AI intelligent logistics warehousing system and method thereof

Abstract

The invention relates to the technical field of logistics storage, and particularly discloses an AI intelligent logistics storage system and a method thereof, and the method comprises the following steps: setting a plurality of speed nodes and a plurality of frame rate nodes, and setting a plurality of experiment groups; judging whether the single experimental group is a normal group based on the definition score, and classifying the normal group; determining the optimal frame rate of the truck at different speeds based on classification; and the real-time speed of the truck is obtained from the driving-in moment, the frame rate of the monitoring equipment is adjusted according to the real-time speed and the optimal frame rate, video data is collected till the driving-out moment is reached, the video data between the driving-in moment and the driving-out moment serves as a real-time video, and whether the truck is loaded with goods or not is judged based on the real-time video. The accuracy of judging whether the truck is loaded or not can be improved.

Description

Technical Field

[0001] This invention relates to the field of logistics and warehousing technology, and specifically to an AI-powered intelligent logistics and warehousing system and method. Background Technology

[0002] In the logistics and warehousing sector, AI has become a core driving force for industry development. AI enables real-time monitoring of truck entry and exit, reducing the need for manual intervention; it also accurately identifies loaded trucks, tracks and numbers them, ensuring that each truck's unloading area is accurate and preventing secondary loading and unloading.

[0003] A Chinese patent application, CN119444047A, discloses an AI-powered intelligent logistics warehousing system. Its solution includes acquiring video data when a truck is within a detection zone, dividing the video data into several image frames, and determining whether the truck is loaded with goods based on the image frames. However, this process neglects the impact of vehicle speed on the image frames. With a fixed frame rate, a high vehicle speed may cause blurring of the image frames, affecting the judgment accuracy. Summary of the Invention

[0004] The purpose of this invention is to provide an AI-powered intelligent logistics warehousing system and method to solve the aforementioned technical problems.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] An AI-powered intelligent logistics warehousing method includes the following steps:

[0007] Several speed nodes and several frame rate nodes are set within the preset speed range and frame rate range, respectively, and several experimental groups are set up based on the speed nodes and frame rate nodes;

[0008] To maintain the speed of the truck and the frame rate of the monitoring equipment to meet the requirements of a single experimental group, video data was collected during the truck's movement. The sharpness parameters of individual image frames in the video data were obtained, and the sharpness score was calculated based on the sharpness parameters.

[0009] When the average sharpness score is greater than the preset threshold, the corresponding experimental group is marked as the normal group, and the normal group is classified. The normal group in the same category has the same speed node.

[0010] For a single category, the normal group corresponding to the frame rate node with the smallest number is taken as the target group. The speed v corresponding to the speed node and the frame rate z corresponding to the frame rate node in the target group are obtained. The optimal frame rate when the truck's speed is v is z.

[0011] The system acquires the vehicle's real-time speed starting from the moment of entry, adjusts the frame rate of the monitoring equipment based on the real-time speed and the optimal frame rate, and collects video data until the moment of exit. The video data between the moment of entry and the moment of exit is used as real-time video, and the system determines whether the truck is loaded with cargo based on the real-time video.

[0012] Preferably, the experimental group includes:

[0013] Starting from the beginning of the speed range, set several speed nodes at preset speed intervals;

[0014] Starting from the beginning of the frame rate range, set several frame rate nodes at preset frame rate intervals;

[0015] A single experimental group includes one speed node and one frame rate node, and the speed node and / or frame rate node are different in different experimental groups.

[0016] Preferably, the calculation of the sharpness score includes:

[0017] Sharpness parameters include motion blur length, blur direction angle, and edge transition width;

[0018] Calculate the sharpness score based on the sharpness parameter and the superior-inferiority distance method.

[0019] Preferably, adjusting the frame rate of the monitoring device includes:

[0020] Get the real-time speed V0 at the moment of entry, and starting from the moment of entry, get the real-time speed at each moment in sequence, and starting from the moment of entry, adjust the frame rate of the monitoring device to the optimal frame rate Z0 corresponding to the real-time speed V0.

