Multi-robot scheduling method, device and equipment in warehouse management and storage medium

Abstract

The invention relates to the field of robots, and discloses a multi-robot scheduling method, device and equipment in warehouse management and a storage medium, and the method is used for realizing efficient conflict-free scheduling of multiple robots in warehouse management through fusion algorithm path planning and conflict prediction and processing. The method comprises the following steps: receiving an operation order, and sorting operation tasks in the operation order according to a priority and an overall path optimal principle to obtain an operation task sequence; assigning the operation tasks in the operation task sequence to the to-be-assigned robot according to the sequence, and generating a plurality of robot operation pairs; and a fusion algorithm formed by an improved artificial potential field method and a space-time A * algorithm of a model prediction mechanism is adopted to carry out path planning on the selected robot in each robot operation pair, a CBS algorithm is adopted to carry out conflict prediction, and when conflicts exist, robot operation paths are updated, and the selected robots are scheduled.

Description

Technical Field

[0001] The present invention relates to the field of robot technology, and in particular to a multi-robot scheduling method, device, equipment and storage medium in warehouse management. Background Art

[0002] The intelligent warehouse management system aims to realize the automated management of goods entering and leaving the warehouse and inventory allocation, and is composed of a business system, a central dispatching system and robots. The typical process is that managers issue tasks through the business system, and the central dispatching system arranges robots and issues tasks, which are then executed by robots. In practice, the storage space is large, and multiple robots need to work together, which requires the coordination ability of the dispatching system.

[0003] The current mainstream rule-based technical solutions in the industry (such as single-robot shortest path planning + traffic rule constraints) have the core advantages of system stability and security, but have significant limitations: Insufficient intelligence: Coordinating multiple robots only through static rules (such as one-way channels and priority avoidance) cannot dynamically respond to changes in the warehouse environment (such as sudden order surges and equipment failures); Efficiency bottleneck: Global resource allocation is not optimized, which may lead to an increase in the robot's idle rate (such as repeated path coverage and uneven task distribution); The cost of scenario migration is high: the rules need to be customized for the specific warehouse layout, and the rule base needs to be redesigned when the shelf density, aisle width or operation process is adjusted.

[0004] The centralized and distributed scheduling solutions proposed by academic research have theoretically broken through traditional limitations, but they face engineering challenges: Centralized solutions (such as constrained global path search) need to deal with NP-hard level computational complexity, and their real-time performance cannot meet the millisecond-level response requirements of warehousing scenarios. Distributed solutions (such as local rules + priority iteration) reduce computing pressure, but may fall into local optimality, especially when the robot density exceeds the threshold, deadlock may occur; Simplified model assumptions: Ignoring the robot's dynamic characteristics (such as acceleration limits, turning radius) and physical collision volume, resulting in a deviation of more than 30% between the simulation results and actual operation.

[0005] Therefore, the existing technology still needs to be improved and developed. Summary of the invention

[0006] The present invention provides a method, device, equipment and storage medium for scheduling multiple robots in warehouse management, which are used to achieve efficient and conflict-free scheduling of multiple robots in warehouse management through fusion algorithm path planning, conflict prediction and processing.

[0007] The first aspect of the present invention provides a multi-robot scheduling method in warehouse management, the multi-robot scheduling method in warehouse management comprising: receiving a job order, and sorting the job tasks in the job order according to the priority and the overall path optimal principle to obtain a job task sequence; obtaining the working status of all robots, and screening out robots with idle working status as robots to be assigned; assigning the job tasks in the job task sequence to the robots to be assigned in order, generating a plurality of robot job pairs, each of the robot job pairs including a job task and a selected robot corresponding to the job task; using a fusion algorithm formed by an improved artificial potential field method and a spatiotemporal A* algorithm of a model prediction mechanism; The selected robot in each robot operation pair performs path planning to obtain the robot operation path, and generates an initial scheduling path set based on all robot operation paths; the CBS algorithm is used to predict the conflicts of the robot operation paths in the initial scheduling path set to determine whether there are conflicts in the robot operation paths; if so, the conflict type of the robot operation path is determined, and based on the conflict type, a fusion algorithm formed by an improved artificial potential field method and a space-time A* algorithm is used to regenerate a conflict-free robot operation path, and the conflict-free robot operation path is added to the scheduling path set to obtain an updated scheduling path set; the selected robot is scheduled based on the robot operation path in the updated scheduling path set.

[0008] Preferably, a job order is received, and the job tasks in the job order are sorted according to the priority and the overall path optimal principle to obtain a job task sequence, including: receiving a job order, sorting the job tasks in the job order according to the priority of the job order to obtain a first-ordered job task; for job tasks of the same priority in the first-ordered job task, calculating the total job moving distance of each job task; based on the total job moving distance of each job task, according to the path optimal principle, sorting the job tasks of the same priority to obtain a second-ordered job task; integrating the first-ordered job task and the second-ordered job task to obtain a job task sequence.

