An automatic goods inspection method and device

Abstract

The application discloses an automatic goods inspection device, which comprises an inspection module, a warehouse matching module and a scheduling module, wherein the inspection module comprises an inspection AMR, a data transmission device, a telescopic support frame, a camera assembly and a motor assembly; the inspection AMR is provided with the telescopic support frame; the inspection AMR is connected with the data transmission device; the data transmission device is further connected with the motor assembly and used for controlling the motor assembly; the telescopic support frame is connected with the camera assembly; the warehouse module comprises a goods shelf, a standardized goods box and a ground grid; and the scheduling module is used for scheduling the inspection AMR. Based on the device, the application further provides an automatic goods inspection method. Compared with the prior art, the application has the beneficial effects that the inspection of goods in a warehouse can be automated, other links in a supply chain can know the state of goods, corresponding decisions and arrangements can be made in time, and therefore the response capability and the synergy of the whole supply chain are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent warehousing, in particular to an automatic goods inspection method and an automatic goods inspection device. BACKGROUND

[0002] With the rapid development of modern industry, the demand for production efficiency and quality control in the manufacturing industry is becoming higher and higher. In the cigarette industry, ensuring the quality and consistency of cigarette products is crucial to maintaining market competitiveness. Goods inspection is an indispensable part of the quality control process. However, traditional manual goods inspection methods have low efficiency, are prone to errors, and are difficult to achieve full coverage. Specifically, due to the neat stacking of standard warehouse boxes, traditional manual inspection has problems such as missed inspection, repeated inspection, and false inspection. In addition, some standard warehouse boxes are placed at high positions, making it difficult for workers to fully cover. Moreover, this tedious and monotonous inspection work also greatly reduces the work efficiency of workers.

[0003] Although the existing automatic goods inspection equipment can liberate manual labor, it also has defects. The main defect is that the real-time adjustment capability of the scheduling scheme of the automatic goods inspection equipment is poor. Once an unexpected situation occurs, it is easy to cause the idle or conflict of the inspection robot, and thus the predetermined work efficiency cannot be achieved. To solve the problem of insufficient adjustment capability of the scheduling scheme, sufficient time is often reserved for each link in actual use, which causes the existing automatic goods inspection equipment to complete the work smoothly, but at the same time, time and efficiency are sacrificed.

[0004] Therefore, there is an urgent need to develop an automatic inspection method and an automatic inspection device that can efficiently and accurately complete the inspection task. This can improve production efficiency and quality control level.

[0005] In view of the above-mentioned defects, the present inventors have finally obtained the present application after long-term research and practice. SUMMARY

[0006] Therefore, in order to achieve efficient inspection, the present application provides an automatic goods inspection method, which comprises the following steps:

[0007] Step S1, a scheduling module generates an inspection task instruction and sends the inspection task instruction to a data transmission device;

[0008] Step S2, the data transmission device sends a control instruction to the inspection AMR, and the data transmission device determines the moving height of the camera according to the number of shelves in the inspection task;

[0009] Step S3, the inspection AMR departs for the designated location, the inspection AMR starts the ground grid positioning to determine whether the designated location is reached, and then the inspection AMR reaches the designated location area;

[0010] Step S4, the data transmission device controls the motor assembly to act, and the motor assembly adjusts the height of the telescopic support frame and the height of the camera;

[0011] Step S5, the camera starts shooting, and the inspection AMR starts moving according to the inspection path information in the inspection task instruction and the two-dimensional code information on the ground grid;

[0012] Step S6, the same-area same-layer priority inspection strategy is executed;

[0013] Step S7, when the same-layer standardized box scanning is completed, the motor assembly starts and adjusts the camera to the height of the next layer to continue shooting and inspection;

[0014] Step S8, when the shelves in the current area are completely scanned, the inspection AMR moves to another area to continue scanning;

[0015] Step S9, steps S5 to S8 are repeated until the shelves and standardized boxes in all areas are completely scanned.

[0016] Preferably, during the execution of the inspection task by the inspection AMR, the scheduling module generates a new constraint and scheduling scheme according to the information from the inspection AMR, and then re-executes step S1.

