An abnormal storage location identification method, a stacking machine, a system, a medium and a product

Abstract

This invention discloses a method, stacker crane, system, medium, and product for identifying abnormal storage locations, relating to the field of logistics scheduling. The method for identifying abnormal storage locations, executed by the stacker crane, includes: obtaining an inventory scheduling instruction; performing a virtual-to-real detection of goods according to the instruction to obtain the detection result; photographing the target storage location space where the virtual-to-real detection result indicates the presence of goods to obtain an image of the storage location to be identified; and determining the abnormal storage location based on the image, the virtual-to-real detection result, and storage location record information from the warehouse management system. The technical solution of this invention can accurately identify storage locations occupied by stagnant goods while balancing cost and efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of logistics scheduling, and in particular to an abnormal storage location identification method, a stacker, a system, a medium and a product. BACKGROUND

[0002] In actual production and operation, the following problems often occur to cause stagnant goods: 1) the surface of the goods box is worn, contaminated, the label is off or the printing quality is poor during the handling and storage process, which causes the bar code or two-dimensional code to be unable to be normally read, forming a "goods without account"; 2) the warehouse scanning code or system input error causes the "identity of the goods box to be inconsistent"; 3) the operation error causes the empty pallet group to be placed in the storage location that should be the goods box, or the goods box to be placed in the storage location that should be the empty pallet group, causing "type inconsistency"; 4) the data communication between the warehouse management system and the stacker, conveying line and other equipment occasionally interrupts or responds overtime, causing "the instruction has been executed, the information is not synchronized", so that the actual position of part of the goods deviates from the system recorded position in the intermediate state; 5) during the warehouse maintenance work such as shelf adjustment and equipment maintenance, the actual position of the goods changes after the operation personnel do not update the manual relocation information in the warehouse management system, so that the system record cannot be synchronized after the actual position of the goods changes, forming "account without location" or "location without account" difference. These stagnant goods occupy the storage location for a long time, causing the inventory data to be distorted, and seriously affecting the accurate execution of the production plan and the lean management of the assets.

[0003] At present, the identification of the storage location occupied by the stagnant goods mainly relies on manual inventory, which has the problems of low efficiency, high cost and poor safety, and cannot realize normal fine management. Some existing technical solutions, such as adding multiple special sensors (such as high-low photoelectric switch, ultrasonic sensor, etc.) to distinguish the material type, will increase the system complexity, cost and maintenance difficulty, and the reliability faces challenges in the dust and vibration environment, and an abnormal storage location identification strategy that can balance efficiency, reliability and cost is urgently needed. SUMMARY

[0004] The present application provides an abnormal storage location identification method, a stacker, a system, a medium and a product to solve the problem that the identification of the storage location occupied by the stagnant goods cannot balance the cost, efficiency and reliability.

[0005] According to an aspect of the present application, an abnormal storage location identification method is provided, which is executed by a stacker and includes: acquiring an inventory scheduling instruction; performing goods virtual-real detection according to the inventory scheduling instruction to obtain a goods virtual-real detection result; photographing a target storage location space where the goods virtual-real detection result is present to obtain a to-be-identified storage location image; Based on the image of the location to be identified, the results of the virtual and real detection of goods, and the location record information of the warehouse management system, abnormal locations are determined.

[0006] According to another aspect of the present invention, an identification device for abnormal storage locations is provided, configured on a stacker crane, comprising: The instruction acquisition module is used to acquire and query scheduling instructions; The virtual-to-real detection module is used to perform virtual-to-real detection of goods according to the inspection and scheduling instructions, and obtain the virtual-to-real detection results of the goods; The image capturing module is used to capture images of the target cargo space where the cargo virtual and real detection results indicate the presence of cargo, thereby obtaining images of the cargo space to be identified; The abnormal storage location determination module is used to determine abnormal storage locations based on the image of the storage location to be identified, the results of virtual and real detection of goods, and the storage location record information of the warehouse management system.

[0007] According to another aspect of the present invention, a stacker crane is provided, the stacker crane comprising: At least one processor; and a memory communicatively connected to said at least one processor; The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the abnormal storage location identification method according to any embodiment of the present invention.

[0008] According to another aspect of the present invention, an automatic identification system for abnormal storage locations is provided, including a stacker crane and an intelligent management and control platform as described in any embodiment of the present invention; The intelligent management and control platform is used to issue inventory and scheduling instructions to the stacker crane, and to create abnormal cargo handling tasks based on the target location of the abnormal cargo location and the priority of the abnormal cargo location type.

[0009] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a processor to execute and implement the abnormal storage location identification method according to any embodiment of the present invention.

[0010] According to another aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the abnormal storage location identification method according to any embodiment of the present invention.

[0011] The technical solution of this invention obtains an inventory and scheduling instruction, then performs a virtual-to-real detection of goods based on the instruction, obtains the detection results, and then photographs the target storage space where goods are found to be present, obtaining an image of the storage space to be identified. Based on the image of the storage space to be identified, the virtual-to-real detection results, and the storage space record information from the warehouse management system, abnormal storage spaces are determined. This solution first performs virtual-to-real detection and collects images of storage spaces containing goods to be identified, then integrates the images of the storage spaces to be identified, the virtual-to-real detection results, and the storage space record information from the warehouse management system. This allows for the most comprehensive and accurate identification of abnormal conditions of goods in storage spaces from multiple dimensions. The identification process of abnormal storage spaces requires no manual intervention and does not rely on a large number of sensors, solving the problem of balancing cost, efficiency, and reliability in identifying storage spaces occupied by stagnant goods. It can accurately identify storage spaces occupied by stagnant goods while balancing cost and efficiency.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart illustrating a method for identifying abnormal cargo locations provided in Embodiment 1 of the present invention; Figure 2 A flowchart illustrating a method for identifying abnormal cargo locations according to Embodiment 2 of the present invention; Figure 3 This is a schematic diagram of the structure of an abnormal cargo location identification device provided in Embodiment 4 of the present invention; Figure 4 A schematic diagram of a stacker crane that can be used to implement an embodiment of the present invention is shown. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0016] It should be noted that the terms "current," "target," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] Example 1 Figure 1 This is a flowchart of an abnormal storage location identification method provided in Embodiment 1 of the present invention. This embodiment is applicable to the accurate and efficient identification of stagnant storage locations. The method can be executed by an abnormal storage location identification device, which can be implemented in hardware and / or software and can be configured in a stacker crane. Figure 1 As shown, the method includes: Step 110: Obtain the disk check and scheduling instructions.

