An agv forklift cooperative scheduling method and system for warehouse management

Abstract

This invention discloses an AGV forklift collaborative scheduling method and system for warehouse management, specifically relating to the field of collaborative scheduling planning and management. It addresses the problems of deviations in the actual release position of AGV forklifts after pallet placement, decreased subsequent retrieval success rates, and lack of targeted correction of retrieval paths during existing warehouse operations. By collecting time-history curves of coasting deceleration under different load levels and different entry conditions, a standard deceleration envelope set is established to generate a coasting deviation vector. A handover deviation mapping model is constructed to predict the expected decoupling release area corresponding to the pallet. Combined with historical handover records, the estimated retrieval success rate is determined. Based on this, reassignment of retrieval vehicles, end-point alignment path correction, and collaborative scheduling to resolve multi-AGV path conflicts are implemented. This improves pallet handover stability, retrieval accuracy, and the collaborative operation efficiency of multiple vehicles within the warehouse area, enhancing the automated management capabilities and task execution stability of the warehouse logistics system.

Description

Technical Field

[0001] This invention relates to the field of collaborative scheduling planning and management technology, and more specifically, to an AGV forklift collaborative scheduling method and system for warehouse management. Background Technology

[0002] With the widespread application of intelligent warehousing systems, AGV forklifts have become crucial equipment in warehousing and logistics, undertaking pallet handling, temporary storage, transfer, and automated handover operations. In scenarios such as large-scale automated warehouses, manufacturing enterprise logistics centers, e-commerce sorting centers, and automated distribution warehouses, a large number of AGV forklifts need to continuously complete pallet placement and retrieval tasks, and rely on temporary storage areas to connect different handling tasks.

[0003] In actual operation, AGV forklifts carrying goods of different weights are easily affected by factors such as load changes, differences in ground conditions, tire wear, drive system response characteristics, and deviations in driving posture during the deceleration and coasting phase before entering the temporary storage area. This can lead to a certain deviation in the final placement position of the pallet corresponding to the same placement instruction. This deviation not only affects the fork alignment accuracy of subsequent AGV forklifts picking up parts, but may also cause problems such as failed picking, repeated alignment, task delays, and even congestion in the temporary storage area.

[0004] Furthermore, when a large number of placement and retrieval tasks occur simultaneously within a storage area, anomalies in the handover of a single temporary storage area can further cause multiple AGV forklifts to wait, detour, or experience path conflicts, reducing overall warehousing efficiency. Existing AGV scheduling solutions typically execute placement and retrieval tasks based on preset pallet coordinates, lacking the ability to predict the actual release location of pallets. This makes it difficult to assess subsequent retrieval risks before placement is completed, and also makes it difficult to dynamically adjust retrieval vehicles, correct retrieval paths, and conduct multi-vehicle collaborative scheduling based on the handover status. Therefore, in high-density, multi-task concurrent warehousing operation scenarios, problems such as insufficient handover stability, decreased scheduling efficiency, and limited equipment collaboration capabilities still exist. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide an AGV forklift collaborative scheduling method and system for warehouse management to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A collaborative scheduling method for AGV forklifts for warehouse management includes the following steps: S1. When the AGV forklift carrying standard pallets of different load levels enters the temporary storage area, the pre-set deceleration section is collected, the sliding deceleration time history curve is collected and the upper and lower envelopes are extracted. The standard deceleration envelope set corresponding to the load level is established according to the different driving condition labels. S2. When the AGV forklift is carrying goods and performing a placement task and enters the same deceleration section, the standard deceleration envelope of the corresponding load level is retrieved, and the current sliding deceleration curve is compared with the standard deceleration envelope in real time to generate a sliding deviation vector. S3. Using the sliding deviation vector as input, construct the handover deviation mapping model, output the expected detent release area of ​​the pallet relative to the handover reference position after this placement, and combine it with historical handover records to determine the corresponding estimated success rate of picking up the item. S4. When the estimated success rate of pickup is lower than the collaborative scheduling threshold, a temporary storage area degradation signal is sent to the pickup module, and pickup vehicles are reassigned based on the current on-site distribution of standby pickup AGV forklifts. S5. Using the displacement vector between the expected geometric center of the derailment release area and the handover reference position as the expected offset vector, the AGV forklift for picking up parts uses the expected offset vector as a feedforward parameter to correct the end alignment path parameter set before entering the temporary storage area. S6. Under the modified end-point alignment path constraint, based on the retrieval tasks corresponding to all temporary storage areas within the storage area, perform path conflict resolution and collaborative scheduling for all retrieval AGV forklifts.

[0007] As a further aspect of the present invention, in step S1, establishing a standard deceleration envelope set corresponding to the load level specifically includes: When an AGV forklift is carrying a standard pallet with a known load level and enters the same temporary storage area before a preset deceleration section, the angle between the actual heading angle of the AGV forklift and the corresponding planned heading angle is obtained, and the driving condition label is divided according to the range of the angle. Collect the coasting deceleration time history curve under the current load level, and align it with the standard deceleration envelope of the unloaded individual according to the deceleration start time. Calculate the deceleration difference between the current coasting deceleration time history curve and the standard deceleration envelope of the unloaded individual for each sampling point, and generate the deceleration correction sequence for the corresponding load level. For all historical taxiing deceleration time history curves under the same entry condition label and the same load level, the deceleration correction sequence calculation is repeatedly performed. The mean of each deceleration correction sequence is statistically analyzed at the corresponding sampling time to generate the load correction sequence corresponding to the current load level. The load correction sequence is superimposed on the standard deceleration envelope of the unloaded individual at each sampling point to generate a standard deceleration envelope subset corresponding to the load level. The standard deceleration envelope subsets corresponding to different load levels of different driving condition labels are combined to form a standard deceleration envelope set.

[0008] As a further aspect of the present invention, in step S2, generating the skid deviation vector specifically includes: Based on the load level, the current coasting deceleration time history curve is compared with the upper and lower envelopes of the corresponding standard deceleration envelope point by point. The positive deviation sampling point sequence that exceeds the upper envelope and the negative deviation sampling point sequence that is lower than the lower envelope are recorded. The deviation amplitudes corresponding to the positive deviation sampling point sequence and the negative deviation sampling point sequence are accumulated respectively, and the sum of the two results is used as the cumulative deviation. The current taxiing deceleration time history curve is dynamically time-normalized with the center line of the standard deceleration envelope. The normalized cumulative matching distance after normalization is calculated as the waveform difference. The difference between the deceleration peak time in the current taxiing deceleration time history curve and the deceleration peak time of the center line of the standard deceleration envelope is located as the peak hysteresis. The cumulative deviation, waveform difference, and peak hysteresis are combined to form the taxiing deviation vector.

