Method for locating labeled storage devices, equipment, and vehicles

Abstract

The invention relates to a computer-implemented method 174 for localizing storage devices 102, 104, 106 provided with labels 108, 110, 134. The localization is based on the labels 108, 110, 134 having identical dimensions 204, 206. The method 174 comprises the following steps: Obtaining 176 image data 164 from a camera 114 arranged on a carrier 112, wherein at least one label 180, 110, 134 is detected in the image data 164; Obtaining 178 a carrier position 130 of the carrier 112 relative to a coordinate origin 128; Obtaining 180 a camera position 126 of the camera 114 relative to the carrier position 130; Detecting 182 the at least one label 108, 110, 134 in the image data 164; Detecting 184 an identifier for each detected label 108, 110, 134; Calculating 186 first label positions 124 of the detected label 108, 110, 134 relative to the camera 114 based on the image data 164 and the dimensions of the labels 204, 206;Calculating 188 second label positions of the detected labels 108, 110, 134 relative to the coordinate origin 128 based on the first label positions 124 and the carrier position 130; storing 190 the second label positions with the respective identifiers as storage means positions 142;

Description

Technical area

[0001] The present invention relates to the locating of marked goods. In particular, the present invention relates to a computer-implemented method for locating storage devices provided with labels. Furthermore, the present invention relates to a data processing device, a computer program product, a computer-readable storage medium, a data carrier signal, and a vehicle. Technical background and task

[0002] In manufacturing companies, pallets (or more generally: storage containers) are often loaded with packaged products and placed in a shipping area for shipping, frequently with the help of a forklift. Since there is always a certain amount of lead time during production, for example before loading onto a truck, a large number of pallets ready for shipping accumulate in the shipping area. When a transport vehicle, such as a truck, arrives to collect pallets and transport them to their next destination, the pallets that are intended for transport by that truck must first be selected from the large number of pallets in the shipping area. A logistics employee then often searches for the corresponding pallets, using identification stickers (also called labels) on the pallets as a guide.

[0003] This process, characterized by manual steps, has several disadvantages. First, a transport vehicle must usually be loaded within a maximum loading time (e.g., 15 minutes). The logistics employee must therefore quickly locate the pallets to be loaded. Depending on how full the shipping area is, how many pallets need to be transferred from the shipping area to the transport vehicle, their dimensions, whether they are stored underneath other pallets, or whether a shift change has taken place between preparation in the shipping area and loading onto the truck, searching for the correct pallets in the shipping area can be difficult. Fines are not uncommon if the maximum loading time cannot be met.

[0004] Secondly, human error cannot be ruled out during the search process. For example, labels may not be correctly recognized and loaded onto the wrong transport vehicle. Such errors are particularly common with long sequences of numbers on labels (e.g., transposed numbers).

[0005] While affected companies are aware of solutions to these problems, these require significant investments in new infrastructure (e.g., automated shelving, etc.) because they are based on automation and digitalization.

[0006] WO 2023 / 193932 A1 discloses methods for determining the position of objects, in particular products on a shelf, by emitting a coded light signal from a light-emitting device. A movable camera captures the scene and generates a digital image. The object position data is generated based on the position of the light signal in the image, the identification information, and the known camera orientation and position in space to represent the position of the object in space.

[0007] DE 102021 126 879 A1 describes a decentralized system for location determination. It comprises an image sensor for capturing environmental data, a stationary marker with position information, and a unit marker with unit information. A controller determines the position coordinate of the unit marker based on the image data, which contains both position and unit information.

[0008] It is therefore the object of the present invention to provide computer-implemented methods for locating labeled storage devices that eliminate at least one of the aforementioned disadvantages. Furthermore, it is the object of the invention to provide a data processing device, a computer program product, a computer-readable storage medium, a data carrier signal, and a vehicle. Disclosure of the invention

[0009] The object is achieved according to the invention by the features of the main claims. Advantageous embodiments can be found in the subclaims.

