Method and setting assistant for knowledge-based setting of operational parameters for uhf rfid systems

Abstract

In the computer-implemented method described herein for the automated setting of parameters of a UHF RFID system (100, 105, 300 to 315) for contactless data exchange between RFID transponders (125, 305 to 315) of objects (101) moving past an RFID reader / writer (120, 200), in particular, it is provided that a knowledge-based setting assistant (400, 500) is provided by means of which a setting of the parameters of a UHF RFID system suitable for an existing UHF RFID system not yet in operation and its intended use is determined on the basis of technical information.

Description

Technical Field

[0001] This invention relates to the field of automated identification or recognition of objects via radio using RFID (Radio Frequency Identification) systems, particularly ultra-high frequency (UHF) RFID systems. Such RFID systems are known to allow contactless data exchange between RFID transponders and RFID writers or readers, with UHF antennas used for contactless reading and / or writing of RFID transponders (tags) disposed on such objects.

[0002] In this technical field, the present invention relates to a method and setup assistant for automatically setting or presetting parameters of such a UHF-based RFID system. Background Technology

[0003] Using UHF technology in RFID systems allows for higher data transmission rates compared to using HF (high frequency) technology alone. This enables faster transaction collection times and quicker data processing, which is particularly advantageous in commercial and industrial environments.

[0004] UHF RFID technology is used in applications such as inventory management, supply chain management, logistics, asset tracking, and access control. This technology is typically used in applications requiring long read distances.

[0005] Before operating an RFID system, commissioning engineers must set up operating parameters to achieve anticipated or even optimal operation. RFID system setup can be categorized into three types: "experience-based," "simulation-based," and "measurement-based." An example of a simulation-based solution is Turck's "Raytracer" software, used for field simulation of complex UHF RFID applications, enabling improved estimations of the accessibility of "data carriers" (i.e., RFID transponders or tags discussed here). An example of an experience-based solution is the RFID Assistant developed by Siemens as part of its so-called "TIA Portal (Total Integrated Automation Portal)," which provides the infrastructure for communication with higher-level agencies. Summary of the Invention

[0006] The underlying technical problem of this invention is that when setting up and deploying a UHF-based RFID system, many influencing factors must be considered, such as configuration, operating environment conditions, and antenna selection.

[0007] However, existing setup and commissioning solutions always require decisions or assessments to be made by UHF RFID system commissioning engineers. Since commissioning engineers are typically not UHF experts, they often choose inappropriate or even incorrect setups. In particular, they often overlook the necessary margins required to account for minor deviations or tolerances in the system or its operating environment.

[0008] The present invention aims to minimize or avoid such incorrect settings caused by commissioning engineers as a source of error, and to achieve expected or even optimal operation.

[0009] Therefore, the object of this invention is to allow for the automated or automated setup and commissioning of UHF RFID systems as easily as possible and with the lowest possible technical complexity. This should also enable commissioning engineers with limited technical knowledge to achieve this objective. In particular, the technical requirements for commissioning engineers should be simplified or reduced as much as possible, especially when dealing with altered or new uses or purposes.

[0010] To achieve this objective and address the corresponding technical issues, a corresponding setup assistant, preferably supported by artificial intelligence (AI), is proposed. This setup assistant considers all technical information known in itself and applicable to the UHF RFID system to be put into operation. This information covers not only the corresponding existing UHF RFID system but also the expected or existing operational and application-related environmental conditions of the UHF RFID system in the corresponding application.

[0011] It should be noted that the setup / commissioning and corresponding pre-configuration of a UHF RFID system require the relevant commissioning engineers to transfer a considerable amount of knowledge in order to translate application-specific knowledge into individual technical configuration values. Therefore, in order to transfer this knowledge, commissioning engineers require substantial technical expertise regarding the specific use or purpose of the UHF RFID system.

[0012] Within the framework of a dialogue with the commissioning engineer, the setup assistant proposed according to the present invention can ask them questions to guide their answers, preferably starting with fundamental technical questions, such as the number of existing RFID transponders (or tags or corresponding data carriers), and... Figure 1 The speed at which an object with an RFID transponder is scanned on the conveyor belt shown is illustrated. However, additionally, for example, Figure 1 The results shown are from UHF-based RFID measurements between the RFID transponder and RFID reader of the object to be scanned, regardless of whether the UHF-based RFID measurements were guided by the setup assistant or performed by the setup assistant itself. Alternatively, the commissioning engineer may also perform such measurements under the guidance of the setup assistant.

[0013] During these UHF-based RFID measurements, a setup assistant or commissioning engineer can perform power scans and / or location scans to obtain accurate information about the reader's antenna location, the corresponding existing antenna type, and / or antenna cable length. Additionally, real-time scans can be performed to determine the reliability of the UHF RFID system discussed herein.

[0014] The purpose of using an AI-supported setup assistant is to allow for the automated setting of parameters, preferably operational parameters, of the UHF RFID system. Based on the aforementioned dialogue with the commissioning engineer, the setup assistant determines or identifies the optimal parameter settings for the appropriate purpose of the UHF RFID system.

[0015] Additionally, a feedback loop can be established within the setup assistant, requiring the commissioning engineer to assess the accuracy and reliability of the UHF RFID system's operation after test runs. This allows the setup assistant to potentially find a better or optimal setup through further iterations. Furthermore, a continuous self-learning algorithm can be developed, constantly improving over time and with increased field usage. Here, the setup assistant is based on feedback from numerous customers / commissioning engineers who have already found optimal settings for given influencing parameters. The setup assistant can then use this knowledge to determine and recommend optimal operating parameters for new settings.

