Method for generating training dataset

Abstract

To accurately determine a position of a measuring device.SOLUTION: A computer-implemented method for generating a training dataset to train an artificial intelligence for operating a measuring device (100), in particular a wall diagnostic device, includes: recording sensor data of at least one sensor unit (101) of the measuring device; determining a position of the measuring device relative to a wall (105); generating position-specific sensor data by a position determination system; labeling the position-specific sensor data taking into account ground truth information; generating labeled sensor data; and aggregating the labeled sensor data into a training dataset. Also disclosed is a method for training an artificial intelligence.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a method for generating a training data set for training an artificial intelligence for a wall diagnostic device.Furthermore, the present invention relates to a method for training an artificial intelligence for a wall diagnostic device. [Background technology]

[0002] From the prior art, diagnostic devices are known for diagnosing walls and for detecting objects arranged in walls. Summary of the Invention [Problem to be solved by the invention]

[0003] It is an object of the present invention to provide an improved method for generating training data sets for training the artificial intelligence of a wall diagnostic device, and an improved method for training such an artificial intelligence. [Means for solving the problem]

[0004] This problem is solved by the methods of the independent claims. Preferred embodiments are the subject of the dependent claims.

[0005] In one aspect, there is provided a computer-implemented method for generating a training data set for training an artificial intelligence for operating a measurement device, in particular a wall diagnostic device, the method comprising: recording sensor data of at least one sensor unit of the measuring device, the sensor data reflecting the wall to be diagnosed; a position determination of the measuring device relative to the wall is performed, and position-related sensor data is generated by a position determination system, wherein position information is assigned to each recorded sensor data, the position information defining a position of the measuring device relative to the wall at which the measuring device was positioned relative to the wall at the time of recording of the sensor data; The method includes labeling the location-related sensor data and generating the labeled sensor data in consideration of ground truth information, wherein the ground truth information includes at least one piece of information regarding the presence of an object in the wall; and The labeled sensor data is combined to form a training data set.

[0006] This provides the technical advantage of providing an improved method for generating training data sets for training an artificial intelligence for operating a measuring device, in particular a wall diagnostic device, by first recording sensor data of at least one sensor unit of the measuring device, in particular the wall diagnostic device, the recorded sensor data reflecting the wall to be diagnosed and, if applicable, an object arranged in the wall.

[0007] Furthermore, a position determination of the measuring device relative to the wall is performed and position-related sensor data is generated by the position determination system, which position-related sensor data comprises information of the recorded sensor data and is additionally extended by position information, which defines the position of the measuring device relative to the wall in which the measuring device was positioned at the time of recording the sensor data.

[0008] Subsequently, the location-related sensor data is labeled, taking into account ground truth information. The ground truth information includes at least one piece of information about the presence of an object in the wall. In this way, the ground truth information describes the actual presence of the object in the wall. The labeling of the location-related sensor data involves identifying the sensor data related to the respective presence or position of the object.

[0009] By labeling in this way, the sensor data is at least identified as to whether or not the respective identified sensor data reflects an object located in the wall. Determining the position of the measuring device at the time of recording the sensor data allows labeling of the sensor data in relation to the presence of an object in the wall.

[0010] The ground truth information describes where within the wall each object is positioned. In order to assign such information to the sensor data recorded by the measurement device, location information of the sensor data is required, which defines where the measurement device was positioned at the time of recording the sensor data.

[0011] The correspondingly labeled sensor data of the training data set allows training of an artificial intelligence of the measuring device, in particular of the wall diagnostic device, which is trained to detect objects located in the wall based on the sensor data of the measuring device, where detection includes recognizing whether an object is located at a position on the wall.

[0012] In one embodiment, the ground truth information includes classification information regarding the object position and / or object type of the object and / or object depth information regarding the object depth inside the wall and / or object extension information regarding the object extension of the object.

[0013] This provides the technical advantage that additional information can be incorporated into the training data set, by taking into account additional classification information regarding the object position, object type, object depth information, and / or object extension of the objects, such that additional information allows for additional training of the artificial intelligence, whereby the artificial intelligence can be trained in particular to determine the object position, object type, object depth, and / or object extension of the detected objects based on the sensor data of the measuring device.

[0014] In such cases, the corresponding ground truth information describes the actual object position of the object within the wall, the actual object type of the object, the actual object depth of the object within the wall, or the actual object extension of the object.

[0015] The above ground truth information can be incorporated into the training data set or recorded sensor data with actual knowledge of objects placed within the wall.

[0016] Ground truth information, in the broadest sense of the present application, describes the actually existing condition of the wall to be inspected.

[0017] In one embodiment, the position determination system includes at least one camera sensor and at least one position marking, the position marking being formed on a surface of a wall, the camera sensor being positioned at a predefined perspective positioning with respect to the wall, and performing a position determination of the measuring device relative to the wall includes: recording camera data of the camera sensor during recording of sensor data of the sensor unit of the measuring device, the camera data reflecting the measuring device positioned at the wall and the position markings arranged on the wall; determining a relative position of the measuring device relative to the position markings based on the camera data; and The position of the measuring device relative to the wall is determined based on the relative position.

[0018] This provides the technical advantage of being able to determine the precise position of the measuring device relative to the wall. To this end, camera data from a camera sensor is first recorded. The camera data reflects the measuring device being positioned on the wall during the recording of the sensor data. Furthermore, the camera data reflects position markings formed on the wall. Based on the camera data, the time-resolved relative position of the measuring device relative to the position markings can be determined, taking the position markings into account.

[0019] Based on this, the time-resolved position of the measuring device relative to the wall can finally be determined. This method allows for accurate positioning of the measuring device relative to the wall. The time stamps of the camera data allow for accurate determination of the position of the measuring device relative to the wall for multiple points in time during the recording of sensor data by the measuring device.

[0020] 10. The method according to claim 9, wherein the position determination is performed further comprising: a time synchronization between the recording of the sensor data of the sensor unit and the recording of the camera data of the camera sensor is performed; and This includes determining position information of the measuring device relative to the wall for the sensor data of each of the sensor units, taking into account the time synchronization.

[0021] This provides the technical advantage of enabling accurate generation of position-related sensor data. Taking into account the timestamps of the camera data, the respective position of the measuring device relative to the wall can be determined for multiple points in time. The timestamps of the sensor data allow the time at which specific sensor data was recorded by the measuring device to be determined. Time-synchronized timestamps of the sensor data and the camera data allow the position of the measuring device relative to the wall at the time the sensor data was recorded to be determined for each recorded sensor data. In this way, accurate position information can be generated for each recorded sensor data, defining in which position the measuring device was positioned during the recording of the respective sensor data.

