Intelligent scheduling method for AGV sorting robots with multi-task parallel processing

Abstract

The present invention provides an intelligent scheduling method for an AGV sorting robot capable of multi-tasking parallel processing, relating to the technical field of logistics sorting scheduling. The method involves reverse parsing the first to-be-palletized information based on a standard stacking template to obtain a logistics box arrival order queue, optimizing logistics task allocation based on the connection time topology and the logistics box arrival order queue to output M parallel logistics task sequences. Logistics conflicts are compensated for N groups of parallel logistics task sequences and M parallel logistics task sequences, outputting M parallel logistics trajectory sequences to schedule M sorting robots, which execute the M parallel logistics task sequences to perform palletizing and delivery at the first sorting port. This method solves the technical problem of frequent path conflicts and sorting congestion in sorting robots in complex scenarios with multiple parallel tasks in the prior art. This method effectively reduces the risk of path conflicts and sorting port congestion, and enhances the adaptability and fault tolerance of sorting robots in complex scenarios with multiple parallel tasks.

Description

Technical Field

[0001] The present invention relates to the technical field of logistics sorting and scheduling, and in particular to an intelligent scheduling method for an AGV sorting robot that processes multiple tasks in parallel. Background Art

[0002] In intelligent warehousing and logistics systems, the collaborative scheduling of multiple AGV sorting robots has long faced the common problems of path planning conflicts and imbalance in task execution efficiency.

[0003] Existing technologies use fixed priority rules or static time window allocation strategies, which are difficult to deal with the congestion problem caused by dynamic path intersection in high-concurrency task scenarios, especially in the sorting area, where the instantaneous convergence of multiple AGVs can easily lead to buffer overflow and task delays.

[0004] At the same time, the traditional task scheduling mechanism lacks a structured connection with palletizing needs, and the temporal and spatial adaptability of the handling order and the palletizing level is poor, which often leads to safety hazards such as the displacement of the center of gravity of the pallet and the misalignment of box stacking.

[0005] In addition, the fragmented scheduling of sorting tasks and palletizing links leads to low resource utilization of the AGV cluster and the inability to achieve load balancing throughout the entire process, which seriously restricts the overall throughput efficiency and operational stability of the warehouse automation system.

[0006] In summary, the existing technology has technical problems such as frequent path conflicts and sorting congestion of AGV sorting robots in complex scenarios with multiple tasks running in parallel, and the disconnection between the execution of handling and sorting tasks and the subsequent palletizing tasks. Summary of the Invention

[0007] The present invention provides an intelligent scheduling method for an AGV sorting robot with multi-task parallel processing, which is used to solve the technical problems in the prior art of AGV sorting robots in complex scenarios with multi-task parallel processing, such as frequent path conflicts and sorting congestion, and the disconnection between the execution of handling and sorting tasks and the subsequent palletizing tasks.

[0008] In view of the above problems, the present invention provides an intelligent scheduling method for an AGV sorting robot with multi-task parallel processing.

[0009] The present invention provides an intelligent scheduling method for an AGV sorting robot with multi-task parallel processing, the method comprising: interactively obtaining first information to be palletized of a first sorting port; reversely parsing the first information to be palletized according to a standard pallet shape template to obtain a logistics box arrival order queue; extracting docking port position information of an automated sorting area from the first information to be palletized; constructing a docking time-consuming topology according to the docking port position information and the first sorting position of the first sorting port; optimizing logistics task allocation according to the docking time-consuming topology and the logistics box arrival order queue, and outputting M parallel logistics task sequences; interactively obtaining N groups of parallel logistics task sequences for the remaining N sorting ports, performing logistics conflict compensation on the N groups of parallel logistics task sequences and the M parallel logistics task sequences, and outputting M parallel logistics trajectory sequences; and scheduling M AGV sorting robots using the M parallel logistics trajectory sequences to execute the M parallel logistics task sequences and perform palletizing delivery at the first sorting port.

[0010] The technical solution provided in the present invention has at least the following technical effects or advantages:

[0011] The method provided by the embodiment of the present invention obtains first information of a first sorting port to be palletized through interaction; reversely analyzes the first information of a first sorting port to be palletized according to a standard pallet shape template to obtain a logistics box arrival order queue; extracts the docking port position information of the automated sorting area from the first information of a first sorting port to be palletized; constructs a docking time-consuming topology according to the docking port position information and the first sorting position of the first sorting port; optimizes logistics task allocation according to the docking time-consuming topology and the logistics box arrival order queue, and outputs M parallel logistics task sequences; interactively obtains N groups of parallel logistics task sequences for the remaining N sorting ports, performs logistics conflict compensation on the N groups of parallel logistics task sequences and the M parallel logistics task sequences, and outputs M parallel logistics trajectory sequences; uses the M parallel logistics trajectory sequences to dispatch M AGV sorting robots to execute the M parallel logistics task sequences to perform palletizing and delivery at the first sorting port. The system ensures that logistics boxes arrive and are stacked accurately in the preset order, significantly improves the stability of the stacking structure and space utilization, effectively reduces the risk of path conflicts and congestion at the sorting port, enhances the adaptability and fault tolerance of AGV sorting robots in complex scenarios with multiple tasks running in parallel, and ensures the reliable operation of high-throughput sorting operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A flow chart of the intelligent scheduling method for the AGV sorting robot with multi-task parallel processing provided by the present invention;