[0016] The method proposed here for automating or pre-setting parameters for such a UHF-based RFID system involves data input from the commissioning engineer (based on the aforementioned dialogue), depending on the specific purpose or intended use of the UHF RFID system. This data is then automatically converted into appropriate configuration data using a configuration assistant based on artificial intelligence (AI), generative artificial intelligence (GenAI), or another knowledge-based approach. This configuration data is then sent to the UHF RFID system for its commissioning.

[0017] According to a first aspect of a computer-implemented method for automating the setting of parameters of a UHF RFID system for contactless data exchange between RFID transponders having RFID transponders and moving past RFID readers / writers, the method includes providing at least one UHF antenna for contactless reading and / or writing of RFID transponders (tags) disposed on these objects. Specifically, a setting assistant supported by artificial intelligence (AI) or another knowledge-based method is configured, with the aid of which, based on technical information, settings of parameters suitable for the existing UHF RFID system to be put into operation and the intended use of the UHF RFID system are determined.

[0018] According to another aspect of the method according to the invention, it is possible to obtain the expected operating conditions with reference to the intended use of the UHF RFID system.

[0019] According to another aspect of the method according to the invention, it is configurable to determine suitable parameters for the UHF RFID system by referring to a dialogue between a setup assistant and a commissioning engineer, wherein the setup assistant is provided with technical information about the UHF RFID system to be commissioned.

[0020] According to another aspect of the method according to the invention, it is configurable that the technical information provided to the setup assistant relates to the number of existing RFID transponders, and / or the possible locations of the RFID transponders, and / or the moving speed of objects having RFID transponders and moving past the RFID reader / writer. It is also configurable that the technical information provided to the setup assistant includes the geometric arrangement (or corresponding layout) of the RFID transponders or corresponding tags or data carriers, antennas, and the environment. Additional information provided to the setup assistant may be the electromagnetic environment of the RFID system. Particularly relevant here are metallic objects, such as metal sheets, doors, machine parts, walls, steel beams, etc.

[0021] According to another aspect of the method according to the invention, it is configurable that a UHF-based RFID measurement is performed by a setup assistant or an operation engineer to obtain or determine the most accurate possible information regarding the location of the UHF antenna of the RFID reader / writer, and / or the appropriate type of existing UHF antenna, and / or the cable length of the UHF antenna, and / or the electromagnetic environment of the UHF RFID system.

[0022] According to another aspect of the method according to the invention, it is configurable that a UHF-based RFID measurement is performed by a setup assistant or an operation engineer to obtain or determine information about the quality of communication or contactless data exchange between the RFID reader / writer and the RFID transponder under changes in the parameters and positioning of the RFID transponder and at least one UHF antenna.

[0023] According to another aspect of the method according to the invention, it is configurable to implement UHF-based RFID measurements using reference power scanning, and / or positioning scanning, and / or real-time scanning.

[0024] According to another aspect of the method according to the invention, it is configurable that a setup assistant performs test runs on an already operational UHF RFID system to assess the accuracy and / or reliability of the UHF RFID system's operation, wherein the setup assistant determines improved settings based on the results of the test runs and with reference to further iterations.

[0025] According to another aspect of the method according to the invention, it is configurable to continuously improve the parameter settings of an already operational UHF RFID system over time and with the use of the UHF RFID system by means of a self-learning AI algorithm.

[0026] According to another aspect of the computer-implemented method of the present invention for automating the setup / commissioning of a UHF RFID system for a specified purpose, it is configurable that the method is implemented using a setup assistant with a large language model (LLM) and based on GenAI, wherein the method includes the following steps:

[0027] - Train a transducer-based setup model by referring to operational data related to UHF RFID systems, or optimize an existing transducer-based setup model by referring to such data;

[0028] - Provide a trained or optimized converter-based setup model via a computer interface, the converter-based setup model including at least one converter component;

[0029] - The commissioning engineer communicates with the setup assistant via a computer interface to collect data on the UHF RFID system for a specific purpose;

[0030] - Use collected data about a specified purpose to prompt a trained transformer-based setup model to output optimized setup data;

[0031] - Release optimized setup data output from a trained transducer-based setup model for automated or pre-setup of UHF RFID systems.

[0032] The aforementioned communication (or dialogue) between the commissioning engineer and the setup assistant can be conducted in natural language, or by referring to a specified list of questions (e.g., a list of multiple-choice questions), or also by referring to diagrams.

[0033] The method according to the invention provides more advantages for commissioning engineers of UHF RFID systems by achieving shorter setup time, fewer sources of error during setup, and immediate operational readiness.

[0034] According to another aspect of the method according to the invention, it is configurable to train the setup model using deep learning and / or reinforcement learning techniques. The use of these learning methods has advantages over simple (or non-generative) machine learning methods, which are best suited for well-defined tasks with structured data; they are also suitable for more complex tasks in which the AI-based computer system must process or semantically understand unstructured data. Deep learning mimics how the human brain processes data and establishes corresponding patterns, recognizing patterns and relationships within datasets of human-generated content. Referring to patterns learned in this way, GenAI generates new content. Generative AI models are typically trained using supervised learning methods. The aforementioned reinforcement learning techniques mimic the human learning process of solving problems or achieving goals through repeated trial and error.

[0035] The present invention also relates to a computer-implemented method for generating setup data suitable for a specified purpose for setting or pre-setting a UHF RFID system using a trained transducer-based setup model as described above, wherein the method includes the following steps:

[0036] - Provides commissioning engineers with access to trained converter-based setup models using computer processors;

[0037] - Receive input data related to a specified purpose via a computer interface;

[0038] - The trained transformer-based setup model analyzes the input data and provides instructions for generating new or revised setup data based on the trained setup model.

[0039] According to another aspect of the method according to the invention, it is configurable to prompt the commissioning engineer to request a trained converter-based setup model to provide machine-readable instructions based on the request made by the commissioning engineer to the setup model via a computer interface.