[0009] As a further aspect of the present invention, in step S3, constructing the handover deviation mapping model specifically includes: Obtain the sliding deviation vector and the corresponding actual pallet offset vector based on the handover reference position for each placement operation in the historical handover record of the temporary storage area. Map the actual pallet offset vector to the three-dimensional coordinate system of the temporary storage area to form an offset point set. Statistically count the spatial clustering range of the offset point set corresponding to the same type of sliding deviation vector, and define the spatial clustering range as the derailment release area. Establish the correspondence between the sliding deviation vector and the derailment release area to form a handover deviation mapping model based on the entry condition label. The aforementioned similar taxiing deviation vectors are taxiing deviation vectors obtained by constructing a three-dimensional deviation vector based on the cumulative deviation, waveform difference, and peak hysteresis under the same entry condition label, and by jointly filtering the three-dimensional deviation vectors based on the included angle between the three-dimensional deviation vectors and the vector magnitude range.

[0010] As a further aspect of the present invention, the estimated success rate of item retrieval in step S3 specifically includes: Obtain the actual pallet offset vector and handover result label for each handover in all historical handover records in the temporary storage area, and map the actual pallet offset vector to the coordinate system of the temporary storage area to form a set of historical offset points. The ratio of the number of successful handover events corresponding to historical offset points within the coverage area of ​​each derailment release zone to the total number of handover events is calculated and used as the success rate benchmark value for the corresponding derailment release zone. A correspondence between the derailment release zone and the success rate benchmark value is established. When the current sliding deviation vector is input, the handover deviation mapping model outputs the corresponding expected derailment release zone. The success rate benchmark value corresponding to the expected derailment release zone is read as the estimated success rate of part retrieval.

[0011] As a further aspect of the present invention, in step S4, reassigning the pickup vehicle specifically includes: The estimated success rate of pickup is compared with the collaborative scheduling threshold. When the estimated success rate of pickup is lower than the collaborative scheduling threshold, a temporary storage area downgrade signal carrying the original temporary storage area identifier is sent to the pickup module. After receiving the temporary storage area degradation signal, the pickup module iterates through all the pickup AGV forklifts that are currently in an idle standby state and selects the vehicle with the shortest path distance between its location and the temporary storage area as the reassigned pickup vehicle.

[0012] As a further aspect of the present invention, in step S5, the correction of the lateral pre-offset and longitudinal pre-stop point of the end alignment path specifically includes: After receiving the expected offset vector, the AGV forklift decomposes the expected offset vector into a longitudinal offset component along the longitudinal direction of the temporary storage area, a lateral offset component along the transverse direction of the temporary storage area, and a placement deflection angle component. The lateral pre-offset amount is obtained by compensating the lateral reference coordinates when the forks enter the temporary storage area in the original end alignment path based on the displacement in the opposite direction of the lateral offset component. The longitudinal reference coordinates at the point where the forks stop extending in the original end alignment path are compensated based on the displacement in the opposite direction of the longitudinal offset component to obtain the longitudinal pre-stop point. The fork alignment direction correction angle is calculated based on the placement deflection angle component, and the target alignment heading angle in the end alignment path is corrected using the fork alignment direction correction angle. Write the lateral pre-offset, longitudinal pre-stop point, and corrected target alignment heading angle into the end alignment path parameter set of the AGV forklift, replacing the corresponding reference coordinates and reference heading angle in the original end alignment path.

[0013] As a further aspect of the present invention, S6, specifically including the collaborative scheduling for resolving path conflicts for all AGV forklifts, includes: Obtain the corrected end-point alignment path of all pickup AGV forklifts in the warehouse area and the temporary storage area identifier corresponding to the current task, and predict the driving trajectory of each pickup AGV forklift in the future scheduling window according to the time sequence. Traverse the path intersection areas in the predicted trajectories of each pickup AGV forklift. When two or more pickup AGV forklifts enter the same path intersection area within the same time window, mark the corresponding path intersection area as a conflict area. Based on the task waiting time, the passage time of each pickup AGV forklift in the conflict area is re-arranged, and the re-arranged passage time is sent to the corresponding pickup AGV forklift for execution.

[0014] On the other hand, the present invention provides an AGV forklift collaborative scheduling system for warehouse management, comprising: The envelope construction module is used to collect the time history curves of the sliding deceleration corresponding to the standard pallets of different load levels, and combine them with the driving condition label to establish the standard deceleration envelope set of the corresponding load level. The deviation analysis module is used to compare and analyze the current coasting deceleration curve with the standard deceleration envelope, extract the cumulative deviation, waveform difference and peak hysteresis, and generate a coasting deviation vector. The handover assessment module is used to construct a handover deviation mapping model based on the sliding deviation vector, output the expected decoupling release area and determine the corresponding estimated success rate of pickup. The pickup module is used to perform temporary storage area downgrading and pickup vehicle reassignment based on the pickup success rate estimate, and to correct the end alignment path parameter set based on the expected offset vector. The collaborative scheduling module is used to predict the driving trajectory of all pickup AGV forklifts, identify path conflict areas and rearrange the passage times to perform collaborative scheduling of multiple AGV forklifts.

[0015] The technical effects and advantages of the AGV forklift collaborative scheduling method and system for warehouse management disclosed in this invention are as follows: This invention establishes standard deceleration envelope sets corresponding to different load levels and different entry conditions, enabling a refined characterization of the AGV forklift's coasting deceleration process. Based on the difference between the current coasting deceleration curve and the standard deceleration envelope, a coasting deviation vector is constructed, improving the accuracy of placement process status identification. Furthermore, this invention utilizes historical handover records to construct a handover deviation mapping model, mapping the coasting deviation vector to the corresponding expected decoupling release area. This allows for advance prediction of the actual release state after pallet placement and, combined with historical handover success data, forms a retrieval success rate prediction mechanism, improving handover risk assessment capabilities. When a high handover risk is detected, this invention can promptly downgrade the temporary storage area and reassign the retrieval vehicle, reducing the impact of abnormal handovers on subsequent operations. Simultaneously, it corrects the end-positioning path of the retrieval AGV forklift based on the expected offset vector, improving the alignment accuracy between the forks and the pallet.