[0017] Preferably, the scheduling module uses a DQN algorithm for recursive solution to obtain a feasible scheduling scheme, wherein the Q(s t ,a t ) calculation method in the DQN algorithm is as follows:

[0018]

[0019] wherein rt represents the instant reward at time t, rt is 50 when the target is reached, rt is -10 when a collision or timeout occurs, represents the probability of transferring to state st+1 under the current state st and action at, represents the maximum reward value that can be obtained by taking different actions at state st+1, γ is a discount factor, representing the influence degree of future rewards on the current action of the inspection AMR; and the update formula of Q(s t ,a t ) is as follows:

[0020]

[0021] Wherein, a is a learning rate, and a value range of a is a [0, 1]; and gamma is a discount factor, and represents a degree of influence of a future reward on a current action of the inspection AMR.

[0022] Preferably, the scheduling module gives an instantaneous reward function of the inspection AMR as follows:

[0023] When the inspection AMR reaches the target point, the value of the instantaneous reward function is C1; when the inspection AMR collides, the value of the instantaneous reward function is C2; and in other cases, the value of the instantaneous reward function is K*cos theta / d, wherein K is a hyperparameter, gamma is an included angle between a vector n1 from a position of the inspection AMR to a position of the target point before the inspection AMR makes an action and a directional vector n2 before and after the action of the inspection AMR, and d is a distance between the position of the inspection AMR and the position of the target point.

[0024] Preferably, the scheduling module adopts a scheduling strategy of waiting at a conflict point in a multi-AMR system, the scheduling strategy simulates an additional operation time of the inspection AMR passing through the conflict point, adopts an Erlang distribution model, and assumes that a probability of collision in the process of inspecting n shelves is beta, and a preset conflict allowance time formula is t0=beta*n*m*y, wherein m is a number of conflict points passed through by the inspection AMR, and y is a mean value of the additional operation time.

[0025] The application also provides an automatic goods inspection device for the automatic goods inspection method, which comprises an inspection module, a warehouse matching module and a scheduling module.

[0026] Preferably, the inspection module comprises an inspection AMR, a data transmission device, a telescopic support frame, a camera assembly and a motor assembly, the inspection AMR is provided with the telescopic support frame, the inspection AMR is connected with the data transmission device, the data transmission device is further connected with the motor assembly, and the data transmission device is used for controlling the motor assembly; the telescopic support frame is connected with the camera assembly; the warehouse module comprises shelves, standardized cases and a ground grid; and the scheduling module is used for decomposing an inspection task to a plurality of inspection AMRs and planning a running route of the inspection AMR.

[0027] Preferably, the motor assembly comprises a first motor and a second motor, the first motor is used for driving the telescopic support frame, and the second motor is used for driving the camera assembly.

[0028] Preferably, the telescopic support frame comprises a frame body, a telescopic support and a first transmission device, the first transmission device is connected with the first motor, and the first transmission device is also connected with the telescopic support; the telescopic support is installed on the frame body, and the frame body is arranged on the inspection AMR; the camera assembly comprises a camera and a second transmission device, and the second transmission device is connected with the second motor and the camera respectively.

[0029] Preferably, the standardized box is placed on the shelf, a first two-dimensional code is arranged on the standardized box, a second two-dimensional code is arranged on the shelf, and a third two-dimensional code is arranged on the ground grid.

[0030] Compared with the prior art, the method and device provided by the application have the following beneficial effects:

[0031] The up-and-down movable camera can shoot the two-dimensional code on the side of the standardized box to obtain the related information of the goods. The standardized boxes can be stacked on the shelf, and the camera can read the information of one layer of boxes in one trip. Every two rows of shelves form a group, and the inspection device can complete the inspection task of one layer of goods of two rows of shelves in one round. After adjusting the height of the camera, the information of other layers of boxes can be read. The scheduling module solves a set of feasible AMR scheduling schemes based on the DQN algorithm for the multi-AMR inspection scene, selects the scheme that can meet the least time, and improves the utilization rate of the AMR as much as possible. The application can automate the inspection of goods in the warehouse, help other links in the supply chain understand the state of the goods, and make corresponding decisions and arrangements in time, thereby improving the response capability and synergy of the entire supply chain. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0033] Figure 1 It is a side view structural schematic diagram of an automatic goods inspection device in the embodiment of the application.

[0034] Figure 2 It is a front view structural schematic diagram of a shelf and a standardized box in the embodiment of the application.