[0040] According to another aspect of the method according to the invention, it is configurable that the prompt includes one or more operational instructions related to one or more operational procedures of the UHF RFID system for a specified purpose. Additionally, it is configurable that the operational instructions include machine-readable instructions for controlling and / or monitoring the operation of the UHF RFID system.

[0041] The present invention also relates to a computer program product comprising computer-readable instructions that, when executed on a computer, prompt the computer to perform the aforementioned method steps.

[0042] Furthermore, the present invention relates to a computer product comprising a computer-readable token for accessing training data for setting up a model and / or for accessing the aforementioned trained or pre-trained setup model.

[0043] The setup assistant's method, tailored to the specific purpose (i.e., application-specific pre-setup or configuration for the UHF RFID system), allows for the automated transmission of the aforementioned setup data. The commissioning engineer, following an intuitively guided process, only needs to provide knowledge of the appropriate use of the expert system upon which the setup assistant relies.

[0044] Therefore, the commissioning engineer only needs to focus on the specific use or purpose of the UHF RFID system and provide the corresponding input data. The setup assistant or the expert system it relies on will guide the engineer through asking basic questions that are typically easy for the commissioning engineer to answer. This data input is automatically supplemented with expert knowledge or setup regarding the use in the method according to the invention and is converted into technical setup data.

[0045] This method systematically and intuitively guides commissioning engineers through a specific dialogue system, allowing them to contribute their knowledge of the technical status of the UHF RFID system's appropriate use. Due to the high degree of automation of the expert system, users advantageously require little or no effort to transfer knowledge.

[0046] Therefore, the commissioning engineer only needs to focus on the setup assistant and provide their input following the assistant's prompts. Thus, specifically for the optimal setup or pre-configuration of the UHF RFID system, there is no need for the commissioning engineer to transfer knowledge or expertise.

[0047] Then, in a single process, technical setup data optimized for a specific purpose can be uploaded to the UHF RFID system discussed here. This method significantly improves the effectiveness and efficiency of setting up or pre-setting the UHF RFID system.

[0048] Therefore, the setup assistant or the method for setup / commissioning proposed according to the present invention advantageously enables the automated or automated (pre) setup of suitable or even optimal operating parameters for the UHF RFID system under discussion herein.

[0049] The setup assistant can be performed locally on the RFID device or set up as an online service. In particular, when using AI or another knowledge-based approach, the assistant benefits advantageously from online experience gained from similar real-world use of the UHF RFID system under discussion.

[0050] Furthermore, the setup assistant or the method for commissioning or setting up proposed according to the present invention enables commissioning engineers to easily operate the corresponding RFID products. Attached Figure Description

[0051] Exemplary embodiments of the present invention are shown in the accompanying drawings and described in more detail in the following description.

[0052] Figure 1 The basic technical structure and typical application areas of the UHF RFID system discussed here are shown according to existing technology.

[0053] Figure 2The simulation results of the propagation of UHF signals from the read / write head according to the prior art are shown (source: https: / / www.turck.de / de / presse-135_simulationssoftware-fuer-uhfrfidapplikationen-1119.php).

[0054] Figure 3a The diagram shows a known power curve between a UHF-based antenna and several objects to be scanned, which is captured using a so-called "power scan".

[0055] Figure 3b The data obtained through a so-called “positioning scan” is shown, which is used to determine the optimal positioning of a UHF-based antenna relative to several objects bearing the aforementioned tags.

[0056] Figure 3c The results of a so-called "real-time scan" are shown, which are used to determine the reliability of the UHF RFID system being discussed here.

[0057] Figure 4 A schematic representation of the operation of the universal settings assistant according to the present invention is shown.

[0058] Figure 5 The flowchart illustrates an exemplary embodiment of a method according to the present invention for automating the setting of operating parameters of the UHF RFID system discussed herein.

[0059] Figure 6 An exemplary topology of a simple neural network for providing a setup assistant according to the present invention is shown.

[0060] Figure 7 It shows Figure 1 The illustrated UHF RFID system is put into operation with the help of a setup assistant according to the invention, which has a large language model and is based on generative AI, for setting or presetting the operating parameters of such a UHF-based RFID system, i.e., referring to the combined flowcharts / block diagrams.

[0061] Figures 8A to 8C illustrate the structure of the converter encoder / decoder architecture discussed here, which is used to generate operating parameter data, i.e., based on the aforementioned converter-based commissioning or setup model. Detailed Implementation

[0062] Figure 1The basic technical structure of the UHF RFID system discussed here and two exemplary typical application areas are shown. The application areas shown are an inventory management system 100 for logistics of goods arriving at a conveyor belt and a receiving control system 105 located at the receiving port.

[0063] An application 115 used by an operations engineer 110 of one of these systems 100, 105 interacts with an RFID reader / writer 120. The RFID reader / writer 120 wirelessly (bidirectionally) exchanges data 135 with a transponder (tag) 125 disposed on an (exemplary) object 101 transported in a container via UHF antennas 102, 103, 104, as shown by way of example. Additionally, the electrically passive transponder 125 is also wirelessly supplied with the necessary operating power 130 by the RFID reader / writer 120 in a manner known per se.

[0064] Figure 2 The illustration specifically shows the propagation behavior 205 of the UHF signal emitted by the read / write head 200, as generated using Raytracer software from Hans Turck GmbH & Co., mentioned at the beginning, from the perspective of the read / write head 200 in this exemplary embodiment. According to the prior art mentioned at the beginning, these field simulations implement complex UHF RFID applications to improve the estimation of the spatial reachability of transponders, such as those discussed here, for example, to avoid shadowing effects (see...). Figure 1 )wait.