[0018] Among them, the inspection and dispatch instruction can be an instruction to inspect stagnant goods in the target area.

[0019] In this embodiment of the invention, the entire warehouse can be checked for stagnant goods based on the inventory and scheduling instructions, or the warehouse can be divided into zones to check for stagnant goods.

[0020] Optionally, the intelligent management and control platform can send inventory and scheduling instructions to stacker cranes during preset warehouse operation off-peak periods (such as nighttime). The intelligent management and control platform can be understood as a "decision-making center" and "learning center" for identifying stagnant goods occupying storage locations, and is deployed on enterprise servers or in the cloud.

[0021] Step 120: According to the inspection and dispatch instructions, conduct cargo virtuality detection and obtain cargo virtuality detection results.

[0022] Cargo virtual-to-real detection can be an operation to identify the existence of goods in a storage location. The result of cargo virtual-to-real detection can be the identification result of the existence of goods in the storage location. The result of cargo virtual-to-real detection can include whether the storage location has goods or not, as well as the coordinates of the storage location, etc.

[0023] In this embodiment of the invention, the cargo location to be inspected can be probed based on the inspection and scheduling instructions to obtain the cargo probe results.

[0024] Step 130: Take a picture of the target cargo space where the cargo virtual and real detection results show that there is cargo, and obtain the cargo space image to be identified.

[0025] The target storage location space can be the storage location space surrounding the entire storage location when viewed directly from the front. The image of the storage location to be identified can be a photograph of the target storage location space.

[0026] In this embodiment of the invention, if the cargo virtual-to-real detection result is that cargo exists, then the target cargo space where cargo exists is photographed, and the photographed photo is used as the cargo space image to be identified.

[0027] Step 140: Based on the image of the location to be identified, the results of the virtual and real detection of goods, and the location record information of the warehouse management system, determine the abnormal location.

[0028] The location record information can be information related to the location recorded in the warehouse management system. This location record information may include, but is not limited to, the type of goods held in the location, the location coordinates, and other relevant information about the goods held in the location, as recorded in the warehouse management system.

[0029] In this embodiment of the invention, a pre-trained, distilled and optimized convolutional neural network model can be used to identify the image of the storage location to be identified, determine the type of goods in the target storage location space, and match the type of goods in the target storage location space and the results of the virtual and real detection of goods with the storage location record information of the warehouse management system to determine whether they are consistent with the storage location record information of the warehouse management system. If they are inconsistent, it indicates that there is a situation where the actual goods are inconsistent with the records, and this storage location is determined to be an abnormal storage location.

[0030] Optionally, the specific anomaly type of goods in the abnormal location can be determined based on the inconsistency type between the image of the location to be identified, the results of the virtual and real detection of goods, and the location record information in the warehouse management system.

[0031] The technical solution of this invention obtains an inventory and scheduling instruction, then performs a virtual-to-real detection of goods based on the instruction, obtains the detection results, and then photographs the target storage space where goods are found to be present, obtaining an image of the storage space to be identified. Based on the image of the storage space to be identified, the virtual-to-real detection results, and the storage space record information from the warehouse management system, abnormal storage spaces are determined. This solution first performs virtual-to-real detection and collects images of storage spaces containing goods to be identified, then integrates the images of the storage spaces to be identified, the virtual-to-real detection results, and the storage space record information from the warehouse management system. This allows for the most comprehensive and accurate identification of abnormal conditions of goods in storage spaces from multiple dimensions. The identification process of abnormal storage spaces requires no manual intervention and does not rely on a large number of sensors, solving the problem of balancing cost, efficiency, and reliability in identifying storage spaces occupied by stagnant goods. It can accurately identify storage spaces occupied by stagnant goods while balancing cost and efficiency.

[0032] Example 2 Figure 2 This is a flowchart of a method for identifying abnormal cargo locations according to Embodiment 2 of the present invention. This embodiment is a specific embodiment based on the above embodiment, and provides specific optional implementation methods for performing cargo virtual-to-real detection according to the inspection and scheduling instructions to obtain the cargo virtual-to-real detection results. Figure 2 As shown, the method includes: Step 210: Obtain the scheduling instruction.

[0033] Step 220: Analyze the inventory and dispatch instructions to determine the location of the cargo storage.

[0034] Among them, the cargo location inspection location can be the location where the cargo location to be inspected is inspected.

[0035] In this embodiment of the invention, the inventory scheduling instruction can be parsed to obtain the location of the cargo storage and the movement path, and the vehicle can move to the location of the cargo storage according to the movement path.

[0036] Step 230: After reaching the cargo location detection position, trigger the cargo virtual / real detection photoelectric switch and obtain the cargo virtual / real detection feedback signal.

[0037] The cargo real-object detection photoelectric switch can be a switch installed on the stacker crane to activate cargo real-object detection. The cargo real-object detection feedback signal can be the feedback signal received after the cargo real-object detection photoelectric switch emits its signal.

[0038] Specifically, once the cargo location is reached, the photoelectric switch for detecting the real or virtual cargo, which is originally set on the stacker crane, can be triggered to obtain the feedback signal of the real or virtual cargo detection.

[0039] Step 240: Determine the cargo real / virtual detection result based on the cargo real / virtual detection feedback signal.

[0040] In this embodiment of the invention, based on the type of cargo virtual-real detection feedback signal, a detection result corresponding to the type of cargo virtual-real detection feedback signal is determined, and the detection result corresponding to the type of cargo virtual-real detection feedback signal is used as the cargo virtual-real detection feedback signal.

[0041] Step 250: Take a picture of the target cargo space where the cargo virtual and real detection results show that there is cargo, and obtain the cargo space image to be identified.

[0042] In an optional embodiment of the present invention, taking a picture of the target cargo space where the cargo virtual-to-real detection result indicates the presence of cargo to obtain an image of the cargo space to be identified may include: determining the current cargo space image taking mode based on the cargo space image recognition accuracy sequence, the mode switching accuracy threshold, and the accuracy fluctuation threshold; wherein, the current cargo space image taking mode includes a scanning mode or a quick shooting mode; taking a picture of the target cargo space where the cargo virtual-to-real detection result indicates the presence of cargo according to the current cargo space image taking mode to obtain an image of the cargo space to be identified.