[0035] Figure 3 It is a plan view of a warehouse.

[0036] Figure 4 It is a flowchart of an automatic goods inspection method in the embodiment of the application.

[0037] Reference signs:

[0038] Inspection AMR 1, data transmission device 2, telescopic support frame 3, telescopic belt 4, first motor 5, second motor 6, camera 7, shelf 8, standardized box 9, first two-dimensional code 10, second two-dimensional code 11 and third two-dimensional code 12. DETAILED DESCRIPTION

[0039] The above and other technical features and advantages of the present application will be more apparent from the following description taken in conjunction with the accompanying drawings.

[0040] In the description of the present application, the terms "first", "second", "third" and the like in the description are used only for distinguishing between similar elements and do not necessarily indicate the number or order of such elements. By the term "a plurality" is meant two or more, unless explicitly stated otherwise.

[0041] In the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting", "fixing" and the like should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or it can be integrated; it can be mechanically connected, or it can be electrically connected or communicate with each other; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the internal communication of two elements or the interaction relationship between two elements, unless otherwise explicitly limited. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0042] In the description of the present application, the description of the terms "one embodiment", "some embodiments", "example", "specific example" or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application.

[0043] In the following description, specific details are set forth in order to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application can be practiced without these specific details. In other instances, well-known systems, devices, circuits and methods have been abridged to avoid unnecessary detail, so as not to obscure the description of the present application.

[0044] In this specification, the illustrative representations of the above-mentioned terms are not necessarily directed to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Furthermore, the person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples, without mutual contradiction.

[0045] In order to illustrate the technical solutions of the present application, the following will be described by specific embodiments.

[0046] Embodiment one

[0047] As Figures 1 to 3 shown, the present application provides an automatic goods inspection device, which comprises an inspection module, a warehouse matching module and a scheduling module.

[0048] The inspection module comprises an inspection AMR1 (autonomous mobile robot), a data transmission device 2, a telescopic support frame 3, a camera assembly and a motor assembly.

[0049] The telescopic support frame 3 is arranged on the inspection AMR1, and the inspection AMR1 is connected with the data transmission device 2. The data transmission device 2 is also connected with the motor assembly for controlling the motor assembly.

[0050] The motor assembly comprises a first motor 5 and a second motor 6. The first motor 5 is used to drive the telescopic support frame 3. The second motor 6 is used to drive the camera assembly.

[0051] The telescopic support frame 3 comprises a frame body, a telescopic support and a first transmission device. The first transmission device is connected with the first motor 5, and the first transmission device is also connected with the telescopic support. The telescopic support is installed on the frame body, and the frame body is arranged on the inspection AMR1. The first motor 5 adjusts the height and attitude of the telescopic support by driving the first transmission device, which can also directly or indirectly adjust the height and attitude of the camera 7. Preferably, the first transmission device is a chain.

[0052] In addition, there are obstacles at high places in some positions of the corridor. After the data transmission device 2 obtains the obstacle position, the height and attitude of the telescopic support frame 3 are adjusted by driving the first motor 5 to avoid touching the obstacle.

[0053] The camera assembly is connected with the telescopic support. The movement of the telescopic support can drive the movement of the camera assembly. The camera assembly comprises a camera 7 and a second transmission device. The second transmission device is connected with the second motor 6 and the camera 7 respectively. The second motor 6 drives the second transmission device to make the camera 7 move up and down. Preferably, the second transmission device is a telescopic belt 4.

[0054] Camera 7 is used to take pictures and identify the two-dimensional code information on the shelves and standardized boxes.

[0055] Preferably, the first motor 5 is arranged on the side of the telescopic support frame 3. The first motor 5 is used to adjust the telescopic support frame.

[0056] Preferably, the second motor 6 is arranged at the bottom of the telescopic support frame 3. The second motor 6 can control the up and down movement of the camera 7 to adapt to different levels of shelves.

[0057] Preferably, the telescopic support frame 3 is located directly above the inspection AMR 1, and the telescopic support frame 3 is telescopic, which is used to avoid obstacles at high places in the warehouse.

[0058] Preferably, the inspection AMR is preferably an embedded AMR, and the endurance time is 6 to 8 hours. The inspection AMR 1 can obtain its current position in the warehouse by identifying the markers embedded on the ground grid. The inspection AMR 1 can adjust its orientation and position.