[0013] Figure 2 A schematic diagram of the flow of obtaining the arrival order queue of logistics boxes in the intelligent scheduling method of the AGV sorting robot with multi-task parallel processing provided by the present invention. DETAILED DESCRIPTION

[0014] The present invention provides an intelligent scheduling method for AGV sorting robots that perform multi-tasking in parallel. This method addresses the technical issues in the prior art, such as frequent path conflicts and sorting congestion, as well as the disconnection between the execution of transport and sorting tasks and the subsequent palletizing tasks, in complex scenarios with multi-tasking in parallel. This method achieves the technical effect of ensuring the precise arrival and stacking of logistics boxes in a preset order, significantly improving the stability and space utilization of the palletizing structure, effectively reducing the risk of path conflicts and congestion at the sorting port, enhancing the adaptability and fault tolerance of AGV sorting robots in complex scenarios with multi-tasking in parallel, and ensuring the reliable operation of high-throughput sorting operations.

[0015] Below, the technical solutions of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention, rather than all, are shown in the accompanying drawings.

[0016] Examples, such as Figure 1 As shown, the present invention provides an intelligent scheduling method for an AGV sorting robot for multi-task parallel processing, the method comprising:

[0017] A100: Interactively obtain the first palletizing information of the first sorting port.

[0018] Specifically, in this embodiment, the first sorting port is any one of the N+1 sorting ports in the automated sorting area. Through an interactive warehouse management system (WMS), first palletizing information for the first sorting port is retrieved. This first palletizing information includes the total number (W) of logistics boxes to be palletized, the number of logistics boxes, the maximum base dimensions, the total weight of the logistics boxes, the current docking port of the automated sorting area where the W logistics boxes to be palletized are located, and information such as the weight distribution and dimensions of each logistics box.

[0019] A200: Reversely analyze the first information to be palletized according to the standard pallet shape template to obtain the arrival order queue of the logistics boxes.

[0020] In one embodiment, the first to-be-palletized information is reversely parsed according to a standard pallet shape template to obtain a logistics box arrival order queue. Prior to this, step A200 of the method provided by the present invention further includes:

[0021] A200-1: Extracting palletizing requirement-related features from the first information to be palletized, wherein the palletizing requirement-related features include the number of logistics boxes, the maximum base size, and the total mass of the logistics boxes.

[0022] A200-2: Match the stacking template according to the associated features of the palletizing requirements to obtain the standard stacking template.

[0023] In one embodiment, Figure 2 As shown, the first information to be palletized is reversely parsed according to the standard pallet shape template to obtain the arrival order queue of the logistics boxes. The method step A200 provided by the present invention also includes:

[0024] A210: Decomposing the standard pallet shape template into a pallet shape structure to obtain H palletizing level feature vectors of H virtual palletizing boxes.

[0025] A220: Extract W logistics box feature vectors of the W logistics boxes to be palletized from the first information to be palletized, where W≤H and HW≤8.

[0026] A230: Use the W logistics box feature vectors to traverse the H palletizing level feature vectors to determine palletizing compatibility, and screen and locate to obtain W palletizing space positions.

[0027] A240: After projecting the W palletizing space positions onto the standard pallet shape template, perform palletizing level descending fitting and output the logistics box arrival order queue.

[0028] Specifically, it should be understood that the number of logistics boxes represents the total number of logistics boxes to be palletized (denoted as W), which is used to determine the total volume of the palletizing space required. The maximum base dimension, defined as the maximum length and width of the logistics box bottom, is used to select suitable palletizing templates to avoid over-palletizing issues. The total logistics box mass is calculated by adding the weight of all logistics boxes to be palletized, and is used to verify whether the load-bearing capacity of the selected template meets safety standards.

[0029] Based on this, this embodiment extracts palletizing demand-related features including the number of logistics boxes, the maximum base size, and the total mass of the logistics boxes from the first information to be palletized.

[0030] At the same time, it should be understood that the data source of the stacking demand association characteristics described in this embodiment is preferably obtained by scanning the RFID tags and barcodes of the logistics boxes during warehousing, and the obtained data is verified with the logistics box specification information in the warehouse database.