[0043] The accuracy sequence for cargo location image recognition can be the sequence of cargo type recognition accuracy in cargo location images within the current time window, i.e., the sequence of cargo type recognition accuracy of cargo location images already captured within a preset time range before the cargo location image to be recognized is captured. The mode switching accuracy threshold can be a cargo location image recognition accuracy threshold indicating the switching of the cargo location image capturing mode. The accuracy fluctuation threshold can be a pre-allowed accuracy error threshold. The current cargo location image capturing mode can be the currently determined capturing mode for the target cargo location space. The scanning mode can be a mode that continuously captures images at a fixed frame rate while reciprocating according to a preset travel distance. The fast capture mode can be a mode that only captures a single frame.

[0044] In this embodiment of the invention, a cargo location image recognition accuracy sequence can be obtained first. The cargo location image recognition accuracy sequence can reflect the recognition accuracy of the cargo type in the cargo location image. Therefore, the mode switching accuracy threshold and the accuracy fluctuation threshold can be used as measurement indicators to determine whether the recognition accuracy of the cargo type in the cargo location image reflected by the cargo location image recognition accuracy sequence meets the standard. When the recognition accuracy does not meet the standard, the current cargo location image shooting mode is set to the scanning mode. When the recognition accuracy meets the standard, the current cargo location image shooting mode is set to the quick shooting mode. Based on the current cargo location image shooting mode, the target cargo location space where the cargo virtual and real detection result indicates the presence of cargo is photographed to obtain the cargo location image to be identified.

[0045] In an optional embodiment of the present invention, determining the current cargo location image capture mode based on the cargo location image recognition accuracy sequence, the mode switching accuracy threshold, and the accuracy fluctuation threshold may include: calculating the average value and standard deviation of the accuracy sequence based on the cargo location image recognition accuracy sequence; when the average value of the accuracy sequence is greater than the mode switching accuracy threshold and the standard deviation of the accuracy sequence is less than the accuracy fluctuation threshold, the current cargo location image capture mode is a quick capture mode; when the average value of the accuracy sequence is less than the mode switching accuracy threshold, or the standard deviation of the accuracy sequence is greater than the accuracy fluctuation threshold, the current cargo location image capture mode is a scanning capture mode.

[0046] The mean of the accuracy sequence can be the average value of the data in the location image recognition accuracy sequence. The standard deviation of the accuracy sequence can be the standard deviation of the data in the location image recognition accuracy sequence.

[0047] In this embodiment of the invention, statistical analysis can be performed on the accuracy of cargo location image recognition based on the accuracy sequence, calculating the average and standard deviation of the accuracy sequence. Further, the average accuracy sequence is compared with a mode switching accuracy threshold, and the standard deviation is compared with an accuracy fluctuation threshold. If the average accuracy sequence is greater than the mode switching accuracy threshold and the standard deviation is less than the accuracy fluctuation threshold, then the current cargo location image capture mode is set to quick capture mode. If the average accuracy sequence is less than the mode switching accuracy threshold, or the standard deviation is greater than the accuracy fluctuation threshold, then the current cargo location image capture mode is set to scan capture mode.

[0048] In a specific example, the intelligent control platform can decide between scan mode and quick scan mode based on the continuous performance of a single frame of the storage location image. The mode switching accuracy threshold can be 0.995, and all stacker cranes switch modes synchronously to ensure strategy consistency. In scan mode, for storage locations with goods, the stacker crane's loading platform performs a scanning motion. Scan trajectory It is an independent simple harmonic motion in the horizontal and vertical directions: X t =X0+A x sin(2πf x t); Y t =Y0+A y sin(2πf y t+ Where (X0, Y0) represents the center coordinates of the cargo location. A x A y These are the scanning amplitudes in the horizontal and vertical directions, respectively. Calculated based on the cargo location opening size and camera field of view, the coverage area typically extends approximately 50mm beyond the cargo location boundary. x f yThese are the scanning frequencies for the horizontal and vertical directions, respectively, typically set to 0.2Hz to 0.5Hz. The phase difference is usually set to 1. This approximates the trajectory as an ellipse to cover a wider range of perspectives. The total scanning time is typically 24 seconds. During this time, the camera acquires images of the cargo location to be identified at a fixed frame rate. seq .

[0049] Step 260: Based on the image of the location to be identified, the results of the virtual and real detection of goods, and the location record information of the warehouse management system, determine the abnormal location.

[0050] In an optional embodiment of the present invention, determining abnormal storage locations based on the image of the storage location to be identified, the results of the virtual and real detection of goods, and the storage location record information of the warehouse management system may include: inputting the image of the storage location to be identified into a goods type recognition model to obtain a goods type recognition result; matching the goods type recognition result and the results of the virtual and real detection of goods with the storage location record information of the warehouse management system to determine the abnormal storage location; wherein, the goods type in the abnormal storage location may include virtual accounts in the system, goods without accounts, type mismatch, identity mismatch, and unreadable identifiers.

[0051] The cargo type recognition model can be used to identify the type of cargo in an image of a cargo location. Cargo type can include container / pallet type or empty pallet type. The cargo type recognition result can be the cargo type output by the recognition model.

[0052] In this embodiment of the invention, the image of the location to be identified can be analyzed by a cargo type identification model to obtain the cargo type identification result. The cargo type identification result and the cargo virtual and real detection result are then matched with the location record information of the warehouse management system. Based on the data matching result, the specific type of the abnormal location is determined.

[0053] Goods type identification model MobileNetV3Small was used as the backbone network, replacing the final classification layer with a fully connected layer of two neurons (corresponding to the cigarette box pallet group and the empty pallet group). The cargo type recognition model was trained using the cross-entropy loss function. In scanning mode, for I... seq Each frame Ii in the process is independently inferred, resulting in a probability vector P_i= =[pi_tobacco,pi_pallet]. pi_tobacco represents the probability of a cigarette box pallet group, and pi_pallet represents the probability of an empty pallet group. The final classification probability output by the cargo type recognition model is obtained by averaging the classification probability vectors of all frames in the image sequence of the cargo location to be identified. The category determination is the category corresponding to the maximum value in the classification probability vector, and the confidence score is the confidence score corresponding to the maximum value in the classification probability vector.

[0054] The cargo type recognition model continuously learns, collecting new training samples after a preset training cycle and training them. The training samples include image features a and true labels b (0 or 1).