[0065] Available Figures 3a to 3c The following three measurements are performed on the UHF RFID system shown. For example, Figure 1 The objects shown as examples (e.g., object 101 explicitly shown) are preferably equipped with passive RFID transponders 305, 310, 315, which are provided by UHF antennas 102 to 104 shown therein, or by... Figures 3a to 3c The UHF antenna 300 shown as an example is automatically read out.

[0066] The (electrical) passive RFID transponders (tags) 305 to 315, schematically shown, each include, in a manner known per se, a microchip with a processor, an antenna, and persistent memory. Unlike active transponders, these tags do not have their own power supply, but instead draw their operating energy from the radio waves emitted by the RFID reader 120 equipped with a UHF antenna 300 (see [link to relevant documentation]). Figure 1 These transponders have read-only memory or read-write EEPROM memory available.

[0067] The passive transponders discussed here are offered with different RFID chips, antennas, and designs for various applications (i.e., the different use cases or purposes mentioned above). Because the application area (including transponder installation) significantly affects transponder performance, corresponding laboratory or field testing with the original object is necessary before procuring RFID system components, and the planned solution must be validated in a proof-of-concept (POC).

[0068] Figure 3a The graph illustrates the probability of a successful read / write operation relative to the transmit power radiated by the UHF antenna 300 of the RFID reader 120. For example, a 50% probability value means that, in a large number of attempts, the corresponding tag can be read in half the cases. It is known that the three passive transponders (tags) 305, 310, and 315 shown as examples each have a receiving antenna for the transmit power radiated by the UHF antenna 300 of the RFID reader 120.

[0069] exist Figure 3a The illustrated empirical "power scan" shows that a linear increase in the electrical energy or power radiated by the UHF antenna 300 produces different slopes or curves 305', 310', and 315' for the electrical energy radiated by the three tags 305 to 315 in various cases. Depending on the different distances between the respective tags 305 to 315 and the UHF antenna 300, the required power for operation is achieved only at a greater distance from the RFID reader 120 with a higher radiated power from the antenna of the RFID reader 120. That is, for example, a lower radiated power 305' is sufficient for tag 305, while a radiated power 310' is required for tag 310. Therefore, even a higher radiated power 315' is needed at tag 315 to achieve the desired power.

[0070] Figure 3b The diagram schematically illustrates the results of a simulated "positioning scan" used to determine the proper or optimal positioning of the UHF antenna 300 relative to the three tags 305, 310, and 315. During this scan, the UHF antenna 300 moves or shifts in the shown vertical x-direction, thereby maintaining its distance from the three tags 305 to 315 in the existing horizontal direction.

[0071] Based on the three curves 305”, 310”, and 315” obtained from the scan, representing the electrical power radiated at each of the three tags 305 to 315, sufficient reliability for approximately 100% read / write operations for all three tags 305 to 315 is achieved only in a relatively narrow or elongated region 'd'325 in the vertical direction of movement (x-direction) of the RFID reader 120 (see...). Figure 3a ).

[0072] As a result of the positioning scan, during the movement of the objects carrying labels 305 to 315 along the conveyor belt in the horizontal direction shown (see... Figure 1 The RFID reader 120 should be arranged within the x region 'd' in the vertical x direction as shown to ensure that all three passive transponders (tags) 305 to 315 are adequately powered as they pass through the RFID reader 120.

[0073] Figure 3c Exemplary measurement results of a so-called "real-time scan" are shown, which allows for the determination of the reliability of the UHF RFID system under discussion here. Here, reliability specifically refers to the most reliable possible detection of objects, each bearing tags 305, 310, and 315, moving past the RFID reader 120 (see [link to documentation]). Figure 1 ).

[0074] exist Figure 3c In the scatter plot shown at the bottom, the successful scan results of the three existing labels 305, 310, and 315 are plotted as rows of points over time 't'. Here, the bottom row '1' represents the scan result for label 305, the middle row '2' represents the scan result for label 310, and the top row '3' represents the scan result for label 315.

[0075] In this example scan, tag 305 was detected six times by UHF antenna 300, tag 310 was detected five times, and tag 315 was also detected six times. According to... Figure 3c The horizontally offset arrangement of tags 305, 310, and 315 shown is illustrated in a timeline as they pass the RFID reader 120 in a leftward direction. Tag 305 or an object with that tag is detected first, followed by tag 310, and then tag 315.

[0076] The limited reliability of the scanning results or the identification of the tagged object (e.g., due to the suboptimal arrangement of the RFID reader 120) is evident from the fact that the three dot rows '1', '2', and '3' have some gaps along the (horizontal) time direction. As an example, gaps 320, 325, 330, and 335 are highlighted here by arrows.

[0077] Figure 4 Exemplary method steps and corresponding interaction steps between the commissioning engineer and the setup assistant 500 discussed here are illustrated.

[0078] The aforementioned interaction begins with basic training 505 of the setup assistant 500's AI (e.g., GenAI). During this training, the commissioning engineer, guided by the setup assistant 500, performs measurements 510 using the described power scan, positioning scan, and / or real-time scan. During this training, the commissioning engineer answers 515 questions posed by the setup assistant 500 regarding the application (i.e., the field of use or purpose) of the UHF RFID system to be set up.

[0079] It should be noted that the dialogue between the commissioning engineer and the setup assistant 500 can be conducted in natural language, through a list of pre-defined questions (e.g., multiple choice) or reference drawings; the commissioning engineer provides the setup assistant 500 with 520 images or drawings of, for example, applications or corresponding UHF RFID systems for training purposes.

[0080] After completing the basic training 505, the setup assistant 500 first outputs 525 a set of parameters that are suitable for the corresponding existing use of the RFID system or are even considered optimal. Additionally, the setup assistant 500 can output 530 suggestions for improving the antenna positioning of the UHF RFID system to be set up. These suggestions for improvement can then be additionally considered by the commissioning engineer in the final setup of the UHF RFID system.

