A multi-agv cooperative scheduling method based on heuristic strategy

Abstract

The application relates to the technical field of assembly platforms, and discloses a multi-AGV cooperative scheduling method based on a heuristic strategy, which comprises the following steps: a multi-AGV cooperative scheduling mathematical model is established based on constraint conditions, the optimization target of which is to minimize the maximum completion time; the multi-AGV cooperative scheduling mathematical model selects an available workpiece set and an available AGV set based on a cost rule, and quickly matches an optimal execution combination and distributes the optimal execution combination to an AGV; and a multi-factor improved A* algorithm is cited. Through a task allocation strategy based on a task priority and a vehicle response cost, the application can realize the quick matching between workpieces and AGVs on the basis of considering the task priority of the workpieces, the available state of the AGVs and the distance cost of the vehicles, and simultaneously adopts the improved A* algorithm to carry out path planning; in the path search process, a turning cost, a deadlock factor and a future road network congestion cost are introduced, so that the AGV planning path is more in line with actual operation requirements.

Description

Technical Field

[0001] This invention relates to the field of AGV collaborative scheduling technology, specifically a multi-AGV collaborative scheduling method based on heuristic strategies. Background Technology

[0002] With the rapid development of intelligent manufacturing and flexible production models, automated guided vehicles (AGVs) have been widely used in flexible manufacturing, intelligent warehousing, and automated logistics. They are gradually becoming important logistics and transportation equipment connecting processing equipment, warehousing units, and assembly units in modern manufacturing systems. Compared with traditional manual handling methods, AGV systems can effectively reduce labor costs, improve material handling efficiency, reduce human error, and enhance the continuity and stability of production system operation. Therefore, the performance of multi-AGV scheduling systems will directly affect the production cycle, equipment utilization rate, and overall operating efficiency of flexible manufacturing systems.

[0003] The multi-AGV collaborative scheduling problem involves how to allocate tasks to vehicles, how vehicles plan routes in a shared road network, how to avoid conflicts, and how to maintain stable system operation. In real-world production environments, when multiple AGVs operate simultaneously, resource competition and path conflicts easily occur at intersections, narrow passages, and shared path areas, potentially leading to vehicle waiting and system congestion. Therefore, establishing an effective collaborative mechanism between task allocation and path planning is a crucial issue in multi-AGV scheduling research. Existing heuristic algorithms offer advantages such as high computational efficiency, clear rules, and ease of implementation in path planning and task allocation; however, traditional methods struggle to fully consider path costs, turning costs, congestion, and conflict risks during vehicle operation. Furthermore, in actual multi-AGV operation, vehicles do not occupy only a single path node. Due to the AGV's size and the need to maintain safe distances during operation, their actual occupied area often covers multiple path nodes or detection points. If the scheduling system still uses a single-point occupancy method for conflict detection, it may underestimate the spatial interference between vehicles, leading to path conflicts, congestion, and deadlock problems. Meanwhile, in flexible operation workshops, there may be disturbances such as machine failures, vehicle derailment, and dynamic changes in tasks, which further increases the complexity of multi-AGV collaborative scheduling.

[0004] Based on this, this invention proposes a multi-AGV cooperative scheduling method based on a heuristic strategy. This method is designed for flexible workshop environments, employing a task allocation strategy based on task priority and vehicle response cost. It combines this with an improved A* path planning method incorporating turning costs, deadlock factors, and future road network congestion costs. Furthermore, it constructs a conflict detection mechanism based on AGV body size and safety distance, as well as a deadlock recovery strategy. This achieves collaborative optimization among task allocation, path planning, conflict detection, and deadlock recovery, thereby improving the operational efficiency and stability of multi-AGV systems under complex dynamic conditions. Summary of the Invention

[0005] This invention provides a multi-AGV collaborative scheduling method based on heuristic strategies, which has the beneficial effects of effectively reducing AGV empty driving distance, reducing path conflicts and deadlock occurrences.

[0006] This invention provides the following technical solution: a multi-AGV cooperative scheduling method based on a heuristic strategy, comprising the following steps:

[0007] S1. Establish a mathematical model for multi-AGV collaborative scheduling based on constraints, with the optimization objective of minimizing the maximum completion time;

[0008] S2. The multi-AGV collaborative scheduling mathematical model selects the set of available workpieces and the set of available AGVs based on cost rules, and quickly matches the optimal execution combination and assigns tasks to AGVs.