[0059] Preferably, the data transmission device is installed on the inspection AMR 1 and is adjacent to the telescopic support frame 3, which is used to drive the motor assembly and send the information taken by the camera, and receive the task requirements sent by the dispatching system.

[0060] The warehouse module includes shelves 8, standardized boxes 9, and ground grids.

[0061] The standardized boxes 9 are placed on the shelves 8, and the second two-dimensional code 11 is arranged on the shelves 8. The second two-dimensional code 11 records the shelf identification, position information, and the specifications of the stored standardized boxes. The standardized boxes 9 are also provided with the first two-dimensional code 10. The first two-dimensional code 10 contains information about the goods. The shelves and standardized boxes are the target objects that the inspection device needs to identify and process. The ground is divided into shelf area and corridor area, and the corridor area is provided with ground grids, and the center of the ground grid is pasted with the third two-dimensional code 12. The third two-dimensional code 12 is used to guide the movement path of the inspection AMR 1. The two-dimensional code in the ground grid can be read by the bottom camera of the inspection AMR 1, so as to realize accurate positioning and navigation. The inspection AMR 1 can use the third two-dimensional code 12 on the ground grid identified by itself to accurately adjust the forward direction and distance.

[0062] Preferably, the second two-dimensional code 11 is arranged on the side of the shelf 8. The first two-dimensional code 10 is arranged on the outside of the standardized box 9. The ground grid is a square grid. The standardized boxes 9 are stored using the shelves 8, and the shelves 8 can be stacked, with a maximum of 3 layers. Each column in each layer of the shelf can place a maximum of 2 standardized boxes, and 3 layers of standardized boxes can be placed on a shelf. The second two-dimensional code 11 is pasted on the support column of the shelf 8.

[0063] Preferably, the standardized boxes have several specifications, each with fixed length, width and height parameters. Boxes of different specifications are stored in different shelves. The standardized boxes, used to store finished goods, are placed on the shelves 8, and the outside fixed height is attached with a two-dimensional code containing the goods information.

[0064] The scheduling module is used to decompose the inspection task to multiple inspection AMRs 1 and plan the running route of the inspection AMRs 1.

[0065] The inspection AMR 1 is the core of the device, which can move autonomously and perform inspection tasks. The inspection AMR 1 is equipped with other components and performs path planning and obstacle avoidance through an autonomous navigation system.

[0066] The camera 7 is used to shoot the two-dimensional code on the shelves and standardized boxes. The position of the camera 7 can be adjusted according to the task.

[0067] The data transmission device 2 is used for data exchange and communication with other systems. Through the data transmission device, the automatic goods inspection device can receive task instructions, send inspection data, etc.

[0068] Preferably, according to the automatic goods inspection device of the present application, the steering accuracy of the inspection AMR 1 is within ±10mm, and the stopping accuracy is within ±20mm.

[0069] Preferably, the standardized box can be placed at most 3 layers, each layer has at most two rows, and each row has at most 5.

[0070] Preferably, the first two-dimensional code 10 on the standardized box 9 has a sticking accuracy within ±5mm.

[0071] Preferably, the telescopic supporting frame has a telescopic accuracy within ±30mm.

[0072] Preferably, the camera 7 on the telescopic supporting frame 3 has an accuracy within ±2mm.

[0073] The working process of the automatic goods inspection device in this embodiment is as follows:

[0074] The automatic goods inspection device starts running, and the scheduling module sends the inventory task instructions to the outside. These task instructions include the coordinates of the starting point and the ending point, and the shelves level to be inventoried (such as 1, 2, 3 layers).

[0075] The data transmission device 2 determines the moving height of the camera 7 according to these task instructions, to ensure that the camera can accurately shoot the required shelf information.

[0076] The data transmission device 2 further conveys the instructions to the autonomous mobile robot (patrol AMR 1), and when the patrol AMR 1 reaches the designated position, the data transmission device 2 adjusts the height of the camera through the driving motor assembly. In this way, the camera can start the shooting task.

[0077] During the whole patrol shooting process, the patrol AMR 1 navigates and positions according to the information of the third two-dimensional code 12 on the ground grid. At the same time, the camera 7 starts shooting the two-dimensional code information on the shelf 8 and the standardized box 9. These two-dimensional codes contain relevant information of the goods, such as product code, batch information, etc.