[0021] When the difference between the newly acquired real-time speed V1 and the real-time speed V0 is less than or equal to the preset speed difference, the frame rate of the monitoring device is kept at the optimal frame rate Z0.

[0022] When the difference between the newly acquired real-time speed V1 and the real-time speed V0 is greater than the preset speed difference, the frame rate of the monitoring device is adjusted to the optimal frame rate Z1 corresponding to the real-time speed V1, and the real-time speed V1 is used as the real-time speed at the entry time. The above steps are repeated until the exit time is reached.

[0023] Preferably, if the real-time speed V1 is not the speed corresponding to the speed node in the target group, then the following steps are performed:

[0024] The speed corresponding to the speed node in the target group is taken as the candidate speed, the candidate speed that is smaller than the real-time speed V1 and is the largest is taken as the selected speed, and the best frame rate corresponding to the selected speed is taken as the best frame rate Z1 of the real-time speed V1.

[0025] Preferably, after classifying the normal group, the following steps are also included:

[0026] Obtain the speed node corresponding to the category, denoted as the normal node, and obtain the speed Vmax corresponding to the normal node. If the real-time speed of the vehicle reaches 0.9Vmax, send a prompt message to the driver indicating that the speed limit is Vmax.

[0027] Preferably, determining whether a truck is loaded with goods includes:

[0028] Acquire real-time video image frames, input the image frames into the target tracking network, and determine whether cargo is loaded.

[0029] An AI-powered intelligent logistics and warehousing system includes:

[0030] Initial module: Set up several speed nodes and several frame rate nodes in the preset speed range and frame rate range respectively, and set up several experimental groups based on the speed nodes and frame rate nodes;

[0031] Analysis module: Maintaining the speed of the truck and the frame rate of the monitoring equipment to meet the requirements of a single experimental group, collect video data during the truck's movement, obtain the sharpness parameters of individual image frames in the video data, and calculate the sharpness score based on the sharpness parameters;

[0032] Calibration module: When the average sharpness score is greater than the preset threshold, the corresponding experimental group is marked as the normal group, and the normal group is classified. The normal group in the same category has the same speed node.

[0033] For a single category, the normal group corresponding to the frame rate node with the smallest number is taken as the target group. The speed v corresponding to the speed node and the frame rate z corresponding to the frame rate node in the target group are obtained. The optimal frame rate when the truck's speed is v is z.

[0034] Judgment Module: Starting from the moment of entry, the real-time speed of the vehicle is acquired. Based on the real-time speed and the optimal frame rate, the frame rate of the monitoring equipment is adjusted and video data is collected until the moment of exit. The video data between the moment of entry and the moment of exit is used as real-time video. Based on the real-time video, it is determined whether the truck is loaded with goods.

[0035] The beneficial effects of this invention compared to the prior art are as follows:

[0036] 1) This invention first collects samples under different speed and frame rate combinations through an experimental group method and selects the normal group with the required clarity. Then, it matches the optimal frame rate for each speed node, so that the monitoring device always works in the preferred clarity range. This overcomes the problem of motion blur caused by fixed frame rate in high-speed scenes, ensures the integrity of edge details and texture information of the driving image, and improves the stability and accuracy of the target tracking network in recognizing the shape and position of the cargo.

[0037] 2) The real-time adaptive frame rate control mechanism dynamically calls the best frame rate generated in the previous stage based on the current speed. It can quickly reconfigure shooting parameters when the speed changes without manual intervention. When the speed change is within an acceptable range, the original frame rate is maintained to avoid frequent switching, reducing hardware load and algorithm computation. When the speed difference is large, timely switching to the new frame rate can prevent the loss of clarity. Therefore, while ensuring video quality, redundant frames are reduced, storage space is saved, and the operating efficiency and energy consumption of the entire logistics monitoring system are improved.

[0038] 3) After determining the maximum speed of a normal node, the system automatically monitors the real-time speed and issues a speed limit warning to the driver when the speed approaches the threshold, effectively reducing image distortion caused by excessive speed. The warning mechanism works in conjunction with dynamic frame rate adjustment to maintain high-quality images even when the vehicle passes through the detection zone at high speed, providing stable input to the target tracking network. This improves the reliability of determining whether goods are loaded, reduces false positives and false negatives, and further supports automated supervision and refined operation of warehouse entry and exit processes. Attached Figure Description

[0039] The invention will now be further described with reference to the accompanying drawings.