[0009] S3. Using the improved A* algorithm with multiple factors, calculate and allocate the AGV planning path, and output the planning path from the current point to the target node;

[0010] S4. Based on the actual physical size conflict detection and deadlock recovery in the planned path, optimize and adjust the planned path to complete the multi-AGV collaborative scheduling.

[0011] As an optional solution to the heuristic-based multi-AGV cooperative scheduling method of the present invention, the method for task allocation based on cost rules includes:

[0012] Real-time identification of the set of available workpieces that are not yet processed and the set of available AGVs that are in a ready state;

[0013] Based on priority, the most urgent workpiece and the candidate AGV that meets the map matching are selected from the available workpiece set to generate a transportation task;

[0014] Calculate the overall cost for the candidate AGV to perform the transportation task;

[0015] Select the AGV with the lowest overall cost to perform the task and update the system status.

[0016] As an optional solution to the heuristic-based multi-AGV cooperative scheduling method described in this invention, the formula for calculating the comprehensive cost is as follows:

[0017]

[0018] in, The distance cost for the AGV to reach the starting point of the task;

[0019] The cost of path risk when the AGV performs this task;

[0020] and These are the weighting coefficients;

[0021] in, + =1.

[0022] As an optional solution to the multi-AGV cooperative scheduling method based on heuristic strategy described in this invention, the multi-factors include determining the starting point and ending point of path planning based on comprehensive cost, the future road network congestion coefficient, the deadlock factor, and the turning penalty coefficient.

[0023] The comprehensive evaluation function calculation formula for the multi-factor improvement A* includes:

[0024]

[0025] in, From the starting point to the current path node The actual path cost;

[0026] This represents the distance to the current candidate road segment;

[0027] This represents the congestion coefficient of the current candidate road segment;

[0028] This is the turning penalty coefficient. As a deadlock factor, Let be the Manhattan distance from the candidate child node to the target node. The deadlock factor weights.

[0029] As an optional solution to the heuristic-based multi-AGV cooperative scheduling method described in this invention, it further includes:

[0030] A path network is established based on the planned path, and the path network includes a number of path nodes;

[0031] Based on the longest side and safe distance of the candidate AGV, generate a set of detection points for each path node;

[0032] When a candidate AGV is located at an occupied path node, multiple detection points within the safe range of that path node are locked simultaneously to prevent spatial interference between candidate AGV entities.

[0033] Determine whether the next path node of the current candidate AGV is occupied by another AGV;

[0034] If the AGV is occupied by another AGV, check for deadlock and perform deadlock recovery.

[0035] As an optional solution to the heuristic-based multi-AGV cooperative scheduling method described in this invention, the deadlock recovery method includes:

[0036] The AGV that triggers the detection is marked as the first AGV, and the AGV that blocks its path is marked as the second AGV;

[0037] Starting from the second AGV, a breadth-first search is used to find an avoidance point, which satisfies the constraints that it is not on the path of the first AGV and does not cause new blockages.

[0038] Control the second AGV to move to the avoidance point, and release the path for the first AGV to pass.

[0039] As an optional solution to the multi-AGV cooperative scheduling method based on heuristic strategy described in this invention, the deadlock recovery method recovers at least one type of deadlock;

[0040] The deadlock types include circular deadlock, circular deadlock with detection points, opposing deadlock, opposing deadlock with detection points, and idle deadlock.

[0041] As an optional solution to the heuristic-based multi-AGV cooperative scheduling method of the present invention, the constraints include:

[0042] The same process for the same workpiece can only be assigned to one machine for processing and one AGV for transportation;

[0043] Only one AGV can occupy a node or an edge at the same time.

[0044] The processing of the workpiece shall not begin earlier than the time when the AGV transports it to the machine.