[0016] This invention combines the task status and driving trajectory execution path conflict resolution of all AGV forklifts for picking up parts with collaborative scheduling, reducing vehicle waiting and path conflict phenomena, improving the collaborative operation efficiency of multiple AGV forklifts in the warehouse area, thereby improving the overall operational stability and automation scheduling level of the warehousing and logistics system. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of an AGV forklift collaborative scheduling method for warehouse management according to the present invention; Figure 2 This is a schematic diagram of the structure of an AGV forklift collaborative scheduling system for warehouse management according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 Figure 1 This invention provides a method and system for collaborative scheduling of AGV forklifts for warehouse management, which includes the following steps: S1. When the AGV forklift carrying standard pallets of different load levels enters the temporary storage area, the pre-set deceleration section is collected, the sliding deceleration time history curve is collected and the upper and lower envelopes are extracted. The standard deceleration envelope set corresponding to the load level is established according to the different driving condition labels. S2. When the AGV forklift is carrying goods and performing a placement task and enters the same deceleration section, the standard deceleration envelope of the corresponding load level is retrieved, and the current sliding deceleration curve is compared with the standard deceleration envelope in real time to generate a sliding deviation vector. S3. Using the sliding deviation vector as input, construct the handover deviation mapping model, output the expected detent release area of ​​the pallet relative to the handover reference position after this placement, and combine it with historical handover records to determine the corresponding estimated success rate of picking up the item. S4. When the estimated success rate of pickup is lower than the collaborative scheduling threshold, a temporary storage area degradation signal is sent to the pickup module, and pickup vehicles are reassigned based on the current on-site distribution of standby pickup AGV forklifts. S5. Using the displacement vector between the expected geometric center of the derailment release area and the handover reference position as the expected offset vector, the AGV forklift for picking up parts uses the expected offset vector as a feedforward parameter to correct the end alignment path parameter set before entering the temporary storage area. S6. Under the modified end-point alignment path constraint, based on the retrieval tasks corresponding to all temporary storage areas within the storage area, perform path conflict resolution and collaborative scheduling for all retrieval AGV forklifts.

[0020] In S1, a standard deceleration envelope set corresponding to the load level is established.

[0021] Historical placement tasks entering the same temporary storage area and then proceeding through a pre-set deceleration section are screened. Standard pallets of uniform specifications, dimensions, and known load ratings are used as sample carriers during screening to avoid additional disturbances introduced by differences in cargo size, center of gravity shifts, or packaging variations. When the AGV forklift travels along the predetermined placement path to the entrance of the pre-set deceleration section, the actual heading angle recorded by the vehicle navigation controller is read, along with the planned heading angle at the corresponding position in the path planning task, and the angle between the two is calculated. To avoid misjudgments of the working condition due to momentary sensor jitter, heading angle data is continuously collected at multiple times before the deceleration section entrance and averaged. The angle between the average heading angle and the planned heading angle is used as the basis for determining the working condition. The angle range is divided according to the statistical results of historical operating data. For example, 0° to 2° is divided into the first entry working condition label, 2° to 5° into the second entry working condition label, and 5° to 8° into the third entry working condition label. The division intervals are obtained based on three consecutive months of operational data from the warehouse site, ensuring a sufficient number of sample records under each working condition label. After marking the entry condition, the AGV forklift's coasting deceleration time history curves were collected after entering the deceleration phase. The deceleration data was obtained from the vehicle's inertial measurement unit and drive controller feedback data, with a fixed sampling frequency of 100Hz. Recording began at the moment the deceleration command was issued and continued until the vehicle stopped and completed its placement. Since the deceleration duration varies for different placement tasks, the moment when the deceleration first continuously decreases for more than three sampling cycles was uniformly used as the deceleration start moment. The coasting deceleration time history curves collected in each instance were aligned with the standard deceleration envelope of the unloaded individual vehicle according to this moment on the time axis.

[0022] For the coasting deceleration time history curve under the current load level, after alignment, the standard deceleration envelope values ​​of the unloaded individuals are read point by point according to the sampling point sequence, and the deceleration difference between the two is calculated. When the measured deceleration at a certain moment is 0.62 m / s², and the corresponding standard deceleration envelope value for the unloaded state is 0.48 m / s², the deceleration difference at that moment is recorded as 0.14 m / s²; when the measured deceleration is 0.41 m / s², and the corresponding standard deceleration envelope value for the unloaded state is 0.52 m / s², the deceleration difference is recorded as -0.11 m / s². After traversing all sampling points in the same way, a complete deceleration correction sequence is formed. The deceleration correction sequence reflects the deceleration change law at each moment relative to the unloaded state after the load is introduced, rather than simply statistically averaging the deceleration, thus preserving the dynamic change characteristics throughout the deceleration process. For historical tasks of the same load level and belonging to the same entry condition label, the above deceleration correction sequence extraction process is repeated. Subsequently, a correspondence was established according to the sampling time. For example, the 50th sampling point corresponds to the 50th sampling point in all historical samples, and the 100th sampling point corresponds to the 100th sampling point in all historical samples. The average value of the deceleration correction value at the corresponding time was then calculated.

[0023] After obtaining the load correction sequence, a sample-by-sample correction is performed on the standard deceleration envelope of the unloaded individual. Specifically, the upper and lower envelopes of the unloaded standard deceleration envelope are read, and the correction values ​​from the load correction sequence are superimposed onto the corresponding positions of the upper and lower envelopes at the corresponding sampling times. For example, if the deceleration corresponding to the upper envelope at a certain sampling time is 0.56 m / s², and the load correction value is 0.08 m / s², then the corrected value of the upper envelope is 0.64 m / s²; if the deceleration corresponding to the lower envelope at the same time is 0.42 m / s², then the corrected value of the lower envelope is 0.50 m / s². This process is repeated for all sampling times to form a subset of the standard deceleration envelope corresponding to the current load level. Subsequently, the above process is repeated for all load levels and all driving condition labels to establish a standard deceleration envelope set containing load level indices and driving condition label indices. In actual storage, a two-level index structure is adopted. The first level index corresponds to the load level, and the second level index corresponds to the entry condition label. Each set of indexes uniquely corresponds to a subset of standard deceleration envelopes. When the actual placement task is executed later, the corresponding standard deceleration envelope is directly retrieved according to the load level and entry condition label of the current task, thereby ensuring that subsequent taxiing deviation analysis is always performed under the same load conditions and the same entry attitude conditions.

[0024] In step S2, a sliding deviation vector is generated.