[0010] According to a first aspect of the invention, a computer-implemented method for locating storage devices provided with labels, wherein the labels have identical dimensions, comprises a step in which image data is obtained from a camera. The camera is arranged on a carrier, and at least one label is detected in the image data.

[0011] Storage equipment refers to means for storing physical goods. Pallets, boxes, containers, and the like are particularly common storage devices.

[0012] A label can be understood as any form of labeling that conveys information about a storage device. Labels are typically attached directly to the storage device, for example, glued, affixed, or printed. QR codes, barcodes, and the like are particularly common forms of labels. A piece of paper with information (e.g., contents, weight, dimensions, or identifier) ​​about the storage device on which it is attached can also be considered a label.

[0013] The labels have identical dimensions. Since a label is typically a nearly two-dimensional structure, with its thickness negligible compared to its height and width, the most important dimensions of a label are its height and width. For example, the dimensions can be described as a known size such as DIN A5 or a direct measurement such as "5cm x 7cm."

[0014] Advantageously, the carrier is a person, a forklift, a pallet truck, or a vehicle. The camera can be mounted, for example, on a person's helmet or at a suitable location on a vehicle such as a forklift. Preferably, locations are suitable from which the camera's field of view can capture the largest possible area around the carrier in its image data.

[0015] In a further step of the process, the carrier's position relative to a coordinate origin is obtained. The relative information to the coordinate origin can be realized, for example, using geographical coordinates such as GPS (Global Positioning System) or GNSS (Global Navigation Satellite System). However, it is also possible to use a more local coordinate system, which refers to a reference such as a gate or the center of a warehouse. The carrier position can then be expressed as a vector.

[0016] In a further step of the method, a camera position relative to the carrier position is obtained. The camera position may be known due to the design, for example, because it is mounted at a fixed location on the carrier. For example, the camera is mounted on the front of a helmet or at a specific location on the roof of a forklift. The position can also include the camera's orientation.

[0017] In a further step of the process, at least one of the labels is detected in the image data. Established image processing methods, such as object detection using artificial neural networks or other machine learning methods, are suitable for this step. Examples of suitable methods include the following: Viola-Jones method, scale-invariant feature transform (SIFT), Speeded Up Robust Features (SURF), histogram of oriented gradients (HOG), region proposals (R-CNN, Fast R-CNN, Faster R-CNN), Detectron, Single Shot MultiBox Detector (SSD), and You Only Look Once (YOLO).

[0018] In a further step of the process, an identifier is assigned to each detected label. The previously mentioned machine learning methods are suitable for this. The identifier can, for example, consist of a number and / or character sequence. It can be written directly on the label or extracted from a pattern such as a barcode or QR code printed on the label.

[0019] In a further step of the process, the initial label positions of the detected label are calculated relative to the camera. This calculation is based on the image data and the dimensions of the labels. The initial label positions can be expressed as a vector, which defines a corresponding displacement relative to the camera.

[0020] Since the dimensions of the labels are known and constant, the position of the detected labels can be determined by taking into account the camera's intrinsic parameters, such as focal length, distortion, and / or pixel size. Since the label positions are calculated from the image data, the positions are initially known relative to the camera. This can be understood as relative to the center of the camera's image sensor. Another point can also serve as a reference if the camera has been calibrated accordingly.

[0021] In an advantageous embodiment, the calculation of the first label positions comprises the following steps: - Detecting corner points of the detected labels, - Determining translations, rotations and scaling factors using intrinsic parameters of the camera such that the identical dimensions are transformed to the corner points, and - Determine the first label positions from the translations.

[0022] In other words, the transformation (consisting of translations, rotations, and scaling) is sought that maps the known dimensions of the labels to the images of the labels contained in the image data. The images of the labels contained in the image data may, for example, appear smaller due to their distance from the camera or distorted (skewed) due to their positioning.