[0045] As an alternative to the heuristic-based multi-AGV cooperative scheduling method described in this invention, for any path node in the path network... The method for calculating the future road network congestion coefficient includes:

[0046] Count the total number of times each of the remaining planned path nodes for all AGVs appears, and the total number of path nodes involved in the remaining planned paths for all AGVs;

[0047] Calculate path nodes The initial congestion rate;

[0048] when season =0.99, based on the corrected The node correction parameters for the path nodes are calculated using the following formula:

[0049]

[0050] Finally, the extension nodes on this road segment were calculated similarly. Node correction parameters ;

[0051] The final formula for the congestion correction factor for this road section is:

[0052]

[0053] The congestion correction coefficient is then used as a weighting factor and incorporated into the comprehensive evaluation function of path planning to dynamically adjust the priority of path selection in order to avoid highly congested areas.

[0054] As an optional solution to the heuristic-based multi-AGV cooperative scheduling method described in this invention, the method for calculating the deadlock factor for any pair of adjacent nodes in the current AGV's planned path includes:

[0055] Get the set of remaining planned paths for all other AGVs. ;

[0056] Traverse each remaining path and determine if there are any adjacent node pairs. satisfy and This means that other AGVs are running in the opposite direction to the current AGV on this section of the road;

[0057] The number of AGVs that meet the above conditions is recorded as the number of AGVs that may have a risk of reciprocal conflict. ;

[0058] Set deadlock factor penalty coefficient ;

[0059] The formula for calculating the deadlock factor of this node pair is:

[0060] .

[0061] As an optional solution to the heuristic-based multi-AGV cooperative scheduling method described in this invention, the turning coefficient is calculated for two consecutive paths from the parent node through path node u to path node v in the current AGV planned path, including:

[0062] Construct the direction vector of the first path segment ;

[0063] Construct the direction vector of the second path segment ;

[0064] Calculate the cross product of two vectors ;

[0065] The turning direction is determined based on the cross product sign, and a turning coefficient is assigned. :

[0066]

[0067] The turning coefficient is then incorporated as an additional term in the path cost into the evaluation function of the improved A* algorithm to suppress frequent turning behavior and improve path smoothness and running efficiency.

[0068] As an optional solution to the heuristic-based multi-AGV cooperative scheduling method described in this invention, it includes the following steps:

[0069] The simulation environment is constructed based on the path network, AGV set, machine set, and workpiece set of the path node set and path edge set.

[0070] During the simulation, the AGV is controlled to run according to kinematic parameters including starting acceleration, constant speed and deceleration in front of the target node, and the current position, running status, target node list, remaining planned path, node occupancy status and detection point occupancy status of the AGV are updated in real time.

[0071] After the simulation, the completion time of each task is recorded, the maximum completion time of each task is calculated as the maximum completion time of the system, and the scheduling results are output to verify the effectiveness of the scheduling method.

[0072] The present invention has the following beneficial effects:

[0073] 1. This heuristic-based multi-AGV collaborative scheduling method, through a task allocation strategy based on task priority and vehicle response cost, can achieve rapid matching between workpieces and AGVs, taking into account workpiece task priority, AGV availability status, and vehicle distance cost. At the same time, it adopts an improved A* algorithm for path planning, introducing turning cost, deadlock factor, and future road network congestion cost during the path search process, making the AGV planned path more in line with actual operation requirements.

[0074] 2. This heuristic-based multi-AGV collaborative scheduling method constructs a conflict detection mechanism based on AGV body size and safety distance, and designs deadlock recovery strategies for circular deadlocks, circular deadlocks with detection points, opposing deadlocks, opposing deadlocks with detection points, and idle deadlocks. This method features clear rules, low implementation complexity, and strong real-time performance. It can effectively reduce AGV empty travel distance, decrease path conflicts and deadlocks, and improve the operating efficiency and stability of multi-AGV systems. Attached Figure Description

[0075] Figure 1 This is a flowchart of the multi-AGV collaborative scheduling method of the present invention.

[0076] Figure 2 This is a schematic diagram of the flexible manufacturing workshop of the present invention.

[0077] Figure 3 This is a schematic diagram of the deadlock recovery process of the present invention. Detailed Implementation

[0078] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0079] Example

[0080] Please see Figures 1-3 One heuristic-based multi-AGV collaborative scheduling method includes the following steps:

[0081] S1. Establish a mathematical model for multi-AGV collaborative scheduling based on constraints, with the optimization objective of minimizing the maximum completion time.

