A cloth type-based transportation path planning method and system

Abstract

The application relates to the technical field of automatic warehousing and discloses a transport path planning method and system based on cloth types, which comprises the following steps: obtaining cloth characteristic information of a target cloth roll and determining the cloth type of the target cloth roll; screening a candidate storage frame set according to the cloth type; constructing an optimization function with the minimum transport energy consumption as a target based on the cloth characteristic information and frame attribute information, and determining an initial transport path; obtaining real-time dynamic information of a storage library, performing local incremental dynamic re-planning on the initial transport path by adopting a hierarchical A* search algorithm, obtaining a final transport path, and controlling a cargo platform to transport. Through cloth type adaptation storage, multi-target energy consumption optimization and dynamic hierarchical re-planning, the application solves the problems of poor storage matching, high transport energy consumption and insufficient dynamic adaptability in cloth warehousing, and realizes intelligent, low-energy-consumption and high-robustness cloth roll transport control.

Description

Technical Field

[0001] This invention relates to the field of automated warehousing technology, and more specifically, to a method and system for planning transportation routes based on fabric type. Background Technology

[0002] In automated warehousing systems for industries such as textiles, dyeing, and garment manufacturing, the transportation and storage of fabric rolls (or fabric rollers) is a crucial component of the logistics process. Traditional methods of transporting fabric rolls often rely on manually driven forklifts or semi-automated guided vehicles (AGVs). Operators need to rely on experience to determine the type of fabric and select storage locations, which is not only inefficient but also prone to human error leading to incorrect storage or physical damage to the fabric.

[0003] With the development of machine vision, automatic control, and path planning technologies, some modern warehousing and logistics systems have begun to introduce automated loading platforms (such as automated guided vehicles and rail shuttles) to handle fabric rolls. Existing path planning methods typically optimize for the shortest transport distance or time, employing static path planning strategies such as Dijkstra's algorithm and A* search algorithm. However, these methods have the following significant drawbacks when applied to fabric warehousing scenarios: First, existing technologies generally ignore the physical characteristics of the fabric itself (such as material softness and surface susceptibility to damage) and the storage requirements of different fabric types. Different fabrics (such as silk, cotton, linen, and synthetic fibers) need to be stored separately to prevent cross-contamination or physical damage, but existing methods lack a mechanism for intelligently selecting storage bins based on fabric type. Second, existing path planning methods often use a single distance or time indicator as the optimization objective, failing to comprehensively consider multiple dimensions such as transport energy consumption, number of turns, and the efficiency improvements in subsequent operations brought about by centralized storage of similar materials. This results in a transport path that, while shortest in distance, does not necessarily have optimal energy consumption and is detrimental to the coordination of subsequent outbound operations. Third, the warehousing environment is a typical dynamic and changing scenario, frequently encountering temporary obstacles, intersecting trajectories of other loading platforms, and localized congestion. Traditional static path planning methods, once the path is determined, cannot cope with dynamic changes. When encountering obstacles or congestion, a global replanning from the starting point to the destination is often required, resulting in high computational costs, slow response times, and difficulty in meeting real-time requirements. Therefore, there is an urgent need for a transportation path planning method that can integrate the physical characteristics of fabrics, optimize transportation energy consumption as the core objective, and possess dynamic environmental adaptability to improve the operational efficiency, safety, and intelligence level of automated fabric roll warehousing systems. Summary of the Invention

[0004] In view of this, the present invention proposes a transportation route planning method and system based on fabric type, aiming to solve one or more of the above problems.

[0005] On the one hand, this invention proposes a transportation route planning method based on fabric type, including: Obtain the fabric feature information of the target fabric roll on the loading platform, and determine the fabric type of the target fabric roll based on the fabric feature information; Based on the fabric type, the storage frames in the storage repository are filtered to determine a set of candidate storage frames for storing the target fabric roll. Based on the fabric feature information of the target fabric roll and the attribute information of each storage box in the candidate storage box set, an optimization function with the goal of minimizing transportation energy consumption is constructed, and the initial transportation path from the loading platform to the candidate storage box is determined based on the optimization function. The system obtains real-time dynamic information of the storage warehouse, and uses a hierarchical A* search algorithm to perform local incremental dynamic replanning on the initial transportation path based on the real-time dynamic information to obtain the final transportation path. Based on the final transportation path, the system controls the loading platform to realize the transportation of the fabric rolls.

[0006] Preferably, the fabric feature information of the target fabric roll on the loading platform is obtained, including: The surface image information of the target fabric roll is collected by a machine vision inspection device. The collected surface image information is input into a pre-trained deep learning image classification model for feature extraction and classification recognition to obtain the fabric feature information of the target fabric roll. The fabric feature information includes: texture feature map, color feature map, structural feature map, roll diameter value and roll width value. The deep learning image classification model is a multi-task learning model based on a convolutional neural network. The multi-task learning model includes a feature extraction backbone network and a size regression branch. The feature extraction backbone network performs multi-layer convolution and pooling operations on the surface image information and outputs the texture feature map, color feature map, and structural feature map of the target fabric roll. The size regression branch outputs the roll diameter and roll width of the target fabric roll based on the structural feature map.

[0007] Preferably, determining the fabric type of the target fabric roll based on the fabric characteristic information includes: The multi-task learning model includes a type classification branch. The texture feature map, color feature map and structural feature map are input into the type classification branch to obtain the fabric type label of the target fabric roll. The fabric type label is matched with a preset fabric type-material attribute mapping table to obtain the material attributes of the target fabric roll, including softness level and surface vulnerability level. At the same time, the fabric type label is cross-validated with the task order information stored in the warehouse management system. If the identification result is inconsistent with the fabric type recorded in the task order, an alarm is triggered and the transportation process is suspended; if they are consistent, the fabric type label is determined to be the fabric type of the target fabric roll.

