stored in the computer system (16) takes into account a usage history of at least one, preferably all, of the optical tool measuring devices (12) and / or the optical tool part and measuring devices (14) of the optical measurement system (10).

16. Localized or distributed computer system (16) comprising at least one processor (40) and at least one data storage device (42) on which a computer program, in particular comprising a hybrid AI system, is stored, which, when executed by the processor (40), is intended to execute the method according to one of the preceding claims.

17. Optical measurement system (10) comprising one or more optical tool measuring devices (12) and / or one or more optical tool part and measuring devices (14), wherein the optical measurement system (10) comprises a localized computer system (16) according to claim 16 or is connected by a data communication link to an externally arranged localized computer system according to claim 16 or to a distributed computer system according to claim 16.

Images