[0082] Specifically:

[0083] Let the set of workpieces i in the factory be . ;

[0084] Process The set is ;

[0085] The first workpiece i Each process is recorded as ;

[0086] The set of machines k is ;

[0087] The set of AGVg is ;

[0088] The set of nodes n is ;

[0089] The path e set is ;

[0090] Where u, f, w, and n represent the number of workpieces, machines, AGVs, and nodes, respectively;

[0091] Process The candidate machine set is ;

[0092] Process The processing time on machine k is ;

[0093] The time of arrival at machine k is ;

[0094] Start processing time is and the end processing time is ;

[0095] Furthermore, the decision variables in the multi-AGV collaborative scheduling problem in a flexible manufacturing environment are defined as follows:

[0096]

[0097]

[0098]

[0099]

[0100]

[0101]

[0102]

[0103]

[0104]

[0105]

[0106] in a A;

[0107] Based on the above, a fitness function is constructed with the objective of minimizing the maximum completion time, as follows:

[0108]

[0109] The constraints in the multi-AGV collaborative scheduling problem in a flexible manufacturing environment are expressed as follows:

[0110]

[0111] The above indicates that the same process for the same workpiece can only be assigned to one machine for processing;

[0112]

[0113] The above indicates that the same process for the same workpiece can only be assigned to one AGV for transportation;

[0114]

[0115] The above indicates that only one AGV can occupy a node at any given time.

[0116]

[0117] The above indicates that only one AGV can occupy an edge at any given time.

[0118]

[0119] The above means that the processing time of the workpiece cannot be earlier than the time when the workpiece arrives at the machine;

[0120] The mathematical model for the multi-AGV collaborative scheduling problem in the flexible manufacturing environment established in this step takes into account multiple constraints such as AGVs, workpieces, processes, and machine processing, which is more in line with the actual situation of AGV scheduling in enterprises.

[0121] A heuristic-based multi-AGV cooperative scheduling method is used to solve the workshop conditions of specific applications. Figure 2 For specific implementation details, please refer to the AGV status table, workpiece example table, and process example table:

[0122] Table 1 AGV Status Table

[0123]

[0124] Table 2. Examples of Workpieces

[0125]

[0126] Table 3. Process Examples

[0127]

[0128] S2. The multi-AGV collaborative scheduling mathematical model selects the set of available workpieces and the set of available AGVs based on cost rules, and quickly matches the optimal execution combination and assigns tasks to AGVs.

[0129] Specifically:

[0130] S201. Determine the current set of allocable workpieces and AGVs:

[0131] Suppose The set of workpieces available at any given time is ;

[0132] The available AGV set is ;

[0133] in Indicates at time A collection of workpieces that are not in the process of processing or transportation and that have not been fully processed. Indicates at time A collection of AGVs that are ready to accept new tasks.

[0134] Based on the above set, first from Select the most urgent tasks based on task priority. , and then from The selection process involves filtering a set of candidate AGVs that meet the criteria for map matching and task allocation. .

[0135] S202. Calculate the overall cost of the AGV performing the task:

[0136] for For all AGVs, calculate the total cost of performing the current task;

[0137] Let AGV exist The location of the moment is ,in workpiece The starting point of the task is The mission endpoint is The distance cost for the AGV to reach the starting point of the task is... ,in This represents the shortest path distance calculated based on the workshop route.

[0138] This represents the path risk cost when the AGV performs this task, calculated from the turning cost, deadlock factor, and congestion coefficient of the improved A* algorithm in S3, representing the comprehensive cost. , and Let be the weighting coefficient, satisfying .

[0139] S203. Determine the task allocation results:

[0140] Select the AGV with the lowest overall cost to perform the workpiece assignment. The transportation task, namely The result of task allocation is represented as , indicating workpiece The transportation task was assigned to the AGV After execution and completion of allocation, update the AGV status, workpiece transportation status, and task pool status.

[0141] S3. Using the improved A* algorithm with multiple factors, calculate and allocate the AGV planning path, and output the planning path from the current point to the target node.

[0142] Specifically:

[0143] S301. Confirm the starting and ending points of the route planning:

[0144] Assume a flexible workshop network as ;

[0145] in Represents a set of path nodes. Represents the set of path edges;

[0146] Let AGV exist The location of the moment is The current target node is determined by the task allocation result of S2. The loading and unloading points of the selected workpieces are added to the AGV's target node list. The target node is the first node in the AGV's target node list, set as... , and perform path planning.