[0031] The stacking template library pre-built in this embodiment includes a variety of template types that have been verified for stability, such as rectangular array, pyramid and staggered stacking structures. Each template is associated with the number of levels, size constraints of each level, load-bearing thresholds and spatial arrangement rules.

[0032] During the matching process, candidate templates with a base area greater than or equal to the maximum base size are first screened out. Based on the total mass of the logistics boxes and the load-bearing threshold of the candidate templates, options with a total mass exceeding the maximum load-bearing capacity of the template are eliminated to avoid instability of the stacking structure. Finally, a template with a stacking structure that can stack the number of logistics boxes H, satisfying H≥W and HW≤8, is selected as the standard stacking template (for example, a rectangular array template may be designed as 3 layers × 4 columns, H=12).

[0033] Through the above matching mechanism, a standard pallet shape template suitable for the current palletizing task is finally output, and its palletizing structure achieves the optimal balance between stability, space utilization and operational feasibility.

[0034] Based on the obtained standard pallet template, a structural analysis is performed on the template, splitting it into H virtual palletizing boxes. Each box corresponds to an independent palletizing position preset in the template and is associated with a palletizing layer feature vector. This feature vector includes the box's three-dimensional spatial positioning coordinates (X, Y, Z) in the template coordinate system; the maximum length, width, and height constraints allowed for the logistics box; the layer load-bearing capacity threshold calculated based on the mechanical model; and the minimum horizontal and vertical spacing rules with adjacent boxes.

[0035] This embodiment converts the standard stack-shaped template into a parametric model consisting of H virtual boxes through the above decomposition process, providing a structured data basis for subsequent logistics box matching.

[0036] W logistics box feature vectors of the W logistics boxes to be palletized are extracted from the first information to be palletized, and the logistics box feature vectors specifically include actual length, width, height dimensional parameters, weight data and center of gravity position coordinates of the logistics boxes to be palletized.

[0037] The total number of logistics boxes, W, satisfies the constraints W ≤ H and HW ≤ 8, where H is the preset capacity of the standard pallet template (i.e., the total number of logistics boxes the template can accommodate). This constraint ensures that the selected template can fully accommodate all logistics boxes (W ≤ H) while limiting the number of unused slots to no more than 8 (HW ≤ 8), thereby maximizing storage space utilization while preventing palletizing failures.

[0038] Furthermore, the W logistics box feature vectors are traversed and matched with the H palletizing level feature vectors in turn, and the suitable palletizing space position is selected through the compatibility scoring algorithm.

[0039] Specifically, for each palletizing container, a matching score is calculated based on the container's feature vector and the palletizing level's feature vector, comparing its size constraints, load-bearing capacity, and adjacent spacing rules with each virtual container. The virtual container with the highest score is then selected as its palletizing target location. If multiple containers compete for the same virtual container, they are prioritized based on their weight and center of gravity stability. Heavy containers or those with a lower tolerance for center of gravity shift are assigned to the bottom container with greater load-bearing capacity. Ultimately, a unique palletizing space is determined for all W containers, and these W palletizing space locations are output.

[0040] After mapping the W stacking space positions to the standard stacking template, the logistics boxes corresponding to the bottom boxes are added to the logistics box arrival order queue first, and then the upper logistics boxes are added in ascending order of the level numbers. For multiple boxes in the same level, the logistics box delivery order is arranged in ascending order of the X-axis (from left to right) and Y-axis (from front to back) of the spatial coordinates, and finally the logistics box arrival order queue is output.

[0041] This embodiment is based on the reverse analysis and feature matching mechanism of the standard stacking template, combined with the compatibility judgment of the physical properties of the logistics box and the template hierarchical constraints, to screen out the stacking space positions suitable for each logistics box, and generate the logistics box arrival order queue through hierarchical descending fitting, thereby achieving the direct technical effect of improving the stability of the stacking structure, and indirectly achieving the indirect technical effect of maximizing the utilization of storage space and improving the overall sorting efficiency while ensuring the stability of the stacking structure.

[0042] A300: Extracting the docking port location information of the automated sorting area from the first information to be palletized.

[0043] Specifically, in this embodiment, the W target sorting port codes of the W logistics boxes to be palletized in the first palletizing information are associated with their corresponding physical coordinates of the docking ports to obtain the docking port position information. The docking port position information includes the two-dimensional plane coordinates (X, Y) of the W docking ports. Since the AGV sorting robot can only dock one logistics box at a time, the same docking port may be repeatedly allocated to multiple logistics boxes, resulting in duplicate coordinate entries in the W docking port position information.

[0044] A400: Construct a docking time-consuming topology based on the docking port position information and the first sorting position of the first sorting port.