[0078] Once the automatic goods patrol device completes the scanning task of the bottom layer, the motor assembly starts and adjusts the camera 7 to the height suitable for the next step of work to continue the shooting work. The camera starts to capture the two-dimensional code on the second layer of shelves and identifies and records it.

[0079] The data transmission device 2 is used to collect and arrange the data shot by the camera. Through the processing of de-duplication, error correction, etc. of the scanning data, the data transmission device 2 can determine the goods information belonging to each shelf and generate the inventory result.

[0080] Finally, the data transmission device 2 returns the inventory result to the scheduling module for warehouse management personnel to view and process. This automatic goods inventory device based on AMR realizes the automatic goods inventory process, improves the efficiency and reduces the occurrence of human errors. At the same time, due to the use of telescopic support frame and camera navigation system, the device can be applied to various complex scenes, bringing convenience and accuracy to warehouse management.

[0081] As shown in Figure 4 The present application also provides an automatic goods patrol method, which comprises the following steps:

[0082] Step S1, the scheduling module generates a patrol task instruction and sends the patrol task instruction to the data transmission device 2;

[0083] Step S2, the data transmission device 2 sends a control instruction to the patrol AMR 1, and the data transmission device 2 determines the moving height of the camera 7 according to the number of shelves in the patrol task;

[0084] Step S3, the patrol AMR 1 sets off to the designated position, the patrol AMR 1 starts the ground grid positioning to determine whether it reaches the designated position, and then the patrol AMR 1 reaches the designated position area;

[0085] Step S4, the data transmission device 2 controls the motor assembly to act, and the motor assembly adjusts the height of the telescopic support frame and the camera height;

[0086] Step S5, the camera 7 starts shooting, and the inspection AMR 1 starts moving according to the inspection path information in the inspection task instruction and the two-dimensional code information on the ground grid;

[0087] Step S6, the same area and the same layer priority inspection strategy is executed;

[0088] Step S7, when the scanning of the standardized cases in the same layer is completed, the motor assembly is started and the camera 7 is adjusted to the height of the next layer to continue shooting and inspection;

[0089] Step S8, when the scanning of all the shelves in the current area is completed, the inspection AMR 1 moves to another area and continues to start scanning;

[0090] Step S9, steps S5 to S8 are repeated until the scanning of all the shelves and standardized cases in all areas is completed.

[0091] Preferably, step S10 is executed after step S9.

[0092] Step S10, the data transmission device 2 processes the scanning data from the camera 7 and returns it to the scheduling module.

[0093] The processing of the scanning data is to process the corresponding goods data of the shelves, including de-duplication, error correction and other operations.

[0094] The scheduling module needs to perform the first scheduling. The first scheduling needs to determine the time required to complete the inspection according to the number of existing shelves in the warehouse, assuming that all shelves are full, according to the conflict-free path, and then add the spare time caused by the interference factor to generate the first time constraint.

[0095] During the execution of the inspection task by the inspection AMR 1, the scheduling module generates new constraints and scheduling schemes according to the information from the inspection AMR 1, and then re-executes step S1. The scheduling module converts the new scheduling scheme into an inspection task instruction and synchronizes it to each inspection AMR 1. The scheduling scheme specifies the starting point, the route of each stage, the end point and the shelves to be counted of each inspection AMR 1. Then, the inspection AMR 1 executes a new task according to the new scheduling scheme.

[0096] In order to deal with the conflict of the trolley, a reasonable spare time is set, and the scheduling module adopts a scheduling strategy of waiting at the conflict point in the multi-AMR system. In order to simulate the additional operation time of the AMR passing through the conflict point, the Erlang distribution model is adopted, and it is assumed that the probability of conflict during the inspection of n shelves is β. The preset conflict spare time formula is: t0=β×n×m×y, wherein m is the number of conflict points passed by the inspection AMR 1, and y is the mean value of the additional operation time.

[0097] Because some shelves are empty, part of the inspection AMR1 inspection speed is faster, if the initial scheme is completely followed, these inspection AMRs will be idle and so on, therefore, the model should be re-solved by modifying the update constraint condition according to the time constraint, and the scheduling scheme with high utilization rate of inspection AMR and meeting the minimum inspection time is selected as the new scheduling scheme.