[0022] In one embodiment, the location determination further comprises: determining a perspective of the camera sensor relative to the wall based on a predefined perspective position at which the camera sensor is positioned relative to the wall; To take into account the perspective position of the camera sensor relative to the wall, a perspective correction is performed to match the relative position of the measuring device relative to the position marking, determined based on the camera data, and the position of the measuring device on the wall is determined based on the perspective correction.

[0023] This provides the technical advantage that the perspective of the camera sensor relative to the wall is taken into account, allowing a perspective correction of the camera data to be performed. The perspective correction allows for perspective-induced errors in determining the position of the measuring device relative to the wall based on the camera data to be eliminated. This allows for accurate position determination relative to the wall based on the camera data.

[0024] In one embodiment, the position markings are formed as ArUco markings or ChArUco markings.

[0025] This provides the technical advantage that the position marking formed by the ArUco or ChArUco marking allows an accurate determination of the relative position of the measuring device relative to the position marking based on camera data. Furthermore, the ArUco or ChArUco marking can be attached to a wall as a correspondingly formed poster, without the need to modify the wall for this purpose.

[0026] As ArUco marking or ChArUco marking, it is possible to use ArUco marking or ChArUco marking known from the prior art.

[0027] In one embodiment, the location markings are formed by markings formed on walls, including wallpaper patterns, light switches, outlets, windows, and furniture.

[0028] This provides the technical advantage that elements already formed on the wall can be used as position markings, thereby eliminating the need for position markings to be applied to the wall from the outside.

[0029] In one embodiment, the camera sensor is disposed on a positioning device through which the camera sensor is positioned at a predetermined perspective position relative to the wall.

[0030] This provides the technical advantage that the camera sensor of the positioning device can be positioned so that it can be accurately positioned at a predetermined perspective position relative to the wall. Because the perspective position is already predefined and set, the required perspective correction of the camera data can be minimized. This simplifies the positioning of the measuring device relative to the wall based on the camera data.

[0031] In one aspect, a training dataset for training artificial intelligence of a measurement device for wall diagnosis is provided, the training dataset being generated by a method for generating a training dataset for training artificial intelligence for operating a measurement device based on one of the above embodiments.

[0032] In one aspect, a computer-implemented method for training artificial intelligence of a measurement device for wall diagnostics is provided, the method comprising: a training dataset is provided by implementing a method for generating a training dataset according to one of the above embodiments; The method includes training an artificial intelligence to perform object recognition of objects configured within the wall based on the training dataset, the object recognition including at least object detection and object classification.

[0033] In one aspect, there is provided a computing unit set up to perform a method for generating a training dataset for training an artificial intelligence for operating a measurement device and / or a method for training an artificial intelligence of a measurement device according to one of the above embodiments.

[0034] In one aspect, a computer program product is provided comprising instructions that, when executed by a data processing unit, instruct the data processing unit to perform a method for generating a training dataset for training an artificial intelligence for operating a measurement device and / or a method for training an artificial intelligence of a measurement device, based on one of the above embodiments.

[0035] Embodiments of the present invention will now be described with reference to the following drawings, in which: [Brief explanation of the drawings]

[0036] [Figure 1] FIG. 1 is a schematic diagram illustrating a measurement device according to one embodiment. [Figure 2] FIG. 10 is another schematic diagram showing a measurement device according to another embodiment. [Figure 3] FIG. 10 is another schematic diagram showing a measurement device according to another embodiment. [Figure 4] FIG. 1 is a schematic diagram illustrating a measurement performed by a measurement device according to one embodiment. [Figure 5] FIG. 10 is another schematic diagram showing a measurement device according to another embodiment. [Figure 6] FIG. 1 is a schematic diagram illustrating a system for generating a training data set according to another embodiment. [Figure 7] FIG. 2 is a schematic diagram showing data recorded by a measurement device. [Figure 8] FIG. 2 is another schematic diagram illustrating a system for generating a training data set according to another embodiment. [Figure 9]1 is a flowchart of a method for generating a training data set according to one embodiment. [Figure 10] 10 is a flowchart of a method for training artificial intelligence of a measurement device according to another embodiment. [Figure 11] FIG. 1 is a schematic diagram illustrating a computer program product. DETAILED DESCRIPTION OF THE INVENTION

[0037] FIG. 1 shows a schematic diagram of a measurement device 100 according to one embodiment.

[0038] The present invention relates to a measuring device, in particular to a wall diagnostic device for inspecting a wall 105 to be processed. Wall diagnostic devices are known in the prior art that are used to detect objects located in the wall. Devices of this kind allow a user to inspect the wall to be processed for the presence of objects located therein and, based on this, to carry out planned operations, such as drilling holes in the wall, in order to avoid damaging the objects located therein.

[0039] In the illustrated embodiment, the measuring device 100 includes a housing 150 having a grip 152 for a user to hold the measuring device 100, a display unit 111 for displaying the wall diagnostic results 109, and an operating member 154 for switching the measuring device 100 into various operating modes.

[0040] According to the invention, the measuring device 100 comprises at least one radar sensor unit 101. The radar sensor unit 101 makes it possible to emit a radar signal in the direction of a wall 105 to be inspected and to receive the radar signal reflected from the wall 105.

[0041] The radar sensor unit 101 may be configured as a narrowband radar detector, for example in the frequency range from 2.4 GHz to 2.4835 GHz, or as an ultra-wideband radar detector, for example in the frequency range from 1.8 GHz to 5.8 GHz.

[0042] The measuring device 100 further comprises a diagnostic module 107 executable in the computing unit 151 of the measuring device 100 for performing a wall diagnosis. The diagnostic module 107 is set up to perform a corresponding diagnosis of the wall to be inspected on the basis of the radar data 103 of the radar sensor unit 101, which reflect the wall 105 to be inspected and possibly an object 113 arranged in the wall 105.

[0043] The wall diagnosis performed by the diagnosis module 107 then includes at least performing object recognition, where object recognition includes object detection and object classification of an object 113 located in the wall 105. The object detection then includes at least determining an object position 115, where the object position represents the positioning of the object located in the wall 105 relative to a reference system defined by the measurement device 100. The object classification of the detected object 113 includes at least determining an object type 117 of the detected object 113.

[0044] The diagnostic result of the wall diagnosis determined in this way, i.e. at least the determined object position 115 and / or the determined object type 117 of the object 113 arranged in the wall 105, is subsequently displayed to the user of the measuring device 100 on a display unit 111 of the measuring device 100. The display unit 111 can be configured, for example, as a suitable display and can display the diagnostic result 109 optically. In addition, the display of the diagnostic result 109 can be supported by acoustic and / or tactile signals. The tactile signals can be embodied, for example, by suitable vibration signals.