[0147] S302, Traversing A* Algorithm Nodes:

[0148] During the A* algorithm's path traversal, let the currently traversed node be... Expand child nodes as .

[0149] S303, Calculate the future road network congestion coefficient:

[0150] node The number of times it appears in all remaining planned paths of AGV is The total number of nodes on all remaining paths of all AGVs is Then the node The congestion rate ,when season Define nodes based on their congestion coefficients. The node correction parameters are Congestion correction coefficient for output nodes Similarly, calculate the extended child nodes. The node congestion correction factor is Select the congestion coefficient for this road segment. .

[0151] S304. Calculate the deadlock factor:

[0152] Let the remaining paths of the other AGVs be... If there are adjacent node pairs in the remaining path ,satisfy , This indicates that other AGVs will move in the opposite direction to the current AGV, posing a risk of conflict. The deadlock factor is defined as follows: ,in This is the deadlock factor penalty coefficient. This represents the number of AGVs that may face the risk of conflict.

[0153] S305. Calculate the turning coefficient: Let the currently traversed node be... The parent node is Then the direction vectors of the two paths are respectively , The turning direction is determined by the cross product of two vectors. ,in , For vectors The x and y coordinate components, where , For vectors The horizontal and vertical coordinate components, and the turning penalty coefficient are .

[0154] S306. Constructing an improved A* comprehensive evaluation function: Based on the above penalty calculation method, the improved A* comprehensive evaluation function is derived as follows: ,in This represents the distance from the starting point to the current node. The actual path cost, Indicates the distance to the current candidate road segment. The congestion coefficient of the current candidate road segment. This is the turning penalty coefficient. Represents the deadlock factor. This represents the Manhattan distance from the candidate child node to the target node. This represents the deadlock factor weight.

[0155] S307. Output the planned path from the current point to the target node.

[0156] S4. Based on the actual physical size conflict detection and deadlock recovery in the planned path, optimize and adjust the planned path to complete the multi-AGV collaborative scheduling.

[0157] Specifically:

[0158] S401. Initialization of the detection point set: Let the length of the longest side of the AGV be... The preset safe distance is During the map initialization phase, each node in the path network... Centered on, with To determine the detection radius, traverse each node in the network and add nodes within the detection radius to the node list. The set of detection points: ,in Represents a node The set of detection points Represents a node To the node The distance. During AGV operation, when the AGV occupies a node. At that time, the set of detection points will be adjusted based on the actual length, width, and current direction of travel of the AGV. The nodes in the AGV undergo a secondary judgment. If a node is within the safe distance range of the AGV, then the AGV occupies that node, and the AGV occupies the node. At the same time, other vehicles should be prevented from entering the testing point.

[0159] S402. Conflict and Deadlock Detection: Assuming an AGV... The next path node is If the node is occupied by other vehicles, then the AGV is considered to be in a state of emergency. There is a risk of operational conflicts. Furthermore, starting from the AGV that caused the conflict, a deadlock detection chain is constructed along the relationships of which AGV occupies the next path node of the current AGV. During deadlock detection, if the detection chain exhibits circular waiting, reciprocal waiting, or a task AGV cannot continue running due to an idle AGV occupying a path node, then a deadlock is determined to have occurred in the system. The deadlock types include at least one of the following: circular deadlock; circular deadlock with detection points; reciprocal deadlock; reciprocal deadlock with detection points; and idle deadlock. Circular deadlock indicates that multiple AGVs form a head-to-tail waiting relationship; reciprocal deadlock indicates that two AGVs are waiting towards each other in a road segment; deadlock with detection points indicates that although an AGV does not directly occupy the target node, the detection point it occupies blocks the operation of other AGVs; and idle deadlock indicates that an idle AGV occupies a node, causing the AGV executing the task to be unable to continue running.

[0160] S403. Determine the deadlock recovery target: When a deadlock is detected, the AGV that triggered the deadlock detection is designated as the first AGV. The occupied AGV corresponding to the next path node of the first AGV is obtained, and this occupied AGV is designated as the second AGV. During the deadlock recovery process, the second AGV is prioritized to perform an avoidance operation, causing it to leave the current blocked position, thereby releasing the running path of the first AGV.

