Storage sorting path collaborative optimization method based on cluster resource dynamic scheduling

Abstract

The invention discloses a warehouse sorting path collaborative optimization method based on cluster resource dynamic scheduling, and belongs to the technical field of logistics automation, and the method comprises the steps: obtaining an order evaluation coefficient based on order cut-off time, order article weight and order article value; acquiring a transfer evaluation coefficient based on the total robot transfer distance, the current robot load and the emergency order proportion in the robot queue; obtaining the busy degree of the workbench based on the queue length of the workbench, the emergency order proportion in the queue and the plugging frequency proportion; obtaining a path evaluation coefficient based on the busy degree of the path, the turning times and the slope road section proportion; obtaining a transfer-workbench-path matching degree based on the order evaluation coefficient, the transfer evaluation coefficient, the workbench busy degree and the path evaluation coefficient; obtaining a target average speed based on the transfer-workbench-path matching degree and the determined average speed of the current robot for transferring the order articles; through multi-dimensional collaborative optimization, the storage sorting efficiency and the resource utilization rate are improved.

Description

Technical Field

[0001] This invention belongs to the field of logistics automation technology, and in particular relates to a collaborative optimization method for warehouse sorting paths based on dynamic scheduling of cluster resources. Background Technology

[0002] With the rapid development of e-commerce logistics, warehousing and sorting systems face challenges such as surging orders and increased timeliness requirements. Existing technologies have the following problems:

[0003] Insufficient order prioritization: Traditional methods prioritize orders based on only one dimension, such as deadline or value, without integrating a comprehensive assessment of weight, timeliness, and value, leading to delays or resource misallocation for high-value orders;

[0004] Resource scheduling and path planning are disconnected: the allocation of transfer robots, selection of workstations, and path planning are often optimized independently, lacking a collaborative mechanism. For example, robot selection does not consider the real-time queue congestion of workstations, and path planning ignores dynamic load changes of robots, leading to local resource conflicts.

[0005] Static speed control: The robot speed is fixed or adjusted only based on the path status, without taking into account the urgency of the order and the matching degree of global resources, which causes high-priority orders to be overdue due to insufficient speed.

[0006] Therefore, there is an urgent need for a collaborative optimization method for warehouse sorting paths based on dynamic scheduling of cluster resources. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a collaborative optimization method for warehouse sorting paths based on dynamic scheduling of cluster resources, thus solving the aforementioned problems.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a collaborative optimization method for warehouse sorting paths based on dynamic scheduling of cluster resources, comprising the following steps:

[0009] An order evaluation model is constructed based on the order deadline, order weight, and order value, and the output order evaluation coefficient is then used.

[0010] A transfer evaluation model is constructed based on the total transfer distance of the robot, the current load of the robot, and the proportion of urgent orders in the robot queue, and the transfer evaluation coefficient is output.

[0011] A workbench busyness model is constructed based on the workbench queue length (number of tasks), the proportion of urgent orders in the workbench queue, and the proportion of times the queue was interrupted, and the workbench busyness is output.

[0012] A path evaluation model is constructed based on path busyness, number of turns, and percentage of gradient sections, and the path evaluation coefficient is output.

[0013] Based on the order evaluation coefficient, the transfer evaluation coefficient, the workbench busyness and the path evaluation coefficient, a transfer-workbench-path matching degree model is constructed and the transfer-workbench-path matching degree is output.

[0014] Based on the matching degree between transfer, workbench, and path, and the current average speed of the robot for the transferred order items, a speed optimization model is constructed to output the target average speed.

[0015] Based on the above technical solutions, the present invention also provides the following optional technical solutions:

[0016] Further technical solution: The steps for constructing an order evaluation model and outputting an order evaluation coefficient based on the order deadline, order weight, and order value are as follows:

[0017] The order deadline index is obtained by taking the ratio of the minimum allowed deadline to the total deadline.

[0018] The weight index and value index are obtained by comparing the weight and value of the ordered items with their maximum allowable values, respectively.

[0019] An order evaluation model is constructed based on the order deadline index, weight index, and value index. The order evaluation model is expressed as follows:

[0020]

[0021] in, This represents the order rating coefficient. Indicates the time decay coefficient. This represents the order deadline index. Indicates the weight index. Indicates the value index. Represents the weight coefficient and The And the larger the value, the higher the priority;

[0022] Import the current order deadline index, current weight index, and current value index into the order evaluation model to output the current order evaluation coefficient.

[0023] Further technical solution: The steps for constructing a transfer evaluation model and outputting transfer evaluation coefficients based on the total transfer distance of the robot, the current load of the robot, and the proportion of urgent orders in the robot queue are as follows:

[0024] The transfer distance index is obtained by comparing the total transfer distance of the robot with the maximum allowable total transfer distance.

[0025] The load index is obtained by comparing the robot's current load with the weight of the ordered items and then with the maximum allowable load.

[0026] A transfer evaluation coefficient is constructed based on the transfer distance index, load index, and the proportion of urgent orders in the robot queue. The transfer evaluation model is expressed as follows:

[0027]

[0028] in, This represents the transshipment evaluation coefficient. Indicates the transit distance index. Indicates the load index. This indicates the proportion of urgent orders in the robot queue. Represents the weight coefficient and The The larger the value, the higher the transfer efficiency;

[0029] Output the transfer evaluation coefficients of multiple robot groups from the current transfer distance index, current load index, and the proportion of urgent orders in the current robot queue. Take the robot with the maximum value among the multiple transfer evaluation coefficients as the order transfer robot.