[0081] Finally, the commissioning engineer provides feedback 540 to the setup assistant 500 regarding the technical requirements described for the reliability of the UHF RFID system to be set up. The AI ​​on which the setup assistant 500 is based can be retrained or further trained 535, as applicable, with reference to data measured during operation, so as to obtain a UHF RFID system that is set up or configured as optimally as possible for the corresponding prior art application or corresponding existing purpose.

[0082] Figure 5 The flowchart illustrates a method according to the present invention for automatically setting operating variables or operating parameters of the UHF RFID system discussed herein. Figure 1 The commissioning engineer interacts with the setup assistant here.

[0083] exist Figure 5In the first step of the method shown, installation is performed, where the commissioning engineer first sets up the 800 UHF RFID system with, for example, an experience-specified infrastructure. In the second step 805, a setup assistant asks the commissioning engineer simple questions. In the third step 810, the commissioning engineer answers the questions asked by the assistant. Afterwards, guided measurements 815 are performed, where the assistant guides the commissioning engineer through a series of measurements. These measurements test the RFID system, already set up with infrastructure 800, to determine whether it delivers scan results that are actually usable for the application or existing purpose discussed herein.

[0084] Based on the data obtained during measurement 815, an improved setup 820 is now determined or specified for the UHF RFID system, where the assistant calculates the optimal setup based on responses from the commissioning engineer and the results of guided measurements. Next, the RFID system is set up 825, where the commissioning engineer implements the optimal setup calculated by the assistant. Following this, an initial final test 830 (according to the aforementioned feedback loop) is performed on the RFID system set up in this manner, where the commissioning engineer uses the RFID system to perform tests and provides corresponding feedback to the assistant. Finally, training 835 is provided, in which the assistant is trained using existing data on the implemented setup.

[0085] The appropriate setting parameters are determined using the data-driven GenAI-based optimization or learning method 405 described below, which is implemented based on, for example, the following model methods used to set the input variables of the assistant:

[0086] - Number of tags

[0087] - Conveyor belt speed for labels

[0088] - Locating, imagering, and drawing metallic objects in the environment ( Figure 4 )

[0089] - Results from the measurements performed:

[0090] The percentage (%) of successful communication with the tag depends on the antenna's positioning (positioning scan) manually adjusted by the commissioning engineer.

[0091] The percentage (%) of successful communication with the tag depends on the power adjusted by the RFID system (power scan).

[0092] o Time series of the detected tags (results of real-time scanning)

[0093] o etc.

[0094] Configure parameters proposed or set by the assistant itself:

[0095] - Antenna transmit power

[0096] - Number of antennas

[0097] - Antenna positioning

[0098] - Filtering parameters used to evaluate the UHF messages of the tag

[0099] - Antenna modulation parameters

[0100] - Antenna polarization

[0101] - wait

[0102] Input variables and parameters can be viewed as vectors of size n and m, respectively. and (Column vector):

[0103] ,as well as ,or ,in ,as well as ,in .

[0104] The assistant is configured to associate parameters with input variables via a multidimensional function f:

[0105]

[0106] According to the first exemplary embodiment, the relationship between the input variables and the parameters can be approximated by a linear function:

[0107]

[0108] Higher-order terms can also be considered here:

[0109]

[0110] in:

[0111] ai is a vector of size m (x = 1…m),

[0112] Bij is a matrix of size m×n (𝑖 = 1…m, j = 1…n), and

[0113] Cijk is a tensor of size m×n×n (𝑖 = 1…m, j = 1…n, k = 1…n).

[0114] Regarding the data inputs / outputs used to train the ML system being discussed here for existing applications or domains of use, the use of vectors I and P can be considered as a data model.

[0115] In this data model, the aforementioned formula P = f(I) is therefore used according to Figure 5 Method step 820.

[0116] In method step 820, the optimal parameter P is determined based on the input variable I collected in method steps 805, 810, and 815. This requires a matrix Bij, Cijk, ..., or a lookup table or a weighted average of an artificial neural network. In method step 825, the optimal parameter P determined in this way is then used to configure the UHF RFID system. In optional method step 830, a final test (i.e., in the form of a feedback loop) determines whether the aforementioned configuration is actually optimal.

[0117] Since the setup assistant preferably operates as an (interactive) online service, the input variable I and parameter P can be sent to the setup assistant, preferably a cloud-based operator. In the cloud, using AI-supported or knowledge-supported learning methods, weighted matrices Bij, Cijk, ..., or lookup tables or artificial neural networks can be learned or relearned from newly sent and existing experience data from similar real-world applications for training purposes (see step 835). The weighted data obtained in this way for matrices Bij, Cijk, ..., or lookup tables or artificial neural networks can then be transferred to multiple or numerous setup assistants deployed outside the cloud.

[0118] According to the second exemplary embodiment, a multidimensional lookup table can be used, and the parameter is determined by interpolation between two table values ​​corresponding to the input parameter.

[0119] The third exemplary embodiment uses a neural network as a machine learning method: here, the input values ​​are used as node values ​​of the input layer 700 of the neural network, and the node values ​​of the output layer 710 are interpreted as parameters.

[0120] exist Figure 6 In this example, n=4 and m=3, and a single hidden layer 705 is chosen. The value of a node in a layer is determined as a weighted sum of all nodes in the left-hand neighboring layer. During the training phase, the weights are determined such that, for a large number of input values, the known optimal parameters are determined as accurately as possible at the output layer, while minimizing any error. If an unknown input vector is used in the operation (also known as "inference"), the optimal parameter vector is provided at the output layer.

[0121] If new optimal allocations between input variables and parameters are obtained through feedback from commissioning engineers, these can be used to retrain or further train the neural network, thereby improving it.