[0008] Preferably, based on the fabric type, the storage frames in the storage repository are filtered to determine a set of candidate storage frames for storing the target fabric roll, including: Type binding tags are pre-configured for storage boxes in the repository, and the type binding tags are used to identify one or more types of fabric that the storage box is allowed to store; Based on the fabric type tag of the target fabric roll, traverse the type binding tags of all storage frames, and filter out the storage frames whose type binding tags contain the fabric type tag of the target fabric roll to form an initial candidate storage frame set. Obtain the real-time status information of each storage frame in the initial candidate storage frame set. The real-time status information includes the current occupancy status and the remaining storage capacity. Remove the storage frames whose current occupancy status is full and whose remaining storage capacity is less than the target fabric roll volume parameter to obtain the candidate storage frame set.

[0009] Preferably, the attribute information of the storage box includes its location coordinates, current occupancy status, adjacent storage density of similar types of fabric, and remaining storage capacity; The adjacent storage density of the same type of fabric is determined by statistically analyzing the ratio of the number of fabric rolls of the same type as the target fabric roll stored within a preset radius around the corresponding storage frame to the total number of stored fabric rolls.

[0010] Preferably, based on the fabric feature information of the target fabric roll and the attribute information of each storage box in the candidate storage box set, an optimization function is constructed with the goal of minimizing transportation energy consumption. Based on the optimization function, the initial transportation path from the loading platform to the candidate storage box is determined, including: Calculate the volume parameters of the target fabric roll, which are obtained based on the roll diameter and roll width. Based on the volume parameters and the remaining storage capacity of the storage frame, the storage adaptation coefficient is determined; Construct an optimization function F=α×D+β×E+γ×S, where D represents the straight-line distance from the current position of the loading platform to the position coordinates of the storage box, E represents the number of turns in the path, S represents the weighted sum of the adjacent storage density and storage fit coefficient of the same type of fabric, and α, β, and γ represent weight coefficients. For each storage box in the candidate storage box set, calculate the optimization function value F, select the storage box with the smallest F value as the target storage location, and use Dijkstra's algorithm to calculate the shortest path from the current position of the loading platform to the target storage location, which is set as the initial transportation path.

[0011] Preferably, the real-time dynamic information includes dynamic obstacle location information, real-time operating trajectory information of other loading platforms, and congestion area identification information.

[0012] Preferably, real-time dynamic information of the storage repository is obtained, and based on the real-time dynamic information, a hierarchical A* search algorithm is used to perform local incremental dynamic replanning on the initial transportation path to obtain the final transportation path, including: The physical space of the storage facility is divided into a global navigation layer and a local obstacle avoidance layer. The global navigation layer constructs a static topology map with the fixed facilities of the storage facility as nodes. The local obstacle avoidance layer, based on the global navigation layer, transforms dynamic obstacles, the real-time running trajectories of other loading platforms, and congested areas in the real-time dynamic information into dynamic cost weights. The initial transportation path is discretized to obtain a path sequence consisting of multiple continuous path segments. Dynamic conflict detection is performed on each path segment in turn. The starting point and ending point of the path segment are used as the starting point and target point of local planning. In the local obstacle avoidance layer, based on real-time dynamic information, it is predicted whether the path segment has a collision risk or enters a congested area within a preset time window. If a collision risk or congestion risk is detected in a path segment, the hierarchical A* search algorithm is triggered to perform local path replanning: In the global navigation layer, global directional guidance for the initial transportation path is maintained. In the local obstacle avoidance layer, the starting point of the current path segment is taken as the starting point, and the ending point of the corresponding path segment is taken as the temporary target point. The cost weights of the predicted occupied area of ​​dynamic obstacles, the trajectory conflict area of ​​other loading platforms, and the congestion area are set to extremely high values. At the same time, a smoothing constraint factor is introduced to reduce the number of turns in the replanned path. The hierarchical A* search algorithm is used to calculate the local alternative path segment that avoids the risk area and has the lowest cost, and then replaces the original path segment; if no risk is detected, the original path segment is kept unchanged. After all path segments have been processed, the connection points between adjacent path segments are smoothly transitioned to form the final transportation path.

[0013] Preferably, when controlling the loading platform based on the final transportation route, the method further includes: The operating speed of the loading platform is adjusted according to the material properties of the target fabric roll; If the softness level is greater than or equal to the preset softness level or the surface vulnerability level is greater than or equal to the preset vulnerability level, then the straight-line running speed of the loading platform will be reduced to 60%-80% of the base speed, and the turning speed will be reduced to 30%-50% of the base speed. Otherwise, maintain the baseline speed.

[0014] Compared with existing technologies, the advantages of this invention are as follows: This invention acquires the fabric characteristic information of the target fabric roll on the loading platform and determines its fabric type, and then filters the storage frames according to the fabric type to determine a set of candidate storage frames. By pre-configuring type binding tags for the frames and removing frames that are full or have insufficient capacity, this technical means directly solves the damage problem caused by mixed storage of fabrics, enabling each fabric roll to be automatically guided to a compatible storage location, achieving refined classification and management of fabrics, and significantly reducing the risk of surface scratches, indentations, or dye migration of the fabric.