[0023] Due to the camera's intrinsic parameters such as focal length, distortion, and pixel size, a correlation between the image data and the label's dimensions can be established using simple geometric considerations. In other words, if an object with known dimensions, namely the label, is detected in the image data, the size in pixels provides information about the distance (and thus the position) from the camera.

[0024] In a further step of the process, second label positions of the detected label are calculated relative to the coordinate origin. This is done based on the first label positions and the carrier position. This corresponds to a conversion of the label position from one coordinate system (that of the camera) to another (for example, a global one such as GPS or a local one such as a warehouse coordinate system). As an intermediate step, the system is first transformed into the carrier coordinate system, since the camera moves with it.

[0025] In a further step of the process, the second label positions are stored with the respective identifiers as storage medium positions. A storage medium is then uniquely identifiable via the identifier and assigned a storage medium position. Storage can be done, for example, in a database and digital table.

[0026] The wearer does not necessarily have to move past the storage devices in a targeted manner, but rather passes them randomly while performing their tasks. The camera consolidates uncontrolled data collections in the form of image data, which are then further processed either on a central processing unit such as a server or cloud, or on a local processing unit located on the wearer (a so-called crowdsourcing approach). Further processing involves recognizing labels in the image data and determining and storing their position based on the known, identical dimensions of the labels.

[0027] One advantage of the disclosed method is that it can be easily integrated into existing logistics systems, which typically already use labels. This avoids significant additional costs, such as requiring investments in new infrastructure such as automated shelving. Another advantage is that the crowdsourcing approach eliminates the need to interfere with established workflows. The storage items do not need to be constantly and specifically recorded, but are recorded "incidentally."

[0028] Advantageously, a digital map is created with the storage location. In other words, the digital map contains the Storage location is marked. The digital map can also include other elements, such as a storage area, foundation walls, or shelves.

[0029] The digital map advantageously also shows the carrier position.

[0030] Advantageously, additional sensor data is used to calculate the initial label positions. Sensors such as time-of-flight cameras, lidar, and radar can provide depth information to the image data, allowing the distance to the label (or to the pixel area associated with the label) to be determined in another way.

[0031] According to a second aspect of the invention, a computer program product comprises instructions which, when executed by a computer, cause the computer to execute a method as described above. The computer program product can be written in a programming language, for example, Python or C++.

[0032] According to a third aspect of the invention, a computer-readable storage medium comprises instructions that, when the program is executed by a computer, cause the computer to execute a method as described above. The computer-readable storage medium can be implemented, for example, as an SSD (solid-state disk) or as a flash memory. The computer-readable storage medium can also store other data, for example, sensor data and / or data that is (temporarily) stored during the execution of the method.

[0033] According to a fourth aspect of the invention, a data carrier signal transmits the computer program product as described above. The data carrier signal can be transmitted via a cable, for example, a CAN-BUS cable (Controller Area Network-BUS), FlexRay, Ethernet, LIN (Local Interconnect Network), or MOST (Media Oriented Systems Transport). However, it can also be transmitted wirelessly via Wi-Fi, Bluetooth, or the like.

[0034] According to a fifth aspect of the invention, a device for locating storage media provided with labels comprises a camera that generates image data. Furthermore, the device comprises a position-determining unit that generates carrier positions. Furthermore, the device comprises an evaluation unit that can execute a method as described above and is communicatively connected to the camera and the position-determining unit.

[0035] The positioning unit can be, for example, a GPS or GNSS device. The evaluation unit can, for example, comprise at least a processor and a memory. It can be located locally on the carrier, for example, in the form of an HPC (high-performance computer), or in the form of a server or a cloud that receives and processes data from the camera and the positioning unit.

[0036] Advantageously, the device has at least one further sensor which is communicatively connected to the evaluation unit and generates further sensor data which are used for calculating the first label positions.

[0037] The at least one additional sensor corresponds to a lidar, a radar, an ultrasound sensor, or a time-of-flight camera. These types of sensors can provide depth information, which can be correlated with the image data using appropriate calibration.