[0025] Based on the load level and entry condition label corresponding to the current task, the corresponding standard deceleration envelope is retrieved from a pre-established set of standard deceleration envelopes. Since the standard deceleration envelopes have been categorized according to load level and entry condition, the current taxiing deceleration time history curve and the standard deceleration envelope share a unified time reference and operating condition. In practice, a synchronous sampling point traversal is performed on the current taxiing deceleration time history curve and the standard deceleration envelope, reading the deceleration value, upper envelope value, and lower envelope value corresponding to the current moment point by point. When the current deceleration value is between the upper and lower envelopes, the sampling point is considered to be in the normal coasting range, and no deviation data is recorded. When the current deceleration value exceeds the upper envelope, the difference between the current deceleration value and the upper envelope is calculated, and the corresponding sampling point number and deviation amplitude are recorded to form a positive deviation sampling point record. When the current deceleration value is below the lower envelope, the absolute value of the difference between the current deceleration value and the lower envelope is calculated, and the corresponding sampling point number and deviation amplitude are recorded to form a negative deviation sampling point record. After traversal, positive and negative deviation sampling point sequences are obtained respectively. To avoid the amplified effect of a single abnormal sampling point on the results, a continuity screening is performed on the deviation sequence before accumulating the deviation amplitude. Only deviation segments with three or more consecutive sampling points are retained, and isolated single-point or double-point deviation records are directly discarded. After screening, all deviation amplitudes in the positive deviation sequence and all deviation amplitudes in the negative deviation sequence are accumulated respectively. After all segments are statistically analyzed, the cumulative positive and negative deviations are obtained. Then, the two are summed directly to obtain the cumulative deviation.

[0026] Using the current coasting deceleration time history curve as the sequence to be matched and the standard deceleration envelope centerline as the reference sequence, a matching matrix consisting of a time axis and sampling point indices is constructed. Each cell in the matrix records the distance cost between the current deceleration value and the reference deceleration value, and the matching path with the minimum total cost is searched step by step according to the dynamic time warping rule. Since the deceleration duration varies for different placement tasks, dynamic time warping allows multiple sampling points to correspond to a single reference point, and also allows a single sampling point to correspond to multiple reference points, thereby eliminating the influence of different deceleration durations. After obtaining the optimal matching path, all matching distances on the path are accumulated and divided by the matching path length to obtain the normalized cumulative matching distance. This normalized cumulative matching distance is defined as the waveform difference degree. To ensure the comparability of results between different tasks, the maximum absolute value normalization method is uniformly used to scale the two curves before calculating the normalized cumulative matching distance. The normalization range is determined based on historical sample statistics, for example, selecting the deceleration range with the highest frequency in all historical deceleration curves as the normalization interval.

[0027] A five-point moving average is applied to the deceleration curve to eliminate the impact of high-frequency jitter on peak identification. Then, the sampling point with the largest absolute value of deceleration is searched, and the corresponding time is recorded as the peak deceleration time of the current curve. The peak deceleration time corresponding to the center line of the standard deceleration envelope is read in the same way. After both are located, the time difference between the peak time of the current curve and the peak time of the standard curve is calculated. For example, if the peak time of the current curve occurs 1.42 seconds after the start of deceleration, and the peak time of the standard curve occurs 1.26 seconds after the start of deceleration, then the peak lag is 0.16 seconds. If the peak time of the current curve occurs before the standard curve, it is recorded as a negative lag; if it occurs after the standard curve, it is recorded as a positive lag. After calculating the peak lag, the cumulative deviation, waveform difference, and peak lag are combined in a fixed order to form a three-dimensional coasting deviation vector. The first dimension corresponds to the cumulative deviation, the second dimension corresponds to the waveform difference, and the third dimension corresponds to the peak lag.

[0028] In S3, a handover deviation mapping model is constructed.

[0029] All valid placement samples are extracted from the historical handover records of the temporary storage area. A valid placement sample refers to a data record that fully documents the coasting deceleration time history curve, load level, entry condition label, and the actual position of the pallet after placement. For each historical sample, its corresponding coasting deviation vector is read, along with the actual offset of the pallet relative to the handover reference position after placement. The actual offset includes not only the displacement of the pallet's center point in the longitudinal and lateral directions within the temporary storage area, but also the deflection angle of the pallet relative to the standard placement posture. To standardize the data representation across different temporary storage areas, a fixed reference coordinate system is established for each area, with the handover reference position as the origin, the fork travel direction defined as the longitudinal axis, the direction perpendicular to the fork travel direction defined as the lateral axis, and the pallet's rotation angle around the vertical direction as the third-dimensional posture coordinate. After coordinate mapping, each historical sample is converted into a three-dimensional spatial point containing longitudinal offset, lateral offset, and deflection angle values. As historical samples accumulate, a set of historical offset points for the corresponding temporary storage area is formed. Since the release patterns of the pallet differ under different entry condition labels, offset point sets are established according to the entry condition labels to avoid mixing of samples under different entry postures.

[0030] After organizing historical samples, taxiing deviation vectors of the same type are filtered. Specifically, pre-classification is performed according to the entry condition label, and comparisons are only made within the dataset corresponding to the same entry condition label. Then, the cumulative deviation, waveform difference, and peak lag for each historical record are used to construct a three-dimensional deviation vector in a fixed order. For the taxiing deviation vector to be analyzed, it is compared sequentially with all historical three-dimensional deviation vectors under the same entry condition label. During the comparison, two constraints are used: vector angle and vector magnitude interval. The vector angle reflects whether the compositional relationship between the three features is consistent. For example, if two three-dimensional deviation vectors have similar magnitudes, but one is mainly composed of cumulative deviation and the other mainly of peak lag, the corresponding angle will be significantly larger, and such samples will not be included in the same sample set. The vector magnitude interval reflects whether the overall deviation degree is similar. The magnitude interval is determined using historical statistical methods, such as sorting all historical sample magnitudes by size and dividing them into multiple continuous intervals, each interval covering a fixed proportion of samples. When historical samples simultaneously meet both the included angle screening condition and the module length interval screening condition, they are classified as similar gliding deviation vectors. This dual-constraint approach ensures that the screening results maintain both a consistent overall degree of deviation and a consistent relationship among the three features, thus avoiding the problem of misclassification of different dynamic characteristics due to reliance solely on similar module lengths. After screening, a set of historical similar samples corresponding to the current gliding deviation vector is obtained.