[0015] This invention reduces transportation costs and improves subsequent operational efficiency through multi-objective energy consumption optimization. It constructs an optimization function aimed at minimizing transportation energy consumption, considering not only straight-line distance D and the number of turns E (directly related to energy consumption), but also innovatively introducing a weighted sum S of the proximity storage density and storage fit coefficient of similar fabrics. By selecting the storage box with the smallest F value as the target location, this invention reduces the energy consumption of the current transportation (reducing travel distance and energy consumption from frequent acceleration, deceleration, and turns) while concentrating similar fabrics in storage. This reduces cross-regional movement of loading platforms during future batch outbound shipments, thereby improving the overall energy efficiency and operational smoothness of the warehousing system from a global perspective.

[0016] This invention employs a hierarchical A* search algorithm for local incremental dynamic replanning of the initial transportation path. By dividing the storage repository into a global navigation layer and a local obstacle avoidance layer, directional guidance is maintained at the global layer, while information such as dynamic obstacles, other loading platform trajectories, and congested areas is transformed into dynamic cost weights at the local layer. When a collision or congestion risk is detected on a path segment, only the current local path segment is replanned, rather than a global recalculation. This technique enables loading platforms to respond quickly to environmental changes, with low computational load and low latency, avoiding control lag problems caused by global replanning, and significantly improving the real-time performance and robustness of transportation path planning.

[0017] On the other hand, this application also provides a transportation route planning system based on fabric type, for applying the above-mentioned transportation route planning method based on fabric type, including: The fabric type identification module is configured to acquire fabric feature information of the target fabric roll on the loading platform, and determine the fabric type of the target fabric roll based on the fabric feature information. The storage frame filtering module is configured to filter the storage frames in the storage repository according to the fabric type to determine a set of candidate storage frames for storing the target fabric roll. The initial transport path determination module is configured to construct an optimization function with the goal of minimizing transport energy consumption based on the fabric feature information of the target fabric roll and the attribute information of each storage box in the candidate storage box set, and determine the initial transport path from the loading platform to the candidate storage box based on the optimization function. The final transportation route determination module is configured to acquire real-time dynamic information of the storage warehouse, perform local incremental dynamic replanning on the initial transportation route based on the real-time dynamic information using a hierarchical A* search algorithm to obtain the final transportation route, and control the loading platform based on the final transportation route to realize the transportation of the fabric rolls.

[0018] It is understood that the methods and systems provided in this application have the same beneficial effects, and will not be described in detail here. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly described below. Obviously, the 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. In the drawings: Figure 1 A flowchart of a transportation route planning method based on fabric type provided in an embodiment of the present invention; Figure 2 This is a functional block diagram of a transportation route planning system based on fabric type, provided for an embodiment of the present invention. Detailed Implementation

[0020] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] In some embodiments of this application, see Figure 1 As shown, this embodiment provides a transportation route planning method based on fabric type, including: Obtain the fabric feature information of the target fabric roll on the loading platform, and determine the fabric type of the target fabric roll based on the fabric feature information; Based on the fabric type, the storage frames in the storage repository are filtered to determine a set of candidate storage frames for storing the target fabric roll. Based on the fabric feature information of the target fabric roll and the attribute information of each storage box in the candidate storage box set, an optimization function with the goal of minimizing transportation energy consumption is constructed, and the initial transportation path from the loading platform to the candidate storage box is determined based on the optimization function. The system obtains real-time dynamic information of the storage warehouse, and uses a hierarchical A* search algorithm to perform local incremental dynamic replanning on the initial transportation path based on the real-time dynamic information to obtain the final transportation path. Based on the final transportation path, the system controls the loading platform to realize the transportation of the fabric rolls.

[0022] By first filtering candidate storage boxes based on fabric type, the scope of path planning is narrowed down, reducing unnecessary computation. Then, an optimization function is constructed with the goal of minimizing transportation energy consumption to obtain the initial path, which can reduce energy consumption in the transportation process from the source and control transportation costs. Finally, combined with real-time dynamic information of the storage warehouse, a hierarchical A* search algorithm is used to perform local dynamic replanning of the initial path, which can avoid unexpected situations such as temporary obstacles and box occupancy in the storage warehouse in a timely manner, ensuring smooth passage of the final transportation path. This not only improves the efficiency of fabric transportation, but also maintains the optimal level of transportation energy consumption, adapting to the dynamic operation scenario of the storage warehouse.

[0023] In some embodiments of this application, obtaining fabric feature information of the target fabric roll on the loading platform includes: The surface image information of the target fabric roll is acquired by a machine vision inspection device. The acquired surface image information is input into a pre-trained deep learning image classification model for feature extraction and classification to obtain the fabric feature information of the target fabric roll. The fabric feature information includes: texture feature map, color feature map, structural feature map, roll diameter value, and roll width value. The deep learning image classification model is a multi-task learning model based on a convolutional neural network. The multi-task learning model includes a feature extraction backbone network and a size regression branch. The feature extraction backbone network performs multi-layer convolution and pooling operations on the surface image information and outputs the texture feature map, color feature map, and structural feature map of the target fabric roll. The size regression branch outputs the roll diameter value and roll width value of the target fabric roll based on the structural feature map.

[0024] Understandably, machine vision and deep learning technologies are used to non-contactly and automatically acquire the physical properties and surface features of the target fabric rolls on the loading platform. This step aims to solve the problem of accurately identifying the texture, color, structure, and other appearance features of the fabric rolls, while simultaneously acquiring their geometric dimensional parameters such as roll diameter and roll width. This provides an accurate and multi-dimensional data foundation for subsequent determination of fabric type, screening and storage frames, and construction of optimization functions, avoiding information loss or errors caused by manual input or single sensors.