[0030] Further technical solution: The steps to construct a workbench busyness model based on the workbench queue length (number of tasks), the proportion of urgent orders in the workbench queue, and the percentage of times queue jumping are as follows:

[0031] The queue length index is obtained by taking the ratio of the workbench queue length to the maximum allowed queue length.

[0032] The ratio of the percentage of cut-in attempts to the percentage of the maximum allowed cut-in attempts is used to obtain the cut-in attempt percentage index;

[0033] A workbench busyness model is constructed based on the queue length index, the percentage of times someone cuts in line index, and the proportion of urgent orders in the workbench queue. This workbench busyness model is expressed as follows:

[0034]

[0035] in, Indicates the workload of the workbench. Indicates the queue length index. This indicates the proportion of urgent orders in the workbench queue. This indicates the percentage of times someone cuts in line. Represents the weight coefficient and The The higher the value, the busier the workbench;

[0036] The queue length index, the percentage of times someone cuts in line index, and the proportion of urgent orders in the workbench queue are used to output the workbench busyness of multiple workbench groups. The workbench with the minimum workbench busyness among the multiple workbench groups is taken as the order processing workbench.

[0037] Further technical solution: The steps for constructing a path evaluation model and outputting path evaluation coefficients based on path busyness, number of turns, and percentage of gradient road sections are as follows:

[0038] The business index and the number of turns are obtained by comparing the route busyness and the number of turns with their maximum allowable values, respectively.

[0039] A route evaluation model is constructed based on the traffic volume index, route turning index, and percentage of sloped road sections. The route evaluation model is expressed as follows:

[0040]

[0041] in, Represents the path evaluation coefficient. Indicates the busyness index. Indicates the path turning index. Indicates the percentage of road sections with slope. Represents the weight coefficient and The The larger the value, the smoother the path;

[0042] The busyness index, turning index, and percentage of slope sections from multiple routes are imported into the route evaluation model to obtain multiple route evaluation coefficients. The route with the maximum value of the multiple route evaluation coefficients is taken as the order item transfer route.

[0043] Further technical solution: Based on the current obstacle density (the number of dynamic obstacles per unit area), the current average robot speed, the current robot queue length, and the robot's time occupancy on the path (the proportion of time a robot occupies the path within the observation period), a path busyness model is constructed to output a path busyness index. The path busyness model is expressed as:

[0044]

[0045] in, Indicates the busyness of the path. Indicates the current obstacle density. Current average robot speed, Indicates the maximum allowed speed for the path. This indicates the current queue length of the robots. Indicates the maximum allowed queue length. This indicates the current time occupancy of the robot on the path. The larger the value, the more congested the path.

[0046] A further technical solution: The transfer-workbench-path matching degree model is represented as follows:

[0047]

[0048] in, Indicates the matching degree of the transfer-workbench-path. This represents the order rating coefficient. This represents the maximum value of the transfer evaluation coefficient among multiple groups of transfer robots. This represents the minimum busy level among multiple worktable groups. This represents the maximum path evaluation coefficient among multiple path groups. Represents the weight coefficient and The The larger the value, the better the system match.

[0049] A further technical solution: The speed optimization model is expressed as follows:

[0050]

[0051] in, Indicates the average speed of the target. Indicates the matching degree of the transfer-workbench-path. Indicates the maximum permissible average speed. Indicates the minimum permissible average speed. Represents the speed smoothing coefficient. Indicates the current average speed. This represents the baseline average speed.

[0052] This invention provides a collaborative optimization method for warehouse sorting paths based on dynamic scheduling of cluster resources, which has the following advantages compared with the prior art:

[0053] 1. This invention breaks through the limitations of traditional single-point optimization by integrating order urgency, robot status, workbench load and path passage efficiency, and achieves optimal matching of global resources. It can dynamically adjust the robot speed based on the global matching degree, speeding up in high matching scenarios to ensure the timeliness of urgent orders, and slowing down to save energy in low matching scenarios.

[0054] 2. In this invention, the workbench busyness and path evaluation are linked in real time, automatically avoiding high queue workbench and congested paths, and realizing intelligent avoidance of congestion hotspots.

[0055] 3. This invention uses order evaluation coefficients to guide global matching, ensuring that high-value, near-expiry orders are allocated resources first, so as to prioritize the protection of urgent orders. Attached Figure Description

[0056] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0058] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0059] Please see Figure 1 The present invention provides a method for collaborative optimization of warehouse sorting paths based on dynamic scheduling of cluster resources, comprising the following steps:

[0060] An order evaluation model is constructed based on the order deadline, order weight, and order value, and the output order evaluation coefficient is then used.

[0061] A transfer evaluation model is constructed based on the total transfer distance of the robot, the current load of the robot, and the proportion of urgent orders in the robot queue, and the transfer evaluation coefficient is output.

[0062] A workbench busyness model is constructed based on the workbench queue length (number of tasks), the proportion of urgent orders in the workbench queue, and the proportion of times the queue was interrupted, and the workbench busyness is output.

[0063] A path evaluation model is constructed based on path busyness, number of turns, and percentage of gradient sections, and the path evaluation coefficient is output.

[0064] Based on the order evaluation coefficient, the transfer evaluation coefficient, the workbench busyness and the path evaluation coefficient, a transfer-workbench-path matching degree model is constructed and the transfer-workbench-path matching degree is output.

[0065] Based on the matching degree between transfer, workbench, and path, and the current average speed of the robot for the transferred order items, a speed optimization model is constructed to output the target average speed.