[0009] Preferably, each of the work tasks includes at least one work assignment, and the total work moving distance of the work task is the sum of the assigned moving distances of all work assignments corresponding to the work task, and the assigned moving distance of the work assignment is expressed in any one of Manhattan distance, Euclidean distance and real distance.

[0010] Preferably, the work tasks in the work task sequence are assigned to the robots to be assigned in order of sequence to generate multiple robot work pairs, each of which includes a work task and a selected robot corresponding to the work task, including: for work tasks of different priority levels, the work tasks are assigned according to the priority levels to obtain multiple work teams; for each work team, the real distance between the robot to be assigned and the starting point of the first work assignment of each work task in the work team is calculated, the global distance is calculated according to the real distance and the total work moving distance of the corresponding work task, and based on the principle of minimizing the global distance, the work tasks are assigned to the corresponding robots to be assigned to generate robot work pairs, each of which includes a work task and a selected robot corresponding to the work task.

[0011] Preferably, the fusion algorithm formed by the improved artificial potential field method and the spatiotemporal A* algorithm of the model prediction mechanism performs path planning for the selected robot in each robot operation pair to obtain the robot operation path, and generates an initial scheduling path set according to all robot operation paths, including: for each of the robot operation pairs, determining the starting position and target operation position of the selected robot in each robot operation pair; creating an open set and a close set, the open set is used to store the node status to be evaluated, the close set is used to record the node status that has been evaluated, the close set is initially set to be empty, and a starting node status is constructed, and the starting node status is placed in the open set, and the starting node status is {(starting position coordinates, initial timestamp): heuristic value}; selecting the node status with the smallest heuristic value from the open set as best, placing best in the close set, and judging whether the node corresponding to best is the end point; if not, executing the adjacent node processing step, the adjacent node processing step includes: obtaining the adjacent node set subs of the node corresponding to best, and judging whether the adjacent node set subs is empty; if not, selecting a node from the adjacent node set subs. point, calculate the motion time from the node corresponding to best to the selected node, and generate the state of the selected node according to the calculated motion time, the state of the selected node is {(coordinates of the selected node, motion time): heuristic value}; determine whether the state of the selected node is in the close set; if not, calculate the heuristic value of the state of the selected node according to the heuristic function of the fusion algorithm formed by the improved artificial potential field method and the space-time A* algorithm, and update the state of the selected node according to the calculation result; put the updated node state into the open set, and delete the selected node from the adjacent node set subs; repeat the adjacent node processing steps until the adjacent node set subs is empty; when the adjacent node set subs is empty, return to execute the node state with the smallest heuristic value from the open set as best, put best into the close set, and determine whether the node corresponding to best is the end point, until the node corresponding to best is the end point, then reversely search the previous state in the close set, and connect the nodes corresponding to the state in the close set in sequence to obtain the robot operation path; generate an initial scheduling path set for all the robot operation paths.

[0012] Preferably, the heuristic function of the fusion algorithm formed by the improved artificial potential field method and the spatiotemporal A* algorithm is expressed as: ; In the formula, Indicates that the robot is The node at the moment, represents the robot's target node, is the gravitational coefficient, is the robot target node pair The gravitational force of the node at the moment; Indicates the obstacles on the robot's current path. is the repulsion coefficient, It's a robot All obstacles on the path at the moment The repulsion of the node at the moment; It's a robot The heuristic function of the node at the time, Indicates that the robot is The node at the moment and the robot's target node The real distance Indicates that the robot is The node at the moment and the obstacles on the robot's current path The real distance.

[0013] Preferably, the time taken to move from the node corresponding to the best to the selected node is calculated, specifically including: when the previous search state is a stationary state, and the node To its neighboring nodes If no rotation occurs, the motion time is expressed as: ; When the previous search state is static and the node To its neighboring nodes If rotation occurs, the motion time is expressed as: ; If the current node and the selected adjacent node coincide, use To determine the duration of the movement; where For Node To its neighboring nodes The translation distance, For Node To its neighboring nodes The rotation angle of is the robot translation acceleration, is the robot rotation speed.

[0014] The second aspect of the present invention provides a multi-robot scheduling device in warehouse management, including: a sorting module, which is used to receive job orders and sort the job tasks in the job orders according to the priority and the overall path optimal principle to obtain a job task sequence; a monitoring module, which is used to obtain the working status of all robots and screen out robots with idle working status as robots to be assigned; a selection module, which is used to select the job tasks in the job task sequence to the robots to be assigned in order, and generate multiple robot job pairs, each of which includes a job task and a selected robot corresponding to the job task; a path planning module, which uses a fusion algorithm formed by an improved artificial potential field method and a spatiotemporal A* algorithm of a model prediction mechanism to select the selected robot in each robot job pair. A robot is dispatched to perform path planning to obtain a robot operation path, and an initial scheduling path set is generated based on all robot operation paths; a conflict search module is used to use a CBS algorithm to perform conflict prediction on the robot operation paths in the initial scheduling path set to determine whether there is a conflict in the robot operation path; a path update module is used to determine the conflict type of the robot operation path when there is a conflict in the robot operation path, and based on the conflict type, a fusion algorithm formed by an improved artificial potential field method and a space-time A* algorithm is used to regenerate a conflict-free robot operation path, and the conflict-free robot operation path is added to the scheduling path set to obtain an updated scheduling path set; a scheduling module is used to schedule the selected robot based on the robot operation path in the updated scheduling path set.