[0122] In principle, other neural network topologies (such as convolutional neural networks) are also possible. Non-linear propagation functions can also be used to compute nodes from left to right. Furthermore, other machine learning methods (such as random forests) are conceivable.

[0123] As a GenAI approach, for example, a GenAI system using neural networks can be provided (refer to below). Figure 7 (In more detail), the GenAI system is used to automate or pre-set the operating parameters of the underlying UHF RFID system.

[0124] Figure 7 The exemplary embodiment of the method steps of the setting assistant 400 according to the present invention is shown here based on a GenAI-based optimization method or corresponding learning method 405, which, by means of the GenAI-based optimization method or corresponding learning method, determines appropriate settings or pre-settings for a specific application scenario or purpose 410 in the field of use of the UHF RFID system, for example, such as Figure 1 As described as an example. In addition, for a specified application scenario 410, corresponding operation-related and / or product-related data 425 are generated from the (pre)set data 420 determined therefrom through data preprocessing.

[0125] The GenAI-based computer interface 430 is primarily used to record and temporarily store the technical information and framework conditions 435 required for the commissioning of the UHF RFID system for the intended purpose. Pre-configuration is specifically implemented for a particular existing application scenario or purpose 410, i.e., through interaction 440 via the corresponding computer interface with the commissioning engineer 445 acting as a setup assistant 400 user in natural language.

[0126] The objective is entered via a prompt issued to the commissioning engineer, whereby the input includes one or more contexts, which refer to additional technical information regarding the specified objective. The context regarding the specified objective can be processed in terms of data based on associated contextual data / metadata.

[0127] Referring to data 435 collected by the commissioning engineer 445 via input 440, the setup assistant 400 trains a specified transducer-based setup model 450 in a database or data storage device for the corresponding existing UHF RFID system and the corresponding purpose. The setup model 450, which has been pre-trained in this manner and has or includes at least one transducer component described below, is provided to the commissioning engineer 445 via a computer interface 430 for interaction 440.

[0128] The existing setup model 450 is retrained using second training data generated based on the updated setup dataset. The converter-based setup model retrained in this manner is then released via computer interface 430 to the appropriate commissioning engineer 445 for automated (pre) setup of the existing UHF RFID system 415 and the corresponding existing purpose.

[0129] The setup model trained in this manner can now be used to generate a pre-setup dataset suitable for the operation of the existing UHF RFID system 415. This is implemented by the commissioning engineer 445 through interaction with a setup assistant 400 based on or in conjunction with a GenAI system. The commissioning engineer 445 accesses the corresponding existing, trained transducer-based setup model via a computer processor. Input data related to the specified purpose is collected via computer interface 430. The existing current setup model 450 is then prompted to analyze the input data and, based on the current setup model, to provide instructions 460 for generating a new or modified pre-setup dataset 455.

[0130] The UHF RFID system 415 can now be automatically configured for a specified purpose using the newly generated or modified (pre)configuration dataset 455. For this purpose, appropriate prompts can be used to request machine-readable instructions from the setup model based on requests made by the commissioning engineer to the setup model via a computer interface.

[0131] The prompts may include one or more operational instructions related to one or more operational procedures of the UHF RFID system 415 for a specific purpose. The operational instructions may include machine-readable instructions for controlling and / or monitoring the operation of the UHF RFID system 415.

[0132] Advantageously, the commissioning engineer 445 does not need to deal with individual technical parameters and their meanings, as is common in the current prior art. Therefore, there is no need for configurations with individual (pre)set values ​​as practiced in the prior art.

[0133] The methods described for automating or pre-setting the UHF RFID system 415 under discussion here also significantly reduce the effectiveness and efficiency of deployment, especially for more complex UHF RFID systems.

[0134] Figures 8a to 8c This illustrates a known structure of the transformer encoder / decoder architecture used here to generate parameter data based on the aforementioned transformer-based setup model.

[0135] Figure 8aAn embodiment of a transducer encoder architecture is illustrated. The transducer encoder includes an encoder input 678, one or more encoder blocks 674, 614, and an encoder output 676. In this exemplary embodiment, the encoder output 676 includes two processing layers: a linear layer 616 and a so-called "softmax" layer 618. Such linear layers are known to be used to connect each input neuron of a neural network to each output neuron. Such softmax layers are known to be commonly used as the last layer in neural network models for classification tasks, and to convert the raw output values ​​(so-called "logit") into probabilities by forming an exponential function for each output and normalizing these values ​​by dividing them by the sum of all values. This allows a vector of K real values ​​to be transformed into a vector of K real values ​​that sum to 1. The input values ​​can be positive, negative, zero, or greater than one, and the softmax function transforms them into values ​​between 0 and 1, such that these values ​​can be interpreted as probability values.

[0136] The converter encoder architecture shown here can be derived from converter encoder-decoder architectures known in the prior art (see [link]). Figure 8c The converter encoder architecture shown can have an additional encoder output to directly connect the encoder block to the decoder of the converter encoder-decoder architecture, such as... Figure 8c As shown.

[0137] Input data is received at encoder input 678, which can apply a so-called "embedding" 602 to the input data. Applying input embedding 602 can refer to continuing to pass the input data through the embedding layer. Encoder input 678 can also apply positioning coding 604. Applying positioning coding 604 can refer to adding positioning factors to the embedded input data. The input data can preferably specify a sequence of elements, wherein the positioning factors provide an indication of the location of the elements within the sequence.

[0138] The embedded input data can be processed by one of the encoder blocks 674 and 614 shown (optional blocks). The embedded input data can be provided via a dedicated data connection to normalization 608 (hereinafter referred to as "layer normalization") implemented for the corresponding layer or layer of the underlying software architecture. Such layers may involve, for example, application layers (or user interfaces), large language model layers, and data layers intended here for training the model setup.