[0066] The order evaluation model is a composite calculation model that integrates order deadline, item weight, and value. It can be implemented using an exponential weighted summation method, such as normalizing each parameter and assigning weight coefficients, to address resource misallocation caused by single-dimensional priority allocation. The transfer evaluation model is a quantitative model for assessing robot transportation efficiency. It can be implemented using a linear combination of transfer distance, load status, and task urgency to balance robot resource utilization and task response timeliness. The workbench busyness model is a dynamic indicator representing the workbench's processing capacity. It can be implemented using a comprehensive calculation of queue length, urgent task ratio, and queue-jumping interference to optimize task allocation and reduce processing delays. The path evaluation model is a composite indicator for assessing path efficiency. It can be implemented using a weighted calculation of path congestion, turning losses, and slope energy consumption to improve the dynamic adaptability of path selection. The transfer-workbench-path matching model is a global resource collaborative optimization indicator used to achieve optimal matching among orders, robots, workbenches, and paths. The speed optimization model refers to the dynamic adjustment mechanism of robot speed. Specifically, it can be implemented by using a mapping function between the matching degree index and the speed range to solve the problem that static speed control cannot adapt to changes in the global state.

[0067] Specifically, the order evaluation model generates standardized indices by comparing order deadlines, weights, and values ​​with preset thresholds. For example, it uses the ratio of the actual deadline to the minimum allowed time as a time decay factor and combines it with an exponential function to calculate time sensitivity. The transfer evaluation model calculates the matching degree between the robot's remaining capacity and the urgency of the task. For example, it weights the transfer distance index and the load index inversely and adds the proportion of urgent orders to generate a transfer efficiency score. The workbench busyness model monitors queue backlog and task queue-jumping frequency. For example, it uses a linear combination of queue length index and queue-jumping frequency index, combined with the proportion of urgent tasks to assess processing capacity. The path evaluation model collects real-time path congestion data and terrain features. For example, it uses the ratio of obstacle density to robot speed as a congestion factor and combines the number of turns and slope percentage to calculate path efficiency. The transfer-workbench-path matching degree model normalizes the scores of each subsystem. For example, it uses the order evaluation coefficient as a global weighting factor and weights and fuses the scores of the optimal transfer robot, the least busy workbench, and the smoothest path. Speed ​​optimization models achieve a balance between transportation efficiency and energy consumption by establishing a nonlinear mapping relationship between matching degree and speed range, such as using the sigmoid function to smoothly adjust the target speed.

[0068] Compared to existing technologies, traditional order priority calculations rely solely on deadlines or single value metrics. This solution, however, integrates a three-dimensional evaluation of time decay, weight loss, and value gain, avoiding misjudgments of priority for high-value, heavy-duty orders. Existing resource scheduling methods independently optimize robot allocation and path planning, while this solution constructs a transfer-workstation-path matching index to achieve cross-subsystem collaborative decision-making. For example, it simultaneously matches high-priority orders with low-load robots, idle workstations, and unobstructed paths. Traditional speed control strategies only respond to local path states, while this solution dynamically adjusts speed based on global matching. For instance, it increases speed to shorten transportation time when system resources are highly matched, and decreases speed to reduce energy consumption when matching is low.

[0069] Through the above technical solutions, this application can effectively solve the problem of multi-dimensional resource collaborative optimization. By employing a composite order evaluation mechanism, resource mismatch caused by single-dimensional assessment is avoided; for example, high-value, heavy orders are prevented from being incorrectly downgraded due to excessively high weight indices. Dynamic matching degree calculation enables collaborative scheduling of robots, workstations, and paths; for example, urgent orders are prioritized for robots with lighter loads and unobstructed paths. A speed optimization model allows the robot's transport speed to adaptively match the global system state; for example, the speed is automatically increased when the path is unobstructed and the workstation is idle to shorten the order processing cycle.

[0070] Preferably, the steps for constructing an order evaluation model and outputting an order evaluation coefficient based on the order deadline, order weight, and order value are as follows:

[0071] The order deadline index is obtained by taking the ratio of the minimum allowed deadline to the total deadline.

[0072] The weight index and value index are obtained by comparing the weight and value of the ordered items with their maximum allowable values, respectively.

[0073] An order evaluation model is constructed based on the order deadline index, weight index, and value index. The order evaluation model is expressed as follows:

[0074]

[0075] in, This represents the order rating coefficient. Indicates the time decay coefficient. This represents the order deadline index. Indicates the weight index. Indicates the value index. Represents the weight coefficient and The And the larger the value, the higher the priority;

[0076] Import the current order deadline index, current weight index, and current value index into the order evaluation model to output the current order evaluation coefficient.

[0077] The order deadline index is the ratio of the minimum processing time allowed by the system to the actual remaining processing time of the order. This can be achieved by calculating the ratio of the remaining time to a preset threshold in real time, quantifying the urgency of the time. The weight index is the ratio of the order item weight to the maximum allowed single-transport weight. This is achieved by obtaining item weight data from sensors and normalizing it against a preset weight limit, reflecting the impact of transportation costs on priority. The value index is the ratio of the order item value to the system's preset maximum single-item value. This is achieved by querying item value data from a database and standardizing it against a preset value limit, representing the economic efficiency of the order. The time decay coefficient k_T is the adjustment parameter of the exponential function. It can be determined through historical data analysis to determine the degree of time sensitivity's impact on priority, controlling the rate of priority increase for orders nearing their deadline. The weight coefficient a_i is the allocation ratio of each evaluation dimension, which can be achieved using a dynamic adjustment algorithm or preset empirical values, balancing the combined effects of time, weight, and value.