[0015] The third aspect of the present invention provides a multi-robot scheduling device for warehouse management, comprising: a memory and at least one processor, the memory storing computer-readable instructions, the memory and the at least one processor being interconnected through lines; the at least one processor calling the computer-readable instructions in the memory so that the multi-robot scheduling device for warehouse management executes the various steps of the multi-robot scheduling method for warehouse management as described above.

[0016] A fourth aspect of the present invention provides a computer-readable storage medium, which stores computer-readable instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute the various steps of the multi-robot scheduling method in warehouse management as described above.

[0017] In the technical solution provided by the present invention, the path planning is performed for the selected robots in each robot operation pair by integrating the improved artificial potential field method and the spatiotemporal A* algorithm of the model prediction mechanism, and the dynamic changes of the robot motion space, the actual motion model and various conflict situations are taken into consideration. It is possible to plan a more reasonable and practical path for the robot in a complex storage environment, reduce the deviation between the path planning and the actual operation, and improve the safety and stability of the robot operation; moreover, the CBS algorithm is used to predict and judge the conflicts of the robot operation paths in the initial scheduling path set, and the fusion algorithm is used to re-plan the paths according to different conflict types, which can timely discover and resolve conflicts in the robot operation paths, ensure that multiple robots do not collide when working collaboratively in the storage space, and improve the reliability of the entire scheduling system. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A flowchart of a multi-robot scheduling method in warehouse management provided by an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a multi-robot scheduling device in warehouse management provided by an embodiment of the present invention; Figure 3 A schematic diagram of the structure of a multi-robot scheduling device in warehouse management provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0020] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 In an embodiment of the present invention, an embodiment of a multi-robot scheduling method in warehouse management includes: S101, receiving a job order, and sorting the job tasks in the job order according to the priority and overall path optimal principle to obtain a job task sequence; S102, obtaining the working status of all robots, and selecting robots with idle working status as robots to be assigned; S103, assigning the work tasks in the work task sequence to the robots to be assigned in order, generating a plurality of robot work pairs, each of which includes a work task and a selected robot corresponding to the work task; S104, using a fusion algorithm formed by an improved artificial potential field method and a spatiotemporal A* algorithm of a model prediction mechanism to perform path planning for the selected robots in each robot operation pair, obtain a robot operation path, and generate an initial scheduling path set based on all robot operation paths; S105, using the CBS algorithm to perform conflict prediction on the robot operation paths in the initial scheduling path set to determine whether there is a conflict in the robot operation paths; S106, if yes, then determine the conflict type of the robot operation path, and based on the conflict type, use a fusion algorithm formed by an improved artificial potential field method and a spatiotemporal A* algorithm to regenerate a conflict-free robot operation path, and add the conflict-free robot operation path to the scheduling path set to obtain an updated scheduling path set; S107. Scheduling the selected robots based on the robot operation paths in the updated scheduling path set.

[0021] It is understandable that the execution subject of the present invention may be a multi-robot scheduling device in warehouse management, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking the server as the execution subject as an example.

[0022] In this embodiment, in step S101, a job order is received, and the job tasks in the job order are sorted according to the priority and the overall path optimal principle to obtain a job task sequence, including: receiving a job order, sorting the job tasks in the job order according to the priority of the job order to obtain a first-ordered job task; for job tasks of the same priority in the first-ordered job task, calculating the total job moving distance of each job task; based on the total job moving distance of each job task, according to the path optimal principle, sorting the job tasks of the same priority to obtain a second-ordered job task; integrating the first-ordered job task and the second-ordered job task to obtain a job task sequence.

[0023] In this embodiment, each operation task includes at least one operation assignment. The assignment moving distance of the operation assignment can be represented by three methods: Manhattan distance, Euclidean distance and real distance. It can be expressed as: or or .

[0024] In the formula, Indicates the first Assignments of work, Indicates the first The starting point of a job assignment, Indicates The first task of The end point of a job assignment, represents the Manhattan distance, represents the Euclidean distance, Represents the true distance, that is, the path distance from the starting point to the target point calculated by the A* algorithm.

[0025] In this embodiment, the total operation moving distance of the operation task is the sum of the assigned moving distances of all the operation assignments of the operation task, which can be expressed as: .

[0026] Where D represents the total moving distance of the task.

[0027] In this embodiment, in step S102, the working status of the robot includes idle, moving, working, waiting for planning, charging and downtime.