[0045] The object 113 can then be displayed on the display, for example, by a corresponding symbol. The object 113 can then be displayed on the display at a corresponding object position 115. The object extension 121 can be visualized by a corresponding size of the displayed symbol. The respective object type 117 of the object 113 can be visualized by a corresponding concept, by color highlighting of the symbol, or by a special shape of the symbol representing the object 113.

[0046] Alternatively, the wall diagnosis may additionally comprise the determination of a wall type 123 in the form of a wall type classification of the wall 105 to be inspected, where the wall type 123 represents the respective type of the wall 105 to be inspected. The wall type may for example be assigned to a corresponding wall type classification, which may include: concrete wall, light / dry wall, masonry wall and / or stacked stone, underfloor heating, wall heating or similar wall types found in buildings.

[0047] In one embodiment, the diagnostic module 107 is further set up to determine an object depth 119 of the object 113 inside the wall 105 based on the radar data 103. The object depth 119 is then defined by the distance from the object configured inside the wall 105 to the surface of the wall 105. This distance may be defined on the object side, for example with respect to the object surface or with respect to the object center point. The distance to the surface of the wall 105 represents the shortest distance defined in a direction perpendicular to the surface of the wall 105.

[0048] In one embodiment, the diagnostic module 107 is further set up to determine an object extension 121 of the object 113 in at least one predefined direction based on the radar data 103. The object extension 121 of the object 113 then represents the spatial extension of the object 113 in at least one spatial direction, preferably in two spatial directions, particularly preferably in three spatial directions. The object 113 can thereby be described as a one-dimensional, two-dimensional or three-dimensional object 113.

[0049] In a typical application, the measuring device 100 is placed on the surface of a wall 105 to be inspected. A radar signal is transmitted through the radar sensor unit 101 in the direction of the wall 105, and a radar signal reflected by the wall 105 or an object 113 located behind it is received. Based on this radar data 103 from the radar sensor unit 101, the diagnostic module 107 performs the wall diagnosis described above and determines a corresponding diagnostic result 109.

[0050] The diagnosis result 109 may include, for example, an object location 115 and / or an object type 117 of an object 113 located in the wall 105. Alternatively or additionally, the diagnosis result 109 may include a wall type 123 of the wall 105, and / or an object depth 119 and / or an object extension 121 of the object 113.

[0051] The diagnostic result 109 configured in this way can subsequently be displayed to the user of the measuring device 100 on a display unit 111 of the measuring device 100. The display unit 111 can be configured, for example, as a suitable display. The diagnostic result 109 can be displayed on the display unit 111 in graphical or textual form.

[0052] In one embodiment, the measuring device 100 further includes a motion detection unit 141. The motion detection unit 141 can detect the motion of the measuring device 100 relative to the wall 105. To this end, the motion detection unit 141 can have, for example, at least one roller member. When the roller member rests on the wall surface of the wall 105, the motion of the measuring device 100 can be detected relative to the wall 105 when the measuring device 100 moves along the motion direction 153 due to the rolling of the roller member. Alternatively, the motion detection unit 141 can have any other configuration capable of detecting the relative motion of the measuring device 100 relative to the wall 105.

[0053] By moving the measuring device 100 relative to the wall 105, the radar data 103 of the radar sensor unit 101 can be recorded for a number of different positionings of the measuring device 100 relative to the wall 105. This allows for the inspection of the wall 105 over a larger spatial area than is given by the range of action of the radar sensor unit 101. This allows for the detection of objects 113 having a spatial extension greater than the range of action of the radar sensor unit 101.

[0054] During the movement of the measuring device 100 along the direction of movement 153, radar data 103 of the radar sensor unit 101 can be continuously recorded. Based on such radar data 103 during the movement of the measuring device 100 along the direction of movement 153, a wall diagnosis can be evaluated by the diagnostic module 107. This allows for a fast wall diagnosis that takes into account the positioning of the measuring device 100 relative to the wall 105.

[0055] In this embodiment, the diagnostic module 107 is configured as a correspondingly trained artificial intelligence 125, which is trained at least to perform the wall diagnostics described above based on the radar data 103 of the radar sensor unit 101 and to determine at least the object position 115 and the object type 117 of an object 113 located inside the wall 105. The determination of the object classification or object type 117 includes assigning the detected object 113 to a predefined object classification.

[0056] The object types may include: metal / non-metallic objects, cables for low voltage, cables carrying single-phase AC signals, cables carrying polyphase AC signals, wooden supports, metal supports, plastic pipes, water-filled plastic pipes, e.g., water pipes, non-water-filled plastic pipes, e.g., sewer pipes, or other components normally attached to building walls.

[0057] Furthermore, the artificial intelligence 125 may be trained to determine the wall type 123 of the wall 105 to be inspected based at least on the radar data 103 of the radar sensor unit 101. The possible wall types 123 may include: concrete walls, light / dry construction walls, masonry walls and / or individual stones in masonry walls, underfloor heating, wall heating, or any other wall type typically installed in buildings.

[0058] In one embodiment, the measurement device 100 may include additional sensors in addition to the radar sensor unit 101, which may detect additional physical quantities. For example, the measurement device 100 may include an inductive sensor and / or an eddy current sensor and / or a capacitance sensor and / or an AC sensor and / or an NMR sensor and / or an ultrasonic sensor, or any other sensor typically mounted in a wall diagnostic device.

[0059] The diagnostic module 107, in particular a correspondingly trained artificial intelligence 125, can then be set up to perform the wall diagnostics described above based on the radar data 103 of the radar sensor unit 101 and taking into account additional sensor information from other sensors. For this purpose, additional information from the aforementioned additional sensors can be used, in particular for object recognition of objects 113 arranged in the wall 105. Through the additional sensor information, an improved detection of the objects 113 and possibly an improved classification of the objects 113 can possibly be achieved.

[0060] In particular, illustratively, the classification of the material of the object 113, for example as a metallic or non-metallic material, can be improved by utilizing additional sensor information.

[0061] FIG. 2 shows another schematic diagram of a measurement device 100 according to another embodiment.

[0062] In the illustrated embodiment, the measuring device 100 includes a pre-processing module 127 in addition to the diagnostic module 107. For wall diagnosis, the measuring device 100 first receives radar data 103 from the radar sensor unit 101. Pre-processing of the received radar data 103 is carried out via the pre-processing module 127. Through pre-processing by the pre-processing module 127, the radar data can be arranged, for example, into a corresponding data structure required for wall diagnosis by the diagnostic module 107.

[0063] As explained above, during the wall diagnosis, the diagnostic module 107 generates the above-mentioned diagnostic result 109. The diagnostic result 109 may then include, for example, the object position 115 and / or object type 117 and / or object depth 119 and / or object extension 121 of an object 113 arranged in the wall 105 to be inspected, and / or the wall type 123 of the wall 105 to be inspected. The correspondingly generated diagnostic result 109 can then be displayed on the display unit 111 of the measuring device 100.