[0078] Specifically, the order evaluation model achieves dynamic priority calculation through the fusion of multidimensional data standardization and nonlinear functions. First, the raw data is processed by ratio analysis to eliminate dimensional differences and transform it into comparable indicators within the [0,1] interval. The time dimension employs an exponential function. This allows orders nearing their deadline to exhibit a non-linear priority growth characteristic, avoiding the problem of insufficient sensitivity of linear models to urgent orders. The weight dimension adopts... The reverse indicator reflects the principle of prioritizing lightweight orders to reduce transportation energy consumption. The value dimension directly adopts... Positive metrics ensure that high-value orders receive higher weighting. Weighting coefficient. The weighting of these three factors is dynamically adjusted using preset or adaptive algorithms. For example, the weight of the value coefficient can be increased during promotional periods, while the weight of the time coefficient can be increased during peak logistics periods. The final output evaluation coefficient... This provides a unified quantitative basis for subsequent resource scheduling, for example, when An emergency dispatch strategy is triggered when the value is >0.8.

[0079] Compared to existing technologies, traditional methods typically prioritize orders based solely on deadline countdowns or a single dimension of value, lacking a comprehensive evaluation using multi-dimensional indicators. For example, some systems only mark orders with less than 2 hours remaining as urgent, without considering whether their weight is suitable for the current idle robot's load capacity, leading to delays for lightweight, high-value orders being replaced by heavy, ordinary orders. This solution introduces standardized processing of weight and value indices, adding transportation costs and economic benefits to the time urgency assessment. For instance, when two orders have the same remaining time, the lighter, higher-value order is prioritized, thus preventing robots from reducing overall efficiency due to transporting overweight items and minimizing the risk of default on high-value orders.

[0080] Through the above technical solutions, this application achieves dynamic multidimensional evaluation of order priority, effectively solving the resource misallocation problem caused by the single-dimensional evaluation in traditional methods. The synergistic effect of the time decay function and the inverse weight index ensures both timely processing of urgent orders and optimizes robot load efficiency. The introduction of the value index enables the warehousing system to prioritize high-profit orders under time constraints. For example, among multiple orders placed simultaneously, priority is given to transporting valuable electronic products rather than ordinary daily necessities, thereby improving economic efficiency per unit time. The adjustability of the weighting coefficient further enhances the system's adaptability; for example, during peak holiday periods, the time factor weight can be increased to strengthen timeliness assurance.

[0081] Preferably, the steps for constructing a transfer evaluation model and outputting transfer evaluation coefficients based on the total transfer distance of the robot, the current load of the robot, and the proportion of urgent orders in the robot queue are as follows:

[0082] The transfer distance index is obtained by comparing the total transfer distance of the robot with the maximum allowable total transfer distance.

[0083] The load index is obtained by comparing the robot's current load with the weight of the ordered items and then with the maximum allowable load.

[0084] A transfer evaluation coefficient is constructed based on the transfer distance index, load index, and the proportion of urgent orders in the robot queue. The transfer evaluation model is expressed as follows:

[0085]

[0086] in, This represents the transshipment evaluation coefficient. Indicates the transit distance index. Indicates the load index. This indicates the proportion of urgent orders in the robot queue. Represents the weight coefficient and The The larger the value, the higher the transfer efficiency;

[0087] Output the transfer evaluation coefficients of multiple robot groups from the current transfer distance index, current load index, and the proportion of urgent orders in the current robot queue. Take the robot with the maximum value among the multiple transfer evaluation coefficients as the order transfer robot.

[0088] The transfer distance index is the ratio of the total transfer distance the robot needs to complete to the system's preset maximum allowable transfer distance. Specifically, it is calculated by collecting real-time path length data from distance sensors and normalizing it using a preset threshold. This index quantifies the impact of transfer path length on efficiency. The load index is the ratio of the robot's current weight plus the weight of newly ordered items to the maximum allowable load. This is achieved by monitoring load data in real-time using weight sensors and calculating remaining carrying capacity. This index dynamically assesses the robot's load status to avoid overload risks. The emergency order ratio is the ratio of orders with an emergency label in the robot's task queue to the total number of tasks. This is calculated by extracting emergency order marker data from the order management system and calculating the ratio. This parameter reflects the robot's ability to handle time-sensitive tasks.

[0089] Specifically, this technical solution achieves dynamic robot allocation by constructing multi-dimensional evaluation indicators. First, the transfer path lengths of different robots are converted into a standardized index through distance ratio processing, eliminating evaluation bias caused by path differences. Second, the remaining transport capacity index is calculated by combining the current load and the weight of new orders, reflecting the robot's available transport capacity in real time. Furthermore, the ability to handle time-sensitive tasks is quantified by statistically analyzing the proportion of urgent orders in the robot's task queue. The above three dimensions of data are input into a linear weighted model, where the weight coefficients can be dynamically adjusted according to the actual scenario; for example, the weight value of the urgent order proportion can be increased during peak periods. By parallel calculation of the transfer evaluation coefficients of all available robots and selecting the robot corresponding to the maximum value, optimal matching is ensured across the three dimensions of path efficiency, load balancing, and urgent task handling.

[0090] Compared to existing technologies, traditional robot scheduling methods typically consider only a single factor, such as selecting only the robot with the shortest path or the lightest load. This solution, however, addresses the local optima problem caused by optimizing a single metric by integrating path distance, real-time load, and task urgency through a collaborative evaluation. Existing scheduling systems using fixed priority strategies cannot dynamically adapt to changes in warehouse workload, while this solution, through a weighted coefficient adjustment mechanism, can flexibly respond to changes in order structure at different times.