[0025] Specifically, firstly, surface images of the target fabric roll are acquired using machine vision inspection devices, such as industrial cameras or 3D scanners, pre-installed at fixed locations on the loading platform or in the warehouse. These surface images should clearly reflect the fabric roll's texture, color distribution, fabric structure, and end-face shape. Then, the acquired surface images are input into a pre-trained deep learning image classification model. This model employs a multi-task learning architecture based on convolutional neural networks, enabling it to simultaneously perform multiple related but distinct prediction tasks.

[0026] Specifically, this multi-task learning model consists of the following components: A feature extraction backbone network: This network sequentially performs multi-layer convolution and pooling operations on the input surface image information, gradually extracting feature maps at different levels of abstraction from the image. After processing by this backbone network, it outputs three independent feature maps: texture feature map, color feature map, and structural feature map. The texture feature map represents the weaving texture or pattern of the fabric surface, the color feature map represents the color distribution and hue information of the fabric, and the structural feature map represents the fabric's surface structure and winding shape. A size regression branch: This branch, as a sub-network of the multi-task model, receives the structural feature map output by the feature extraction backbone network as input. Based on the geometric structure information in this structural feature map, the size regression branch calculates the diameter and width of the target fabric roll through regression, thereby obtaining the radial and axial dimensions of the fabric roll in physical space.

[0027] Specifically, the machine vision inspection device includes at least one industrial camera and is equipped with an auxiliary lighting source. The auxiliary lighting source automatically adjusts the intensity of the supplementary light according to the ambient light brightness to ensure the clarity of the acquired image.

[0028] In some embodiments of this application, determining the fabric type of the target fabric roll based on the fabric feature information includes: the multi-task learning model includes a type classification branch; the texture feature map, color feature map, and structural feature map are input into the type classification branch to obtain a fabric type label for the target fabric roll; the fabric type label is matched with a preset fabric type-material attribute mapping table to query and obtain the material attributes of the target fabric roll, the material attributes including softness level and surface vulnerability level; simultaneously, the fabric type label is cross-validated with the task order information stored in the warehouse management system; if the identification result is inconsistent with the fabric type recorded in the task order, an alarm is triggered and the transportation process is suspended; if they are consistent, the fabric type label is determined to be the fabric type of the target fabric roll.

[0029] Understandably, the purpose of this step is to automatically identify the specific type of the target fabric roll based on the acquired fabric feature information, using the branching structure of the same deep learning model, and to obtain its material properties by looking up a table, while simultaneously cross-validating it with the task order information of the warehouse management system.

[0030] Specifically, firstly, the multi-task learning model includes a type classification branch. The texture, color, and structural feature maps output by the feature extraction backbone network are input into this branch. The type classification branch performs fusion analysis and classification calculations on these three feature maps, outputting a fabric type label for the target fabric roll. For example, this label could be a specific category such as "silk," "cotton," "linen," or "synthetic fiber." Then, the obtained fabric type label is matched against a pre-built and stored fabric type-material attribute mapping table. This mapping table records the material attributes corresponding to each fabric type, including at least a softness level and a surface vulnerability level. The softness level characterizes the fabric's flexibility, and the surface vulnerability level characterizes the degree to which the fabric surface is easily scratched or damaged when subjected to friction or pressure. By looking up the table, the softness level and surface vulnerability level of the target fabric roll can be obtained; these parameters will be used for subsequent adaptive control of the loading platform's operating speed. Simultaneously, the identified fabric type label is cross-validated with the task order information stored in the warehouse management system. The task order information pre-records the expected fabric type of the fabric roll to be transported. The system compares the visual recognition result with the task order record: if they do not match, it indicates that there may be an error in the visual recognition, an error in the task order information, or a discrepancy between the fabric roll on the loading platform and the task order. In this case, the system triggers an alarm signal and suspends the current transportation process, awaiting manual intervention or reconfirmation; if they match, the visual recognition result is confirmed as correct, the fabric type label is identified as the final fabric type of the target fabric roll, and the subsequent transportation process is allowed to continue.

[0031] Specifically, the fabric type label includes at least one of the following: silk, satin, cotton, linen, denim, canvas, velvet, synthetic fiber, blended fabric, or stretch fabric.

[0032] In some embodiments of this application, the storage frames in the storage repository are filtered according to the fabric type to determine a set of candidate storage frames for storing the target fabric roll. This includes: pre-configuring type binding tags for the storage frames in the storage repository, the type binding tags being used to identify one or more fabric types that the storage frames are allowed to store; based on the fabric type tag of the target fabric roll, traversing all the type binding tags of the storage frames, filtering out the storage frames whose type binding tags contain the fabric type tag of the target fabric roll to form an initial set of candidate storage frames; obtaining the real-time status information of each storage frame in the initial set of candidate storage frames, the real-time status information including the current occupancy status and remaining storage capacity, and removing storage frames whose current occupancy status is full and whose remaining storage capacity is less than the volume parameter of the target fabric roll to obtain the set of candidate storage frames.

[0033] Understandably, based on the determined fabric type, a set of candidate storage boxes that are both type-compatible and meet the storage requirements in terms of capacity are selected from all storage boxes in the repository. The aim is to automatically and quickly eliminate invalid storage boxes that are type-mismatched, full, or have insufficient remaining space from a large number of storage boxes, thereby narrowing down the target range for subsequent route planning. This provides an effective set of alternative targets for calculating the energy consumption optimization function and avoids planning transportation routes to storage boxes that cannot store the fabric roll.