[0045] In one embodiment, a docking time-consuming topology is constructed based on the docking port position information and the first sorting position of the first sorting port. Step A400 of the method provided by the present invention further includes:

[0046] A410: Decompose the docking port position information to obtain W docking port position information of W logistics boxes to be stacked.

[0047] A420: Construct W docking movement trajectories based on the first sorting position and the W docking port position information.

[0048] A430: Output W benchmark docking times based on the benchmark AGV displacement speed and the W docking movement trajectories.

[0049] A440: Construct W leaf nodes based on the W docking port location information, and construct a central node based on the first sorting position.

[0050] A450: Based on the W benchmark connection time-consuming quantified topological connections, construct in-degree connections from the W leaf nodes to the central node, and complete the construction of the connection time-consuming topology.

[0051] Specifically, in this embodiment, the docking port location information is decomposed through the mapping relationship between the logistics box ID and the docking port ID to determine the docking port location of each logistics box (X p ,Y p ), and obtain the location information of W docking ports of W logistics boxes to be stacked.

[0052] Based on the sorting port coordinates corresponding to the first sorting position and the W docking port coordinates, the A* path planning algorithm is used to generate W docking movement trajectories.

[0053] It should be understood that each docking movement trajectory consists of a series of path points. The coordinates of the path points are generated through a rasterized map or a continuous coordinate space interpolation algorithm, and avoid fixed obstacles in the warehouse (such as shelves and charging stations). Then, W (theoretical) benchmark docking times are output based on the benchmark AGV displacement speed and the W docking movement trajectories.

[0054] The W docking ports are abstracted as leaf nodes of the topological network. Each node's attributes include the port coordinates and benchmark docking time. The first sorting port is used as the central node, and its attributes include the sorting port coordinates, processing rate, and buffer capacity. The connections between nodes are defined by the physical paths from the docking ports to the sorting ports, forming a star topology with the sorting ports as the convergence center. The W benchmark docking times are used as the initial edge weights from the leaf nodes to the central node, and in-degree connections are established to complete the construction of the docking time topology.

[0055] This implementation builds a connection time-consuming topology based on logistics trajectories and logistics time consumption, achieving the technical effect of providing a reference for subsequent parallel logistics task allocation.

[0056] A500: Optimize logistics task allocation based on the docking time-consuming topology and the logistics box arrival order queue, and output M parallel logistics task sequences.

[0057] In one embodiment, logistics task allocation is optimized based on the docking time topology and the logistics box arrival order queue, and M parallel logistics task sequences are output. Step A500 of the method provided by the present invention further includes:

[0058] A510: Based on the M AGV sorting robots, the logistics box arrival order queue is decomposed to obtain multiple adjacent local arrival order queues.

[0059] A520: Based on the multiple local arrival order queues, perform connection delay superposition and decomposition on the connection time-consuming topology to obtain multiple update time-consuming sub-topologies.

[0060] A530: Based on the M AGV sorting robots, split the multiple update time-consuming sub-topologies to obtain M logistics task time limit sequences.

[0061] A540: Add the W docking movement trajectory mappings to the M logistics task time limit sequences, and output the M parallel logistics task sequences, where W ≥ 25M.

[0062] In one embodiment, based on the multiple local arrival order queues, the connection time-consuming topology is subjected to connection delay superposition and decomposition to obtain multiple update time-consuming sub-topologies. Step A520 of the method provided by the present invention further includes:

[0063] A521: Based on the mapping relationship between the multiple local arrival order queues and the W logistics boxes to be palletized in the docking time-consuming topology, the docking time-consuming topology is decomposed into multiple task time-consuming sub-topologies.

[0064] A522: Interactively obtain the arrival time interval constraint of the first sorting port.

[0065] A523: Based on the multiple local arrival order queues, the arrival time interval constraint is superimposed on the multiple task time-consuming sub-topology mappings to obtain multiple update time-consuming sub-topologies.

[0066] In one embodiment, the W docking movement trajectory mappings are added to the M logistics task time limit sequences, and the M parallel logistics task sequences are output. Step A540 of the method provided by the present invention further includes:

[0067] A541: The mapping relationship between the M parallel logistics task sequences and the W logistics boxes to be stacked is combined with the W docking movement trajectories to obtain M docking task trajectory sequences.

[0068] A542: Store the M docking task trajectory sequences and the M logistics task time limit sequences in a temporally and spatially aligned manner to generate the M parallel logistics task sequences.

[0069] Specifically, in this embodiment, in order to adapt to the parallel processing capabilities of M AGV sorting robots, the logistics box arrival order queue is divided into multiple adjacent local arrival order queues in chronological order.

[0070] The queue decomposition rule is based on the number of AGV sorting robots M and the preset single AGV task cycle length, ensuring that the number of logistics boxes contained in each local arrival order queue matches the AGV processing capacity. For example, if the total number of tasks W = 100 and M = 4, the queue is divided into 25 local arrival order queues (each containing 4 logistics boxes), and the time windows of adjacent queues are seamlessly connected to avoid task gaps.