[0064] In one embodiment, in addition to the radar data 103 of the radar sensor unit 101, the above-described additional sensor information of the additional sensors can be taken into account in the wall diagnosis of the diagnostic module 107. A corresponding pre-processing of the additional sensor information can be performed accordingly by the pre-processing module 127.

[0065] In the illustrated embodiment, the diagnostic module 107 includes a wall-type classification module 129 and an object recognition module 131. The preprocessing module 127 includes a first preprocessing module 135 and a second preprocessing module 137. The first preprocessing module 135 includes an S-matrix reduction 155. The second preprocessing module 137 includes a background correction 157, an inverse fast Fourier transform 159, and focusing and migration 161. The preprocessing of the radar data 103 by the preprocessing module 127 begins with the preprocessing of the radar data 103 by the first preprocessing module 135 and the S-matrix reduction 155 included therein.

[0066] The first pre-processing module 135 then generates input data 133 based on the radar data 103. The input data 133 serves as input data for the wall type classification module 129. The wall type classification module 129 then performs wall type classification of the wall 105 to be inspected based on the input data 133 to generate wall type information 139. The wall type information 139 includes the wall type 123 of the wall 105 to be inspected determined in the wall type classification.

[0067] Subsequently, a second pre-processing module 137 performs pre-processing based on the radar data 103 and the wall type information 139. At this time, a background correction 157 of the radar data 103 is performed taking into account the wall type 123 contained in the wall type information 139. Depending on the wall type 123 of the wall 105 to be inspected, different effects on the radar data 103 may occur.

[0068] A background correction 157 can correct such effects that may affect object recognition, which mainly depend on the respective wall type 123. After the background correction has been performed, another preprocessing can be performed, such as an inverse fast Fourier transform 159 or focusing and migration 161, to prepare new input data 133 for the object recognition module 131. Based on the input data 133 provided by the second preprocessing module 137, the object recognition module 131 performs object recognition of objects 113 located in the wall 105 to be inspected, and determines at least the object position 115 and the object type 117 of each object 113. In addition, the object recognition module 131 can determine the object depth 119 and the object extension 121.

[0069] In one embodiment, the diagnostic module is further set up to determine an object depth of an object inside the wall based on the radar data, the object depth being defined as a distance from an object configured within the wall to a surface of the wall.

[0070] Preprocessing is optional here. Depending on the algorithm applied to the diagnostic module 107, completely unprocessed radar echoes of various frequencies can be used as radar data 103 and as input data for the diagnostic module 107. Alternatively, radar data 103 that has been processed in advance through several steps can be used. These preprocessing steps include, for example, transforming the signal from frequency space to time or distance space, background removal, denoising and normalizing the signal, etc. For radar data 103 that exists in the form of complex numbers, only absolute values ​​can be processed. Alternatively or additionally, phase information can be considered.

[0071] FIG. 3 shows another schematic diagram of a measurement device 100 according to another embodiment.

[0072] In the illustrated embodiment, the diagnostic module 107 includes multiple processing paths 102 running in parallel. Each processing path 102 includes a diagnostic module 107 including a pre-processing module 127, e.g., a wall-type classification module 129 and / or an object recognition module 131 according to the embodiment of FIG. 2, and a post-processing module 163.

[0073] 3, radar data 103 is primarily shown as input data for wall diagnosis. However, in addition to the illustrated radar data, additional information from additional sensors can also serve as input data for wall diagnosis. In this case, different information from different sensor types can be processed in different parallel processing paths 102, and corresponding wall diagnosis can be performed separately based on the different sensor information. After wall diagnosis is completed, the individual partial analysis results can be integrated through an integration module and combined into a diagnosis result 109 for wall diagnosis.

[0074] Alternatively or additionally, different processing paths 102 may perform different aspects of wall diagnostics based on the same sensor information.

[0075] The individual processing paths 102 can then process different radar data 103, which have been recorded for different positions of the measuring device 100 relative to the wall 105, for example during the movement of the measuring device 100 relative to the wall 105. The radar data 103, which have been recorded successively in time during the movement of the measuring device 100 relative to the wall 105 and which reflect different areas of the wall 105, can then be processed in the various processing paths 102 by the modules shown.

[0076] The various processing paths then perform independent wall diagnostics, which include at least determining the object location 115 and / or object type 117 of an object 113 located within the wall 105 .

[0077] The integration module 165 allows the integration of the partial results of the independent wall diagnosis of different areas of the wall 105 provided in the individual processing paths 102 into an associated diagnostic result 109, which then describes the wall diagnosis of the relevant spatial area reflected by the corresponding radar data 103 passed through and recorded during the movement of the measuring device 100 relative to the wall 105. This parallel processing of the radar data 103 or additional sensor information 104 of additional sensor elements in the various processing paths 102 allows for accelerated wall diagnosis.

[0078] Alternatively, different functions of the wall diagnostics may be performed in different processing paths 102. For example, a processing path 102 may perform wall type classification of a wall 105 to be inspected and determination of the wall type 123. Another processing path 102 may perform object recognition of an object 113 located in the wall. Object detection together with determination of object location 115 and object classification together with determination of object type 117 may then be performed in one processing path 102.

[0079] Alternatively, object detection and object classification can be performed in two separate processing paths 102. The separate processing paths 102 can each trigger an object depth determination, i.e., a determination of the object depth 119 and / or a determination of the object extension 121. In an integration module 165, various partial results of the wall diagnosis can be integrated into a corresponding diagnosis result 109.

[0080] The diagnostic module 107 may then be divided into different artificial intelligences 125, as already shown in the embodiment of Fig. 2. The diagnostic module 107 may then include, for example, a wall type classification module 129 and an object recognition module 131. The object recognition module may be further divided into an object detection module and an object classification module. The diagnostic module 107 may further include an object depth determination module and an object extension module, which are set up to determine the object depth 119 and the object extension 121, respectively.

[0081] Each corresponding module may be configured as an independent artificial intelligence 125, for example as a neural network, or alternatively, the various modules may form parts of an overall artificial neural network, which are connected to form the overall neural network according to structures known from the prior art.

[0082] FIG. 4 shows a measurement schematic diagram of the measurement device 100 according to one embodiment.

[0083] For preprocessing, the radar data 103 or the additional sensor information 104 of other sensors can be normalized, in particular for numerical stabilization of the subsequent steps performed by the diagnostic module 107 during wall diagnosis. For this purpose, for example, amplitude and / or offset compensation can be performed. Furthermore, the radar data 103 can be filtered to reduce disturbances, and the corresponding sensor data can be downsampled to reduce the data rate. Furthermore, the radar data 103 or the additional sensor information 104 can be transformed into the required frequency domain or time domain, respectively. For this purpose, methods known from the prior art can be applied.