[0034] Specifically, firstly, a type binding tag is pre-configured for each storage frame in the repository. This type binding tag is a data identifier field used to identify that the corresponding storage frame is allowed to store one or more types of fabric. For example, the type binding tag for one frame could be "pure cotton, linen," indicating that the frame only allows storing rolls of pure cotton or linen fabric; another frame could be labeled "silk," indicating that the frame only allows storing rolls of silk fabric. Then, based on the fabric type label of the target fabric roll determined in the previous steps, the system iterates through the type binding tags of all storage frames in the repository. For each frame, it determines whether its type binding tag contains the fabric type label of the target fabric roll. If it does, the frame is selected; otherwise, it is excluded. After iteration and selection, all storage frames whose type binding tags contain the target fabric type label constitute the initial candidate storage frame set. Next, the real-time status information of each storage frame in the initial candidate storage frame set is obtained. This real-time status information includes at least the current occupancy status and remaining storage capacity. The current occupancy status indicates whether the storage frame is full, while the remaining storage capacity indicates how much available space remains in the frame. Finally, a second filtering process is performed based on the real-time status information: frames currently marked "full" are removed, as are frames with remaining storage capacity less than the target fabric roll volume parameter. The target fabric roll volume parameter can be calculated based on the roll diameter and width values ​​obtained in the preceding steps. After these removal operations, the remaining storage frames constitute the final candidate storage frame set. These frames are compatible in type and have sufficient remaining space to store the target fabric roll.

[0035] Specifically, storage boxes with different types of binding tags are arranged in zones or mixed in the warehouse geographic space.

[0036] In some embodiments of this application, the attribute information of the storage frame includes location coordinates, current occupancy status, neighboring storage density of the same type of fabric, and remaining storage capacity; the neighboring storage density of the same type of fabric is determined by statistically analyzing the ratio of the number of fabric rolls of the same type as the target fabric roll stored within a preset radius around the corresponding storage frame to the total number of stored fabric rolls.

[0037] In some embodiments of this application, based on the fabric feature information of the target fabric roll and the attribute information of each storage frame in the candidate storage frame set, an optimization function with the goal of minimizing transportation energy consumption is constructed. The initial transportation path from the loading platform to the candidate storage frame is determined based on the optimization function, including: calculating the volume parameters of the target fabric roll, which are obtained based on the roll diameter and roll width; determining the storage adaptation coefficient based on the volume parameters and the remaining storage capacity of the storage frame; constructing the optimization function F = α × D + β × E + γ × S, where D represents the straight-line distance from the current position of the loading platform to the location coordinates of the storage frame, E represents the number of turns in the path, S represents the weighted sum of the adjacent storage density of the same type of fabric and the storage adaptation coefficient, and α, β, and γ represent weighting coefficients; calculating the optimization function value F for each storage frame in the candidate storage frame set, selecting the storage frame with the smallest F value as the target storage location, and using Dijkstra's algorithm to calculate the shortest path from the current position of the loading platform to the target storage location, which is then set as the initial transportation path.

[0038] Understandably, based on the attribute information of each storage box in the candidate storage box set, a comprehensive evaluation function is constructed with minimizing transportation energy consumption as the core optimization objective. This function selects the optimal target storage location from multiple candidate storage boxes and plans the initial transportation path to that location. The aim is to quantitatively integrate multiple factors such as distance, number of turns, the global benefits of storing similar fabrics together, and the adaptability of storage box space, so that the selected target location and planned path minimize energy consumption in this transportation process, while also improving the overall efficiency of subsequent outbound operations.

[0039] Specifically, firstly, the attribute information of the storage frame is defined. This attribute information includes location coordinates, current occupancy status, neighboring storage density of similar fabrics, and remaining storage capacity. Among these, the neighboring storage density of similar fabrics is a key parameter, calculated as follows: Using the current storage frame as the center, count the number of fabric rolls of the same type as the target fabric roll among all stored fabric rolls within a preset radius. Divide this number by the total number of fabric rolls stored within the aforementioned radius; the resulting ratio is the neighboring storage density of the same type of fabric for that frame. A higher density value indicates a greater concentration of similar fabrics around the frame.