[0040] Figure 1 This is a flowchart illustrating an AI-powered intelligent logistics warehousing method according to the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0042] Please see Figure 1 As shown, this invention is an AI-powered intelligent logistics warehousing method, comprising the following steps:

[0043] Step 1: Set up several speed nodes and several frame rate nodes in the preset speed range and frame rate range respectively, and set up several experimental groups based on the speed nodes and frame rate nodes.

[0044] In a preferred embodiment of the present invention, the experimental group includes:

[0045] Starting from the beginning of the speed range, first determine the upper and lower limits of the entire range, such as the minimum and maximum speeds allowed in the monitoring scene. Then, increment the speeds one by one according to the pre-set speed intervals, and record each increment point as a speed node. For example, if the speed range is 0 to 50 and the speed interval is set to 5, then 0, 5, 10 up to 50 are all marked as nodes, and any subsequent vehicle speed will fall into a certain node or between two adjacent nodes. The same applies to the frame rate range, which will not be elaborated here.

[0046] Each speed node is combined with each frame rate node by performing a Cartesian product. Each combination is an experimental group. For example, a speed of 10 and a frame rate of 15 is an independent experimental group. An experimental group contains only one pair of speed and frame rate because keeping the combination of a single variable allows for accurate tracing of the cause of ambiguity during testing.

[0047] Step 2: Maintain the speed of the truck and the frame rate of the monitoring equipment to meet the requirements of a single experimental group, collect video data during the truck's movement, obtain the sharpness parameters of individual image frames in the video data, and calculate the sharpness score based on the sharpness parameters.

[0048] In a specific embodiment, after setting the vehicle speed node and frame rate node of a certain preset experimental group as the current target, the driver is first instructed to slowly adjust the vehicle speed to the corresponding node using the vehicle instrument or an external speed measuring device. For example, if the current node requires 30, the vehicle speed is first accelerated to slightly higher than 30 and then slightly reduced to 30, so that the vehicle speed pointer stays close to the target value for a long time. At the same time, the frame rate is fixed to the frame rate specified in the experimental group, such as 20 frames, through the setting interface of the monitoring device. After the adjustment is completed, the viewfinder is observed, and after confirming that there is no significant jump in exposure and brightness, the formal recording begins. After the recording is completed, the entire video is sequentially split into individual frames.

[0049] In a preferred embodiment of the present invention, the calculation of the sharpness score includes:

[0050] For each frame, edge detection is performed to extract obvious straight lines or high-contrast contours. The transition length of the grayscale change of adjacent pixels along these contours can be used to obtain the edge transition width. The shorter the transition, the sharper the details.

[0051] Analyze the energy distribution of the entire frame spectrum to determine whether there is a trailing phenomenon with the same direction. If a trailing phenomenon is detected, measure the pixel distance from the main peak to the end of the trailing phenomenon along the trailing direction as the motion blur length.

[0052] Simultaneously, the tilt angle of the trail relative to the horizontal plane is recorded as the fuzzy direction angle.

[0053] After normalization, the three parameters represent image sharpness, motion blur, and directional consistency, respectively. Then, the comprehensive distance between each frame and the ideal sharp image and the extremely blurred image is converted into a dimensionless sharpness score by the superior-inferior solution distance method.

[0054] Step 3: When the average sharpness score is greater than the preset threshold, the corresponding experimental group is marked as the normal group, and the normal group is classified. The normal groups in the same category have the same speed node.

[0055] After completing the clarity assessment, all experimental groups with average scores exceeding the threshold were uniformly labeled "normal," and then they were grouped into the same pocket according to the speed node. Although the members in the pocket had different frame rates, they shared the same speed value.

[0056] Step 4: For a single category, take the normal group corresponding to the frame rate node with the smallest number as the target group, obtain the speed v corresponding to the speed node and the frame rate z corresponding to the frame rate node in the target group, and denote the optimal frame rate when the truck's speed is v as z.