[0055] The cargo type identification model update is achieved by minimizing the following loss function: = CrossEntropy (a),b)+ CrossEntropy is the cross-entropy loss function. For regularization terms, The parameter vector for the cargo type identification model (such as learning to adjust all weights and biases). This represents the version of the cargo type identification model at time t. For example, the Fisher information matrix penalty term in the elastic weight consolidation can be used. F i ( - ) 2 , as a regularization term, to prevent catastrophic forgetting of old knowledge. F i For parameters Fisher information, where {t, i} is a pair of coordinates, t locates the time version, and i locates the position of the parameters. Together they located the value of a specific parameter in a specific version of the model. This represents the regularization strength coefficient. The updated cargo type recognition model is evaluated on a predefined central validation set. Only when the difference between the accuracy of the updated cargo type recognition model and the accuracy of the original model is greater than a minimum improvement margin (e.g., 0.002) is the model packaged and deployed to the edge nodes. The edge nodes load the new model in the background and test it in parallel on a small number of cargo locations. Once confirmed to be error-free, they switch to the latest cargo type recognition model, while the old model is retained as a rollback backup.

[0056] Optionally, the optimal frame can be selected from the sequence of images of the location to be identified using the following formula: Q(I)=w1S Laplacian(I) +w2(1-|μ-128| / 128)+w3(1-E blur(I) )+w4C code(I) .

[0057] Among them, S Laplacian(I) E represents the variance of the Laplacian operator in the image, measuring sharpness. μ is the average gray value of the image, measuring brightness; the ideal value is approximately 128. blur(I) This represents the blur estimate based on the image gradient, ranging from [0, 1]. C code(I)This represents the confidence level of the pre-running barcode area detector output, ranging from [0, 1]. w1, w2, w3, and w4 are weighting coefficients, calibrated experimentally, and w1 + w2 + w3 + w4 = 1. In scanning mode, from I... seq The highest quality frame or the first K frames are selected for decoding. In snapshot mode, a single frame is decoded directly.

[0058] The system's virtual inventory logic is as follows: the current location's inventory detection result is no goods, but the warehouse management system records the current location as having goods. The inventory-but-not-account logic is as follows: the goods type is a cigarette box pallet group, and the warehouse management system records the current location as empty. The type mismatch logic is as follows: ((the current location's goods type is an empty pallet group type)). (The current storage location is recorded as empty in the warehouse management system.) (The current cargo type at this location is a cigarette box pallet set.) (The current storage location recorded in the warehouse management system is empty); the identity mismatch logic is (the goods in the current storage location are of the cigarette box pallet group type). (The identification of the goods is different from the identification recorded in the warehouse management system.) (The identifier is a valid string); the logic for an unreadable identifier is (the goods in the current storage location are of the cigarette box pallet group type). (The goods' identification string is invalid). If any of the above conditions are met, the item is considered abnormal; otherwise, it is considered normal. Indicates that, It indicates "or".

[0059] In an optional embodiment of the present invention, after determining the abnormal storage location based on the image of the storage location to be identified, the results of the virtual and real detection of goods, and the storage location record information of the warehouse management system, the method may further include: reporting the target storage location of the abnormal storage location to the intelligent control platform, obtaining the abnormal goods handling task created by the intelligent control platform based on the target storage location of the abnormal storage location and the priority of the abnormal storage location type; and handling the goods in the abnormal storage location to the target verification location according to the abnormal goods handling task.

[0060] The target storage location can be the location of an abnormal storage location. The abnormal storage location type priority can be a pre-set priority for handling goods in the abnormal storage location. The abnormal goods handling task can be a task to move goods in the abnormal storage location. The target verification location can be the verification location of the goods in the abnormal storage location.

[0061] In this embodiment of the invention, after the target location of the abnormal storage location is reported to the intelligent management and control platform, the intelligent management and control platform can generate a cargo handling path for the abnormal storage location according to the target location and the priority of the abnormal storage location type, and create an abnormal cargo handling task based on the cargo handling path of the abnormal storage location. The abnormal cargo handling task is then sent to the stacker crane, which moves the cargo from the abnormal storage location to the target verification location based on the abnormal cargo handling task.

[0062] The technical solution of this invention obtains and parses an inventory scheduling command to determine the location for cargo detection. Upon reaching the location, it triggers a photoelectric switch for cargo virtual-to-real detection and acquires a feedback signal. Then, it captures an image of the target cargo location where the virtual-to-real detection result indicates the presence of cargo, thus obtaining an image of the cargo location to be identified. This solution first performs virtual-to-real detection and acquires an image of the cargo location containing cargo. It then integrates the image of the cargo location to be identified, the cargo virtual-to-real detection result, and the location record information from the warehouse management system. This allows for the most comprehensive and accurate identification of abnormal cargo conditions from multiple dimensions. The identification process for abnormal cargo locations requires no manual intervention and does not rely on a large number of sensors. This solves the problem of balancing cost, efficiency, and reliability in identifying cargo locations occupied by stagnant cargo, and achieves a balance between these factors. Example 3 Embodiment 3 of the present invention provides an automatic identification system for abnormal cargo locations. Technical terms that are the same as or correspond to those in the above embodiments will not be repeated here.

[0063] The abnormal storage location automatic identification system includes a stacker crane and an intelligent management and control platform. The intelligent management and control platform is used to issue inventory and scheduling instructions to the stacker crane, and to create abnormal cargo handling tasks based on the target storage location and the priority of the abnormal storage location type.

[0064] The intelligent management and control platform, serving as the system's decision-making brain and learning center, is deployed on enterprise servers or in the cloud. The platform includes a check and scheduling module, an anomaly detection and rule engine module, a handling task optimization and scheduling module, an evidence chain reasoning module, and a model management center. The model management center is responsible for the aggregation and management of training samples, as well as the retraining, evaluation, version management, and update distribution of the cargo type recognition model.

[0065] The intelligent management and control platform is deployed in the data center and integrates with the existing warehouse management system through an enterprise service bus to obtain a logical view of inventory. Its software adopts a microservice architecture, including inventory scheduling, anomaly detection, handling optimization, and evidence chain reasoning services. The inventory scheduling service, based on the warehouse's 3D model and work plan, intelligently generates periodic full-warehouse inventory tasks and decomposes and optimizes these tasks into execution sequences for each stacker crane. The anomaly detection service uses a built-in rule engine to receive perception results (status, type, identity) reported by the stacker cranes, compares them in real-time with records in the warehouse management system, executes judgment logic, and triggers location locking. The handling optimization service models abnormal locations as dynamic handling tasks and uses operations research algorithms for real-time scheduling, aiming to minimize total operation time and task latency. The evidence chain reasoning service builds a multi-evidence (time, location, visual features) matching model for unidentified cigarette boxes, inferring the most likely identity from historical data.