[0091] Through the above technical solutions, this application effectively solves the resource mismatch problem in the robot allocation process, improving the processing capacity of high-time-sensitive orders while ensuring transportation efficiency. By dynamically evaluating the transfer distance index, it avoids selecting robots with excessively long paths, reducing ineffective transportation time; by calculating the load index, it ensures the rational utilization of robot capacity, preventing overloaded downtime or wasted empty loads; and by statistically analyzing the proportion of urgent orders, it prioritizes the allocation of robots with strong high-time-sensitive task processing capabilities, reducing the risk of order delays. The synergistic effect of these three indicators enables the system to select the transfer robot with the best overall performance in real time, improving the overall operational efficiency of the warehousing and sorting system.

[0092] Preferably, the steps for constructing a workbench busyness model based on the workbench queue length (number of tasks), the proportion of urgent orders in the workbench queue, and the percentage of times queue jumping occur are as follows:

[0093] The queue length index is obtained by taking the ratio of the workbench queue length to the maximum allowed queue length.

[0094] The ratio of the percentage of cut-in attempts to the percentage of the maximum allowed cut-in attempts is used to obtain the cut-in attempt percentage index;

[0095] A workbench busyness model is constructed based on the queue length index, the percentage of times someone cuts in line index, and the proportion of urgent orders in the workbench queue. This workbench busyness model is expressed as follows:

[0096]

[0097] in, Indicates the workload of the workbench. Indicates the queue length index. This indicates the proportion of urgent orders in the workbench queue. This indicates the percentage of times someone cuts in line. Represents the weight coefficient and The The higher the value, the busier the workbench;

[0098] The queue length index, the percentage of times someone cuts in line index, and the proportion of urgent orders in the workbench queue are used to output the workbench busyness of multiple workbench groups. The workbench with the minimum workbench busyness among the multiple workbench groups is taken as the order processing workbench.

[0099] The queue length index is the ratio of the current number of tasks to the system's capacity limit. It can be dynamically calculated using a real-time counter and a preset threshold, quantifying the impact of task backlog on processing efficiency. The queue insertion frequency index is the ratio of the number of times a sudden high-priority task is inserted to the system's maximum allowed insertion frequency. It can be implemented using event log statistics and a sliding window algorithm, reflecting the stability fluctuations of the workbench due to task scheduling conflicts. The emergency order ratio is the proportion of tasks with time-sensitive constraints in the current queue, implemented using tag recognition and real-time queue scanning technology, assessing the workbench's ability to handle highly sensitive tasks. The weighting coefficients are the adjustment parameters for each indicator in the model, dynamically adjusted using the analytic hierarchy process (AHP) combined with historical data to balance the impact of different factors on workload assessment.

[0100] Specifically, by collecting real-time workbench operation data, the system normalizes three dynamic parameters—queue length, queue-jumping events, and urgent tasks—to eliminate the impact of dimensional differences on the evaluation results. The queue length index avoids evaluation distortion caused by single task accumulation by limiting the maximum allowable value; the queue-jumping frequency ratio index identifies potential scheduling risks by monitoring abnormal queue-jumping behavior; and the urgent order ratio updates priority distribution by dynamically scanning task tags. During model calculation, a weighted summation method is used to fuse the three dimensions, with the queue length index constrained by a minimum function to prevent evaluation bias in extreme cases. The final generated comprehensive busyness value is mapped to a standard range, and the workbench with the lowest busyness is selected as the processing node through global comparison, achieving load balancing and optimized resource allocation.

[0101] Compared to existing technologies, traditional methods allocate workbench tasks solely based on static queue length or fixed priority, failing to detect processing delays caused by queue-jumping operations and efficiency degradation due to the aggregation of urgent tasks. This solution introduces a queue-jumping frequency index to capture dynamic scheduling interference and combines it with the proportion of urgent orders to adjust the busyness assessment weight, forming a multi-dimensional real-time load monitoring mechanism. Compared to a single-indicator evaluation system, this model can more accurately reflect the actual processing capacity of the workbench and avoid resource misallocation caused by sudden changes in local task characteristics.

[0102] Through the above technical solution, this application can effectively identify local resource conflicts caused by sudden queue-jumping tasks and the aggregation of urgent orders, and dynamically select the workbench with the lightest load and the best stability. This mechanism can reduce queuing time during task processing, reduce the risk of delays in high-priority orders due to workbench overload, and improve the adaptability of the warehouse sorting system to complex task flows.

[0103] Preferably, the steps for constructing a path evaluation model and outputting path evaluation coefficients based on path busyness, number of turns, and percentage of gradient road sections are as follows:

[0104] Based on the current obstacle density (the number of dynamic obstacles per unit area), the current average robot speed, the current robot queue length, and the robot's time occupancy on the path (the proportion of time a robot occupies the path within the observation period), a path busyness model is constructed to output a path busyness index. The path busyness model is expressed as follows:

[0105]

[0106] in, Indicates the busyness of the path. Indicates the current obstacle density. Current average robot speed, Indicates the maximum allowed speed for the path. This indicates the current queue length of the robots. Indicates the maximum allowed queue length. This indicates the current time occupancy of the robot on the path. The larger the value, the more congested the path;

[0107] The business index and the number of turns are obtained by comparing the route busyness and the number of turns with their maximum allowable values, respectively.