[0084] Furthermore, the recorded radar data 103 or additional sensor information 104 can be divided into temporal or spatial windows 167. The temporal windows 167 can be generated by recording the radar data 103 or additional sensor information or preprocessed radar data 103 over a fixed time interval, whereas the spatial windows 167 can be generated by assigning the radar data 103 or additional sensor information 104 to the position of the measuring device 100 relative to the wall 105 along the direction of movement 153.

[0085] Diagram a) of Figure 4 shows such a data matrix resulting from the steps described above. The data matrix in window 167 shown in diagram a) shows multiple sensor data, which may include, for example, radar data 103 or additional sensor information 104 from other sensors, plotted along a frequency channel axis 171 or a space / time axis 169.

[0086] The width of the time window 167 can then be selected to compensate for the different sampling rates of the sensors and to provide new windows 167 sufficiently frequently, so that the display of the wall diagnostic results 109 on the display unit 111 can be performed without excessive time delay while the measurement is being carried out or immediately after the measurement of the measuring device 100 has been completed.

[0087] For this purpose, a rate of 2 to 20 windows per second of data recording of sensor data may be preferred. For spatial windows, the spatial sampling rate can be selected to achieve the desired position accuracy. A sampling rate of 1 mm to 1 cm may then be preferred. This means that for every 1 mm to 1 cm of movement of the measuring device 100 along the direction of movement 153, the corresponding sensor data is recorded.

[0088] The width of the spatial window 167 can be selected so that the window contains information related to the object 113. A width of 1 cm to 20 cm for each spatial window 167 may then be preferred, resulting in 4 to 100 measurements per window 167. This allows for more efficient algorithmic processing of the correspondingly recorded radar data 103 or additional sensor information by the diagnostic module 107.

[0089] The next temporal or spatial window 167 can then be provided as soon as one or more sampling points become available.

[0090] The diagnostic module 107 may be configured to record as input data for each processing path 102, for example in the embodiment of Fig. 3, a matrix corresponding to the window size of the respective spatial or temporal window 167. In this case, the corresponding input data may comprise preprocessed sensor data, i.e. radar data 103 and additional sensor information 104 of the additional sensor, in accordance with the embodiment of Fig. 2.

[0091] As explained above, the wall diagnosis by the diagnostic module 107 can be performed on the basis of an appropriately trained artificial intelligence. Alternatively, the various processing paths can also be calculated by rule-based algorithms. Within the processing path 102, a combination of artificial intelligence and rule-based algorithms is also possible in the form of a parallel circuit or a concatenation.

[0092] The wall diagnosis result 109 can be expressed as a numerical value, a vector, or a matrix. Furthermore, for object detection, the probability of detection can be displayed, or for wall type classification or object classification, the probability of the displayed object type or wall type type. The same can also be applied to the position determination and / or depth determination, which can also display corresponding probability values.

[0093] If, in addition to radar data 103, additional sensor information from another sensor type is also processed in processing path 102, these can be integrated within artificial intelligence 125 or combined by rule-based combination.

[0094] In the post-processing of each processing path 102 of the embodiment of Figure 3, the results of multiple algorithms based on multiple windows 167 can be combined by a combination module 165. Such combination can be realized by, among other things, majority formation, sum formation, or multiplication of successive probability values.

[0095] Furthermore, clustering of multiple results, for example of multiple objects detected close to each other, allows recognition of which objects are the same object, so that they are not mistakenly recognized multiple times.

[0096] Similarly, a weighting function 177 can be applied by multiplication when combining results from multiple windows 167. Preferably, the diagnostic sub-results 175 corresponding to corresponding data points in space can then be weighted with respect to the positioning of the diagnostic sub-results 175 relative to the center point of the respective window 167. This is shown by way of example in diagram b), where each diagnostic sub-result 175 is weighted with respect to the center point of the respective window 167 according to the weighting function 177 shown.

[0097] In one embodiment, the results of one processing path 102 can influence the expansion of other processing paths 102 after post-processing 163. Weighting parameters can then be adapted and for each window the weighting parameters can be made dependent on the respective results coming from the processing paths 102.

[0098] For example, using the results of object classification that defines the object type 117 of an object located inside a wall 105, the weight of wall type classification that determines the wall type 123 of each wall 105 can be increased in post-processing in areas where there are no objects 113. This is because the radar data 103 in such areas is less affected by reflections from the objects 113.

[0099] FIG. 5 shows another schematic diagram of a measurement device 100 according to another embodiment.

[0100] Diagrams a) and b) of FIG. 5 show two different options for joint data processing of radar data 103 and additional sensor information 104 by the diagnostic module 107.

[0101] Diagram a) shows the joint processing of radar data 103 and additional sensor information 104 from an additional sensor by a diagnostic module 107. To this end, the radar data 103 and the additional sensor information 104 are jointly used as input data for a diagnostic module 107 configured as an artificial intelligence, in particular as an artificial neural network. The diagnostic module 107 includes a number of folding layers 108 and a number of tightly coupled layers 106. The radar data 103 and the additional sensor information 104 are jointly processed as input data via the folding layers 108 and the tightly coupled layers 106. The output data of the diagnostic module 107, on the basis of which the above-described diagnostic result 109 is generated.

[0102] In contrast, in diagram b), radar data 103 and additional sensor information 104 are used as independent input data for a diagnostic module 107. The diagnostic module 107 is made up of multiple processing paths 102, each of which includes multiple folding layers 108 and at least one tightly coupled layer 106. In the various processing paths 102, wall diagnoses are generated by the diagnostic module 107 independently of each other, based on the radar data 103 or the additional sensor information 104.

[0103] In an additional concatenation layer 148, the partial results of the partial diagnoses of the various processing paths 102 are integrated and fed to the final tightly coupled layer 106. The output data of the diagnosis module 107 corresponds to the diagnosis result 109 described above.

[0104] A correspondingly configured diagnostic module 107 is set up to perform the wall diagnostics described above based on the radar data 103 and additional sensor information 104 and with the configuration requirements described above.

[0105] In the illustrated embodiment, the diagnostic module 107 is configured as an artificial neural network, in particular as a folding network. Corresponding network architectures with a folding layer 108, a densely connected layer 106, and a connection layer 148 are well known in the prior art.

[0106] FIG. 6 shows a schematic diagram of a system 600 for generating a training data set according to one embodiment.

[0107] According to the invention, to generate the training data set 143, firstly, sensor data 103, 104 of at least one sensor unit 101 of the measuring device 100 are recorded. For this purpose, multiple measurements of one wall 105 or of several different walls 105 can be carried out by one measuring device 100 or by multiple measuring devices 100. During the measurements 105, the measuring device 100 can be moved along the wall 105 to be inspected in a movement direction 153, as explained above, and corresponding sensor data 103, 104 can be recorded, including the wall 105 to be inspected and possibly including an object 113 arranged therein. The sensor data 103, 104 can include radar data 103 and / or additional sensor information 104.