[0028] In this embodiment, real-time status data (such as position, power, and task progress) is obtained through the robot communication module, and the robots to be assigned are screened out according to preset rules (such as sufficient power and no faults).

[0029] Example: Robot 1 is marked as "Charging" because its battery level is less than 20%, while Robot 2 is selected because its status is "Idle".

[0030] In this embodiment, in step S103, the work tasks in the work task sequence are assigned to the robots to be assigned in order to generate robot work pairs, including: for work tasks of different priority levels, the work tasks are assigned according to the priority levels to obtain multiple work teams; for each work team, the actual distance between the robot to be assigned and the starting point of the first work assignment of each work task in the work team is calculated, the global distance is calculated according to the actual distance and the total work moving distance of the corresponding work task, and based on the principle of minimizing the global distance, the work tasks are assigned to the corresponding robots to be assigned to generate robot work pairs.

[0031] In this embodiment, a robot operation pair refers to a pairing combination between a specific operation task and a robot responsible for executing the task in the warehouse management system. Each operation pair clearly includes an operation task and a selected robot responsible for executing the task, ensuring that each task has a corresponding execution subject.

[0032] In this embodiment, by giving priority to high-priority tasks (such as red warning orders and VIP customer needs), the on-time completion rate of key operations is ensured (SLA compliance rate is increased by 15%-25%1). For example: in medical supplies warehousing, emergency medicine orders automatically receive the highest priority. For another example, in e-commerce warehousing, orders that promise to be delivered within 2 hours have a higher priority than ordinary orders.

[0033] It is understandable that each job task has an implicit time window requirement (such as the delivery deadline and the delivery time commitment), and priority sorting is essentially a process of quantifying the urgency of the time window.

[0034] In this embodiment, in step S104, the robot conflicts (or obstacles) include four types: local dynamic node conflicts, local dynamic path conflicts, local semi-dynamic node conflicts, and static node conflicts. At the moment, a node on the robot's motion path The obstacle appears on the path and disappears after a limited time; the local dynamic path conflict refers to the At this moment, a certain section of the robot's motion path An obstacle appears on the node and disappears after a limited time. The local semi-dynamic node conflict refers to the At the moment, a node on the robot's motion path An obstacle appears on the robot and does not disappear; static node conflict refers to a node on the robot's motion path. There are always obstacles.

[0035] Obstacles on the map when searching with the spatiotemporal A* algorithm Introducing the time parameter, the local dynamic node conflict can be described as , the local dynamic path conflict can be described as , the local semi-dynamic node conflict can be described as , static node conflict can be described as If the search is performed in space and also associated with time, different types of conflicts can be handled. Therefore, the design form of the heuristic function introduces the time parameter for expansion, that is, .

[0036] In the formula, For the moment The estimated cost from the starting node to the robot's target node; for The actual distance from the starting node at the moment to the node where the robot is currently located; for The estimated distance from the robot's current node to the target node at the moment. This design can preliminarily solve the path planning problem when the motion space changes dynamically: the path search is performed in the time-space dimension. If a conflict occurs at a certain moment in the search process, the search is not performed at this timestamp, but starts from the previous timestamp. However, this design may fail to search under extreme conditions (such as narrow road avoidance scenarios) or solve an incorrect search path. The reason is that in extreme scenarios, a single node has fewer adjacent nodes that can be expanded and adjacent nodes often conflict. Therefore, it is necessary to introduce an improved artificial potential field to improve the heuristic function.

[0037] The improved artificial potential field method intervenes in the robot's movement in advance by designing a virtual force: the robot's target area will have an attractive force on it, and conversely, the obstacle area will have a repulsive force on the robot. Finally, the two forces are combined to help the robot complete the search for the optimal path.

[0038] In this embodiment, the attraction and repulsion of the artificial potential field are directly embedded in the heuristic function of the space-time A*. Therefore, the heuristic function of the fusion algorithm formed by the improved artificial potential field method and the space-time A* algorithm is expressed as: .

[0039] In the formula, Indicates that the robot is The node at the moment, represents the robot's target node, is the gravitational coefficient, is the robot target node pair The gravitational force of the node at the moment; Indicates the obstacles on the robot's current path. is the repulsion coefficient, It's a robot All obstacles on the path at the moment The repulsion of the node at the moment; It's a robot The heuristic function of the node at the time, Indicates that the robot is The node at the moment and the robot's target node The real distance Indicates that the robot is The node at the moment and the obstacles on the robot's current path The real distance.

[0040] Understandably, in the spatiotemporal A* algorithm, due to the static motion space assumption, the heuristic function is often designed in the form of Euclidean distance or Manhattan distance. However, in the composite spatiotemporal model, since the motion space will change dynamically, the Euclidean distance or Manhattan distance cannot truly describe the distance estimation from the robot's node to the target node at a certain moment of change, which will cause the state search to enter an infinite loop. Therefore, in this embodiment, the current true distance (True Distance), that is, the path distance from the starting point to the target point calculated by the spatiotemporal A* algorithm, is introduced as the distance estimate.