[0108] A route evaluation model is constructed based on the traffic volume index, route turning index, and percentage of sloped road sections. The route evaluation model is expressed as follows:

[0109]

[0110] in, Represents the path evaluation coefficient. Indicates the busyness index. Indicates the path turning index. Indicates the percentage of road sections with slope. Represents the weight coefficient and The The larger the value, the smoother the path;

[0111] The busyness index, turning index, and percentage of slope sections from multiple routes are imported into the route evaluation model to obtain multiple route evaluation coefficients. The route with the maximum value of the multiple route evaluation coefficients is taken as the order item transfer route.

[0112] The path congestion index is a quantitative indicator of path congestion generated by combining dynamic obstacle density, robot real-time speed, queue length, and time occupancy. It is specifically calculated by weighting data collected from visual sensors, speed monitoring modules, and queue statistics modules to dynamically reflect changes in path traffic efficiency. The number of path turns refers to the number of points of directional change in the path, which can be achieved through path topology analysis or robot motion trajectory recording, and is used to assess the impact of path operational complexity on transportation efficiency. The percentage of sloped sections is the ratio of the length of road sections with an inclination angle exceeding a set threshold to the total path length, which can be achieved using 3D map data or robot posture sensor measurements, and is used to measure the impact of slope on robot energy consumption and driving stability. The congestion index is the ratio of path congestion to a preset maximum allowable value. Normalization can be used to convert path congestion of different dimensions into a comparable index to eliminate the interference of parameter dimension differences on the comprehensive evaluation. The path turn index is the ratio of the number of path turns to a preset maximum allowable value, which can be normalized to map the number of turns to a unified range, and is used to quantify the negative impact of the number of turns on path smoothness.

[0113] Specifically, the path congestion model reflects the interference of dynamic obstacles on traffic efficiency through obstacle density, combines the ratio of the robot's current speed to its maximum speed to reflect actual operating efficiency, and introduces queue length and time occupancy to quantify path occupancy, thus dynamically characterizing the path congestion status. The path congestion and the number of turns are ratio-processed to generate a congestion index and a path turning index. After eliminating dimensional differences, these are input into the path evaluation model along with the percentage of sloped road sections. The path evaluation model adjusts the contribution of these three factors through weighting coefficients. The congestion index reflects the degree of dynamic congestion, the path turning index measures operational complexity, and the percentage of sloped road sections correlates with energy consumption and stability, ultimately generating a comprehensive path evaluation coefficient. By selecting the path with the highest evaluation coefficient, the optimal path decision is achieved, balancing dynamic congestion, operational efficiency, and safety.

[0114] Compared to existing technologies, traditional path planning methods typically only consider static path length or fixed obstacle distribution, failing to integrate dynamic parameters such as obstacle density, robot real-time speed, and queuing status, leading to delayed path congestion assessment. Existing technologies often use independent threshold judgments for handling turn counts and slope, lacking a collaborative optimization mechanism with path busyness, which can easily cause localized congestion or excessive energy consumption. This solution constructs a path busyness model by dynamically monitoring obstacle density, robot speed, and queuing status, and combines this with standardized processing of turn counts and slope to achieve a comprehensive evaluation of multi-dimensional parameters, effectively improving the real-time performance and accuracy of path selection.

[0115] Through the above technical solution, this application solves the problem of path congestion assessment bias caused by ignoring dynamic obstacle density, robot real-time speed, and queuing status in path planning, and overcomes the limitations of traditional methods that rely on a single threshold for judging the number of turns and slope sections. By dynamically fusing the real-time status of the path with static structural features, it achieves accurate assessment of path smoothness, optimizes robot transfer path selection, and reduces the risk of transportation delays and increased energy consumption caused by path congestion, frequent turns, or slope sections.

[0116] Preferably, the transfer-workbench-path matching degree model is expressed as:

[0117]

[0118] in, Indicates the matching degree of the transfer-workbench-path. This represents the order rating coefficient. This represents the maximum value of the transfer evaluation coefficient among multiple groups of transfer robots. This represents the minimum busy level among multiple worktable groups. This represents the maximum path evaluation coefficient among multiple path groups. Represents the weight coefficient and The The larger the value, the better the system match.

[0119] The order evaluation coefficient is a priority indicator calculated by comprehensively considering order deadline, item weight, and value. It can be implemented using an exponential weighted model. Its function is to quantify the timeliness, transportation cost, and economic value of orders, providing a benchmark weight for subsequent resource allocation. The maximum transfer evaluation coefficient is the evaluation coefficient corresponding to the robot with the highest transfer efficiency among multiple robots. It is calculated in real-time based on the robot's load, distance, and proportion of urgent orders. Its function is to select the robot with the optimal overall transfer capacity under the current condition, avoiding overall delays caused by inefficient transfers. The minimum workbench busyness is the busyness indicator corresponding to the individual with the lowest queue pressure among multiple workbenches. It is dynamically calculated using queue length, number of interruptions, and proportion of urgent orders. Its function is to prioritize allocating orders to workbenches with less processing capacity, alleviating local congestion. The maximum path evaluation coefficient is the evaluation indicator corresponding to the path with the highest smoothness among multiple optional paths. It is calculated by comprehensively considering path busyness, number of turns, and slope ratio. Its function is to select the path with the shortest transportation time and lowest energy consumption, improving overall transfer efficiency.