[0040] Based on the aforementioned attribute information and the fabric characteristics of the target fabric roll, the specific steps for determining the initial transportation path are as follows: First, calculate the volume parameters of the target fabric roll. Using the roll diameter and width values ​​obtained in the previous steps, calculate the volume parameters of the target fabric roll using the cylinder volume formula. This parameter is used for subsequent comparison with the remaining capacity of the storage frame. Second, determine the storage adaptation coefficient based on the volume parameters and the remaining storage capacity of the storage frame. Specifically, the storage adaptation coefficient reflects the degree of matching between the volume of the target fabric roll and the remaining capacity of the storage frame. For example, the ratio of the target fabric roll volume to the remaining capacity of the storage frame can be used as the storage adaptation coefficient. The closer the ratio is to a certain ideal range (e.g., 0.7 to 0.9), the more reasonable the space utilization; a ratio that is too small indicates wasted space, while a ratio that is too large indicates overcrowding. Third, construct the optimization function F. The form of this function is: F = α × D + β × E + γ × S. Where D represents the straight-line distance from the current position of the loading platform to the coordinates of the storage box; the larger this value, the higher the transportation energy consumption. E represents the number of turns in the path from the current position of the loading platform to the storage box; the more turns, the greater the additional energy consumption from acceleration, deceleration, and turning. S represents the weighted sum of the proximity storage density and storage fit coefficient of similar types of fabrics; the larger this value, the better the fabric roll can be stored in the storage box, both for convenient subsequent outbound storage and for good matching with the remaining space in the storage box. α, β, and γ are the weighting coefficients of distance, number of turns, and combination term S, respectively, and can be adjusted according to the emphasis of the actual warehousing scenario. For example, if distance is more important, increase α; if reducing turns is more important, increase β. Fourth step: For each storage box in the candidate storage box set, calculate its corresponding optimization function value F. Compare the F values ​​of all candidate boxes and select the storage box with the smallest F value as the target storage location for this transportation. This box performs best in terms of distance, turning energy consumption, and overall storage efficiency. The fifth step involves using Dijkstra's algorithm to calculate the shortest path from the current location of the loading platform to the target storage location. Since Dijkstra's algorithm guarantees finding the shortest path in a weighted graph, applying it to the warehouse's path topology map yields the initial transportation path. This path is geometrically the shortest, and combined with the previously selected target storage location already optimized for energy consumption, this minimizes overall transportation energy consumption.

[0041] In some embodiments of this application, the real-time dynamic information includes dynamic obstacle location information, real-time operating trajectory information of other loading platforms, and congestion area identification information.

[0042] In some embodiments of this application, real-time dynamic information of the storage facility is obtained, and a hierarchical A* search algorithm is used to perform local incremental dynamic replanning on the initial transportation path based on the real-time dynamic information to obtain the final transportation path. This includes: dividing the physical space of the storage facility into a global navigation layer and a local obstacle avoidance layer; the global navigation layer constructs a static topology map using the fixed facilities of the storage facility as nodes; the local obstacle avoidance layer, based on the global navigation layer, converts dynamic obstacles, real-time operating trajectories of other loading platforms, and congested areas in the real-time dynamic information into dynamic cost weights; the initial transportation path is discretized to obtain a path sequence composed of multiple continuous path segments; dynamic conflict detection is performed on each path segment in sequence, and the starting point and ending point of the path segment are used as the starting point and target point of local planning; in the local obstacle avoidance layer, the path is predicted based on the real-time dynamic information. The system checks whether a segment faces collision risk or enters a congested area within a preset time window. If a collision risk or congestion risk is detected, a hierarchical A* search algorithm is triggered for local path replanning: At the global navigation layer, global directional guidance for the initial transport path is maintained. At the local obstacle avoidance layer, the starting point of the current path segment is used as the starting point, and the ending point of the corresponding path segment is used as the temporary target point. The predicted occupancy area of ​​dynamic obstacles, the trajectory conflict area of ​​other loading platforms, and the cost weight of congested areas are set to extremely high values. Simultaneously, a smoothing constraint factor is introduced to reduce the number of turns in the replanned path. The hierarchical A* search algorithm calculates the local alternative path segment that avoids the risk area and has the lowest cost, replacing the original path segment. If no risk is detected, the original path segment remains unchanged. After processing all path segments, the connection points of adjacent path segments are smoothly transitioned to form the final transport path.

[0043] Understandably, by utilizing real-time dynamic information from the storage repository and employing a hierarchical A* search algorithm to perform local incremental dynamic replanning of the initial transportation path, a final transportation path is generated that avoids dynamic obstacles, conflicts with other loading platform trajectories, and congested areas. This approach aims to address the problem that statically planned paths cannot adapt to real-time changes in the warehousing environment. By replanning only risky local path segments while maintaining global directional guidance, computational complexity is reduced, and the real-time performance and response speed of path planning are improved. Simultaneously, smooth transition processing ensures the executability of the path.

[0044] In some embodiments of this application, when controlling the loading platform based on the final transportation path, the method further includes: adjusting the running speed of the loading platform according to the material properties of the target fabric roll; if the softness level is greater than or equal to a preset softness level or the surface vulnerability level is greater than or equal to a preset vulnerability level, then the straight-line running speed of the loading platform is reduced to 60%-80% of the reference speed, and the speed when turning is reduced to 30%-50% of the reference speed; otherwise, the reference speed is maintained.

[0045] Understandably, during the control of the loading platform according to the final transportation path, the platform's operating speed is adaptively adjusted based on the material properties of the target fabric roll (i.e., softness level and surface vulnerability level). This aims to address the issue of different fabrics having varying tolerances to inertial forces, vibrations, and impacts during transportation. By proactively reducing the transportation speed for soft or vulnerable fabrics, especially by significantly slowing down during turns, the platform's acceleration / deceleration or centrifugal force can prevent fabric roll deformation, surface scratches, or loosening, thus achieving protective transportation of high-value or fragile fabrics.