[0038] According to a sixth aspect of the invention, a vehicle comprises a device as described above. Summary of the characters

[0039] The invention is explained in more detail below using exemplary embodiments and figures. The figures show: Fig. 1: A schematic overview sketch of the invention from a bird's eye view; Fig. 2: A similar schematic overview sketch as Fig. 1 from a top view; Fig. 3: Sketch for determining a label position relative to a camera; Fig. 4: An embodiment of a device for locating storage means provided with labels as a block diagram; Fig. 5: An embodiment of a method for locating labeled storage devices as a flow chart; Fig. 6: An embodiment of a method step from Fig. 1 as a flow chart; Fig. 7: A schematic sketch of the process step from Fig. 6; and Fig. 8: An example of a vehicle. Detailed description of the characters

[0040] Fig. Figure 1 shows a schematic overview of the invention from a bird's eye view. It shows a storage area 100 in which a first group of storage means 102, a second group of storage means 104, and a storage means 106 are stored. In the first group of storage means 102, a total of eight boxes are stacked on a pallet, while in the second group of storage means 104, five boxes are loosely stacked and not on a pallet. The storage means 106 is secured to a pallet with a strap.

[0041] A first label 108 is attached to one of the storage means of the first group of storage means 102, a second label 110 is attached to a storage means from the second group of storage means 104. All other storage means 106 are also provided with labels, but for the sake of clarity Fig. 1 not marked.

[0042] A carrier 112 in the form of a forklift truck moves into Fig. 1 along a track 122. The track 122 is in Fig. 1 is shown only schematically and can take many different forms depending on the tasks of the carrier 112. For example, the guideway 122 looks different than in Fig. 1 when the carrier, in the performance of his duties, picks up and transports the pallets from the storage area 100.

[0043] A camera 114 is arranged on the carrier 112, which records image data 164 from its field of view 116. Initially, only certain labels 108, 110, 134 are captured, for example, the first label 108. However, the second label 110 is still concealed by the first group of storage devices 102 and is only captured by the camera 114 once the carrier 112 has moved along the travel path 122.

[0044] The carrier 112 has in the example of Fig. 1 also has communication means 118, with the help of which it can determine its own position and send data such as the image data 164 of the camera 114 to a computing unit. A carrier coordinate system 120 defines a coordinate system which is fixed to the carrier 112. In the example of Fig. 1, the carrier coordinate system 120 is defined centrally on its rear axle. The carrier coordinate system 120 serves as a reference for defining the position of the carrier.

[0045] In addition, Fig. 1 also shows a storage coordinate system 128. It serves as a reference coordinate system for which the positions of the storage means 134 are to be recorded.

[0046] Since the camera 114 is rigidly attached to the support 112, a camera position can be defined relative to the support coordinate system 126. This is a constant vector because both the camera 114 and the support coordinate system 120 remain constant relative to the support 112. The vector can, for example, point from the origin of the support coordinate system 120 to a point on the housing of the camera 114 or the center of an image sensor of the camera 114.

[0047] If at least one label 108, 110, 134 is detected in the image data 164 of the camera 114, a label position relative to the camera 124 can be calculated. In addition, an identifier is determined for each detected label 108, which uniquely identifies the storage medium on which a label is arranged. This can be, for example, a number, a numerical code, a letter code, or the like. The identifier can be printed directly on the label 108, 110, 134 or coded, for example, in the form of a QR code or barcode. Fig. 3, Fig. 6 and Fig. 7, with the help of which such a calculation is discussed.

[0048] The carrier 112 can determine its own position using the communication means 118 in the bearing coordinate system 128. If the data from the communication means 118 originates from a GPS or GNSS device, it can be easily converted to the bearing coordinate system 128. A carrier position relative to the bearing coordinate system 130 is then known. Converting the label position relative to the camera 124 into a label position relative to the bearing coordinate system 128 is just as easy by adding the relevant vectors 124, 126, and 130.