[0041] When there is a conflict on the robot's planned path, the heuristic function gives different responses according to the different forms of the conflict: for local dynamic node conflicts and local dynamic path conflicts, the heuristic function guides the robot to search for a path in the direction of avoiding the conflict when the conflict occurs, and recovers when the conflict disappears; for local semi-dynamic node conflicts, the heuristic function guides the robot to search for a path that completely avoids the conflict after the conflict occurs, forcing the robot to turn to the direction of avoiding the conflict after the conflict occurs and search for a path without returning; for static node conflicts, the heuristic function will guide the robot to search for a path that completely avoids the conflict, avoiding collision with it from beginning to end.

[0042] Furthermore, in order to improve the applicability, the fusion algorithm formed by improving the artificial potential field method and the space-time A* algorithm of the model prediction mechanism introduces the model prediction mechanism to expand the discrete time into continuous time. Specifically, in the process of searching the adjacent nodes of each spatial node, the motion prediction is performed according to the distance between the adjacent nodes, whether the adjacent nodes rotate, the current motion of the robot and the actual motion model of the robot.

[0043] Motion prediction based on the distance between adjacent nodes, whether rotation occurs between adjacent nodes, the current motion of the robot, and the actual robot motion model includes the following situations: When the previous search state is a uniform motion state, and the node To its neighboring nodes If no rotation occurs, the motion time is expressed as: .

[0044] When the previous search state is a uniform motion state, and the node To its neighboring nodes If rotation occurs, the motion time is expressed as: .

[0045] When the previous search state is static and the node To its neighboring nodes If no rotation occurs, the motion time is expressed as: .

[0046] When the previous search state is static and the node To its neighboring nodes If rotation occurs, the motion time is expressed as: .

[0047] When the previous search state is static and the node To its neighboring nodes No displacement occurs, that is When , the movement time is expressed as: .

[0048] In the formula, For Node To its neighboring nodes The translation distance, For Node To its neighboring nodes The rotation angle of is the robot translation speed, is the robot translation acceleration, is the robot rotation speed.

[0049] In this embodiment, a fusion algorithm formed by an improved artificial potential field method and a spatiotemporal A* algorithm of a model prediction mechanism is used to perform path planning for the selected robots in each robot operation pair, obtain the robot operation path, and generate an initial scheduling path set based on all robot operation paths, including: For each robot operation pair, determining a starting position and a target operation position of the selected robot in each robot operation pair; Create an open set and a close set. The open set is used to store the node states to be evaluated, and the close set is used to record the node states that have been evaluated. The close set is initially set to empty, and the starting node state is constructed and placed in the open set. The starting node state is {(starting position coordinates, initial timestamp): heuristic value}; Select the node state with the smallest heuristic value from the open set as the best, put the best into the close set, and determine whether the node corresponding to the best is the end point; If not, then execute the adjacent node processing step, which includes: obtaining the adjacent node set subs of the node corresponding to best, and judging whether the adjacent node set subs is empty; if not, then select a node from the adjacent node set subs, calculate the motion time from the node corresponding to best to the selected node, and generate the state of the selected node according to the calculated motion time, the state of the selected node is {(coordinates of the selected node, motion time): heuristic value}; judge whether the state of the selected node is in the close set; if not, calculate the heuristic value of the state of the selected node according to the heuristic function of the fusion algorithm formed by the improved artificial potential field method and the space-time A* algorithm, and update the state of the selected node according to the calculation result; put the updated node state into the open set, and delete the selected node from the adjacent node set subs; Repeat the adjacent node processing steps until the adjacent node set subs is empty; When the adjacent node set is empty, return to execute and select the node state with the smallest heuristic value from the open set as the best, put the best into the close set, and determine whether the node corresponding to the best is the end point. Until the node corresponding to the best is the end point, reverse search the previous state in the close set, and connect the nodes corresponding to the state in the close set in order to obtain the robot operation path; Generate an initial scheduling path set from all robot operation paths.

[0050] In this embodiment, the heuristic value in the node state is calculated according to the heuristic function of the fusion algorithm formed by the improved artificial potential field method and the spatiotemporal A* algorithm.

[0051] In this embodiment, when the path planning search begins, since it starts from the starting node, it can be considered that the previous search state is a static state. At this time, the prediction formula is selected based on whether rotation occurs between adjacent nodes. As described above: When the previous search state is static and the node To its neighboring nodes If no rotation occurs, the motion time is expressed as: .

[0052] When the previous search state is static and the node To its neighboring nodes If rotation occurs, the motion time is expressed as: .

[0053] If the current node and the selected adjacent node overlap, In the case of, regardless of the previous search status, directly use To determine the duration of the exercise.