[0108] During recording of these sensor data 103, 104 by the measuring devices 100, position information 172 is determined for each measuring device 100 by the position determination system 145. The position information 172 then describes the position of the measuring device 100 relative to the wall 105 in which the measuring device 100 was positioned during recording of the sensor data 103, 104.

[0109] Subsequently, location information 172 can be incorporated into the sensor data 103, 104, thereby generating location-specific sensor data.

[0110] For each of the sensor data 103, 104, labeling of the sensor data 103, 104 can be performed based on position information 172 that defines the position of the measurement device 100 relative to the wall 105 at the time of recording of the respective sensor data 103, 104. When labeling the sensor data 103, 104, ground truth information is assigned to each of the sensor data 103, 104. The ground truth information relates to at least one piece of information that indicates whether or not the respective sensor data 103, 104 reflects an object 113 located inside the wall 105.

[0111] Such ground truth information depends on the one hand on whether an object 113 is located in the inspected wall reflected by the respective sensor data or not, and further on whether the object 113 located in the wall 105 is located at an object position 115 that at least partially coincides with the position of the measurement device 100 relative to the wall 105 at the time of recording the respective sensor data 103, 104.

[0112] Additionally, the ground truth information may further include information regarding the actual object position 115 of the object 113, and / or regarding the object type 117 of the object 113, and / or regarding the object depth 119 of the object 113, and / or regarding the object extension 121 of the object 113. Additionally, the ground truth information may include information regarding the wall type 123 of the inspected wall 105.

[0113] Labeling the sensor data 103, 104 with respect to the ground truth information allows each data to be identified in relation to the ground truth information, thereby at least characterizing whether an object 113 located within the wall 105 is reflected by the respective sensor data 103, 104.

[0114] The labeling of the sensor data 103, 104 then corresponds to identifying the sensor data 103, 104 with respect to the corresponding ground truth information.

[0115] The position information 172 of the position determination system 145 makes it possible to clearly define for each sensor data 103, 104 where the measuring device 100 was positioned relative to the wall 105 at the time of recording the respective sensor data 103, 104. This allows, with knowledge of the actual object position 115 of the object 113 located within the wall 105, a clear and unambiguous labelling of the sensor data 103, 104 in relation to the object position 115, i.e. whether or not the object 113 located within the wall is actually reflected by the respective sensor data 103, 104.

[0116] The sensor data 103, 104 labeled in this way are subsequently integrated into a corresponding training data set 143.

[0117] The labelling of the sensor data 103, 104, as well as the consideration of the location information 172 and the generation of the training data set 143, can be performed by an external computation unit 170.

[0118] The position determination system 145 may include, for example, a position sensor disposed on the measuring device 100. Through the position sensor, the positioning of the measuring device 100 relative to the wall may thus be determined.

[0119] The position determining system 145 preferably includes a camera sensor positioned externally relative to the measuring device 100. Through the camera sensor, camera data can be recorded that reflects the measuring device 100 during the recording of the sensor data 103, 104. Through the camera data, the positioning of the measuring device 100 relative to the wall 105 during the recording of the sensor data 103, 104 can thus be determined.

[0120] Through the time stamps of the camera data, the position of the measuring device 100 relative to the wall 105 can be determined in a time-resolved manner. In this way, the respective position of the measuring device 100 relative to the wall 105 can be determined for multiple points in time. Through the time stamps of the sensor data 103, 104, the time at which the respective sensor data 103, 104 was recorded by the measuring device 100 can be determined.

[0121] To determine the location information 172, the camera data of the camera sensor may further be synchronized in time with the sensor data 103, 104 of the measurement device 100. By synchronizing the timestamps of the respective sensor data 103, 104 with the timestamps of the camera data, it may be possible to assign the correct location information 172 to each of the sensor data 103, 104.

[0122] FIG. 7 shows another schematic diagram of a system 600 for generating a training data set according to another embodiment.

[0123] Diagrams a) and b) show, by way of example, two camera data, i.e., two images or two frames, of the camera sensor 147 of the position determination system 145. The images shown in diagrams a) and b) show a wall 105. A location marking 149 in the form of an ArUco / ChArUco poster 158 is arranged on the wall 105. Furthermore, a light switch 162, also defined as a location marking 149, is positioned on the wall. Next to the wall 105 there is a door 160, also defined as a location marking 149.

[0124] Additionally, in diagram a) a measuring device 100 within the meaning of the present invention is arranged, which can be moved along a direction of movement 153 in order to record corresponding sensor data 103, 104, thus performing a wall diagnostic of the spatial region of a wall 105 traversed by the measuring device 100 during its movement.

[0125] In one embodiment, the position information 172, i.e. the determination of the position of the measuring device 100 relative to the wall 105 during the recording of the sensor data, can be achieved through the use of the position marking 149 shown in diagram a). The position marking 149 and the measuring device 100 are then detected in the camera data, allowing the relative position of the measuring device 100 to be determined with respect to the position marking 149. Through the relative position thus determined between the measuring device 100 and the position marking 149, the position of the measuring device 100 relative to the wall 105 can be determined. On this basis, the position information 172 can be generated accordingly.

[0126] Furthermore, in diagram b) ground truth information regarding an object 113 located in the wall is graphically displayed in the area of ​​the location marking 149 in the form of an ArUco / ChArUco poster 158 positioned in diagram a), where the ground truth information includes at least the positioning of the object 113 in the wall 105.

[0127] Diagram b) suggests a pipe as the object 113, thereby suggesting that the corresponding information, namely the object location 115, the object type 117, and other information about the object 113, is known as ground truth information.

[0128] Through the displayed ground truth information, the sensor data 103, 104 recorded by the measuring device 100 in the area of ​​the ArUco / ChArUco poster 158 shown in diagram a) can be appropriately labeled to generate an appropriate training data set.

[0129] FIG. 8 shows a schematic diagram of the data recorded by the measurement device 100.

[0130] Diagrams a), b), and c) show various measurement planes 164 that define the spatial regions where sensor data 103, 104 are recorded by moving the measurement device 100 relative to the wall 105. The measurement planes 164 can here illustratively represent the surfaces of the ArUco / ChArUco poster 158.

[0131] Within the measurement plane 164, the measurement device 100 moves in a number of measurement trajectories 166 for recording the sensor data 103, 104. Along the measurement trajectory 166, the sensor data 103, 104 are recorded at different measurement points 168, respectively.

[0132] The measurement tracks 166 may be arranged largely parallel to one another, as shown in diagram a), in addition the measurement tracks may be arranged in a crossing configuration, as shown in diagram b).