[0071] According to the logistics box ID in the local arrival order queue, the corresponding docking port node is located from the docking time-consuming topology, and these nodes and their connecting edges are extracted to form a local sub-topology, namely the task time-consuming sub-topology.

[0072] Each task-time-consuming sub-topology covers all paths from docking ports to sorting ports involved in the corresponding local arrival order queue, retaining the weight calculation rules of the original topology but limiting its scope to tasks within the current time window. This embodiment decomposes the docking-time-consuming topology into multiple task-time-consuming sub-topologies, ensuring that subsequent optimization targets only local tasks, reducing the complexity of global calculations.

[0073] Interactively obtain the arrival time interval constraint of the first sorting port, which includes the minimum safe time difference (such as 5 seconds) between adjacent AGVs arriving at the sorting port and the buffer capacity limit (such as the maximum number of logistics boxes processed simultaneously).

[0074] In the task time-consuming subtopology, time interval constraints are dynamically added based on the ranking order of tasks in the local arrival order queue. Specifically, the ranking of each task in the local queue (e.g., 1st, 2nd, etc.) determines its time interval compensation value. For a task ranked k (k=1, 2, ..., K), its edge weight update formula is:

[0075] .in, is the inter-arrival time constraint, It is the sequential number of the tasks in the local queue. For example, the baseline time of the task ranked first in a local queue is 30 seconds, and the updated time is 30+1×5=35 seconds. The time of the task ranked second is 30+2×5=40 seconds, and so on. This mechanism ensures that the tasks on the same path gradually increase the time intervals in the order of arrival, avoiding instantaneous congestion at the sorting port due to the concentrated arrival of multiple tasks. Based on this, according to the multiple local arrival order queues, the arrival time interval constraints are superimposed on the multiple task time-consuming sub-topologies to obtain multiple updated time-consuming sub-topologies. It should be understood that the updated time-consuming sub-topology is used to limit the logistics transportation speed of the AGV sorting robot to ensure that the order of the logistics boxes that finally arrive at the sorting port meets the logistics box arrival order queue.

[0076] A greedy algorithm is used to split multiple time-consuming sub-topologies into M logistics task time-limit sequences. Each sequence corresponds to a task list of an AGV, including the logistics box ID to be processed, the connection path, the start time and the latest completion time.

[0077] When splitting, tasks with similar time consumption and low path conflict are assigned to the same AGV sorting robot first, and the total amount of tasks of each AGV sorting robot is consistent (for example, if the total number of tasks W = 100 and M = 4, the queue is divided into 25 local arrival order queues, then the task volume of each AGV sorting robot is 25, and each task comes from the 25 update time-consuming sub-topologies corresponding to the 25 local arrival order queues).

[0078] Based on each AGV's task list (i.e., its assigned container IDs) in the M parallel logistics task sequences, the W docking trajectories are reorganized into a sequence of M docking task trajectories by AGV number. For example, if AGV1's task list includes containers B1, B5, B9, and so on, the path segments corresponding to these containers are extracted from the W trajectories and assembled into a complete docking trajectory for AGV1 in chronological order of task initiation.

[0079] The coordinates (X, Y) of the path points in each docking task trajectory sequence are compared with the corresponding task time limit (such as the start time t start , deadline t end ) and writes it to the scheduling database. For example, the trajectory point (X1, Y1) of an AGV corresponds to the time 08:00:00, and the point (X2, Y2) corresponds to 08:00:05, with an error tolerance of ±300 milliseconds.

[0080] Based on this, the M parallel logistics task sequences are generated by spatially and temporally aligning and storing the M docking task trajectory sequences with the M logistics task time limit sequences.

[0081] This embodiment reorganizes the docking movement trajectory according to the AGV task list to generate continuous and conflict-free path instructions for each AGV. It then stores them in time and space aligned with the task execution time limit to ensure that the path execution time strictly matches the task window.

[0082] A600: Interactively obtain N groups of parallel logistics task sequences for the remaining N sorting ports, perform logistics conflict compensation on the N groups of parallel logistics task sequences and the M parallel logistics task sequences, and output M parallel logistics trajectory sequences.

[0083] In one embodiment, N groups of parallel logistics task sequences for the remaining N sorting ports are interactively obtained, logistics conflicts are compensated for the N groups of parallel logistics task sequences and the M parallel logistics task sequences, and M parallel logistics trajectory sequences are output. Step A600 of the method provided by the present invention further includes:

[0084] A610: After time-series alignment of the N groups of parallel logistics task sequences and the M parallel logistics task sequences, spatial movement trajectory mapping is performed to locate multiple trajectory conflicting spatiotemporal nodes.