[0054] In this embodiment, in step S105, in this embodiment, the conflict tree (CT) search method in the CBS algorithm is adopted, that is, based on the best-first search strategy, a priority mechanism is introduced, and according to the urgency of the task, the conflict branch where the robot corresponding to the task with a high priority is located is searched first.

[0055] In this embodiment, when the CBS algorithm is used to predict the conflicts of the robot operation paths in the initial scheduling path set, it is determined whether the robot operation paths have the same position nodes in the same time period. If so, it is determined that there is a conflict in the robot operation paths. If not, it is determined that there is no conflict in the robot operation paths.

[0056] The CBS algorithm simulates the operation of each robot along its own working path. It divides time into discrete time steps, and in each time step, checks the position of all robots. If at a certain moment, two or more robots are at the same position node, or their paths are about to cross, this detects a conflict. For example, robot A reaches the coordinate (10, 10) in the 5th time step, and robot B also plans to reach the coordinate in the 5th time step, which constitutes a conflict. By checking the position and path direction of all robots in this way, the CBS algorithm can comprehensively predict the robot's working path and obtain the operation status of the robot's working path. The operation status of the robot's working path records in detail the time, location, and robot number involved in the possible conflict.

[0057] In this embodiment, the conflict types include local dynamic node conflict, local dynamic path conflict, local semi-dynamic node conflict and static node conflict. When there is a conflict on the robot's planned path, the heuristic function gives different responses according to the different forms of the conflict: for local dynamic node conflict and local dynamic path conflict, the heuristic function guides the robot to search for a path in the direction of avoiding the conflict when the conflict occurs, and recovers when the conflict disappears; for local semi-dynamic node conflict, the heuristic function guides the robot to search for a path that completely avoids the conflict after the conflict occurs, forcing the robot to turn to the direction of avoiding the conflict after the conflict occurs and search for a path without returning; for static node conflict, the heuristic function will guide the robot to search for a path that completely avoids the conflict, avoiding collision with it from beginning to end.

[0058] Specifically, for different types of conflicts, the fusion algorithm formed by the improved artificial potential field method and the space-time A algorithm starts from the node near the location where the conflict occurs, and continuously explores new nodes, calculates the movement time, and evaluates the cost of the new path according to the heuristic function and search strategy. During the search process, various factors are comprehensively considered, such as obstacles, the movement trajectories of other robots, etc., and finally a new conflict-free path is generated, that is, a conflict-free robot operation path.

[0059] In this embodiment, in step S107, the robot operation path in the updated scheduling path set is sent to the robot, and the execution process is monitored.

[0060] Specifically, the path sequence (coordinates + timestamp) is transmitted to the robot through the communication module, and the robot's position, speed, and task progress are tracked to dynamically handle emergencies (such as the addition of obstacles).

[0061] If the robot deviates from the path or fails, the replanning process is triggered. For example, if robot C suspends its mission due to low battery, the system reassigns another robot to take over.

[0062] The present embodiment provides a multi-robot scheduling method in warehouse management, which performs path planning for the selected robots in each robot operation pair by integrating the space-time A* algorithm of the improved artificial potential field method and the model prediction mechanism, taking into account the dynamic changes of the robot motion space, the actual motion model and various conflict situations, and can plan a more reasonable and practical path for the robot in a complex warehouse environment, reduce the deviation between the path planning and the actual operation, and improve the safety and stability of the robot operation; moreover, the CBS algorithm is used to predict and judge the conflicts of the robot operation paths in the initial scheduling path set, and the fusion algorithm is used to re-plan the paths according to different conflict types, which can timely discover and resolve conflicts in the robot operation paths, ensure that multiple robots do not collide when working together in the warehouse space, and improve the reliability of the entire scheduling system.

[0063] The above describes the multi-robot scheduling method in warehouse management in the embodiment of the present invention. The following describes the device in the embodiment of the present invention. Figure 2 , the implementation of the multi-robot scheduling device in warehouse management in the embodiment of the present invention includes: The sorting module 201 is used to receive the job order and sort the job tasks in the job order according to the priority and the overall path optimal principle to obtain a job task sequence; The monitoring module 202 is used to obtain the working status of all robots and select robots with idle working status as robots to be assigned; The selection module 203 is used to select the work tasks in the work task sequence to the robots to be assigned in order, and generate a plurality of robot work pairs, each of which includes a work task and a selected robot corresponding to the work task; The path planning module 204 is used to perform path planning for the selected robots in each robot operation pair by using a fusion algorithm formed by the improved artificial potential field method and the spatiotemporal A* algorithm of the model prediction mechanism, obtain the robot operation path, and generate an initial scheduling path set according to all robot operation paths; The conflict search module 205 is used to use the CBS algorithm to perform conflict prediction on the robot operation paths in the initial scheduling path set to determine whether there is a conflict in the robot operation paths; The path updating module 206 is used to determine the conflict type of the robot operation path when there is a conflict in the robot operation path, and regenerate a conflict-free robot operation path based on the conflict type by using a fusion algorithm formed by an improved artificial potential field method and a spatiotemporal A* algorithm, and add the conflict-free robot operation path to the scheduling path set to obtain an updated scheduling path set; The scheduling module 207 is used to schedule the selected robots based on the robot operation paths in the updated scheduling path set.