[0057] The normal groups within each pocket are arranged in ascending order of frame rate node. The item at the head of the queue is selected as the target group for that pocket because it has the lowest frame rate, the lightest bandwidth and storage pressure, and its clarity has been verified as qualified. Its speed node v and frame rate node z are recorded, thus forming a mapping of "the lowest qualified frame rate available when the vehicle speed is v is z".

[0058] In another preferred embodiment of the present invention, after classifying the normal group, the method further includes the following steps:

[0059] Each pocket's speed node is individually identified and named a normal node. By traversing all normal nodes, the one with the largest value is selected and denoted as Vmax. This represents the highest safe vehicle speed within the verified range that still provides a clear image. When the vehicle is running, the real-time speed is continuously read and compared with 0.9Vmax. Once the instrument reading touches this 90% threshold, a voice or text prompt is immediately triggered, pushing the message "Speed ​​limit not exceeding Vmax" to the driver's cab to remind the driver to slow down appropriately. This ensures that the image clarity does not drop drastically and also preserves sufficient image quality for subsequent target recognition.

[0060] Step 5: Starting from the moment the vehicle enters the vehicle, acquire the real-time speed, adjust the frame rate of the monitoring equipment according to the real-time speed and the optimal frame rate, and collect video data until the moment of departure is reached.

[0061] In another preferred embodiment of the present invention, adjusting the frame rate of the monitoring device includes:

[0062] The real-time speed V0 at the moment of entry is obtained. The speed-frame rate comparison table obtained through previous experiments has been completely stored in the control unit. Therefore, the optimal frame rate Z0 can be found directly in the table with V0. The reason for choosing this value is that each row in the table corresponds to the minimum acceptable frame rate at which the image just gets rid of the trailing and can still guarantee brightness at this speed. Therefore, sufficient clarity can be obtained with the most economical exposure cost.

[0063] Send a switching command to the camera to adjust the frame rate to Z0 and start recording immediately, ensuring that the first frame meets the stopping requirements of the current speed.

[0064] A stable sampling period continues to monitor the vehicle speed to obtain a new value V1. The difference is first calculated with the reference speed V0. If the difference falls within the pre-set allowable fluctuation range, this is usually a small jolt caused by the driver's slight throttle adjustment, which will not significantly worsen the motion blur. Therefore, maintaining the current frame rate Z0 of the camera can avoid the hardware from being constantly reinitialized.

[0065] If the difference exceeds the allowable range, it means that the vehicle has accelerated or decelerated significantly. If the old frame rate is used again, the trailing or exposure problem may reappear. In this case, the table is consulted again to find the new best frame rate Z1 with V1 and switch immediately. At the same time, V1 is replaced with the new reference speed V0 to ensure that subsequent comparisons are always based on the latest operating conditions.

[0066] This process repeats until the rear of the vehicle crosses the exit trigger line, at which point the system records the exit moment and stops adjusting. The entire recording then forms a series of continuous images captured at different speeds using their respective most suitable frame rates, ensuring that the image clarity always adapts to the vehicle speed to maintain optimal performance and avoiding the accumulation of unnecessary data.

[0067] It is worth noting that if the real-time speed V1 is not the speed corresponding to the speed node in the target group, the following steps are performed:

[0068] When a new speed V1 cannot find a completely equal speed node in the comparison table, the speed nodes of all target groups in the mapping table are taken out one by one and arranged into an ordered list. Since these nodes are the benchmark speeds for image clarity that have been verified in the experimental phase, they can be used as candidates for subsequent inference.

[0069] The candidate data is compared with the real-time value V1 item by item, and all elements less than V1 are selected. The principle behind this is that frame rate and clarity are positively correlated. Selecting nodes slightly lower than the current vehicle speed can ensure that the existing frame rate still has a margin to resist trailing at the actual speed.

[0070] The candidate with the largest value is selected from those that meet the conditions as the "selection speed". For example, if the candidates are 10, 20, and 30 and V1 is 27, then the candidate with the highest value is 20, which is the next highest value before 30. This is done in order to get as close as possible to the real-time speed without crossing the risk point.