[0066] The task optimization and scheduling module assigns m abnormal cargo handling tasks to n stacker cranes. Information related to the abnormal cargo handling tasks includes the target verification location (x_j, y_j) and processing time p for each abnormal cargo handling task j. j Priority weight w j Each stacker crane i has an initial position (X_i, Y_i) and a moving speed v. Define the decision variable x. {ij} ∈{0,1} indicates whether the abnormal cargo handling task j is assigned to stacker crane i; C {ij} Let be the completion time of the abnormal goods handling task j on stacker crane i. The optimization objective of the handling task optimization scheduling module is to minimize the weighted completion time: Minimize( max i (C i )+ Σ j w j Tardiness j ), where C i This indicates the completion time of the last task on stacker crane i. (Tardiness) j =max(0,C {ij} d j ), Tardiness j d represents the delay time for abnormal cargo handling task j. j The expected completion time for the task (can be set to the task generation time plus a tolerance value). , The weighting coefficients are used. The constraints of the optimization objective include that each task must be assigned to exactly one stacker crane, and the processing order of tasks on each stacker crane must satisfy the logic of cumulative movement time. The mixed-integer programming problem involved in the optimization objective is solved using an improved genetic algorithm. For example, chromosome encoding uses a two-stage approach: the first stage is the sequence of tasks assigned to stacker cranes, and the second stage is the processing order of tasks on each stacker crane. The fitness function is the reciprocal of the objective function. Iterative optimization is performed through operations such as tournament selection, sequential crossover, and exchange mutation.

[0067] For cigarette boxes with unknown identities, the evidence chain reasoning model uses an evidence set E = {L, T, F}, where L is the source location area, T is the last reliable entry timestamp of the cigarette box, and F is the deep visual features extracted from the last reliable entry image of the cigarette box (e.g., the output of the penultimate layer of a pre-trained convolutional neural network). A candidate set C is retrieved from the cargo knowledge base, satisfying the time constraint. ( (A time window, such as 12 hours). Goods Calculate the matching score Score_i = γ1K(T, Ti); )+γ2sim_loc(L,Li)+γ3cos_sim(F,Fi); where: K(T,Ti; ) is a Gaussian kernel function used to evaluate temporal proximity. Ti represents the timestamp of the cigarette box in the current storage location, and Fi represents the image of the cigarette box in the current storage location to be identified. `Score_i` represents the bandwidth parameter of the Gaussian kernel. `sim_loc(L,Li)` is the location similarity function; if cargo location L and cargo location Li are in the same lane and adjacent, the score is higher. `cos_sim` represents the cosine similarity, which measures the similarity of visual features. `γ1`, `γ2`, and `γ3` are learnable evidence weights, and `γ1+γ2+γ3=1`. Sort by `Score_i` in descending order and output the Top N candidates and their scores.

[0068] The model management center is the core of the system's evolution. It is responsible for collecting training samples (image feature vectors + real labels) from various verification workstations and maintaining a centralized sample library. The model training pipeline is periodically launched, using incremental learning and other techniques to iteratively optimize the cargo type recognition model, and rigorously evaluating the performance of new models through A / B testing. New models that meet performance standards are packaged into update packages and distributed to all online enhanced stacker cranes via a secure channel. Visual monitoring service: Provides managers with a global visualization interface. This interface displays the real-time status of all storage locations in the high-bay warehouse in a two-dimensional plan view. Each storage location corresponds to an interactive graphical element on the map, and its display attributes (color, icon, border, etc.) are dynamically updated based on the current status.

[0069] The stacker crane, acting as the intelligent sensor and agile arm of the automatic identification system for abnormal storage locations, is an upgrade from the standard stacker crane. Key upgrades include the integration of the original cargo-based photoelectric switches (typically diffuse reflection type) on the loading platform, a new visual acquisition module consisting of an industrial camera and a high-brightness ring light, and a new edge computing unit with neural network acceleration capabilities, featuring a built-in artificial intelligence inference engine and model update interface. The upgraded control system software supports configurable area scanning motion control. The stacker crane's camera optical axis undergoes precise laser calibration to ensure strict parallelism with the fork movement plane, and its field of view center coincides with the centering position of the loading platform, guaranteeing complete coverage of materials within the storage location. The installation position and sensitivity of the cargo-based photoelectric switches require recalibration during system deployment to ensure reliable detection of even the lowest-height cigarette pallets.

[0070] The automatic identification system for abnormal storage locations can also include a verification workstation (target verification location). As a "human-machine collaboration hub," the verification workstation is located near the warehouse entrance and exit, equipped with an industrial touchscreen computer and necessary auxiliary scanning equipment. Its software interface clearly displays abnormal information and system inference results, and allows for convenient input of manual confirmation information, while automatically triggering the generation and uploading of training samples. The verification workstation uses an industrial communication network, such as a hybrid architecture of highly reliable industrial Ethernet and wireless networks, to ensure real-time, stable, and secure transmission of instructions and data (especially image data and model update packages) between the platform, stacker crane, and workstation. The intelligent control platform, stacker crane, and verification workstation are connected via a high-speed industrial communication network.

[0071] Inside the control box of the loading platform, an embedded industrial computer (edge ​​computing unit) with artificial intelligence acceleration capabilities is integrated. This unit is connected to the camera and the main controller via a high-speed bus and is responsible for running a lightweight cargo type recognition model. It realizes localized real-time processing of image acquisition, preprocessing, model inference, and result reporting, which greatly reduces network bandwidth pressure and reduces recognition latency.

[0072] A new "intelligent inventory" function block is embedded in the original programmable logic controller or motion controller of the stacker crane. This function block can parse the inventory scheduling instructions issued by the platform, and after completing the routine positioning, decide whether to start the scanning motion based on the current image capture mode of the cargo location, and coordinate the synchronous work of the camera, supplementary light, and edge computing unit.

[0073] Verification workstations are typically located in the warehouse's inbound / outbound or maintenance areas, equipped with dustproof industrial computers. Their software client communicates in real-time with the intelligent management platform, automatically receiving and listing information on abnormal materials moved to the station. The interface intuitively displays the system's raw data (such as captured images), identification results, and (for cigarette boxes with "unreadable labels") a candidate list derived from the chain of evidence. Operators make final confirmations by scanning the physical barcode (if readable) or based on experience. After clicking confirm, the result is immediately transmitted back to the platform, driving the warehouse management system to repair data and generate training samples.

[0074] For example, the initialization phase mode is set to scan mode, and the initial cargo type recognition model is loaded. The cargo type recognition model has been trained on historical datasets and can achieve a high multi-frame fusion recognition accuracy (e.g., 98.5%), but the single-frame recognition rate may be insufficient (e.g., 92%).