[0031] After obtaining a set of similar samples, all historical offset points in the set are read, and spatial clustering analysis is performed in the three-dimensional coordinate system of the temporary storage area. Specifically, the actual distribution range of all historical offset points in three-dimensional space is statistically analyzed. First, the distribution boundaries of each offset point in the longitudinal, lateral, and deflection angle directions are calculated. Then, outliers exceeding three times the median absolute deviation from the overall distribution center are removed to eliminate the impact of extreme samples caused by accidental collisions, ground obstacles, and sensor anomalies. After outlier removal, a spatial envelope is constructed for the remaining offset points. The spatial envelope covers the distribution range of samples accounting for more than 95% of the remaining offset points and is defined as the decoupling release zone. Since the decoupling release zone is established in three-dimensional space, it reflects not only the possible location range of the pallet but also the possible attitude range of the pallet. Subsequently, the correlation between the current sliding deviation vector and the corresponding decoupling release zone is established, and this correlation is written into the handover deviation mapping model. When a new coasting deviation vector appears, the corresponding model branch is first determined based on the driving condition label, and then the corresponding sample set of the same type is found by the joint screening method of included angle and model length. Finally, the corresponding decoupling release area is output.

[0032] In S3, the estimated success rate of package pickup is given.

[0033] The historical handover records of the temporary storage area are compiled. Each historical handover record includes the actual pallet offset status after the placement task is completed and the subsequent retrieval result. The actual pallet offset status is recorded using a coordinate expression method consistent with the aforementioned handover deviation mapping model, including the longitudinal offset, lateral offset, and pallet attitude deflection angle of the pallet center point relative to the handover reference position. The handover result label is determined based on the actual execution result of the subsequent retrieval task. When the retrieval AGV forklift completes the entire process of fork insertion, pallet lifting, and leaving the temporary storage area in a single retrieval action without triggering a repositioning command, the corresponding handover record is marked as a successful handover event. When the retrieval process encounters issues such as the forks failing to enter the pallet fork holes, the pallet slipping after fork insertion, multiple repositioning actions failing to complete the retrieval, or the retrieval module reassigning the retrieval task, the corresponding record is marked as a failed handover event. After adopting the above judgment criteria, the handover result labels in the historical handover records all have a unified definition. Subsequently, the actual pallet offset status in all historical records is read and mapped to the unified coordinate system of the temporary storage area to form a set of historical offset points. To ensure data consistency across different time periods, the coordinate system of the temporary storage area is uniformly calibrated before establishing the historical offset point set. The calibration benchmark uses the fixed center position of the storage location and the standard pallet reference posture. If storage location adjustments or equipment relocations occur during the period when historical records are generated, the corresponding offset coordinates are recalculated according to the adjusted storage location parameters before being written into the offset point set. After all data conversions are completed, each historical record corresponds to a three-dimensional offset point containing longitudinal, lateral, and angular coordinates, along with a corresponding handover result label. As historical tasks accumulate, a historical offset point set covering the long-term operational status of the entire temporary storage area is formed.

[0034] The process involves reading all the decoupling release zones already established in the aforementioned handover deviation mapping model. Each decoupling release zone corresponds to a defined spatial coverage area. Then, all offset points in the historical offset point set are traversed to determine whether each offset point falls within the coverage area of ​​its corresponding decoupling release zone. During this determination, the longitudinal coordinate, lateral coordinate, and attitude angular coordinate are compared simultaneously. Only when all three coordinates are within the boundary of the corresponding decoupling release zone is the offset point considered to belong to the current decoupling release zone. After determining the attribution of all offset points, all historical handover events within the coverage area of ​​the current decoupling release zone are counted, and the number of successful handover events and the total number of handover events are counted. For example, if a certain decoupling release zone corresponds to a total of 218 historical samples, with 191 successful handover events, then the baseline success rate for that decoupling release zone is the result of 191 divided by 218. To avoid statistical distortion due to an insufficient sample size, a minimum sample size constraint is added during the statistical process. For example, it is stipulated that a decoupling release zone corresponds to no less than 30 historical events. If the statistical result is less than 30, it is automatically expanded to adjacent decoupling release zones for joint statistics; or release zones with connected boundaries and similar release states are merged according to spatial adjacency and the success rate benchmark value is recalculated. After a new placement task completes the calculation of the coasting deviation vector, the current coasting deviation vector is input into the handover deviation mapping model. The handover deviation mapping model first determines the corresponding mapping branch based on the driving condition label, and then obtains the corresponding expected decoupling release zone based on the coasting deviation vector matching. After obtaining the expected decoupling release zone, the success rate benchmark value of the decoupling release zone in the corresponding relationship table is read and used as the estimated retrieval success rate value for the current placement task.

[0035] In step S4, the pickup vehicle is reassigned.

[0036] Using all temporary storage area handover tasks within the last three consecutive months as the statistical object, the correlation between the estimated success rate of each handover task and the actual retrieval result is calculated. The estimated value corresponding to the inflection point where the actual retrieval success rate begins to show a significant decline is selected as the collaborative scheduling threshold. For example, during the statistical process at a certain warehouse site, when the estimated value falls below 0.82, the frequency of actual retrieval failure events begins to increase continuously, so 0.82 is determined as the collaborative scheduling threshold. After the threshold is determined, it is written into the scheduling configuration parameter table and updated again according to a fixed period. After calculating the estimated success rate of the current task, the collaborative scheduling threshold is read and compared. When the estimated retrieval success rate reaches or exceeds the collaborative scheduling threshold, the original handover process continues to be executed; when the estimated retrieval success rate is lower than the collaborative scheduling threshold, the pallet corresponding to the current temporary storage area is determined to be in a high-risk handover state. At this time, a temporary storage area degradation signal is generated. The temporary storage area degradation signal not only records the original temporary storage area identifier, but also records the current handover task number, pallet identifier, generation time, and corresponding estimated success rate. To avoid duplicate scheduling caused by multiple degradation events occurring consecutively in the same temporary storage area within a short period, the current status record of the temporary storage area is further queried before sending a degradation signal. If the temporary storage area is already in a degradation state, a new degradation signal is not generated again; instead, the current task information is appended to the existing degradation record. If the temporary storage area is in a normal state, its status flag is updated and a degradation signal is sent.