[0046] Specifically, the system acquires the material properties of the previously determined target fabric roll, including at least a softness grade and a surface vulnerability grade. The softness grade quantifies the fabric's flexibility; a higher grade indicates a softer fabric that is more easily deformed. The surface vulnerability grade quantifies the fabric's resistance to friction and abrasion; a higher grade indicates a surface that is more prone to scratches or damage. The system then compares the acquired softness grade with a preset softness grade threshold, and simultaneously compares the surface vulnerability grade with a preset vulnerability grade threshold. The preset softness grade and preset vulnerability grade can be set according to the definition standards for fragile fabrics in actual production. If either of the following two conditions is met: the softness grade is greater than or equal to the preset softness grade, or the surface vulnerability grade is greater than or equal to the preset vulnerability grade, then the current target fabric roll is considered a fragile or highly flexible material, requiring protective transportation measures. At this point, the system adjusts the speed of the loading platform by reducing it: during straight-line travel, the platform's speed is reduced to 60% to 80% of the baseline speed; during turning, because centrifugal force has a more significant impact on the fabric roll during turns, the speed is further reduced to 30% to 50% of the baseline speed. The baseline speed refers to the standard operating speed of the loading platform when transporting ordinary fabric (i.e., fabric with both softness and surface vulnerability grades below the threshold). If the softness grade is less than the preset softness grade and the surface vulnerability grade is less than the preset vulnerability grade, it indicates that the current target fabric roll is of ordinary material and is not sensitive to inertial forces and vibrations during transportation. In this case, the loading platform operates normally at the baseline speed without requiring speed reduction adjustment.

[0047] On the other hand, see Figure 2 As shown, this application also provides a transportation route planning system based on fabric type, for applying the above-mentioned transportation route planning method based on fabric type, including: The fabric type identification module is configured to acquire fabric feature information of the target fabric roll on the loading platform and determine the fabric type of the target fabric roll based on the fabric feature information.

[0048] The storage frame filtering module is configured to filter the storage frames in the storage repository according to the fabric type to determine a set of candidate storage frames for storing the target fabric roll.

[0049] The initial transport path determination module is configured to construct an optimization function with the goal of minimizing transport energy consumption based on the fabric feature information of the target fabric roll and the attribute information of each storage box in the candidate storage box set, and determine the initial transport path from the loading platform to the candidate storage box based on the optimization function.

[0050] The final transportation route determination module is configured to acquire real-time dynamic information of the storage warehouse, perform local incremental dynamic replanning on the initial transportation route based on the real-time dynamic information using a hierarchical A* search algorithm to obtain the final transportation route, and control the loading platform based on the final transportation route to realize the transportation of the fabric rolls.

[0051] This invention first narrows the scope of path planning by selecting suitable candidate storage boxes based on the type of the target fabric roll, avoiding invalid traversals and improving planning efficiency. Then, it constructs an optimization function with the goal of minimizing transportation energy consumption to obtain the initial path. Combining the real-time dynamic information of the storage warehouse, it completes local dynamic replanning through a hierarchical A* search algorithm to obtain the final path. This ensures that the path planning can adapt to the real-time traffic conditions within the storage warehouse, while reducing the computational load of global replanning. It balances transportation energy consumption control with the real-time performance and accuracy of path planning, and can adapt to the storage and transportation needs of different types of fabrics, thereby improving the overall operational efficiency of the fabric storage and transfer process.

[0052] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0053] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0054] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0055] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0056] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for planning transportation routes based on fabric type, characterized in that, include: Obtain the fabric feature information of the target fabric roll on the loading platform, and determine the fabric type of the target fabric roll based on the fabric feature information; Based on the fabric type, the storage frames in the storage repository are filtered to determine a set of candidate storage frames for storing the target fabric roll; Based on the fabric feature information of the target fabric roll and the attribute information of each storage box in the candidate storage box set, an optimization function with the goal of minimizing transportation energy consumption is constructed, and the initial transportation path from the loading platform to the candidate storage box is determined based on the optimization function. The system obtains real-time dynamic information of the storage warehouse, and uses a hierarchical A* search algorithm to perform local incremental dynamic replanning on the initial transportation path based on the real-time dynamic information to obtain the final transportation path. Based on the final transportation path, the system controls the loading platform to realize the transportation of the fabric rolls.

2. The transportation route planning method based on fabric type according to claim 1, characterized in that, Obtain the fabric feature information of the target fabric roll on the loading platform, including: The surface image information of the target fabric roll is collected by a machine vision inspection device. The collected surface image information is input into a pre-trained deep learning image classification model for feature extraction and classification recognition to obtain the fabric feature information of the target fabric roll. The fabric feature information includes: texture feature map, color feature map, structural feature map, roll diameter value and roll width value. The deep learning image classification model is a multi-task learning model based on a convolutional neural network. The multi-task learning model includes a feature extraction backbone network and a size regression branch. The feature extraction backbone network performs multi-layer convolution and pooling operations on the surface image information and outputs the texture feature map, color feature map, and structural feature map of the target fabric roll. The size regression branch outputs the roll diameter and roll width of the target fabric roll based on the structural feature map.

3. The transportation route planning method based on fabric type according to claim 2, characterized in that, Determining the fabric type of the target fabric roll based on the fabric feature information includes: The multi-task learning model includes a type classification branch. The texture feature map, color feature map and structural feature map are input into the type classification branch to obtain the fabric type label of the target fabric roll. The fabric type label is matched with a preset fabric type-material attribute mapping table to obtain the material attributes of the target fabric roll, including softness level and surface vulnerability level. At the same time, the fabric type label is cross-validated with the task order information stored in the warehouse management system. If the identification result is inconsistent with the fabric type recorded in the task order, an alarm is triggered and the transportation process is suspended; if they are consistent, the fabric type label is determined to be the fabric type of the target fabric roll.