[0071] Finally, the optimal frame rate corresponding to the "selected speed" in the mapping table is extracted and assigned to Z1, so that the camera can adjust immediately according to Z1. The basis for this is that the mapping table records the lowest qualified frame rate that is most resource-efficient yet sufficient to ensure clarity for each speed node. Therefore, this replacement can complete parameter matching without increasing the probability of blur and leave sufficient buffer space for the next round of speed sampling to maintain the continuity and usability of the entire image.

[0072] Step Six: Use the video data from the entry time to the exit time as real-time video, and determine whether the truck is loaded with goods based on the real-time video.

[0073] In another preferred embodiment of the present invention, determining whether a truck is loaded with goods includes:

[0074] After the vehicle enters, the entire video (real-time video) is split into consecutive frames in chronological order while keeping the original timestamps unchanged, and then each frame is fed into the trained target tracking network in sequence.

[0075] The network first searches for visual blocks within the carriage area that match the shape, texture, and brightness distribution of the cargo, and then continuously tracks the outline positions of these blocks between adjacent frames. If the outline area and center coordinates of the same block change very little within several consecutive frames, the block is determined to be real cargo rather than background noise. Subsequently, the appearance and disappearance times of the cargo blocks are recorded along the time axis. If it is found that the block is continuously missing after a certain moment and does not reappear in all subsequent frames, it is considered that unloading has occurred at this time, and all subsequent frames are marked as empty.

[0076] Finally, only the last stable time window before leaving the detection area (e.g., the last few consecutive frames) is summarized: if the cargo block can still be tracked within this time window, the conclusion "loaded" is output; if the cargo block has disappeared stably, the conclusion "unloaded" is output. This process accurately segments the unloading event midway by using the three-stage trajectory of "appearance-continuity-disappearance".

[0077] An AI-powered intelligent logistics and warehousing system includes:

[0078] Initial module: Set up several speed nodes and several frame rate nodes in the preset speed range and frame rate range respectively, and set up several experimental groups based on the speed nodes and frame rate nodes.

[0079] Analysis module: Maintaining the speed of the truck and the frame rate of the monitoring equipment to meet the requirements of a single experimental group, video data is collected during the truck's movement. The sharpness parameters of individual image frames in the video data are obtained, and a sharpness score is calculated based on the sharpness parameters.

[0080] Calibration module: When the average sharpness score is greater than the preset threshold, the corresponding experimental group is marked as the normal group, and the normal group is classified. The speed nodes corresponding to the normal groups in the same category are the same.

[0081] For a single category, the normal group corresponding to the frame rate node with the smallest number is taken as the target group. The speed v corresponding to the speed node and the frame rate z corresponding to the frame rate node in the target group are obtained. The optimal frame rate when the speed of the truck is v is z.

[0082] Judgment Module: Starting from the moment of entry, the real-time speed of the vehicle is acquired. Based on the real-time speed and the optimal frame rate, the frame rate of the monitoring equipment is adjusted and video data is collected until the moment of exit. The video data between the moment of entry and the moment of exit is used as real-time video. Based on the real-time video, it is determined whether the truck is loaded with goods.

[0083] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. An AI-powered intelligent logistics warehousing method, characterized in that, Includes the following steps: Several speed nodes and several frame rate nodes are set within the preset speed range and frame rate range, respectively, and several experimental groups are set up based on the speed nodes and frame rate nodes; To maintain the speed of the truck and the frame rate of the monitoring equipment to meet the requirements of a single experimental group, video data was collected during the truck's movement. The sharpness parameters of individual image frames in the video data were obtained, and the sharpness score was calculated based on the sharpness parameters. When the average sharpness score is greater than the preset threshold, the corresponding experimental group is marked as the normal group, and the normal group is classified. The normal group in the same category has the same speed node. For a single category, the normal group corresponding to the frame rate node with the smallest number is taken as the target group. The speed v corresponding to the speed node and the frame rate z corresponding to the frame rate node in the target group are obtained. The optimal frame rate when the truck's speed is v is z. The system acquires the vehicle's real-time speed starting from the moment of entry, adjusts the frame rate of the monitoring equipment based on the real-time speed and the optimal frame rate, and collects video data until the moment of exit. The video data between the moment of entry and the moment of exit is used as real-time video, and the system determines whether the truck is loaded with cargo based on the real-time video.