[0120] Specifically, by using order evaluation coefficients as global weighting factors, high-priority orders are prioritized for matching with the optimal resource combination. In robot selection, the robot with the highest transfer evaluation coefficient is chosen to ensure its load capacity, transfer distance, and emergency order processing capabilities are at their highest efficiency. In workbench allocation, workbenches with the lowest busyness are selected to reduce processing delays caused by queue backlogs. In path planning, the path with the highest rating coefficient is selected to reduce time losses caused by turns, slopes, and congestion. The contribution ratios of robots, workbenches, and paths are dynamically adjusted through weighting coefficients. For example, during peak order periods, the weighting coefficient of workbench busyness can be increased to alleviate congestion, and in complex path scenarios, the weighting coefficient of path rating can be increased to optimize traffic efficiency. Finally, a product relationship is established to link order priority with resource matching, forming a closed-loop optimization mechanism that achieves dynamic collaboration among the three.

[0121] Compared to existing technologies, traditional methods typically optimize robot scheduling, workbench allocation, and path planning independently. For example, they may allocate tasks based solely on the robot's current position without considering the real-time queue status of the target workbench, or ignore the impact of the robot's dynamic load on turning speed when planning paths. This solution integrates all three into a unified evaluation system through a multi-dimensional matching degree model, dynamically calculating the optimal combination using real-time data, thus resolving the local resource conflict problem caused by the disconnect between resource scheduling and path planning.

[0122] Through the above technical solution, this application can automatically match the optimal combination of robot, workbench, and path according to order priority, reducing transfer delays and path congestion caused by resource mismatch. For example, during peak order periods, high-value orders that are close to their deadlines can be prioritized for allocation to the robot with the highest transfer efficiency and guided to the workbench with the shortest queue and the path with the fewest turns, thereby achieving a balance between resource utilization and timeliness at the global level.

[0123] Preferably, the speed optimization model is expressed as:

[0124]

[0125] in, Indicates the average speed of the target. Indicates the matching degree of the transfer-workbench-path. Indicates the maximum permissible average speed. Indicates the minimum permissible average speed. Represents the speed smoothing coefficient. Indicates the current average speed. This represents the baseline average speed.

[0126] Among them, the transfer-workbench-path matching degree refers to the comprehensive collaborative evaluation index of order priority, robot transfer efficiency, workbench busyness and path smoothness. Specifically, it can be calculated through a multi-dimensional model. Its function is to quantify the global resource status into matching degree parameters and provide a dynamic basis for speed adjustment.

[0127] The speed smoothing coefficient is a parameter that controls the rate of speed change. It can be implemented by using the adjustment factor in the exponential function. Its function is to avoid robot energy consumption or path conflicts caused by sudden speed changes.

[0128] The reference average speed refers to the system's preset steady-state operating speed, which can be determined through historical operating data or experimental calibration. Its function is to provide a reference benchmark for speed adjustment and maintain system stability.

[0129] Specifically, this technical solution dynamically adjusts robot speed by combining global resource matching with real-time status. First, the model uses a minimum permissible speed as a benchmark to ensure the speed does not fall below a safety threshold. Second, by using matching parameters, the collaborative results of order priority, robot transfer efficiency, workbench busyness, and path unobstructedness are mapped to the speed adjustment range, resulting in a significant speed increase at high matching degrees, prioritizing the timeliness of urgent orders. Furthermore, speed changes are smoothly adjusted using an exponential function, where the speed smoothing coefficient controls the adjustment rate. The deviation between the current speed and the benchmark value reflects the real-time operating status, and the target speed is dynamically corrected based on the matching degree, responding to real-time demands while avoiding sudden speed changes.

[0130] Compared to existing technologies, traditional methods employ fixed speeds or adjust speeds solely based on path status, failing to consider the dynamic correlation between order urgency and global resource matching. For example, existing technologies rely solely on path congestion for robot speed adjustment, while this solution introduces a matching parameter to comprehensively incorporate order priority, workbench busyness, and path status into speed decisions, achieving multi-dimensional collaborative optimization. Furthermore, existing technologies lack a smooth transition mechanism for speed adjustment, while this solution uses an exponential function to achieve gradual speed changes, reducing energy consumption and conflict risks.

[0131] Through the above technical solution, this application solves the problem of order delays caused by static robot speed control. By dynamically matching global resource status with real-time operating parameters, higher speeds are prioritized for high-priority orders, while avoiding the impact of sudden speed changes on system stability. For example, when the matching degree is high, the target speed is close to the maximum allowable value, shortening the transfer time of urgent orders; when the matching degree is low, the speed is maintained at the baseline level, balancing efficiency and safety requirements. Thus, while ensuring the timeliness of high-priority orders, the overall stability and energy consumption controllability of the system are maintained.

[0132] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0133] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A collaborative optimization method for warehouse sorting paths based on dynamic scheduling of cluster resources, characterized in that, Includes the following steps: An order evaluation model is constructed based on the order deadline, order weight, and order value, and the output order evaluation coefficient is then used. A transfer evaluation model is constructed based on the total transfer distance of the robot, the current load of the robot, and the proportion of urgent orders in the robot queue, and the transfer evaluation coefficient is output. A workbench busyness model is constructed based on the workbench queue length, the proportion of urgent orders in the workbench queue, and the proportion of times orders are cut off, and the workbench busyness is output. A path evaluation model is constructed based on path busyness, number of turns, and percentage of gradient sections, and the path evaluation coefficient is output. Based on the order evaluation coefficient, the transfer evaluation coefficient, the workbench busyness and the path evaluation coefficient, a transfer-workbench-path matching degree model is constructed and the transfer-workbench-path matching degree is output. Based on the matching degree between transfer, workbench, and path, and the current average speed of the robot for the transferred order items, a speed optimization model is constructed to output the target average speed.