[0085] A620: Extract multiple connecting movement trajectories and multiple logistics task time limits from the M parallel logistics task sequence mappings based on the multiple trajectory conflict spatiotemporal nodes.

[0086] A630: Taking the time limits of the multiple logistics tasks as constraints, locally update the multiple connecting movement trajectories according to the multiple trajectory conflict spatiotemporal nodes, and output multiple updated movement trajectories, wherein the multiple updated movement trajectories have multiple updated AGV displacement speeds.

[0087] A640: Based on the multiple trajectory conflicting spatiotemporal nodes, the multiple updated moving trajectories and the multiple updated AGV displacement speeds are covered to the M parallel logistics task sequences, and the M parallel logistics trajectory sequences are output.

[0088] In one embodiment, the method step A630 of the present invention further includes: using the time limits of the multiple logistics tasks as constraints, locally updating the multiple connecting movement trajectories based on the multiple trajectory conflict spatiotemporal nodes, and outputting multiple updated movement trajectories.

[0089] A631: Frame multiple trajectory compensation areas at the multiple trajectory conflict time and space nodes.

[0090] A632: Using the multiple trajectory compensation areas as compensation space limitations, locally updating the multiple connecting movement trajectories according to the multiple trajectory conflict spatiotemporal nodes, and outputting multiple sets of alternative movement trajectories.

[0091] A633: Interactively obtain the maximum AGV displacement speed.

[0092] A634: Calculate multiple sets of alternative AGV displacement speeds based on the multiple logistics task time limits and multiple sets of alternative movement trajectories.

[0093] A635: Filter out the multiple updated movement trajectories from the multiple groups of candidate movement trajectories according to the deviation scales between the multiple groups of candidate AGV displacement speeds and the limit AGV displacement speed.

[0094] In one embodiment, Q times the reference AGV displacement speed is the limit AGV displacement speed, Q∈[1.2,2.6].

[0095] In this embodiment, in order to ensure that the AGV sorting robot strictly executes the task within the time limit, the corresponding logistics speed control sequence is calculated based on the path trajectory and time window in the logistics task sequence of each AGV.

[0096] Specifically, for each path segment in a task sequence (such as the trajectory from the docking port to the sorting port), the AGV's displacement speed is dynamically set based on the ratio of the path length to the time window. For example, if a path segment is 30 meters long and the time window is 20 seconds, the AGV's theoretical speed in this segment is 1.5 meters per second. For N sets of parallel logistics task sequences at other sorting ports, the same speed control logic is used to generate corresponding logistics speed control sequences, resulting in N sets of logistics speed control sequences.

[0097] By unifying timestamps and the global coordinate system, the M main sorting port task sequences are spatiotemporally aligned with the N groups of other sorting port task sequences. Based on the aligned task data, spatial trajectory mapping is performed to detect spatiotemporal nodes where different AGVs' paths intersect or areas overlap within the same time period. For example, if AGV1 is scheduled to pass through coordinates (X1, Y1) at 08:00:10, and AGV2 passes through the same coordinates at 08:00:12, this area is marked as a potential conflict node.

[0098] For detected trajectory conflict spatiotemporal nodes, the affected connecting movement trajectories and their associated logistics task time limits are extracted from the M main sorting port task sequences. For example, if a conflict node involves AGV1's task T1 (path trajectory P1, time window 08:00:00-08:00:30), the path point set and time constraint data of P1 are extracted.

[0099] The trajectory compensation area is defined as a certain range (such as a circular area with a radius of 2 meters) extending outward from the conflict time and space node as the center. This area covers the time window (such as 08:00:10-08:00:15) when the conflict occurs and the spatial range (such as coordinates (X 1±2 ,Y 1±2 )), and serves as the operational boundary for trajectory updates.

[0100] The size of the compensation zone is dynamically set based on the AGV's braking distance and maximum steering angle to ensure the physical feasibility of the detour path. For example, if the AGV's maximum braking distance is 1.5 meters, the compensation zone radius must be at least 2 meters to provide a safety margin.

[0101] Using the multiple trajectory compensation areas as compensation space constraints, the multiple connecting movement trajectories are locally updated based on the multiple trajectory conflict spatiotemporal nodes, and multiple sets of alternative movement trajectories are output to resolve the conflicts. Each set of trajectories is generated using the A* algorithm combined with a dynamic obstacle map, avoiding conflicting paths of other AGVs, with the path length increment not exceeding 20% ​​of the original path, and the steering angle meeting the mechanical limitations of the AGV sorting robot (e.g., a maximum steering angle of 45 degrees).

[0102] The limit AGV displacement speed is obtained interactively, where Q times of the reference AGV displacement speed is the limit AGV displacement speed, Q∈[1.2,2.6].