2. The AI-powered intelligent logistics warehousing method according to claim 1, characterized in that, The experimental groups were set up as follows: Starting from the beginning of the speed range, set several speed nodes at preset speed intervals; Starting from the beginning of the frame rate range, set several frame rate nodes at preset frame rate intervals; A single experimental group includes one speed node and one frame rate node, and the speed node and / or frame rate node are different in different experimental groups.

3. The AI-powered intelligent logistics warehousing method according to claim 1, characterized in that, The calculation of the sharpness score includes: Sharpness parameters include motion blur length, blur direction angle, and edge transition width; Calculate the sharpness score based on the sharpness parameter and the superior-inferiority distance method.

4. The AI-powered intelligent logistics warehousing method according to claim 1, characterized in that, Adjusting the frame rate of monitoring equipment includes: Get the real-time speed V0 at the moment of entry, and starting from the moment of entry, get the real-time speed at each moment in sequence, and starting from the moment of entry, adjust the frame rate of the monitoring device to the optimal frame rate Z0 corresponding to the real-time speed V0. When the difference between the newly acquired real-time speed V1 and the real-time speed V0 is less than or equal to the preset speed difference, the frame rate of the monitoring device is kept at the optimal frame rate Z0. When the difference between the newly acquired real-time speed V1 and the real-time speed V0 is greater than the preset speed difference, the frame rate of the monitoring device is adjusted to the optimal frame rate Z1 corresponding to the real-time speed V1, and the real-time speed V1 is used as the real-time speed at the entry time. The above steps are repeated until the exit time is reached.

5. The AI-powered intelligent logistics warehousing method according to claim 4, characterized in that, If the real-time speed V1 is not the speed corresponding to the speed node in the target group, then perform the following steps: The speed corresponding to the speed node in the target group is taken as the candidate speed, the candidate speed that is smaller than the real-time speed V1 and is the largest is taken as the selected speed, and the best frame rate corresponding to the selected speed is taken as the best frame rate Z1 of the real-time speed V1.

6. The AI-powered intelligent logistics warehousing method according to claim 1, characterized in that, After classifying the normal group, the following steps are also included: Obtain the speed node corresponding to the category, denoted as the normal node, and obtain the speed Vmax corresponding to the normal node. If the real-time speed of the vehicle reaches 0.9Vmax, send a prompt message to the driver indicating that the speed limit is Vmax.

7. The AI-powered intelligent logistics warehousing method according to claim 1, characterized in that, Determining whether a truck is carrying cargo includes: Acquire real-time video image frames, input the image frames into the target tracking network, and determine whether cargo is loaded.

8. An AI-powered intelligent logistics and warehousing system, characterized in that, include: Initial module: Set up several speed nodes and several frame rate nodes in the preset speed range and frame rate range respectively, and set up several experimental groups based on the speed nodes and frame rate nodes; Analysis module: Maintaining the speed of the truck and the frame rate of the monitoring equipment to meet the requirements of a single experimental group, collect video data during the truck's movement, obtain the sharpness parameters of individual image frames in the video data, and calculate the sharpness score based on the sharpness parameters; Calibration module: When the average sharpness score is greater than the preset threshold, the corresponding experimental group is marked as the normal group, and the normal group is classified. The normal group in the same category has the same speed node. For a single category, the normal group corresponding to the frame rate node with the smallest number is taken as the target group. The speed v corresponding to the speed node and the frame rate z corresponding to the frame rate node in the target group are obtained. The optimal frame rate when the truck's speed is v is z. Judgment Module: Starting from the moment of entry, the real-time speed of the vehicle is acquired. Based on the real-time speed and the optimal frame rate, the frame rate of the monitoring equipment is adjusted and video data is collected until the moment of exit. The video data between the moment of entry and the moment of exit is used as real-time video. Based on the real-time video, it is determined whether the truck is loaded with goods.

Images