[0098] The scheduling module uses the DQN algorithm to recursively solve according to the inspection time of goods in each warehouse, the number of inspection AMRs and other constraint conditions, and obtains a feasible scheduling scheme to meet these constraint conditions.

[0099] For the DQN algorithm, the calculation method of Q(s t ,a t ) is based on the Bellman equation, which adds the maximum Q value of the next state to the current reward to update the Q value of the current state. By continuously iterating and updating the Q value table, the agent gradually learns the optimal behavior strategy, so as to maximize the benefits in a complex environment. The calculation method of Q(s t ,a t ) is as follows:

[0100]

[0101] Where r t represents the immediate reward at time t, r t is 50 when the target is reached, r t is -10 when collision or timeout occurs, represents the probability of moving to state s t under the current state s t and action a t+1 , and represents the maximum reward value that can be obtained by taking different actions in state s t+1 . γ is the discount factor, which represents the degree of influence of future rewards on the current action of the mobile robot. The update formula of Q(s t ,a t ) is as follows:

[0102] Where α is the learning rate, and its value range is α∈[0,1]; γ is the discount factor, which represents the degree of influence of future rewards on the current action of the mobile robot.

[0103] The instant reward function is an important part of the path planning based on reinforcement learning, and its setting directly affects the effect of the inspection planning and the convergence of the network. In the reward function of the traditional DQN algorithm, the inspection AMR 1 can obtain a reward value when reaching the target point and when planning fails, but cannot obtain instant and effective feedback in other states. Although the inspection AMR 1 can still plan the path through training, the process usually needs a large number of rounds, and the planned path is often not optimal. Therefore, the application redesigns the reward function. When the inspection AMR 1 reaches the target point, the reward obtained is C1, when the inspection AMR 1 collides, the reward obtained is C2, and in other cases, the robot obtains a reward of K*cos theta / d. Wherein K is a hyperparameter, theta is the included angle between the vector n1 from the position of the AMR to the position of the target point before the AMR acts and the direction vector n2 before and after the action of the robot, and d is the distance between the position of the AMR and the position of the target point. The greater the deviation of the direction from the target point, the smaller the reward; the farther the distance from the target point, the smaller the reward.

[0104] In step S6, the same area and same layer priority inspection strategy includes:

[0105] In step S60, if there are standardized boxes 9 to be inspected on the shelves in the same area and the same layer, the inspection AMR 1 preferentially inspects the shelves in the same area and the same layer, without adjusting the height of the camera;

[0106] In step S61, if the standardized boxes in the same layer are scanned, but there are other shelves in the same layer, the inspection AMR 1 rounds to another row to continue to inspect the shelves in the layer.

[0107] The automatic goods inspection method and device provided by the application have the following beneficial effects:

[0108] The camera 7 is arranged on the support frame and can move up and down. The camera 7 is used for shooting the two-dimensional code on the side of the standardized box to obtain the related information of the goods. The standardized boxes 9 can be stacked on the shelves 8, and the camera can read the information of the boxes in one layer at a time; every two rows of shelves form a group, and the inspection device can complete the inspection task of the goods in two rows of shelves in one layer by one round; after adjusting the height of the camera, the information of the boxes in other layers can be read; the scheduling module solves a set of feasible AMR scheduling schemes based on the DQN algorithm for the multi-AMR inspection scene, selects the scheme that can meet the least time, and improves the utilization rate of the AMR as much as possible. The application can automate the inspection of the goods in the warehouse, help other links in the supply chain understand the state of the goods, make corresponding decisions and arrangements in time, and thus improve the response ability and cooperativeness of the entire supply chain.

[0109] Those skilled in the art will appreciate that embodiments of the present application can be devised for a method, a system, or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer-readable program code thereon for use by or in connection with an instruction execution system. For the purposes of this description, a computer-usable or computer readable storage medium can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