[0037] The system reads the real-time status information of all AGV forklifts within the current storage area, retaining only vehicles in the "idle and ready" state for screening. Vehicles in a task execution state, fault state, charging state, or locked by other scheduling tasks are directly excluded. After screening, a candidate vehicle set is obtained. Then, for each AGV forklift in the candidate set, the current position coordinates are read, along with the corresponding temporary storage area position coordinates from the degradation signal. The path distance is not calculated using the straight-line distance between two points, but rather based on the actual navigable road network of the storage area. In practice, the storage aisle centerline, turning nodes, intersections, and one-way traffic areas are constructed as a storage path network diagram, with the edge lengths determined by actual measured aisle lengths. Then, using the node corresponding to the current position of the candidate AGV forklift as the starting point and the target temporary storage area entrance node as the ending point, a shortest path search is performed along the storage path network to obtain the actual drivable path distance from the corresponding vehicle to the target temporary storage area. All candidate vehicles are traversed in the same way to form a path distance set. If two or more candidate vehicles have the same path distance, the time of their most recent task completion is read, and the vehicle with the longer idle waiting time is prioritized. If the waiting times are still the same, the vehicles are selected according to their vehicle numbers. After all comparisons are completed, the AGV forklift with the shortest path distance and meeting the priority criteria is selected as the reassigned pickup vehicle.

[0038] In S5, the lateral pre-offset and longitudinal pre-stop point of the end alignment path are corrected.

[0039] After the handover deviation mapping model outputs the expected decoupling release area corresponding to the current pallet, the geometric center of this expected decoupling release area is calculated. Since the aforementioned decoupling release area is a spatial cluster formed by historical similar samples in the three-dimensional coordinate system of the temporary storage area, it contains three parts of information: longitudinal position distribution, lateral position distribution, and pallet attitude angle distribution. In specific implementation, the longitudinal coordinate values, lateral coordinate values, and attitude angle values ​​of all historical offset points within the coverage area of ​​the decoupling release area are read, and the center positions in the three dimensions are calculated respectively. The longitudinal center position corresponds to the average position of the longitudinal coordinates of all offset points within the decoupling release area, the lateral center position corresponds to the average position of the lateral coordinates, and the attitude angle center position corresponds to the average position of the attitude angles. After statistical analysis, the geometric center of the decoupling release area is obtained. Then, the handover reference position corresponding to the temporary storage area is read. The handover reference position is the position and attitude coordinates corresponding to the standard pallet when it is placed according to the standard placement requirements. The geometric center of the decoupling release area is compared with the handover reference position, and the longitudinal displacement difference, lateral displacement difference, and attitude angle difference between the two are calculated to form the expected offset vector. The three components in the expected offset vector correspond to the longitudinal offset component, the lateral offset component, and the placement deflection angle component, respectively.

[0040] The system reads the original end-point alignment path parameter set corresponding to the current pickup task. This set records the lateral reference coordinates when the forks enter the storage area and the longitudinal reference coordinates when the forks stop extending. Then, it reads the lateral offset component from the expected offset vector. If the expected lateral offset component indicates that the pallet has shifted to the right relative to the handover reference position, compensation is performed in the opposite direction during path correction, i.e., the original lateral reference coordinates are moved to the left by the corresponding distance; if the expected lateral offset component indicates that the pallet has shifted to the left, the original lateral reference coordinates are moved to the right by the corresponding distance. The compensated coordinates are then determined as the lateral pre-offset. The reason for using the reverse compensation method is that the pallet has already shifted, so the forks need to actively correct the alignment path towards the current position of the pallet. After completing the lateral compensation, the system reads the longitudinal offset component from the expected offset vector. If the expected longitudinal offset component indicates that the pallet has shifted relative to the standard position in the fork's forward direction, the original longitudinal stop position is corrected in the opposite direction; if the expected longitudinal offset component indicates that the pallet has shifted behind the forks, the original longitudinal stop position is corrected forward. The corrected longitudinal position is then used as the longitudinal pre-stop point.

[0041] After calculating the lateral pre-offset and longitudinal pre-stop point, the placement deflection angle component is further processed. Specifically, the placement deflection angle component in the expected offset vector is read and used as the expected rotation angle of the pallet relative to its standard orientation. When the placement deflection angle component is clockwise, the fork alignment direction correction angle is calculated counter-clockwise; conversely, when the placement deflection angle component is counter-clockwise, the fork alignment direction correction angle is calculated clockwise. This reverse correction ensures that the forks maintain consistency with the pallet's current orientation when entering the pallet fork holes. Subsequently, the target alignment heading angle in the original end-of-line alignment path is read, and the fork alignment direction correction angle is superimposed on the target alignment heading angle to form the corrected target alignment heading angle. After completing all parameter corrections, the lateral pre-offset, longitudinal pre-stop point, and corrected target alignment heading angle are written into the end-of-line alignment path parameter set corresponding to the current pickup task, while simultaneously replacing the corresponding lateral reference coordinates, longitudinal reference coordinates, and target heading angle in the original path parameter set. After the path replacement is completed, the pickup AGV forklift performs the final approach and pickup action according to the updated end alignment path, so that the actual approach position of the forks is consistent with the pallet state corresponding to the expected fork release area.

[0042] In step S6, path conflict resolution and collaborative scheduling are performed on all AGV forklifts for picking up parts.

[0043] The system reads information from all AGV forklifts currently executing pickup tasks within the storage area, including their current position, speed, heading angle, corresponding temporary storage area identifier, and corrected end-point alignment path parameter set. Since different vehicles correspond to different target temporary storage areas, the current path needs to be concatenated with subsequent end-point alignment paths to form a complete pickup path. A standard traffic network is pre-established within the storage area, containing all main lanes, branch lanes, intersections, turning areas, and temporary storage area entrances. Each lane corresponds to its actual measured length and allowed traffic direction. Then, trajectory prediction is performed on each pickup AGV forklift in chronological order. During trajectory prediction, a simple straight-line prediction method is not used; instead, the future position is calculated segment by segment based on the current vehicle position, real-time speed, and target path length. For straight sections in the path, the estimated arrival time is calculated based on the current speed and corresponding segment length; for turning areas, compensation calculations are performed based on the vehicle's steering mechanism's allowed turning radius and the average turning duration in historical operation records. This method yields the time-series trajectory set for each vehicle within the future scheduling window. The scheduling window is set based on the task density at the warehouse site. For example, it's set to 30 seconds in a warehouse scenario with a large number of vehicles, and 60 seconds in a scenario with a small number of vehicles. After predicting all vehicle trajectories, spatial overlap analysis is performed on each trajectory. Specifically, any two predicted trajectories are traversed, and when two trajectories overlap at the same path node, the same intersection, or the same passage section, their corresponding arrival times are further compared. When the time difference between two or more AGV forklifts entering the area is less than the preset conflict determination time, an actual traffic conflict is identified in that area. The conflict determination time is determined based on both vehicle braking distance and scheduling response time, for example, using the 95% safe passage time from continuous historical operation data statistics as the determination criterion. Once the condition is met, the corresponding area is marked as a conflict area, and the vehicle identification, expected entry time, and expected departure time of the vehicles involved in the conflict are recorded. For the same vehicle participating in multiple conflict areas, a conflict event chain is formed according to the expected entry time order, thereby obtaining all potential conflict information within the future scheduling window of the warehouse area.