[0049] Finally, a digital map 132 of the warehouse is created, in which the positions of the storage devices 102, 104, and 106, linked to their respective identifiers, are stored and displayed. The carrier 112 itself is also recorded on the digital map 132.

[0050] Fig. 2 shows a similar schematic overview sketch as Fig. 1 from a top view. In addition to the Fig. 1 known components are in Fig. 2 positioning satellites 136 are shown, with the help of which the communication means 118 can determine the position of the carrier 112. The communication means 118 are also used to send the image data 164 to a carrier-external evaluation unit 138. The evaluation unit 138 evaluates the image data 164, detects labels 108, 110, 134 with the associated identifiers and stores this in a database 140. In the example of Fig. 2, the identifiers are stored as numerical values ​​in the "ID" column of the database 140. The database 140 can contain further information, such as the contents of the storage device 102, 104, 106, the shipping address, the delivery date, and the like. Furthermore, a digital map 132 is generated, in which digital positions 142 of the storage devices 102, 104, 106 are displayed. In addition, the position of the carrier 112 is also displayed in the digital map 132 as the digital carrier position 114.

[0051] Fig. Figure 3 shows a diagram for determining a label position relative to a camera 124. An object, for example, a label 108, 110, 134, is located at a distance 154 from the aperture of a camera 114. An image corresponding to the object is captured on an image sensor at focal length 146. An object size 148, for example, the label dimensions, is known (since all labels 180, 110, 134 are the same size). The focal length 146 of the camera 114 is also a known parameter. From the captured image data 164, an image size 150 of the image can be calculated (for example, via the pixel size). Due to the identical angle 152 and the known sizes 146, 148, 150, the distance 154 can be calculated as: Distance=object size / image size×focal length.

[0052] Fig. 4 shows an embodiment of a device 156 for locating storage means 102, 104, 106 provided with labels 108, 110, 134 as a block diagram.

[0053] The device 156 has a camera 114 that captures image data 164. Furthermore, the device 156 has an evaluation unit 138, which in turn is formed with a processor 158 and a memory 160. The latter has a computer program product 162 that has instructions that, when the program is executed by the processor 158, cause the processor to process the image data 164 as disclosed by the invention.

[0054] Furthermore, the device 156 has a position determination unit 166, with which the carrier 112 can determine its position. In addition, the device 156 has further sensors 168, in the example of Fig. 4, a radar 170 and a lidar 172. These sensors can measure distances and support the determination of the label position relative to the camera 124. Data from the positioning unit 166 and any additional sensors 168 are processed by the computer program product 162 as disclosed by the invention.

[0055] Fig. 5 shows an embodiment of a method 174 for locating storage means 102, 104, 106 provided with labels 108, 110, 134 as a flow chart.

[0056] In a first acquisition step 176, image data 164 from a camera 114 is obtained. The camera 114 is arranged on a carrier 112. At least one label 108, 110, 134 is detected in the image data 164.

[0057] In parallel, a carrier position 130 of the carrier relative to a coordinate origin 128 is obtained in a second preservation step 178. However, this step can also occur before or after the first preservation step 176. It only needs to occur before a first calculation step 186, which is explained further below.

[0058] Parallel to the second acquisition step 178, a camera position 126 of the camera 114 relative to the carrier position 130 is acquired in a third acquisition step 180. This step can also occur before or after the second acquisition step 178. It only needs to occur before a second calculation step 188, which is explained further below.

[0059] In a first recognition step 182, at least one label 108, 110, 134 is recognized in the image data 164. This can be done, for example, using established means of object recognition.

[0060] In a second recognition step 184, an identifier is recognized for each recognized label 108, 110, 134. This can be done, for example, using established object recognition methods. The identifier can consist of a character or letter sequence, a QR code, or a barcode, for example.

[0061] In the first calculation step 186, the first label positions 124 of the detected labels 108, 110, 134 are calculated relative to the camera 114 based on the image data 164 and the dimensions of the labels 204, 206. Since the dimensions 204, 206 are known and unchangeable, as exemplified by Fig. 3, the first label positions 124 are extracted from the image data 164. Additional information such as intrinsic parameters of the camera 114, for example, the focal length 146, may be required for the first calculation step 186.