[0110] The foregoing is merely illustrative of the principles of the application and various modifications can be made by persons skilled in the art. The present application is not intended to be limited to the embodiments shown, but is to be accorded the full scope that resides in the art thereof. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. An automatic goods inspection method based on an automatic goods inspection device, characterized by, The automatic goods inspection device comprises an inspection module, a storage module and a scheduling module, the inspection module comprises an inspection AMR, a data transmission device, a telescopic support frame, a camera assembly and a motor assembly, the telescopic support frame is arranged on the inspection AMR, the inspection AMR is connected with the data transmission device, the data transmission device is further connected with the motor assembly and is used for controlling the motor assembly, the telescopic support frame is connected with the camera assembly, the storage module comprises a goods shelf, a standardized box and a ground grid, the scheduling module is used for decomposing an inspection task to a plurality of inspection AMRs and planning a running route of the inspection AMR, the standardized box is arranged on the goods shelf, the standardized box is provided with a first two-dimensional code, the goods shelf is provided with a second two-dimensional code, and the ground grid is provided with a third two-dimensional code. The automatic goods inspection method comprises the following steps: Step S1, the scheduling module generates an inspection task instruction and sends the inspection task instruction to the data transmission device; Step S2, the data transmission device sends a control instruction to the inspection AMR, and the data transmission device determines the moving height of the camera according to the shelf layer in the inspection task; Step S3, the inspection AMR departs to a specified position, the inspection AMR starts ground grid positioning to determine whether the specified position is reached, and then the inspection AMR reaches the specified position area; Step S4, the data transmission device controls the motor assembly to act, and the motor assembly adjusts the height of the telescopic support frame and the camera height; Step S5, the camera starts shooting, and the inspection AMR starts moving according to the inspection path information in the inspection task instruction and the two-dimensional code information on the ground grid; Step S6, a same-area same-layer priority inspection strategy is executed; Step S7, when the scanning of the standardized boxes in the same layer is completed, the motor assembly is started and the camera is adjusted to the height of the next layer to continue shooting and inspection; Step S8, if the shelves in the current area are completely scanned, the inspection AMR moves to another area to continue scanning; Step S9, steps S5 to S8 are repeated until the scanning of all the shelves and standardized boxes in all the areas is completed; Step S10, the data transmission device processes the scanning data from the camera and returns the scanning data to the scheduling module; The scheduling module uses a DQN algorithm to recursively solve and obtain a feasible scheduling scheme, wherein in the DQN algorithm The calculation method is as follows: ； Where, r t Represents the immediate reward at time t, where r represents the reward when the goal is reached. t The value is 50, and the value is r in the event of a collision or timeout. t -10, Indicates the current state s t and action a t Transition to state s under certain conditions t+1 The probability, Indicates that in state s t+1 The maximum reward value that can be obtained by taking different actions at different times. This is a discount factor, representing the degree to which future rewards affect the current actions of the inspection AMR. The update formula is: ； wherein, is a learning rate, which has a value range of ; The instant reward function of the scheduling module for the inspection AMR is: C1 when the inspection AMR reaches the target point, C2 when the inspection AMR collides, and 0 otherwise where K is a hyperparameter, is the angle between the vector n1 from the location of the inspection AMR to the target point before the inspection AMR takes an action and the direction vector n2 of the inspection AMR before and after the action, and d is the distance between the location of the inspection AMR and the target point. The scheduling module adopts a scheduling strategy of collision point waiting in a multi-AMR system, the scheduling strategy simulates the additional operation time of the inspection AMR passing through the collision point, adopts an Erlang distribution model, and assumes that the probability of collision in the process of inspecting n shelves is β, and a preset collision allowance time formula is t0=β×n×m×y, wherein m is the number of collision points passed through by the inspection AMR, and y is the mean value of the additional operation time.

2. The automatic goods inspection method based on the automatic goods inspection device according to claim 1, characterized by, In the process of executing the inspection task by the inspection AMR, the scheduling module generates a new constraint and scheduling scheme according to the information from the inspection AMR, and then re-executes step S1.

3. The automatic goods inspection method based on the automatic goods inspection device according to claim 1, characterized by, The motor assembly includes a first motor and a second motor, wherein the first motor is used to drive the retractable support frame and the second motor is used to drive the camera assembly.

4. The automatic goods inspection method based on the automatic goods inspection device according to claim 3, characterized by, The retractable support frame includes a frame body, a retractable bracket, and a first transmission device. The first transmission device is connected to the first motor and is also connected to the retractable bracket. The retractable bracket is mounted on the frame body, and the frame body is mounted on the inspection AMR. The camera assembly includes a camera and a second transmission device, which is connected to the second motor and the camera, respectively.

Images