[0075] The intelligent control platform initiates an inventory check and dispatch command. Based on this command, the stacker crane accesses storage location A001, triggering the photoelectric switch for cargo detection. The detection result is empty, and the record is completed instantly. Accessing storage location A002, the detection result is present, initiating a scan and acquiring 15 image frames. After analysis by the edge computing unit, it is determined to be an empty pallet group with a 96% confidence level. The warehouse management system records this as an empty pallet group, consistent with the status, indicating normal operation.

[0076] Accessing storage location B105, the goods detection results indicate that goods are present. After scanning, the goods type identification model determines it to be a cigarette box pallet with a 93% confidence level. Decoding attempt failed (identifier unreadable). The warehouse management system records that this storage location should contain cigarette boxes, but the identification is "X", which is determined as "identifier unreadable". The storage location is locked, and an abnormal goods handling task is generated.

[0077] The intelligent control platform assigns the task for pallet location B105 to an idle stacker crane, which then moves the pallet to the verification workstation. At the workstation, the operator verifies the cigarette box's true identity as "Y" based on the printed information, updates the warehouse management system, binds the identity "Y" to the pallet, and unlocks it. Simultaneously, the verified "cigarette box" label and its corresponding 15 image frames are uploaded to the intelligent control platform as a set of training samples.

[0078] The model management center collects thousands of new samples weekly for incremental training. After three months, the single-frame recognition accuracy of the new cargo type identification model reached 99.6% on the independent test set, and remained stable above 99.5% for four consecutive weeks. The intelligent control platform's decision module was triggered, issuing instructions to all stacker cranes to switch their operating mode to snapshot mode.

[0079] During subsequent inspections, for location C201 (with goods), only one image frame was captured after the loading platform came to a complete stop. The new cargo type recognition model directly identified it as a cigarette box pallet with 99% confidence, decoding was successful, and the comparison was normal. The entire processing time was reduced from approximately 5 seconds to less than 1 second.

[0080] To provide a complete management loop and traceability, a structured inventory report is automatically generated after each inventory task is completed, and detailed inventory logs are continuously recorded. These reports and logs are stored in the database of the intelligent management and control platform for managers to query, analyze, and audit.

[0081] The inventory check report mainly includes a report summary, anomaly summary, identification performance, and handling recommendations. The report summary includes the inventory check task identifier, execution time range, involved lanes / areas, total number of storage locations, and number of locations completed. Efficiency statistics include total time consumed, average processing time per storage location, percentage of empty storage locations, and percentage of non-empty storage locations. The anomaly summary includes the total number of anomalies found, categorized by anomaly type (system fictitious entries, items present but not recorded, type mismatch, identity mismatch, unreadable identifier). Identification performance includes the average confidence level and identification success rate (based on feedback from subsequent manual verification) for cargo type identification during this inventory check. Handling recommendations are preliminary handling suggestions or priority prompts based on the anomaly summary.

[0082] The inventory log records more granular process data. Each inventory location's inspection result generates a log record containing the following fields: log identifier, associated inventory task, inspection timestamp, inventory location coordinates (aisle, row, floor), cargo virtual / real detection photoelectric switch signal (0 / 1), current inventory location image capture mode, cargo type (empty / cigarette box / empty pallet group), identification confidence level, barcode decoding result (success / failure), decoded identity information, inventory location status recorded by the warehouse management system, material type recorded by the warehouse management system, identity information recorded by the warehouse management system, abnormality type of cargo in the identified abnormal inventory location (if no abnormality, it is "normal"), action taken (none / marked abnormal / generated handling task), and associated image storage path or feature vector index.

[0083] Inventory reports are displayed on the intelligent management platform interface in the form of visual charts (such as pie charts and bar charts) and tables, and can be exported as PDF or table format. Inventory logs support filtering and querying by multiple dimensions such as time, location, and anomaly type, providing a data foundation for inventory analysis, equipment performance evaluation, and system optimization.

[0084] The intelligent management and control platform's visual monitoring service provides an intuitive, real-time updated two-dimensional warehouse location monitoring screen. Based on a warehouse layout diagram, this screen uses different graphical elements and color coding to display the real-time status of each location. This feature provides managers with a global perspective, greatly improving the visualization of the inventory process and the timeliness of anomaly response.

[0085] The core functions and display logic of the monitoring screen are as follows: 1) Color coding for the status of storage locations. Each storage location is rendered as a rectangle or cell on the two-dimensional plan view. Its fill color is dynamically determined according to the current status, following the mapping rules below: Gray indicates that the storage location has not yet been checked (pending processing). Green indicates that the goods in the storage location have been checked and the inventory matches the records, and the status is normal. Red indicates that the goods in the storage location have an "item present but not recorded" anomaly. Orange indicates that the goods in the storage location have a "system fictitious record" anomaly. Yellow indicates that the goods in the storage location have a "type mismatch" anomaly. Purple indicates that the goods in the storage location have an "identity mismatch" anomaly. Blue indicates that the goods in the storage location have an "unreadable identifier" anomaly. Through this clear color differentiation, managers can clearly grasp the distribution and types of anomalies throughout the warehouse. 2) The monitoring screen reflects the execution progress of the inventory check task in real time. As the stacker crane scans, the color of the accessed storage location will change from "gray" (not checked) to other colors according to the judgment result. The top or side of the screen can display statistical information, such as "Total Number of Storage Locations Checked" and "Quantity of Various Anomalies," along with a progress bar to visualize the inventory progress. 3) Users can hover the mouse over or click on any storage location graphic to bring up an information card displaying detailed information about that location, including: location coordinates, warehouse management system records (material type, identification), actual perception results (type, confidence level, decoded identification), anomaly type, lock status, and last inspection time. For abnormal storage locations, the information card provides quick operation buttons, such as "View Details," "Jump to Processing Order," and "Manually Unlock" (requires permissions).

[0086] On the 2D plan view, the stacker crane currently performing an inventory check is displayed with a dynamic icon featuring a directional arrow, showing its real-time location and direction of movement. This allows managers to intuitively understand the equipment's operating status and current position. The monitoring screen provides a filtering toolbar, allowing users to dynamically filter the displayed storage locations by criteria such as "anomaly type," "aisle," and "whether it has been checked," facilitating focused attention on specific problem areas. It supports searching by storage location coordinates or cargo identity to quickly locate a specific storage location.