[0044] The system reads the waiting time of each AGV forklift's corresponding pickup task within the conflict area. The task waiting time is defined as the time difference between the current moment and the moment the task enters the pending execution state. This time accumulates from the moment the task is created and enters the scheduling queue, not from the moment the vehicle starts moving, thus accurately reflecting the task backlog. The waiting times of all vehicles within the conflict area are then sorted. The task with the longest waiting time is given priority passage through the current conflict area. When two vehicles have the same waiting time, the remaining path length from the target temporary storage area is compared, with the vehicle with the shorter remaining path length passing first. If the remaining path lengths are still the same, the final priority is determined according to the task creation order. After priority determination, the passage time for all vehicles within the conflict area is recalculated. Specifically, the original planned entry time of the priority vehicle remains unchanged, while the remaining vehicles move backward according to the safety interval order. The safety interval is determined based on the length of vehicles in the storage area, the average braking distance, and the length of the intersection area. For example, the highest frequency time interval is taken from historical safe passage records as the standard safety interval. For delayed vehicles, instead of directly modifying their final destination path, the target arrival times for each segment before reaching the conflict area are adjusted, thus forming a new passage time sequence. After rescheduling, the new passage times for the corresponding vehicles are written into the task execution parameters, and the estimated entry times for subsequent conflict areas are updated simultaneously. If an adjustment causes a new conflict, the next round of conflict detection and rescheduling continues until there are no unresolved conflict events in the current scheduling window. Finally, the rescheduled passage times are sent to the corresponding AGV forklifts, and each vehicle executes its pickup task according to the new timing plan, thereby achieving collaborative scheduling and path conflict resolution for multiple AGV forklifts within the warehouse area.

[0045] Example 2 The difference between Embodiment 2 and Embodiment 1 is that this embodiment introduces an AGV forklift collaborative scheduling system for warehouse management.

[0046] Figure 2 A schematic diagram of an AGV forklift collaborative scheduling system for warehouse management is provided according to the present invention. The AGV forklift collaborative scheduling system for warehouse management includes: The envelope construction module is used to collect the time history curves of the sliding deceleration corresponding to the standard pallets of different load levels, and combine them with the driving condition label to establish the standard deceleration envelope set of the corresponding load level. The deviation analysis module is used to compare and analyze the current coasting deceleration curve with the standard deceleration envelope, extract the cumulative deviation, waveform difference and peak hysteresis, and generate a coasting deviation vector. The handover assessment module is used to construct a handover deviation mapping model based on the sliding deviation vector, output the expected decoupling release area and determine the corresponding estimated success rate of pickup. The pickup module is used to perform temporary storage area downgrading and pickup vehicle reassignment based on the pickup success rate estimate, and to correct the end alignment path parameter set based on the expected offset vector. The collaborative scheduling module is used to predict the driving trajectory of all pickup AGV forklifts, identify path conflict areas and rearrange the passage times to perform collaborative scheduling of multiple AGV forklifts.

[0047] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0048] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0049] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0050] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0051] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0052] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0053] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0054] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0055] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A collaborative scheduling method for AGV forklifts used in warehouse management, characterized in that, Includes the following steps: S1. When the AGV forklift carrying standard pallets of different load levels enters the temporary storage area, the pre-set deceleration section is collected, the sliding deceleration time history curve is collected and the upper and lower envelopes are extracted. The standard deceleration envelope set corresponding to the load level is established according to the different driving condition labels. S2. When the AGV forklift is carrying goods and performing a placement task and enters the same deceleration section, the standard deceleration envelope of the corresponding load level is retrieved, and the current sliding deceleration curve is compared with the standard deceleration envelope in real time to generate a sliding deviation vector. S3. Using the sliding deviation vector as input, construct the handover deviation mapping model, output the expected detent release area of ​​the pallet relative to the handover reference position after this placement, and combine it with historical handover records to determine the corresponding estimated success rate of picking up the item. S4. When the estimated success rate of pickup is lower than the collaborative scheduling threshold, a temporary storage area degradation signal is sent to the pickup module, and pickup vehicles are reassigned based on the current on-site distribution of standby pickup AGV forklifts. S5. Using the displacement vector between the expected geometric center of the derailment release area and the handover reference position as the expected offset vector, the AGV forklift for picking up parts uses the expected offset vector as a feedforward parameter to correct the end alignment path parameter set before entering the temporary storage area. S6. Under the modified end-point alignment path constraint, based on the retrieval tasks corresponding to all temporary storage areas within the storage area, perform path conflict resolution and collaborative scheduling for all retrieval AGV forklifts.

2. The AGV forklift collaborative scheduling method for warehouse management according to claim 1, characterized in that, In step S1, establishing the standard deceleration envelope set corresponding to the load level specifically includes: When an AGV forklift is carrying a standard pallet with a known load level and enters the same temporary storage area before a preset deceleration section, the angle between the actual heading angle of the AGV forklift and the corresponding planned heading angle is obtained, and the driving condition label is divided according to the range of the angle. Collect the coasting deceleration time history curve under the current load level, and align it with the standard deceleration envelope of the unloaded individual according to the deceleration start time. Calculate the deceleration difference between the current coasting deceleration time history curve and the standard deceleration envelope of the unloaded individual for each sampling point, and generate the deceleration correction sequence for the corresponding load level. For all historical taxiing deceleration time history curves under the same entry condition label and the same load level, the deceleration correction sequence calculation is repeatedly performed. The mean of each deceleration correction sequence is statistically analyzed at the corresponding sampling time to generate the load correction sequence corresponding to the current load level. The load correction sequence is superimposed on the standard deceleration envelope of the unloaded individual at each sampling point to generate a standard deceleration envelope subset corresponding to the load level. The standard deceleration envelope subsets corresponding to different load levels of different driving condition labels are combined to form a standard deceleration envelope set.