[0064] In this embodiment, the path planning of the selected robot in each robot operation pair is carried out by integrating the space-time A* algorithm which improves the artificial potential field method and the model prediction mechanism, taking into account the dynamic changes of the robot motion space, the actual motion model and various conflict situations, so that a more reasonable and practical path can be planned for the robot in a complex storage environment, the deviation between the path planning and the actual operation can be reduced, and the safety and stability of the robot operation can be improved; moreover, the CBS algorithm is used to predict and judge the conflicts of the robot operation paths in the initial scheduling path set, and the fusion algorithm is used to re-plan the paths according to different conflict types, so that the conflicts in the robot operation paths can be discovered and resolved in time, ensuring that multiple robots will not collide when working together in the storage space, thereby improving the reliability of the entire scheduling system.

[0065] Figure 2 The structure of the multi-robot scheduling device in warehouse management shown does not constitute a limitation of the multi-robot scheduling device in warehouse management, and can implement the steps of the multi-robot scheduling method in warehouse management provided by the above-mentioned method embodiments.

[0066] above Figure 2 The multi-robot scheduling device for warehouse management in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The multi-robot scheduling device for warehouse management in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0067] Figure 3 It is a structural diagram of a multi-robot scheduling device in warehouse management provided by an embodiment of the present invention. The device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage devices) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 can be short-term storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the device 300. Furthermore, the processor 310 can be configured to communicate with the storage medium 330 and execute a series of instruction operations in the storage medium on the device 300.

[0068] The device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc.

[0069] An embodiment of the present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of a multi-robot scheduling method in warehouse management.

[0070] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0071] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0072] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-robot scheduling method in warehouse management, characterized in that: The multi-robot scheduling method in warehouse management includes: Receive job orders, and sort the job tasks in the job orders according to the priority and overall path optimal principle to obtain a job task sequence; Get the working status of all robots and select robots with idle working status as robots to be assigned; Assigning the work tasks in the work task sequence to the robots to be assigned in order to generate a plurality of robot work pairs, each of which includes a work task and a selected robot corresponding to the work task; The fusion algorithm formed by the improved artificial potential field method and the spatiotemporal A* algorithm of the model prediction mechanism is used to plan the path of the selected robot in each robot operation pair, obtain the robot operation path, and generate the initial scheduling path set based on all robot operation paths; The CBS algorithm is used to predict the conflicts of the robot operation paths in the initial scheduling path set to determine whether there are conflicts in the robot operation paths; If so, the conflict type of the robot operation path is determined, and based on the conflict type, a fusion algorithm formed by the improved artificial potential field method and the space-time A* algorithm is used to regenerate a conflict-free robot operation path, and the conflict-free robot operation path is added to the scheduling path set to obtain an updated scheduling path set; The selected robots are scheduled based on the robot operation paths in the updated scheduling path set.

2. The multi-robot scheduling method in warehouse management according to claim 1, characterized in that: Receive the job order and sort the job tasks in the job order according to the priority and overall path optimization principle to obtain the job task sequence, including: Receive the job order, sort the job tasks in the job order according to the priority of the job order, and obtain a sorted job task; For the tasks with the same priority in a sorted task, calculate the total moving distance of each task; Based on the total moving distance of each task, the tasks of the same priority are sorted according to the optimal path principle to obtain secondary sorted tasks; Integrate the first-order sorting job tasks and the second-order sorting job tasks to obtain a job task sequence.

3. The multi-robot scheduling method in warehouse management according to claim 2 is characterized in that: Each of the work tasks includes at least one work assignment, and the total work moving distance of the work task is the sum of the assigned moving distances of all work assignments corresponding to the work task, and the assigned moving distance of the work assignment is represented by any one of Manhattan distance, Euclidean distance and real distance.

4. The multi-robot scheduling method in warehouse management according to claim 1, characterized in that: The process of assigning the work tasks in the work task sequence to the robots to be assigned in order to generate a plurality of robot work pairs, each of which includes a work task and a selected robot corresponding to the work task, comprises: For tasks with different priority levels, the tasks are assigned according to the priority levels to obtain multiple work teams; For each work team, the actual distance between the robot to be assigned and the starting point of the first work assignment of each work task in the work team is calculated, and the global distance is calculated according to the actual distance and the total work moving distance of the corresponding work task. Based on the principle of minimizing the global distance, the work task is assigned to the corresponding robot to be assigned to generate a robot work pair, each of which includes a work task and a selected robot corresponding to the work task.