[0139] For input data embedded in this manner, a so-called "multi-head self-attention" mechanism can be implemented 606, which allows the underlying model to focus on different parts of the input data, similar to how people focus on specific words when understanding sentences. This mechanism allows the model to determine which parts of the input are relevant to a given task, making it extremely flexible and powerful. Therefore, this processing step involving the transformer is used to improve the expressiveness and modeling capabilities of the neural network at the bottom of the AI ​​system. This allows the transformer to capture various types of dependencies and relationships between words or elements in the input data sequence. This processing step can also be understood as a filtering applied to the embedded input data. By applying this filtering to the embedded input data, elements associated with the embedded input data that contribute to the generation of output data can be identified for use in generating the output data. Therefore, this filtering can represent the degree of contribution of the elements associated with the embedded input data to the generated output data.

[0140] For the embedded input data, in addition to the multi-head self-attention 606, a so-called "feedforward" layer 610 and a corresponding layer normalization 612 may be provided. With the help of such a feedforward layer, a feedforward neural network is formed in a manner known per se, consisting of an input layer, one or more hidden layers, and an output layer. Therefore, data flows in only one direction, i.e., from the input layer through the hidden layers to the output layer.

[0141] Figure 8b An embodiment of a converter decoder architecture is illustrated. The converter decoder includes a decoder input 684, one or more decoder blocks 680, 632, and a decoder output 682. In this exemplary embodiment, the decoder output 682 also includes two processing levels: a linear layer 634 and a so-called "softmax" layer 636. Existing converter decoder architectures can be found in the prior art and Figure 8c The transformer encoder-decoder architecture is shown. The transformer decoder architecture can correspond to the decoder architecture associated with the encoder-decoder architecture, regardless of whether the encoder of the transformer encoder-decoder acquires one or more hidden states. Numerous such transformer decoder architectures are available in the prior art, for example, the universally known and already trained "GPT" transformer.

[0142] The decoder input 684 can apply the existing input data embedding 620 and positioning coding 622, which is consistent with... Figure 8a The input embedding 602 and localization coding 604 described in the corresponding context are similar. Decoder block 680 may include corresponding layer normalization 626, masked multi-head self-attention 624, a so-called "feedforward" layer 628, and corresponding layer normalization 630, which is provided by corresponding layer normalization 626 via a dedicated data connection.

[0143] Multi-head self-attention masking 624 can also be applied to embedded input data; this multi-head self-attention masking 624 is essentially the same as... Figure 8a The masking in the context corresponds to multi-head self-attention. However, it includes an additional mask for the portion of the embedded input data associated with elements in the sequence that appear later than the element to be generated.

[0144] Additionally or alternatively, the portion of the input data associated with elements in the sequence that appear later than the element to be generated may not be available or received, and / or may not be converted into embedded input data. Therefore, the transformer encoder may be configured for classification tasks, while the transformer decoder may be configured only for text generation.

[0145] Figure 8c The illustration shows the cooperation and interaction between the described converter encoder and converter decoder.

[0146] The transducer encoder-decoder may include an encoder input 688, one or more encoder blocks 686, 664, a decoder input 694, a decoder block 690, and a decoder output 692. In this exemplary embodiment, the decoder output 692 also includes two processing layers: a linear layer 644 and a so-called "softmax" layer 646. The encoder input 688 may be connected to... Figure 8a The encoder input 678 corresponds to this. One or more encoder blocks 686, 664 can be used with... Figure 8a One or more encoder blocks 674, 614 correspond to this. Decoder input 694 can be associated with... Figure 8b This corresponds to the decoder input 684 in the code. The decoder block 690 may have a masked multi-head self-attention 670, layer normalization 672, the aforementioned feedforward layer 638, and layer normalization 640. These can be associated with, for example... Figure 8b The mask multi-head self-attention 624, layer normalization 626, 630, and feedforward layer 628 described in the context are configured in a similar manner. The decoder block 690 may also include multi-head self-attention 650 and layer normalization 648.

[0147] and Figure 8b Similar to the description, the context tensor can be obtained from mask multi-head self-attention 670 and layer normalization 672. Layer normalization 648 can be applied to the context vector obtained from multi-head self-attention 650 and the hidden states of one or more encoder blocks 686, 664. The context vector generated from layer normalization 648 can be compared with... Figure 8b The description is similarly processed via the feedforward layer or layer 638 using the aforementioned layer normalization 640.

[0148] The context vector generated from layer normalization 640 can be provided to another decoder block 642 similar to decoder block 690. The context vector obtained from one or more decoder blocks 690, 642 can be provided to decoder output 692. Decoder output 692 can be... Figure 8b The decoder output corresponds to 682.

[0149] Using the described architecture, the converter encoder-decoder can obtain or receive input data at encoder input 688 and one or more encoder blocks 686, 664, and decoder block 690 and decoder output 692, and process them as described. Based on the input data, the converter encoder-decoder can sequentially generate output data. The output data generated sequentially in this manner can be provided to decoder input 694, one or more decoder blocks 690, 642, and decoder output 692, and / or processed by them.

[0150] The sequence can preferably be provided to encoder input 688. After at least a portion of the output data has been generated, decoder input 694 can then be supplied to at least a portion of the elements of the already generated output data. This allows for the generation of subsequent elements of the output data with greater accuracy, taking into account both the input data and the generated output data.

Claims

1. A computer-implemented method for automatically setting parameters of a UHF RFID system (100, 105, 300 to 315) for contactless data exchange between RFID transponders (125, 305 to 315) on an object (101) having RFID transponders and moving past RFID readers / writers (120, 200), wherein at least one UHF antenna (102 to 104, 300) of the RFID transponders (125, 305 to 315) arranged on these objects (101) is provided for contactless reading and / or writing, the computer-implemented method being characterized by providing a knowledge-based setup assistant (400, 500) by means of which, based on technical information, the UHF transponders are determined to be suitable for the existing UHF RFID system (100, 105, 300 to 315) to be put into operation and its intended use. The settings of the parameters of the RFID system (100, 105, 300 to 315).