[0087] The monitoring screen is developed based on web technology, and the backend maintains a long-term connection with the core services of the intelligent management and control platform. Whenever the status of a storage location changes (such as completing an inventory check, identifying an anomaly, or the anomaly being handled and repaired), the intelligent management and control platform immediately pushes the status update event to the frontend. The frontend then updates the color and attributes of the corresponding storage location accordingly, achieving near real-time global status synchronization.

[0088] To ensure system performance, the following key steps must be performed during deployment: a) Sensor installation and calibration: Industrial cameras are calibrated using a laser calibrator to ensure the camera's optical axis is perpendicular to the loading platform plane and parallel to the fork extension direction. Camera intrinsic parameters (focal length, distortion coefficient) are calibrated. The brightness and angle of supplementary lighting are adjusted to ensure uniform, non-reflective illumination even at the deepest point of the storage location. The photoelectric switch for cargo detection uses a standard test block (simulating the height of a cigarette box) to calibrate its detection distance and sensitivity, ensuring stable detection of single-layer cigarette box pallets and no false alarms for empty storage locations. b) Manually control the stacker crane and measure the actual opening width and depth of the storage location. Calculate the camera's single-frame field of view based on the camera's field of view angle and object distance. Set the scanning amplitude A. x (Deepness of the storage bay opening in the horizontal view of the camera) / 2 + margin, A y (Width of the storage location opening under the camera's horizontal field of view) / 2 + margin, where margin is a safety margin (e.g., 20mm). c. Control the stacker crane to acquire images of storage locations in various typical conditions in the warehouse (different lighting, different brands of cigarette boxes, new and old empty pallet groups), and manually annotate them precisely to construct a high-quality initial training dataset. d. During the trial operation, adjust the confidence threshold and mode switching accuracy threshold based on the actual recognition results to balance recognition speed, accuracy, and system stability.

[0089] A creative "photoelectric initial screening" strategy was adopted, utilizing existing sensors with millisecond-level response to instantly process a large number of empty storage locations. Complex visual computing resources were then concentrated on non-empty storage locations that truly needed differentiation, resulting in a qualitative leap in overall inventory efficiency while ensuring the accuracy of key identifications. Breaking through the limitations of traditional systems that are "deployed and then finalized," a complete "perceptual learning" closed loop was designed, enabling the system to continuously optimize its core artificial intelligence model using real data generated during production. This not only steadily improves recognition accuracy but, more importantly, drives the entire recognition process to evolve from complex "multi-frame scanning" to efficient "single-frame snapshots," achieving automatic upgrades in long-term operational efficiency—something static systems cannot match. The existing, proven positioning system and photoelectric sensors of the stacker crane were reused to the greatest extent possible, with newly added vision and computing modules being mature products from the industry. This "retain the core, enhance intelligence" transformation model involves minimal hardware modifications, low implementation risk, and controllable transformation costs, making it highly suitable for rapid deployment in existing warehouses. The recognition logic covers all common anomaly types, including information loss, incorrect identity, and misplacement of types. From automatic discovery, intelligent classification, optimized scheduling and handling to assisted verification and repair, the entire process is automated, requiring only low-complexity confirmation at key decision points, forming a highly intelligent management closed loop. This directly releases inventory funds and storage space locked up by stagnant materials, improving inventory turnover and asset utilization. By ensuring material traceability and supply continuity, it indirectly improves production efficiency and product quality stability. Fully automated inventory checks and processing significantly reduce labor costs and safety risks, enhancing the digitalization and intelligence of warehouse management.

[0090] Example 4 Figure 3 This is a schematic diagram of the structure of an abnormal cargo location identification device provided in Embodiment 4 of the present invention. Figure 3 As shown, the device includes: The instruction acquisition module 310 is used to acquire the query and scheduling instructions; The virtual-to-real detection module 320 is used to perform virtual-to-real detection of goods according to the inspection and scheduling instructions, and obtain the virtual-to-real detection results of goods; The image capturing module 330 is used to capture images of the target cargo space where the cargo virtual-real detection result indicates the presence of cargo, thereby obtaining an image of the cargo space to be identified; The abnormal storage location determination module 340 is used to determine abnormal storage locations based on the image of the storage location to be identified, the results of the virtual and real detection of goods, and the storage location record information of the warehouse management system.

[0091] The technical solution of this invention obtains an inventory and scheduling instruction, then performs a virtual-to-real detection of goods based on the instruction, obtains the detection results, and then photographs the target storage space where goods are found to be present, obtaining an image of the storage space to be identified. Based on the image of the storage space to be identified, the virtual-to-real detection results, and the storage space record information from the warehouse management system, abnormal storage spaces are determined. This solution first performs virtual-to-real detection and collects images of storage spaces containing goods to be identified, then integrates the images of the storage spaces to be identified, the virtual-to-real detection results, and the storage space record information from the warehouse management system. This allows for the most comprehensive and accurate identification of abnormal conditions of goods in storage spaces from multiple dimensions. The identification process of abnormal storage spaces requires no manual intervention and does not rely on a large number of sensors, solving the problem of balancing cost, efficiency, and reliability in identifying storage spaces occupied by stagnant goods. It can accurately identify storage spaces occupied by stagnant goods while balancing cost and efficiency.

[0092] Optionally, the virtual-real detection module 320 is used to parse the inspection and scheduling instruction, determine the cargo location detection position; after arriving at the cargo location detection position, it triggers the cargo virtual-real detection photoelectric switch and obtains the cargo virtual-real detection feedback signal; and determines the cargo virtual-real detection result based on the cargo virtual-real detection feedback signal.

[0093] Optionally, the image capturing module 330 is used to determine the current cargo location image capturing mode based on the cargo location image recognition accuracy sequence, mode switching accuracy threshold, and accuracy fluctuation threshold; wherein, the current cargo location image capturing mode includes a scanning mode or a quick capture mode; according to the current cargo location image capturing mode, the target cargo location space where the cargo virtual and real detection result indicates the presence of cargo is captured to obtain the cargo location image to be identified.

[0094] Optionally, the image capturing module 330 includes a shooting mode determination unit, used to calculate the average value and standard deviation of the accuracy sequence based on the accuracy sequence of the cargo location image recognition; when the average value of the accuracy sequence is greater than the mode switching accuracy threshold and the standard deviation of the accuracy sequence is less than the accuracy fluctuation threshold, the current cargo location image shooting mode is a quick capture mode; when the average value of the accuracy sequence is less than the mode switching accuracy threshold, or the standard deviation of the accuracy sequence is greater than the accuracy fluctuation threshold, the current cargo location image shooting mode is a scan capture mode.