[0103] Based on the multiple logistics task time limits and the multiple sets of alternative movement trajectories, multiple sets of alternative AGV displacement speeds are calculated. Based on the deviation scales of the multiple sets of alternative AGV displacement speeds from the limit AGV displacement speed, the multiple updated movement trajectories with the largest deviation scales from the limit AGV displacement speed are screened from the multiple sets of alternative movement trajectories.

[0104] According to the multiple trajectory conflicting spatiotemporal nodes, the multiple updated moving trajectories and the multiple updated AGV displacement speeds are overlaid onto the M parallel logistics task sequences to replace the local path segments and speed instructions corresponding to the conflicting nodes, and the M parallel logistics trajectory sequences are output.

[0105] This embodiment achieves the technical effects of improving the collaborative ability of multiple AGV sorting robots to perform multi-task parallel processing, reducing sorting path conflicts, and improving the efficiency of sorting task execution.

[0106] A700: Use the M parallel logistics trajectory sequences to schedule M AGV sorting robots to execute the M parallel logistics task sequences to perform palletizing and delivery at the first sorting port.

[0107] Specifically, in this embodiment, the scheduling module will generate the M parallel logistics trajectory sequences and send them to the M AGV sorting robots. The M AGV sorting robots will transport the logistics boxes from the docking port to the sorting port in a preset order based on the path point coordinates, timestamps and speed instructions in the corresponding M parallel logistics trajectory sequences.

[0108] When the AGV sorting robot arrives at the sorting port, it uses a robotic arm or conveyor to accurately stack the logistics boxes according to the levels and coordinates of the standard stacking template. The bottom box is placed first and stacked up layer by layer. At the same time, the stacking position is corrected in real time through pressure sensors and visual positioning systems to prevent the stack from tilting or collapsing. M AGV sorting robots work together at the sorting port to complete the delivery of all logistics boxes, forming a stable stacking structure with optimal space utilization.

[0109] This embodiment achieves the technical effect of ensuring that logistics boxes arrive and are stacked accurately in the preset order, significantly improving the stability of the stacking structure and space utilization, effectively reducing the risks of path conflicts and congestion at the sorting port, enhancing the adaptability and fault tolerance of AGV sorting robots in complex scenarios with multiple tasks running in parallel, and ensuring the reliable operation of high-throughput sorting operations.

[0110] Any of the methods or steps described above may be stored as computer instructions or programs in various types of computer memories, and the computer instructions or programs may be recognized by various types of computer processors to implement any of the methods or steps described above.

[0111] Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by those skilled in the art without departing from the principles of the present invention shall fall within the scope of patent protection of the present invention.

Claims

1. An intelligent scheduling method for AGV sorting robots with multi-task parallel processing, characterized in that: The method comprises: Interactively obtain the first to-be-palletized information of the first sorting port; Reversely analyze the first to-be-palletized information according to the standard pallet shape template to obtain the logistics box arrival order queue; Extracting the docking port location information of the automated sorting area from the first information to be palletized; Constructing a docking time-consuming topology according to the docking port position information and the first sorting position of the first sorting port; The logistics task allocation is optimized based on the docking time-consuming topology and the logistics box arrival order queue, and M parallel logistics task sequences are output, including: Based on M AGV sorting robots, the logistics box arrival order queue is decomposed to obtain multiple adjacent local arrival order queues; According to the multiple local arrival order queues, performing connection delay superposition and decomposition on the connection time-consuming topology to obtain multiple update time-consuming sub-topologies; Splitting the multiple time-consuming sub-topologies based on the M AGV sorting robots to obtain M logistics task time limit sequences; Add W docking movement trajectory mappings to the M logistics task time limit sequences, and output the M parallel logistics task sequences, where W ≥ 25M; Interactively obtain N groups of parallel logistics task sequences for the remaining N sorting ports, perform logistics conflict compensation on the N groups of parallel logistics task sequences and the M parallel logistics task sequences, and output M parallel logistics trajectory sequences, including: After the N groups of parallel logistics task sequences and the M parallel logistics task sequences are time-series aligned, spatial movement trajectory mapping is performed to locate multiple trajectory conflicting spatiotemporal nodes; Extracting a plurality of connecting movement trajectories and a plurality of logistics task time limits from the M parallel logistics task sequence mappings according to the plurality of trajectory conflict spatiotemporal nodes; Taking the multiple logistics task time limits as constraints, locally updating the multiple docking movement trajectories according to the multiple trajectory conflict spatiotemporal nodes, and outputting multiple updated movement trajectories, wherein the multiple updated movement trajectories have multiple updated AGV displacement speeds; According to the multiple trajectory conflict spatiotemporal nodes, overlay the multiple updated movement trajectories and the multiple updated AGV displacement speeds onto the M parallel logistics task sequences, and output the M parallel logistics trajectory sequences; The M parallel logistics trajectory sequences are used to schedule M AGV sorting robots to execute the M parallel logistics task sequences to perform palletizing and delivery at the first sorting port.