[0062] In the second calculation step 188, second label positions of the detected labels 108, 110, 134 relative to the coordinate origin 128 are calculated based on the first label positions 124 and the carrier position 130. Since the carrier position 130 is already defined relative to the coordinate origin 128, as well as the first label positions 124 relative to the camera 114, and the camera position 126 of the camera 114 relative to the carrier position 130 is also known, the second label positions follow from a simple addition of these position vectors 124, 130, 126.

[0063] In a storage step 190, the second label positions are stored together with the respective identifiers as storage medium positions.

[0064] Finally, in a creation step 192, a digital map 132 is created in which the storage means positions and optionally also the carrier position 130 are marked.

[0065] Fig. 6 shows an embodiment of a method step 186 from Fig. 1 as a flow chart.

[0066] In a third recognition step 194, 164 corner points of the recognized labels 108, 110, 134 are detected in the image data. The goal is to compare the corner points with the known dimensions 204, 206 in order to calculate the first label positions 124.

[0067] In a first determination step 196, various geometric transformations are applied to the dimensions 204, 206. The goal is to transform the dimensions 204, 206 so that they correspond to the corner points. The transformations consist of translations, rotations, and scaling factors using intrinsic parameters of the camera 114, such as focal length and / or pixel size.

[0068] In a second determination step 198, the first label positions 124 are then determined from the translations. The method 174 is then Fig. 5 with transition to the second calculation step 188.

[0069] Fig. 7 shows a schematic sketch of process step 186 from Fig. 6.

[0070] Fig. 7a) shows a label coordinate system 200 with a label edge 202 corresponding to the label dimensions 204, 206. The label dimensions 204, 206 have a label width 204 and a label height 206. Although a label 108, 110, 134 with rectangular dimensions 204, 206 is discussed here as an example, it is also possible to perform the method 174 with other dimensions, for example, a round label 108, 110, 134. A radius would then have to be identified, which is transformed into the detected label 108, 110, 134 in the image data 164 using transformations. The relationship between the label coordinate system 200 and the image data 164 is known via intrinsic parameters of the camera 114, such as pixel size and focal length.

[0071] Fig. 7b) shows a first transformation, which corresponds to a rotation around an axis. A further transformation in the form of a translation 208 is shown in Fig. 7c).

[0072] Fig. 7d) shows the image data 164 in which a storage medium 106 is recorded. The corner points of the label 134 detected therein serve as a target for the transformations from Fig. 7b) and c), so that the dimensions 204,206 coincide with the corner points (marked between Fig. 7c) and d) as dashed lines).

[0073] The Fig. The translation 208 applied in 7c) now indicates the distance at which the label 134 is located. The distance is relative to the camera 114, since the image data 164 were recorded with the camera 114.

[0074] In addition, a centroid of a polygon can be determined from the vertices in the image data 164. Using the optical center of the image sensor of camera 114 and the focal length 146 of the lens used in camera 114, the location of this centroid in 3D space is calculated. In other words, a pixel deprojection is performed. The result is the 3D coordinates of the centroid relative to camera 114, or in other words, the relative position of label 134 to camera 114.