[0095] Optionally, the abnormal storage location determination module 340 is used to input the image of the storage location to be identified into the cargo type recognition model to obtain the cargo type recognition result; and to match the cargo type recognition result and the cargo virtual-to-real detection result with the storage location record information of the warehouse management system to determine the abnormal storage location; wherein, the types of abnormal storage locations include virtual accounts in the system, items without records, type mismatch, identity mismatch, and unreadable identifiers.

[0096] Optionally, the abnormal storage location identification device further includes a verification and handling module, which is used to report the target storage location of the abnormal storage location to the intelligent management and control platform, obtain the abnormal goods handling task created by the intelligent management and control platform based on the target storage location of the abnormal storage location and the abnormal storage location type priority; and handle the goods in the abnormal storage location to the target verification location according to the abnormal goods handling task.

[0097] The abnormal storage location identification device provided in this embodiment of the invention can execute the abnormal storage location identification method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0098] Example 5 Figure 4 A schematic diagram of a stacker crane that can be used to implement embodiments of the present invention is shown. The stacker crane is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The stacker crane can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0099] like Figure 4 As shown, the stacker crane 10 includes at least one processor 11 and a memory, such as ROM 12, RAM 13, etc., communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded into the RAM 13 from the storage unit 18. The RAM 13 can also store various programs and data required for the operation of the stacker crane 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An I / O interface 15 is also connected to the bus 14. The ROM 12 is a read-only memory, the RAM 13 is a random access memory, and the I / O interface 15 is an input / output interface.

[0100] Multiple components in the stacker crane 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, optical disk, etc.; and a communication unit 19, such as a network card, modem, wireless transceiver, etc. The communication unit 19 allows the stacker crane 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0101] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as methods for identifying abnormal storage locations.

[0102] In some embodiments, the method for identifying abnormal storage locations may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the stacker crane 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for identifying abnormal storage locations described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the method for identifying abnormal storage locations by any other suitable means (e.g., by means of firmware).

[0103] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0104] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0105] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, RAM, ROM, erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0106] To provide user interaction, the systems and techniques described herein can be implemented on a stacker crane, which includes: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the stacker crane. Other types of devices can also be used to provide user interaction; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0107] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0108] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS servers, such as high management difficulty and weak business scalability.

[0109] This application also discloses a computer program product, which includes a computer program that, when executed by a processor, implements the abnormal storage location identification method provided in any embodiment of this application. This program product and the abnormal storage location identification methods disclosed in the embodiments of this application belong to the same inventive concept, and therefore will not be described in detail here.

[0110] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0111] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for identifying abnormal storage locations, characterized in that, Performed by a stacker crane, including: Obtain the query and dispatch instructions; According to the inspection and dispatch instructions, a cargo virtuality detection is performed to obtain the cargo virtuality detection results; The target cargo space where the virtual and real detection results indicate the presence of cargo is photographed to obtain an image of the cargo space to be identified. Based on the image of the location to be identified, the results of the virtual and real detection of the goods, and the location record information of the warehouse management system, the abnormal location is determined.

2. The method according to claim 1, characterized in that, According to the aforementioned inspection and dispatch instructions, a cargo verification process is performed to obtain the cargo verification results, including: Analyze the inventory and dispatch instructions to determine the location of the cargo to be inspected; After reaching the cargo location detection position, the cargo virtual and real detection photoelectric switch is triggered, and the cargo virtual and real detection feedback signal is obtained; The cargo virtuality / reality detection result is determined based on the cargo virtuality / reality detection feedback signal.

3. The method according to claim 1, characterized in that, The target cargo location space where the virtual and real detection results indicate the presence of cargo is photographed to obtain an image of the cargo location to be identified, including: Based on the image recognition accuracy sequence, mode switching accuracy threshold, and accuracy fluctuation threshold, the current image capture mode of the storage location is determined; wherein, the current image capture mode of the storage location includes either a scanning mode or a quick capture mode. According to the current cargo location image shooting mode, the target cargo location space where the cargo virtual and real detection results indicate the presence of cargo is photographed to obtain the cargo location image to be identified.

4. The method according to claim 3, characterized in that, Based on the image recognition accuracy sequence, mode switching accuracy threshold, and accuracy fluctuation threshold, the current image capture mode for the storage location is determined, including: Based on the accuracy sequence of the location image recognition, calculate the average value and standard deviation of the accuracy sequence; When the average value of the accuracy sequence is greater than the mode switching accuracy threshold and the standard deviation of the accuracy sequence is less than the accuracy fluctuation threshold, the current cargo location image capture mode is the quick capture mode. When the average value of the accuracy sequence is less than the mode switching accuracy threshold, or the standard deviation of the accuracy sequence is greater than the accuracy fluctuation threshold, the current cargo location image capture mode is the scanning mode.

5. The method according to claim 1, characterized in that, Based on the image of the location to be identified, the results of the virtual and real detection of the goods, and the location record information of the warehouse management system, abnormal locations are determined, including: The image of the cargo location to be identified is input into the cargo type recognition model to obtain the cargo type recognition result; The cargo type identification result and the cargo virtual-to-real detection result are matched with the cargo location record information of the warehouse management system to determine the abnormal cargo location; The types of goods in the abnormal storage locations include those in the system's virtual accounts, those with goods but no record, those of mismatched types, those of mismatched identities, and those with unreadable identifiers.

6. The method according to claim 1, characterized in that, After determining the abnormal storage location based on the image of the storage location to be identified, the results of the virtual and real detection of the goods, and the storage location record information of the warehouse management system, the process further includes: After reporting the target location of the abnormal storage location to the intelligent management and control platform, the platform obtains the abnormal cargo handling task created by the intelligent management and control platform based on the target location of the abnormal storage location and the priority of the abnormal storage location type. According to the abnormal cargo handling task, the cargo in the abnormal cargo location is moved to the target verification location.

7. A stacker crane, characterized in that, The stacker crane includes: At least one processor, and a memory communicatively connected to said at least one processor; The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the method for identifying abnormal storage locations as described in any one of claims 1-6.

8. An automatic identification system for abnormal storage locations, characterized in that, Includes the stacker crane as described in claim 7 and the intelligent control platform; The intelligent management and control platform is used to issue inspection and scheduling instructions to the stacker crane, and to create abnormal cargo handling tasks based on the target location of the abnormal cargo location and the priority of the abnormal cargo location type.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method for identifying abnormal storage locations as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method for identifying abnormal storage locations according to any one of claims 1-6.

Images