[0133] The position information 172 can be combined with the sensor data 103, 104 recorded at the measurement points 168 along the measurement trajectories 166 so that for each sensor data 103, 104 at each measurement point 168 of the various measurement trajectories 166, a respective position of the measuring device 100 for recording the respective sensor data 103, 104 can be determined. For intersecting measurement trajectories 166, the already determined position information 172 can be reused. In addition, for measurement points 168 that are closely adjacent to one another, the corresponding position information can be estimated based on the position information of the measurement points 168 that are located nearby.

[0134] Such a fine gridding of the measurement surface 164 in the illustrated measurement trajectory 166 allows an additional background correction to be generated on the basis of the sensor data 103, 104 due to the dense spacing of the measurement points 168. The background correction then takes into account measurement signals that are reflected only from the wall 105 and that are not influenced by objects located within the wall 105. In this way, an object determination can be performed, in which changes in the recorded sensor data 103, 104 of the various measurement points 168 are determined.

[0135] FIG. 9 shows another schematic diagram of a system 700 for generating a training data set 143 according to another embodiment.

[0136] In the illustrated embodiment, the position determination system 145 includes a camera sensor 147. The camera sensor 147 is arranged in a position device 156. The position device 156, and in particular the camera sensor 147, is arranged at a distance relative to the wall 105. A position marking 149 in the form of an ArUco / ChArUco poster 158 is arranged on the wall 105. Furthermore, Figure 9 shows a measuring device 100 on the wall 105 in the area of ​​the ArUco / ChArUco poster 158.

[0137] FIG. 9 shows that perspective correction must be performed on the camera data recorded by the camera sensor 147 in order to determine the position of the measuring device 100 relative to the wall 105 based on the distance from the camera sensor 147 to the wall 105.

[0138] The perspective correction can be performed by the mathematical relationship given below.

number

[0139] where E is the identity matrix, K is the camera matrix, which contains the embedded camera data of camera sensor 147, and T is the transformation vector describing the transformation from the world coordinate system to the camera coordinate system of camera sensor 147. The entries u and v represent pixel elements in the image plane. The components Xw, Yw, and Zw here represent world coordinates in the world coordinate system. The components Xc, Yc, and Zc represent coordinates in the camera coordinate system of camera sensor 147 shown in FIG. 9.

[0140] Taking into account the matrix notation, the following occurs:

number

[0141] where f is the focal length of the camera sensor 147, r is an element of the rotation matrix R, and t is an element of the translation vector. Afterwards, a transformation can be calculated to get the actual position of the camera sensor 147 in the world coordinate system, where the world coordinate system is represented by ArUco / ChArUco poster 158. To convert a vector from the camera system to the world system, the following transformation relationship is needed:

number

[0142] where R is the rotation matrix. This relationship makes it possible to extract real-world position and orientation information of the camera sensor 147 relative to the ArUco / ChArUco poster 158 from the camera data. For this purpose, the entire video or just individual frames of camera data can be used. Going further, another vector-based transformation relationship can be used to convert to the position and orientation of the measuring device 100 reflected by the camera data. Projection into 3D creates an image-based vector relationship that must be resolved. For this purpose, the positions, i.e., the camera data points, must be converted to positions relative to the world coordinate system, from which the required vectors can subsequently be calculated. This is done using the following equation:

number

[0143] The end result is a vector x that represents the transformation from the position of the measuring device 100 projected by the camera data to the actual position of the measuring device 100 in space, i.e., relative to the wall 105. This relationship is graphically represented in FIG.

number

[0144] where Cw are the camera coordinates of the camera sensor 147 in the world coordinate system and Dw are the device coordinates of the measuring device 100 in the world coordinate system, the relationship shown representing the transformation of the camera coordinates into the device coordinates of the measuring device 100. The accuracy of determining the position of the measuring device 100 relative to the wall 105 in this manner depends critically on the quality of the calibration of the camera sensor 147 and on the unambiguous recognition of the position markings. Therefore, to ensure that the recognition of the position markings is highly reliable, it may be preferable to calibrate the camera sensor 147 and carry out sufficient test shots before implementing the method.

[0145] By implementing the method described above, it is possible to determine at any instant the position of the measuring device 100 relative to the wall 105. This can be used to assign sensor data 103, 104 recorded by the measuring device during measurements to these positions of the measuring device 100 relative to the wall 105. In particular, several measurements can be combined into one measurement sequence and preferably processed jointly.

[0146] FIG. 10 shows a flowchart of a method 300 for generating a training data set 143 according to one embodiment.

[0147] In order to generate a training data set 143 for training the artificial intelligence 125 for operating the measuring device 100, in particular the wall diagnostic device, in a first method step 301, firstly the sensor data 103, 104 of at least one sensor unit 101 of the measuring device 100 are recorded, whereby the sensor data 103, 104 reflect the wall 105 to be diagnosed.

[0148] In the next method step 303, a position determination of the measuring device 100 relative to the wall 105 is performed and position-related sensor data 103, 104 are generated by the position determination system 145. In the position determination, each recorded sensor data 103, 104 is assigned position information 172. The position information 172 defines the position of the measuring device 100 relative to the wall 105 in which the measuring device 100 was positioned relative to the wall 105 at the time of the respective sensor data 103, 104.

[0149] In one embodiment, the position determination system 145 includes at least one camera sensor 147 and at least one position marking 149. The position marking 149 is formed on a surface of the wall 105, and the camera sensor 147 is positioned at a predefined perspective positioning relative to the wall 105.

[0150] Thus, firstly in method step 309, camera data of the camera sensor 147 is recorded by the measuring device 100 during the recording of the sensor data 103, 104, and the camera data reflects the measuring device 100 positioned on the wall 105 and the position marking 149 placed on the wall 105.

[0151] In method step 311, the relative position of the measuring device relative to the position marking 149 is determined on the basis of the camera data.

[0152] In method step 315 , a time synchronization is subsequently carried out between the recorded sensor data 103 , 104 of the measuring device 100 and the recorded camera data of the camera sensor 147 .

[0153] Subsequently, in method step 317, position information 172 defining the position of the measuring device 100 relative to the wall 105 at the time of recording of the sensor data 103, 104 is determined for each sensor data, taking into account time synchronization.

[0154] This is followed in method step 313 by determining the position of the measuring device 100 relative to the wall 105 .

[0155] To that end, in method step 319 , the perspective of the camera sensor 147 relative to the wall 105 is determined based on a predefined perspective position of the camera sensor 147 relative to the wall 105 .

[0156] Subsequently, in method step 321, a perspective correction is performed to take into account the perspective position of the camera sensor 147 relative to the wall 105, in accordance with the relative position of the measuring device 100 relative to the position marking 149, determined based on the camera data.