[0075] Fig. 8 shows an embodiment of a vehicle 112. The vehicle 112, in Fig. 8 in the form of a forklift truck, has a camera 114, communication means 118, a position determination unit 166, and an evaluation unit 138. The vehicle 112 does not have to be in the form of a forklift truck, but can also be designed, for example, as a pallet truck or as a person. List of reference symbols 100 storage area 102 First group of storage devices 104 Second group of storage devices 106 storage aids 108 First label 110 Second label 112 carriers 114 Camera 116 field of view 118 means of communication 120 Carrier coordinate system 122 Track 124 Label position relative to the camera 126 Camera position relative to the carrier coordinate system 128 bearing coordinate system 130 Carrier position relative to the bearing coordinate system 132 Digital Map 134 label 136 positioning satellites 138 Evaluation unit 140 database 142 Digital Position Storage Devices 144 Digital carrier position 146 focal length 148 Item size 150 image size 152 angles 154 Distance 156 Device 158 processor 160 storage 162 Computer program product 164 image data 166 Positioning Unit 168 Additional Sensors 170 radars 172 Lidar 174 procedures 176 First conservation step 178 Second conservation step 180 Third conservation step 182 First recognition step 184 Second recognition step 186 First calculation step 188 Second calculation step 190 storage step 192 Creation step 194 Third recognition step 196 First investigative step 198 Second investigation step 200 Label coordinate system 202 Label margin 204 label width 206 label height 208 Distance from Translation

Claims

[1] A computer-implemented method (174) for locating storage means (102, 104, 106) provided with labels (108, 110, 134), the labels (108, 110, 134) having identical dimensions (204, 206), the method (174) comprising the following steps: a) obtaining (176) image data (164) from a camera (114) arranged on a carrier (112), wherein at least one label (180, 110, 134) is detected in the image data (164), b) obtaining (178) a carrier position (130) of the carrier (112) relative to a coordinate origin (128), c) obtaining (180) a camera position (126) of the camera (114) relative to the carrier position (130), d) detecting (182) the at least one label (108, 110, 134) in the image data (164), e) recognizing (184) an identifier for each recognized label (108,110,134), f) calculating (186) first label positions (124) of the detected label (108, 110, 134) relative to the camera (114) based on the image data (164) and the dimensions of the labels (204, 206), g) calculating (188) second label positions of the recognized label (108, 110, 134) relative to the coordinate origin (128) based on the first label positions (124) and the carrier position (130), and h) storing (190) the second label positions with the respective identifiers as storage means positions (142). [2] Computer-implemented method according to claim 1, characterized by that a digital map (132) is created with the storage means positions (142). [3] Computer-implemented method according to claim 2, characterized by that the digital card (132) also has the carrier position (130). [4] Computer-implemented method according to one of the preceding claims, characterized bythat the carrier (112) is a person, a forklift truck, a pallet truck or a vehicle. [5] Computer-implemented method according to one of the preceding claims, characterized by that the calculation (186) of the first label positions (124) comprises the following steps: a) Detecting (194) corner points of the detected labels (108,110,134), b) determining (196) translations, rotations and scaling factors using intrinsic parameters of the camera (112) such that the identical dimensions (204, 206) are transformed to the corner points, and c) Determining (198) the first label positions (124) from the translations. [6] Computer-implemented method according to one of the preceding claims, characterized by that further sensor data are used for the calculation (186) of the first label positions (124). [7] A computer program product (162) comprising instructions which, when executed by a computer, cause the computer to carry out a method (174) according to any one of the preceding claims. [8] A computer-readable storage medium (160) comprising instructions which, when executed by a computer, cause the computer to carry out a method (174) according to any one of claims 1 to 6. [9] A data carrier signal which transmits the computer program product (162) according to claim 7. [10] Device (156) for locating storage means (102,104,106) provided with labels (108,110,134), comprising: a) A camera (114) which generates image data (164), b) A position determination unit (166) which generates carrier positions (130), and c) An evaluation unit (138) which can carry out a method (174) according to one of claims 1 to 6 and is communicatively connected to the camera (114) and the position determination unit (166). [11] Device according to claim 10, characterized in that it has at least one further sensor (168,170,172) which is communicatively connected to the evaluation unit (138) and generates further sensor data which are used for the calculation (186) of the first label positions (124). [12] Device according to claim 11, characterized in that the at least one further sensor (168, 170, 172) corresponds to a lidar (172), a radar (170), an ultrasound or a time-of-flight camera. [13] Vehicle (112) comprising a device (156) according to any one of claims 10 to 12.

Images