3. The AGV forklift collaborative scheduling method for warehouse management according to claim 1, characterized in that, In step S2, generating the gliding deviation vector specifically includes: Based on the load level, the current coasting deceleration time history curve is compared with the upper and lower envelopes of the corresponding standard deceleration envelope point by point. The positive deviation sampling point sequence that exceeds the upper envelope and the negative deviation sampling point sequence that is lower than the lower envelope are recorded. The deviation amplitudes corresponding to the positive deviation sampling point sequence and the negative deviation sampling point sequence are accumulated respectively, and the sum of the two results is used as the cumulative deviation. The current taxiing deceleration time history curve is dynamically time-normalized with the center line of the standard deceleration envelope. The normalized cumulative matching distance after normalization is calculated as the waveform difference. The difference between the deceleration peak time in the current taxiing deceleration time history curve and the deceleration peak time of the center line of the standard deceleration envelope is located as the peak hysteresis. The cumulative deviation, waveform difference, and peak hysteresis are combined to form the taxiing deviation vector.

4. The AGV forklift collaborative scheduling method for warehouse management according to claim 1, characterized in that, In S3, constructing the handover deviation mapping model specifically includes: Obtain the sliding deviation vector and the corresponding actual pallet offset vector based on the handover reference position for each placement operation in the historical handover record of the temporary storage area. Map the actual pallet offset vector to the three-dimensional coordinate system of the temporary storage area to form an offset point set. Statistically count the spatial clustering range of the offset point set corresponding to the same type of sliding deviation vector, and define the spatial clustering range as the derailment release area. Establish the correspondence between the sliding deviation vector and the derailment release area to form a handover deviation mapping model based on the entry condition label. The aforementioned similar taxiing deviation vectors are taxiing deviation vectors obtained by constructing a three-dimensional deviation vector based on the cumulative deviation, waveform difference, and peak hysteresis under the same entry condition label, and by jointly filtering the three-dimensional deviation vectors based on the included angle between the three-dimensional deviation vectors and the vector magnitude range.

5. The AGV forklift collaborative scheduling method for warehouse management according to claim 4, characterized in that, In S3, the estimated success rate of package pickup specifically includes: Obtain the actual pallet offset vector and handover result label for each handover in all historical handover records in the temporary storage area, and map the actual pallet offset vector to the coordinate system of the temporary storage area to form a set of historical offset points. The ratio of the number of successful handover events corresponding to historical offset points within the coverage area of ​​each derailment release zone to the total number of handover events is calculated and used as the success rate benchmark value for the corresponding derailment release zone. A correspondence between the derailment release zone and the success rate benchmark value is established. When the current sliding deviation vector is input, the handover deviation mapping model outputs the corresponding expected derailment release zone. The success rate benchmark value corresponding to the expected derailment release zone is read as the estimated success rate of part retrieval.

6. The AGV forklift collaborative scheduling method for warehouse management according to claim 1, characterized in that, In step S4, the reassignment of the pickup vehicle specifically includes: The estimated success rate of pickup is compared with the collaborative scheduling threshold. When the estimated success rate of pickup is lower than the collaborative scheduling threshold, a temporary storage area downgrade signal carrying the original temporary storage area identifier is sent to the pickup module. After receiving the temporary storage area degradation signal, the pickup module iterates through all the pickup AGV forklifts that are currently in an idle standby state and selects the vehicle with the shortest path distance between its location and the temporary storage area as the reassigned pickup vehicle.

7. The AGV forklift collaborative scheduling method for warehouse management according to claim 1, characterized in that, In step S5, the correction of the lateral pre-offset and longitudinal pre-stop point of the end alignment path specifically includes: After receiving the expected offset vector, the AGV forklift decomposes the expected offset vector into a longitudinal offset component along the longitudinal direction of the temporary storage area, a lateral offset component along the transverse direction of the temporary storage area, and a placement deflection angle component. The lateral pre-offset amount is obtained by compensating the lateral reference coordinates when the forks enter the temporary storage area in the original end alignment path based on the displacement in the opposite direction of the lateral offset component. The longitudinal reference coordinates at the point where the forks stop extending in the original end alignment path are compensated based on the displacement in the opposite direction of the longitudinal offset component to obtain the longitudinal pre-stop point. The fork alignment direction correction angle is calculated based on the placement deflection angle component, and the target alignment heading angle in the end alignment path is corrected using the fork alignment direction correction angle. Write the lateral pre-offset, longitudinal pre-stop point, and corrected target alignment heading angle into the end alignment path parameter set of the AGV forklift, replacing the corresponding reference coordinates and reference heading angle in the original end alignment path.

8. The AGV forklift collaborative scheduling method for warehouse management according to claim 1, characterized in that, In step S6, the collaborative scheduling for resolving path conflicts for all AGV forklifts specifically includes: Obtain the corrected end-point alignment path of all pickup AGV forklifts in the warehouse area and the temporary storage area identifier corresponding to the current task, and predict the driving trajectory of each pickup AGV forklift in the future scheduling window according to the time sequence. Traverse the path intersection areas in the predicted trajectories of each pickup AGV forklift. When two or more pickup AGV forklifts enter the same path intersection area within the same time window, mark the corresponding path intersection area as a conflict area. Based on the task waiting time, the passage time of each pickup AGV forklift in the conflict area is re-arranged, and the re-arranged passage time is sent to the corresponding pickup AGV forklift for execution.

9. An AGV forklift collaborative scheduling system for warehouse management, used to implement the AGV forklift collaborative scheduling method for warehouse management as described in any one of claims 1-8, characterized in that, include: The envelope construction module is used to collect the time history curves of the sliding deceleration corresponding to the standard pallets of different load levels, and combine them with the driving condition label to establish the standard deceleration envelope set of the corresponding load level. The deviation analysis module is used to compare and analyze the current coasting deceleration curve with the standard deceleration envelope, extract the cumulative deviation, waveform difference and peak hysteresis, and generate a coasting deviation vector. The handover assessment module is used to construct a handover deviation mapping model based on the sliding deviation vector, output the expected decoupling release area and determine the corresponding estimated success rate of pickup. The pickup module is used to perform temporary storage area downgrading and pickup vehicle reassignment based on the pickup success rate estimate, and to correct the end alignment path parameter set based on the expected offset vector. The collaborative scheduling module is used to predict the driving trajectory of all pickup AGV forklifts, identify path conflict areas and rearrange the passage times to perform collaborative scheduling of multiple AGV forklifts.

Images