4. The transportation route planning method based on fabric type according to claim 3, characterized in that, Based on the fabric type, the storage frames in the repository are filtered to determine a set of candidate storage frames for storing the target fabric roll, including: Type binding tags are pre-configured for storage boxes in the repository, and the type binding tags are used to identify one or more types of fabric that the storage box is allowed to store; Based on the fabric type tag of the target fabric roll, traverse the type binding tags of all storage frames, and filter out the storage frames whose type binding tags contain the fabric type tag of the target fabric roll to form an initial candidate storage frame set. Obtain the real-time status information of each storage frame in the initial candidate storage frame set. The real-time status information includes the current occupancy status and the remaining storage capacity. Remove the storage frames whose current occupancy status is full and whose remaining storage capacity is less than the target fabric roll volume parameter to obtain the candidate storage frame set.

5. The transportation route planning method based on fabric type according to claim 4, characterized in that, The storage box's attribute information includes its location coordinates, current occupancy status, adjacent storage density of similar fabrics, and remaining storage capacity. The adjacent storage density of the same type of fabric is determined by statistically analyzing the ratio of the number of fabric rolls of the same type as the target fabric roll stored within a preset radius around the corresponding storage frame to the total number of stored fabric rolls.

6. The transportation route planning method based on fabric type according to claim 5, characterized in that, Based on the fabric feature information of the target fabric roll and the attribute information of each storage box in the candidate storage box set, an optimization function is constructed with the goal of minimizing transportation energy consumption. Based on this optimization function, the initial transportation path from the loading platform to the candidate storage box is determined, including: Calculate the volume parameters of the target fabric roll, which are obtained based on the roll diameter and roll width. Based on the volume parameters and the remaining storage capacity of the storage frame, the storage adaptation coefficient is determined; Construct an optimization function F=α×D+β×E+γ×S, where D represents the straight-line distance from the current position of the loading platform to the position coordinates of the storage box, E represents the number of turns in the path, S represents the weighted sum of the adjacent storage density and storage fit coefficient of the same type of fabric, and α, β, and γ represent weight coefficients. For each storage box in the candidate storage box set, calculate the optimization function value F, select the storage box with the smallest F value as the target storage location, and use Dijkstra's algorithm to calculate the shortest path from the current position of the loading platform to the target storage location, which is set as the initial transportation path.

7. The transportation route planning method based on fabric type according to claim 6, characterized in that, The real-time dynamic information includes the location information of dynamic obstacles, the real-time operating trajectory information of other loading platforms, and the identification information of congested areas.

8. The transportation route planning method based on fabric type according to claim 7, characterized in that, Obtain real-time dynamic information of the storage repository, and based on this real-time dynamic information, use a hierarchical A* search algorithm to perform local incremental dynamic replanning on the initial transportation path to obtain the final transportation path, including: The physical space of the storage facility is divided into a global navigation layer and a local obstacle avoidance layer. The global navigation layer constructs a static topology map with the fixed facilities of the storage facility as nodes. The local obstacle avoidance layer, based on the global navigation layer, transforms dynamic obstacles, the real-time running trajectories of other loading platforms, and congested areas in the real-time dynamic information into dynamic cost weights. The initial transportation path is discretized to obtain a path sequence consisting of multiple continuous path segments. Dynamic conflict detection is performed on each path segment in turn. The starting point and ending point of the path segment are used as the starting point and target point of local planning. In the local obstacle avoidance layer, based on real-time dynamic information, it is predicted whether the path segment has a collision risk or enters a congested area within a preset time window. If a collision risk or congestion risk is detected in a path segment, the hierarchical A* search algorithm is triggered to perform local path replanning: In the global navigation layer, global directional guidance for the initial transportation path is maintained. In the local obstacle avoidance layer, the starting point of the current path segment is taken as the starting point, and the ending point of the corresponding path segment is taken as the temporary target point. The cost weights of the predicted occupied area of ​​dynamic obstacles, the trajectory conflict area of ​​other loading platforms, and the congestion area are set to extremely high values. At the same time, a smoothing constraint factor is introduced to reduce the number of turns in the replanned path. The hierarchical A* search algorithm is used to calculate the local alternative path segment that avoids the risk area and has the lowest cost, and then replaces the original path segment; if no risk is detected, the original path segment is kept unchanged. After all path segments have been processed, the connection points between adjacent path segments are smoothly transitioned to form the final transportation path.

9. The transportation route planning method based on fabric type according to claim 8, characterized in that, When controlling the loading platform based on the final transportation route, the method further includes: The operating speed of the loading platform is adjusted according to the material properties of the target fabric roll; If the softness level is greater than or equal to the preset softness level or the surface vulnerability level is greater than or equal to the preset vulnerability level, then the straight-line running speed of the loading platform will be reduced to 60%-80% of the reference speed, and the turning speed will be reduced to 30%-50% of the reference speed. Otherwise, maintain the baseline speed.

10. A transportation route planning system based on fabric type, used to apply the transportation route planning method based on fabric type as described in any one of claims 1-9, characterized in that, include: The fabric type identification module is configured to acquire fabric feature information of the target fabric roll on the loading platform, and determine the fabric type of the target fabric roll based on the fabric feature information. The storage frame filtering module is configured to filter the storage frames in the storage repository according to the fabric type to determine a set of candidate storage frames for storing the target fabric roll. The initial transport path determination module is configured to construct an optimization function with the goal of minimizing transport energy consumption based on the fabric feature information of the target fabric roll and the attribute information of each storage box in the candidate storage box set, and determine the initial transport path from the loading platform to the candidate storage box based on the optimization function. The final transportation route determination module is configured to acquire real-time dynamic information of the storage warehouse, perform local incremental dynamic replanning on the initial transportation route based on the real-time dynamic information using a hierarchical A* search algorithm to obtain the final transportation route, and control the loading platform based on the final transportation route to realize the transportation of the fabric rolls.

Images