2. The intelligent scheduling method for the AGV sorting robot with multi-task parallel processing according to claim 1 is characterized in that: Reversely analyzing the first to-be-palletized information according to the standard pallet shape template to obtain a logistics box arrival order queue, the method comprising: Decomposing the standard pallet shape template into a pallet shape structure to obtain H palletizing level feature vectors of H virtual palletizing boxes; Extracting W logistics box feature vectors of W logistics boxes to be palletized from the first information to be palletized, where W≤H, and HW≤8; The W logistics box feature vectors are used to traverse the H palletizing level feature vectors to determine palletizing compatibility, and W palletizing space positions are obtained by screening and positioning; After the standard pallet shape template is projected onto the W palletizing space positions, palletizing level descending fitting is performed to output the logistics box arrival order queue.

3. The intelligent scheduling method for the AGV sorting robot with multi-task parallel processing according to claim 1, characterized in that: Constructing a docking time-consuming topology based on the docking port position information and the first sorting position of the first sorting port, the method comprising: Decomposing the docking port position information to obtain W docking port position information of W logistics boxes to be stacked; Constructing W docking movement trajectories based on the first sorting position and the W docking port position information; According to the benchmark AGV displacement speed and the W docking movement trajectories, W benchmark docking times are output; Construct W leaf nodes according to the W docking port location information, and construct a central node according to the first sorting position; According to the W benchmark connection time-consuming quantified topological connections, in-degree connections from the W leaf nodes to the central node are constructed to complete the construction of the connection time-consuming topology.

4. The intelligent scheduling method for the AGV sorting robot with multi-task parallel processing according to claim 3 is characterized in that: Based on the multiple local arrival order queues, performing connection delay superposition and decomposition on the connection time-consuming topology to obtain multiple update time-consuming sub-topologies, the method comprising: Decomposing the docking time-consuming topology into multiple task-consuming sub-topologies based on a mapping relationship between the multiple local arrival order queues and the W logistics boxes to be palletized in the docking time-consuming topology; Interactively obtain the arrival time interval constraint of the first sorting port; According to the multiple local arrival order queues, the arrival time interval constraint is superimposed on the multiple task time-consuming sub-topologies to obtain multiple update time-consuming sub-topologies.

5. The intelligent scheduling method for the AGV sorting robot with multi-task parallel processing according to claim 1, characterized in that: Adding the W docking movement trajectory mappings to the M logistics task time limit sequences, and outputting the M parallel logistics task sequences, the method includes: The mapping relationship between the M parallel logistics task sequences and the W logistics boxes to be stacked is combined with the W docking movement trajectories to obtain M docking task trajectory sequences; The M docking task trajectory sequences are temporally and spatially aligned with the M logistics task time limit sequences and stored to generate the M parallel logistics task sequences.

6. The intelligent scheduling method for the AGV sorting robot with multi-task parallel processing according to claim 3, characterized in that: Taking the multiple logistics task time limits as constraints, locally updating the multiple connecting movement trajectories according to the multiple trajectory conflict spatiotemporal nodes, and outputting multiple updated movement trajectories, the method includes: Frame a plurality of trajectory compensation areas at the plurality of trajectory conflict spatiotemporal nodes; Using the multiple trajectory compensation areas as compensation space limits, locally updating the multiple connecting movement trajectories according to the multiple trajectory conflict spatiotemporal nodes, and outputting multiple sets of alternative movement trajectories; Interactively obtain the ultimate AGV displacement speed; Calculating multiple sets of alternative AGV displacement speeds based on the multiple logistics task time limits and multiple sets of alternative movement trajectories; The multiple updated movement trajectories are selected from the multiple groups of candidate movement trajectories according to deviation scales between the multiple groups of candidate AGV displacement speeds and the limit AGV displacement speed.

7. The intelligent scheduling method for the AGV sorting robot with multi-task parallel processing according to claim 6, characterized in that: Q times the reference AGV displacement speed is the limit AGV displacement speed, Q∈[1.2,2.6].

8. The intelligent scheduling method for the AGV sorting robot with multi-task parallel processing according to claim 2, characterized in that: Reversely analyzing the first to-be-palletized information according to the standard pallet shape template to obtain the logistics box arrival order queue, the method previously includes: Extracting palletizing requirement-related features from the first to-be-palletized information, wherein the palletizing requirement-related features include the number of logistics boxes, the maximum base size, and the total weight of the logistics boxes; The stacking shape template is matched according to the associated features of the palletizing requirements to obtain the standard stacking shape template.

Images