2. The method according to claim 1, characterized in that, The knowledge-based setup assistants (400, 500) are supported by or based on artificial intelligence (AI) or experiential knowledge.

3. The method according to claim 1 or 2, characterized in that, The expected operating conditions are derived with reference to the intended use of the UHF RFID systems (100, 105, 300 to 315).

4. The method according to one or more of the preceding claims, characterized in that, The appropriate parameters for the UHF RFID system (100, 105, 300 to 315) are determined by referring to the dialogue between the setup assistant (400, 500) and the commissioning engineer (110, 445), wherein the setup assistant (400, 500) is provided with the technical information about the UHF RFID system (100, 105, 300 to 315) to be commissioned.

5. The method according to claim 4, characterized in that, The technical information provided to the setup assistant (400, 500) relates to the number of existing RFID transponders (125, 305 to 315), and / or the possible location of the RFID transponders (125, 305 to 315), and / or the moving speed of the object (101) having an RFID transponder and moving past the RFID reader / writer (120, 200), and / or the number and / or type and / or location and / or cable length of the at least one UHF antenna (102 to 104, 300), and / or the electromagnetic environment of the UHF RFID system (100, 105, 300 to 315).

6. The method according to one or more of the preceding claims, characterized in that, The setup assistant (400, 500) or the commissioning engineer (110, 445) performs UHF-based RFID measurements to obtain or determine information about the quality of communication or contactless data exchange between the RFID reader / writer (120, 200) and the RFID transponder (125, 305 to 315) in the event of parameter and positioning variations of the RFID transponder (125, 305 to 315).

7. The method according to claim 6, characterized in that, Reference power scan, and / or positioning scan, and / or real-time scan implementation.

8. The method according to one or more of the preceding claims, characterized in that, The setup assistants (400, 500) perform test runs on the already operational UHF RFID systems (100, 105, 300 to 315) to assess the accuracy and / or reliability of the operation of the UHF RFID systems (100, 105, 300 to 315), wherein the setup assistants (400, 500) determine improved settings based on the results of the test runs and with reference to further iterations.

9. The method according to one or more of the preceding claims, characterized in that, The parameter settings of the UHF RFID systems (100, 105, 300 to 315) that have already been put into operation are continuously improved over time and with the use of the UHF RFID systems (100, 105, 300 to 315) by leveraging self-learning AI algorithms.

10. The method according to one or more of the preceding claims, characterized in that, The method is implemented using a setup assistant (400, 500) with a large language model (LLM) and based on GenAI (405), wherein the method includes the following steps: - Train (505, 535) a transducer-based setup model by referring to the operation-related data (425) of the UHF RFID system (100, 105, 300 to 315), or optimize an existing transducer-based setup model by referring to such data; - Provide a trained or optimized converter-based setup model via a computer interface, the converter-based setup model including at least one converter component; - The commissioning engineer (110, 445) communicates with the setup assistant (400, 500) (515, 520) via the computer interface to collect data about the UHF RFID system (100, 105, 300 to 315) for a specified purpose; - Use collected data about the specified purpose to prompt (525) the trained transformer-based setup model to output optimized setup data; - Release the optimized setup data output by the trained transducer-based setup model for the automated setup or pre-setting of the UHF RFID system (100, 105, 300 to 315).

11. The method according to claim 10, characterized in that, The training (505, 535) of the configured model is carried out using deep learning and / or reinforcement learning techniques.

12. The method according to claim 10 or 11, characterized in that, The prompts are used to prompt the commissioning engineer (110, 445) to request the trained converter-based setup model to provide machine-readable instructions based on the request made by the commissioning engineer (110, 445) to the setup model via the computer interface.

13. The method according to one or more of claims 10 to 12, characterized in that, The prompts include one or more operational instructions related to one or more operational procedures of the UHF RFID system (100, 105, 300 to 315) for the specified purpose. Additionally, the operational instructions may include machine-readable instructions for controlling and / or monitoring the operation of the UHF RFID system (100, 105, 300 to 315).

14. The use of a trained transducer-based setup model for generating setup data suitable for a specified purpose for setting up a UHF RFID system (100, 105, 300 to 315), characterized in that: step: - Provides commissioning engineers (110, 445) with access to a trained converter-based setup model using a computer processor; - Receive input data related to the specified purpose via a computer interface (515, 520); - Tip (525) The trained transformer-based setup model analyzes the input data and provides instructions for generating new or modified setup data based on the trained setup model.

15. A setup assistant (400, 500) for automatically setting parameters of a UHF RFID system (100, 105, 300 to 315) for contactless data exchange between RFID transponders (125, 305 to 315) on an object (101) having RFID transponders and moving past RFID readers / writers (120, 200), wherein a UHF antenna (102 to 104, 300) arranged on the RFID transponders (125, 305 to 315) on these objects (101) is provided for contactless reading and / or writing, the setup assistant being characterized in that, according to the method of one or more of the preceding claims, the UHF antenna is determined based on technical information regarding the existing UHF RFID system (100, 105, 300 to 315) to be put into operation and its intended use, using empirical knowledge or artificial intelligence (AI). Appropriate settings of the parameters of the RFID system (100, 105, 300 to 315).

16. The setup assistant (400, 500) according to claim 15, characterized in that, The expected operating conditions of the UHF RFID systems (100, 105, 300 to 315) are obtained with the aid of artificial intelligence and reference to the intended use of the UHF RFID systems (100, 105, 300 to 315).

Images