5. The multi-robot scheduling method in warehouse management according to claim 1, characterized in that: The fusion algorithm formed by the improved artificial potential field method and the spatiotemporal A* algorithm of the model prediction mechanism performs path planning for the selected robots in each robot operation pair to obtain the robot operation path, and generates an initial scheduling path set according to all robot operation paths, including: For each of the robot operation pairs, determining a starting position and a target operation position of the selected robot in each robot operation pair; Create an open set and a close set. The open set is used to store the node states to be evaluated, and the close set is used to record the node states that have been evaluated. The close set is initially set to be empty, and a starting node state is constructed. The starting node state is placed in the open set. The starting node state is {(starting position coordinates, initial timestamp): heuristic value}; Select the node state with the smallest heuristic value from the open set as the best, put the best into the close set, and determine whether the node corresponding to the best is the end point; If not, then execute the adjacent node processing step, which includes: obtaining the adjacent node set subs of the node corresponding to best, and judging whether the adjacent node set subs is empty; if not, selecting a node from the adjacent node set subs, calculating the movement time from the node corresponding to best to the selected node, and generating the state of the selected node according to the calculated movement time, the state of the selected node is {(coordinates of the selected node, movement time): heuristic value}; judging whether the state of the selected node is in the close set; if not, calculating the heuristic value of the state of the selected node according to the heuristic function of the fusion algorithm formed by the improved artificial potential field method and the space-time A* algorithm, and updating the state of the selected node according to the calculation result; putting the updated node state into the open set, and deleting the selected node from the adjacent node set subs; Repeat the adjacent node processing steps until the adjacent node set subs is empty; When the adjacent node set subs is empty, return to execute to select the node state with the smallest heuristic value from the open set as the best, put the best into the close set, and determine whether the node corresponding to the best is the end point, until the node corresponding to the best is the end point, then reversely search the previous state in the close set, and connect the nodes corresponding to the states in the close set in order to obtain the robot operation path; All the robot operation paths are used to generate an initial scheduling path set.

6. The multi-robot scheduling method in warehouse management according to claim 5, characterized in that: The heuristic function of the fusion algorithm formed by the improved artificial potential field method and the space-time A* algorithm is expressed as: ; In the formula, Indicates that the robot is The node at the moment, represents the robot's target node, is the gravitational coefficient, is the robot target node pair The gravitational force of the node at the moment; Indicates the obstacles on the robot's current path. is the repulsion coefficient, It's a robot All obstacles on the path at the moment The repulsion of the node at the moment; It's a robot The heuristic function of the node at the time, Indicates that the robot is The node at the moment and the robot's target node The real distance Indicates that the robot is The node at the moment and the obstacles on the robot's current path The real distance.

7. The multi-robot scheduling method in warehouse management according to claim 6, characterized in that: Calculate the movement time from the best corresponding node to the selected node, including: When the previous search state is static and the node To its neighboring nodes If no rotation occurs, the motion time is expressed as: ; When the previous search state is static and the node To its neighboring nodes If rotation occurs, the motion time is expressed as: ; If the current node and the selected adjacent node coincide, use To determine the duration of the movement; where For Node To its neighboring nodes The translation distance, For Node To its neighboring nodes The rotation angle of is the robot translation acceleration, is the robot rotation speed.

8. A multi-robot scheduling device in warehouse management, characterized in that: include: A sorting module is used to receive job orders and sort the job tasks in the job orders according to the priority and overall path optimization principle to obtain a job task sequence; The monitoring module is used to obtain the working status of all robots and select robots with idle working status as robots to be assigned; A selection module is used to select the work tasks in the work task sequence to the robots to be assigned in order, and generate a plurality of robot work pairs, each of which includes a work task and a selected robot corresponding to the work task; The path planning module is used to plan the path of the selected robot in each robot operation pair by using a fusion algorithm formed by the improved artificial potential field method and the spatiotemporal A* algorithm of the model prediction mechanism, obtain the robot operation path, and generate an initial scheduling path set based on all robot operation paths; The conflict search module is used to use the CBS algorithm to predict the conflicts of the robot operation paths in the initial scheduling path set to determine whether there are conflicts in the robot operation paths; The path updating module is used to determine the conflict type of the robot operation path when there is a conflict in the robot operation path, and regenerate a conflict-free robot operation path based on the conflict type by using a fusion algorithm formed by an improved artificial potential field method and a spatiotemporal A* algorithm, and add the conflict-free robot operation path to the scheduling path set to obtain an updated scheduling path set; The scheduling module is used to schedule the selected robots based on the robot operation paths in the updated scheduling path set.

9. A multi-robot scheduling device in warehouse management, characterized in that: comprising a memory and at least one processor, wherein the memory has computer-readable instructions stored therein; The at least one processor calls the computer-readable instructions in the memory to execute the various steps of the multi-robot scheduling method in warehouse management as described in any one of claims 1-7.

10. A computer-readable storage medium having computer-readable instructions stored thereon, characterized in that: When the computer-readable instructions are executed by a processor, the various steps of the multi-robot scheduling method in warehouse management as described in any one of claims 1-7 are implemented.

Images