2. The warehouse sorting path collaborative optimization method based on dynamic scheduling of cluster resources according to claim 1, characterized in that, The speed optimization model is expressed as follows: ； in, Indicates the average speed of the target. Indicates the matching degree of the transfer-workbench-path. Indicates the maximum permissible average speed. Indicates the minimum permissible average speed. Represents the speed smoothing coefficient. Indicates the current average speed. This represents the baseline average speed.

3. The warehouse sorting path collaborative optimization method based on dynamic scheduling of cluster resources according to claim 1, characterized in that, The transfer-workbench-path matching degree model is represented as follows: ； in, Indicates the matching degree of the transfer-workbench-path. This represents the order rating coefficient. This represents the maximum value of the transfer evaluation coefficient among multiple groups of transfer robots. This represents the minimum busy level among multiple worktable groups. This represents the maximum path evaluation coefficient among multiple path groups. Represents the weight coefficient and The The larger the value, the better the system match.

4. The warehouse sorting path collaborative optimization method based on dynamic scheduling of cluster resources according to claim 1 or 3, characterized in that, The steps to build an order evaluation model and output the order evaluation coefficient based on the order deadline, order weight, and order value are as follows: The order deadline index is obtained by taking the ratio of the minimum allowed deadline to the total deadline. The weight index and value index are obtained by comparing the weight and value of the ordered items with their maximum allowable values, respectively. An order evaluation model is constructed based on the order deadline index, weight index, and value index. The order evaluation model is expressed as follows: ； in, This represents the order rating coefficient. Indicates the time decay coefficient. This represents the order deadline index. Indicates the weight index. Indicates the value index. Represents the weight coefficient and The And the larger the value, the higher the priority; Import the current order deadline index, current weight index, and current value index into the order evaluation model to output the current order evaluation coefficient.

5. The warehouse sorting path collaborative optimization method based on dynamic scheduling of cluster resources according to claim 1 or 3, characterized in that, The steps to construct a transfer evaluation model and output the transfer evaluation coefficient based on the total transfer distance of the robot, the current load of the robot, and the proportion of urgent orders in the robot queue are as follows: The transfer distance index is obtained by comparing the total transfer distance of the robot with the maximum allowable total transfer distance. The load index is obtained by comparing the robot's current load with the weight of the ordered items and then with the maximum allowable load. A transfer evaluation coefficient is constructed based on the transfer distance index, load index, and the proportion of urgent orders in the robot queue. The transfer evaluation model is expressed as follows: ； in, This represents the transshipment evaluation coefficient. Indicates the transit distance index. Indicates the load index. This indicates the proportion of urgent orders in the robot queue. Represents the weight coefficient and The The larger the value, the higher the transfer efficiency; Output the transfer evaluation coefficients of multiple robot groups from the current transfer distance index, current load index, and the proportion of urgent orders in the current robot queue. Take the robot with the maximum value among the multiple transfer evaluation coefficients as the order transfer robot.

6. The warehouse sorting path collaborative optimization method based on dynamic scheduling of cluster resources according to claim 1 or 3, characterized in that, The steps to construct a workbench busyness model based on the workbench queue length, the proportion of urgent orders in the workbench queue, and the percentage of times orders were interrupted are as follows: The queue length index is obtained by taking the ratio of the workbench queue length to the maximum allowed queue length. The ratio of the percentage of cut-in attempts to the percentage of the maximum allowed cut-in attempts is used to obtain the cut-in attempt percentage index; A workbench busyness model is constructed based on the queue length index, the percentage of times someone cuts in line index, and the proportion of urgent orders in the workbench queue. This workbench busyness model is expressed as follows: ； in, Indicates the workload of the workbench. Indicates the queue length index. This indicates the proportion of urgent orders in the workbench queue. This indicates the percentage of times someone cuts in line. Represents the weight coefficient and The The higher the value, the busier the workbench; The queue length index, the percentage of times someone cuts in line index, and the proportion of urgent orders in the workbench queue are used to output the workbench busyness of multiple workbench groups. The workbench with the minimum workbench busyness among the multiple workbench groups is taken as the order processing workbench.

7. The warehouse sorting path collaborative optimization method based on dynamic scheduling of cluster resources according to claim 1 or 3, characterized in that, The steps to construct a path evaluation model based on path busyness, number of turns, and percentage of gradient sections, and output path evaluation coefficients are as follows: The business index and the number of turns are obtained by comparing the route busyness and the number of turns with their maximum allowable values, respectively. A route evaluation model is constructed based on the traffic volume index, route turning index, and percentage of sloped road sections. The route evaluation model is expressed as follows: ； in, Represents the path evaluation coefficient. Indicates the busyness index. Indicates the path turning index. Indicates the percentage of road sections with slope. Represents the weight coefficient and The The larger the value, the smoother the path; The busyness index, turning index, and percentage of slope sections from multiple routes are imported into the route evaluation model to obtain multiple route evaluation coefficients. The route with the maximum value of the multiple route evaluation coefficients is taken as the order item transfer route.

8. The warehouse sorting path collaborative optimization method based on dynamic scheduling of cluster resources according to claim 7, characterized in that, The method for obtaining the path busyness is as follows: A path busyness model is constructed based on the current obstacle density, current average robot speed, current robot queue length, and robot time occupancy on the path, outputting a path busyness index. The path busyness model is expressed as follows: ； in, Indicates the busyness of the path. Indicates the current obstacle density. Current average robot speed, Indicates the maximum allowed speed for the path. This indicates the current queue length of the robots. Indicates the maximum allowed queue length. This indicates the current time occupancy of the robot on the path. The larger the value, the more congested the path.

Images