[0161] S404. Search for Available Exit Points: Starting from the current position of the second AGV, a breadth-first search method is used to find available exit points in the path network. During the search, nodes that meet the following conditions are added to the exit point queue: not on the remaining path of the first AGV; not a detection point of the first AGV; not a detection point of the second AGV; not in a deadlock loop generated during deadlock detection; not a previously selected avoidance point of the second AGV; will not continue to block the first AGV due to detection point occupation; and meets the AGV safety detection range constraints. If no exit point is found, the first AGV is used as the second AGV, and the search operation is repeated.

[0162] S405, Perform deadlock recovery: After the avoidance point is determined, plan an avoidance path for the second AGV from its current position to the avoidance point, and control the second AGV to move along the avoidance path to the avoidance point.

[0163] S5. Construct a simulation environment based on the path network, AGV set, machine set, and workpiece set of the path node set and path edge set;

[0164] During the simulation, the AGV is controlled to run according to kinematic parameters including starting acceleration, constant speed and deceleration in front of the target node, and the current position, running status, target node list, remaining planned path, node occupancy status and detection point occupancy status of the AGV are updated in real time.

[0165] After the simulation, the completion time of each task is recorded, the maximum completion time of each task is calculated as the maximum completion time of the system, and the scheduling results are output to verify the effectiveness of the scheduling method.

[0166] Specifically:

[0167] S501. Constructing the Simulation Environment: In the simulation platform, construct a flexible workshop path network, a set of workpieces, a set of machines, and a set of AGVs. The path network includes a set of path nodes and a set of path edges. The machine set includes the position and operating status of each machine. The AGV set includes the initial position, operating speed, acceleration, deceleration, vehicle size, and task status of each AGV.

[0168] S502. Simulated AGV movement process: During the simulation, the AGV operates according to preset kinematic parameters, including acceleration during the start-up phase, running speed during the constant speed phase, and deceleration before reaching the target node. As the AGV moves along the planned path, its current position, running status, target node list, remaining planned path, node occupancy status, and detection point occupancy status are updated in real time.

[0169] S503. Output simulation test results: After the simulation is completed, record the completion time of each workpiece task. Assume the workpiece... The completion time is Then the maximum completion time of the system is The scheduling results are then output to verify the effectiveness of the method.

[0170] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0171] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A multi-AGV cooperative scheduling method based on heuristic strategies, characterized in that, Includes the following steps: S1. Establish a mathematical model for multi-AGV collaborative scheduling based on constraints, with the optimization objective of minimizing the maximum completion time; S2. The multi-AGV collaborative scheduling mathematical model selects the set of available workpieces and the set of available AGVs based on cost rules, and quickly matches the optimal execution combination and assigns tasks to AGVs. S3. Using the improved A* algorithm with multiple factors, calculate and allocate the AGV planning path, and output the planning path from the current point to the target node; S4. Based on the actual physical size conflict detection and deadlock recovery in the planned path, optimize and adjust the planned path to complete the multi-AGV collaborative scheduling.

2. The multi-AGV cooperative scheduling method based on heuristic strategy according to claim 1, characterized in that, The method for task allocation based on cost rules includes: Real-time identification of the set of available workpieces that are not yet processed and the set of available AGVs that are in a ready state; Based on priority, the most urgent workpiece and the candidate AGV that meets the map matching are selected from the available workpiece set to generate a transportation task; Calculate the overall cost for the candidate AGV to perform the transportation task; Select the AGV with the lowest overall cost to perform the task and update the system status.

3. The multi-AGV cooperative scheduling method based on heuristic strategy according to claim 2, characterized in that, The formula for calculating the comprehensive cost is as follows: in, The distance cost for the AGV to reach the starting point of the task; The cost of path risk when the AGV performs this task; and These are the weighting coefficients; in, + =1.

4. The multi-AGV cooperative scheduling method based on heuristic strategy according to claim 2, characterized in that: The multiple factors include determining the starting and ending points of path planning based on comprehensive costs, the future road network congestion coefficient, deadlock factor, and turning penalty coefficient; The comprehensive evaluation function calculation formula for the multi-factor improvement A* includes: in, From the starting point to the current path node The actual path cost; This represents the distance to the current candidate road segment; This represents the congestion coefficient of the current candidate road segment; This is the turning penalty coefficient. As a deadlock factor, Let be the Manhattan distance from the candidate child node to the target node. The deadlock factor weights.