[0157] In a next method step 305, the location-related sensor data 103 is subsequently labeled taking into account ground truth information to generate labeled sensor data 103, 104, wherein the ground truth information comprises at least one piece of information regarding the presence of at least one object 113 in the wall 105.

[0158] In the next method step 307 , the labelled sensor data 103 , 104 are combined to form a training data set 143 .

[0159] FIG. 11 shows another flowchart of a method 300 for generating a training data set 143 according to another embodiment.

[0160] To train the artificial intelligence 125 of the measuring device 100 for wall diagnosis, in method step 401, a training data set 143 is first provided by executing the method 300 for generating a training data set 143 based on the embodiment described above.

[0161] In a next method step 403, the artificial intelligence 125 is trained to perform object recognition of objects 113 configured in the wall 105 based on the training data set 143, the object recognition including at least one object detection.

[0162] FIG. 12 shows a schematic diagram of a computer program product 500 comprising commands that, when executed by a data processing unit, direct it to perform the method 300 for generating a training data set 143 .

[0163] The computer program product 500 is stored in the illustrated embodiment on a storage medium 501, which may be any storage medium known in the art. [Explanation of symbols]

[0164] 100 Measuring Device 101 Radar sensor unit 103 Sensor Data 104 Sensor Data 105 Wall 107 Diagnostic Module 109 Diagnosis Results 111 Display unit 113 Object 115 Object position 117 Object Types 119 Object Depth 121 Object extension 123 Wall Type 125 Artificial Intelligence 127 Pre-processing module 129 Wall Type Classification Module 131 Object Recognition Module 133 Input Data 135 First Pre-processing Module 137 Second Pre-processing Module 139 Wall Type Information 141 Motion Detection Unit 143 training datasets 145 Positioning System 147 Camera Sensor 149 Position Marking 151 computing units 156 Positioning device 158 ArUco / ChArUco Poster 162 Light Switch 172 Location information 300 ways 301 Sensor Data Recording 303 Performing Position Fix 305 Labeling 307 Integration 309 Camera Data Recording 311 Determining Position 313 Determining Position 315 Performing Synchronization 317 Determining Location Information 319 Perspective Judgment 321 Performing Perspective Correction 400 ways 401 Providing a training dataset 403 Training 500 Computer Program Products

Claims

1. 1. A computer-implemented method (300) for generating a training data set (143) for training an artificial intelligence (125) for operating a measurement device (100), in particular a wall diagnostic device, comprising: The method includes recording (301) sensor data (103) of at least one sensor unit (101) of the measuring device (100), said sensor data (103) reflecting a wall (105) to be diagnosed; a position determination (303) of the measuring device (100) relative to a wall (105) is performed, and position-related sensor data (103) is generated by a position determination system (145), wherein the position determination assigns position information (172) to each recorded sensor data (103), the position information (172) defining the position of the measuring device (100) relative to the wall (105) at the time of recording of the sensor data (103); The method includes labeling (305) the location-related sensor data (103) and generating the labeled sensor data (103) taking into account ground truth information, the ground truth information including at least one piece of information regarding the presence of at least one object (113) in the wall (105); and A method (300) comprising aggregating (307) the labeled sensor data (103) to form a training data set (143).

2. The method (300) of claim 1, wherein the ground truth information includes classification information regarding the object position (115) of the object (113) and / or the object type (117) of the object (113) and / or object depth information regarding the object depth (119) inside the wall (105) and / or object extension information regarding the object extension (121) of the object (113).

3. The position determination system (145) includes at least one camera sensor (147) and at least one position marking (149), the position marking (149) being formed on a surface of the wall (105), the camera sensor (147) being positioned in a predefined perspective positioning with respect to the wall (105), and performing (303) a position determination of the measuring device (100) relative to the wall (105) comprises: recording (309) camera data of the camera sensor (147) during the recording of sensor data (103) of the sensor unit (101) of the measuring device (100), the camera data reflecting the measuring device (100) positioned on a wall (105) and the position marking (149) arranged on the wall (105); determining (311) a relative position of the measuring device (100) relative to the position marking (149) based on the camera data; and 3. The method (300) of claim 1 or 2, comprising determining (313) a position of the measuring device (100) relative to a wall (105) based on the relative position.

4. The position determination execution (303) further comprises: a time synchronization (315) is performed between the recording of the sensor data (103) of the sensor unit (101) and the recording of the camera data of the camera sensor (147); and 4. The method (300) according to claim 1, further comprising determining (317) position information (172) of the measuring device (100) relative to a wall (105) for each sensor data (103) of the sensor units (101) taking into account time synchronization.

5. The determination of the location (313) further includes: determining (319) a perspective of the camera sensor (147) relative to the wall (105) based on a predefined perspective position of the camera sensor (147) relative to the wall (105); 5. The method (300) according to claim 1, further comprising: performing (321) a perspective correction to take into account the perspective position of the camera sensor relative to the wall, the perspective correction being determined based on camera data (147) in accordance with the relative position of the measuring device (100) relative to the position marking (149); and determining the position of the measuring device (100) on the wall (105) based on the perspective correction.

6. The method (300) of any one of claims 1 to 5, wherein the position markings (149) are formed as ArUco markings (158) or ChArUco markings (158).

7. 7. The method (300) of any one of claims 1 to 6, wherein the location markings (149) are formed by markings formed on a wall (105), including wallpaper patterns, light switches (162), electrical outlets, windows, and furniture.

8. 8. The method (300) according to any one of claims 1 to 7, wherein the camera sensor (147) is disposed on a positioning device (156) through which the camera sensor (147) is positioned at a predetermined perspective position relative to the wall (105).

9. A training dataset (143) for training an artificial intelligence (125) of a measuring device (100) for wall diagnostics, the training dataset (143) being generated by a method (300) for generating a training dataset (143) for training an artificial intelligence (125) for operating a measuring device (100) based on any one of claims 1 to 8.

10. A computer-implemented method (400) for training artificial intelligence (125) of a measurement device (100) for wall diagnostics, comprising: a training data set (143) is provided (401) by implementing a method (300) for generating a training data set (143) according to any one of claims 1 to 8; The method (400) includes training (403) an artificial intelligence (125) to perform object recognition of objects (113) configured in a wall (105) based on the training dataset (143), the object recognition including at least object detection and object classification.

11. A computing unit (151) set up to perform a method (300) for generating a training data set (143) for training an artificial intelligence (125) for operating a measuring device (100) according to any one of claims 1 to 8 and / or a method (400) for training an artificial intelligence (125) of a measuring device (100) according to claim 10.

12. A computer program product (500) comprising instructions which, when executed by a data processing unit, cause the data processing unit to perform a method (300) for generating a training data set (143) for training an artificial intelligence (125) for operating a measurement device (100) according to any one of claims 1 to 8 and / or a method (400) for training an artificial intelligence (125) of a measurement device (100) according to claim 10.

Images