5. The multi-AGV cooperative scheduling method based on heuristic strategy according to claim 4, characterized in that, Also includes: A path network is established based on the planned path, and the path network includes a number of path nodes; Based on the longest side and safe distance of the candidate AGV, generate a set of detection points for each path node; When a candidate AGV is located at an occupied path node, multiple detection points within the safe range of that path node are locked simultaneously to prevent spatial interference between candidate AGV entities. Determine whether the next path node of the current candidate AGV is occupied by another AGV; If the AGV is occupied by another AGV, check for deadlock and perform deadlock recovery.

6. The multi-AGV cooperative scheduling method based on heuristic strategy according to claim 5, characterized in that, The deadlock recovery method includes: The AGV that triggers the detection is marked as the first AGV, and the AGV that blocks its path is marked as the second AGV; Starting from the second AGV, a breadth-first search is used to find an avoidance point, which satisfies the constraints that it is not on the path of the first AGV and does not cause new blockages. Control the second AGV to move to the avoidance point, and release the path for the first AGV to pass; The deadlock recovery method shall recover at least one type of deadlock. The deadlock types include circular deadlock, circular deadlock with detection points, opposing deadlock, opposing deadlock with detection points, and idle deadlock; The constraints include: The same process for the same workpiece can only be assigned to one machine for processing and one AGV for transportation; Only one AGV can occupy a node or an edge at the same time. The processing of the workpiece shall not begin earlier than the time when the AGV transports it to the machine.

7. The multi-AGV cooperative scheduling method based on heuristic strategy according to claim 4, characterized in that, For any path node in the path network The method for calculating the future road network congestion coefficient includes: Count the total number of times each of the remaining planned path nodes for all AGVs appears, and the total number of path nodes involved in the remaining planned paths for all AGVs; Calculate path nodes initial congestion rate ; when season =0.99, based on the corrected congestion rate Calculate the node correction parameters of the path nodes. The calculation formula is: Similarly, calculate the extended nodes on this path. Node correction parameters ; The final formula for the congestion correction factor for this road section was determined. The calculation formula is: The congestion correction coefficient is then used as a weighting factor and incorporated into the comprehensive evaluation function of path planning to dynamically adjust the priority of path selection in order to avoid highly congested areas.

8. The multi-AGV cooperative scheduling method based on heuristic strategy according to claim 4, characterized in that, Methods for calculating the deadlock factor for any pair of adjacent nodes in the current AGV's planned path include: Get the set of remaining planned paths for all other AGVs. ; Traverse each remaining path and determine if there are any adjacent node pairs. satisfy and This means that other AGVs are running in the opposite direction to the current AGV on this section of the road; The number of AGVs that meet the above conditions is recorded as the number of AGVs that may have a risk of reciprocal conflict. ; Set deadlock factor penalty coefficient ; The formula for calculating the deadlock factor of this node pair is: 。 9. The multi-AGV cooperative scheduling method based on heuristic strategy according to claim 4, characterized in that: For the two continuous paths in the current AGV planning path from the parent node through path node u to path node v, calculate the turning coefficient, including: Construct the direction vector of the first path segment ; Construct the direction vector of the second path segment ; Calculate the cross product of two vectors ; The turning direction is determined based on the cross product sign, and a turning coefficient is assigned. : The turning coefficient is then incorporated as an additional term in the path cost into the evaluation function of the improved A* algorithm to suppress frequent turning behavior and improve path smoothness and running efficiency.

10. The multi-AGV cooperative scheduling method based on heuristic strategies according to any one of claims 1-6, characterized in that: Includes the following steps: The simulation environment is constructed based on the path network, AGV set, machine set, and workpiece set of the path node set and path edge set. During the simulation, the AGV is controlled to run according to kinematic parameters including starting acceleration, constant speed and deceleration in front of the target node, and the current position, running status, target node list, remaining planned path, node occupancy status and detection point occupancy status of the AGV are updated in real time. After the simulation, the completion time of each task is recorded, the maximum completion time of each task is calculated as the maximum completion time of the system, and the scheduling results are output to verify the effectiveness of the scheduling method.

Images