A multi-agv partition scheduling system

Abstract

This invention relates to the field of AGV scheduling technology, specifically disclosing a multi-AGV zone scheduling system, including a central controller, zone controllers, workstations, a monitoring terminal, and an AGV integrated control module. The central controller communicates wiredly with the zone controllers within each zone and also communicates with the monitoring terminal; the central controller is responsible for issuing overall tasks. Zone controllers communicate wirelessly with the AGVs within their respective zones and also communicate with the workstations within those zones; the zone controllers are responsible for the specific implementation of tasks, autonomously allocating task priorities, and obtaining the AGV operating status. The AGV integrated control module is responsible for executing specific tasks according to the task requirements of the zone controllers, including autonomous positioning and navigation, path planning, and picking up and delivering designated goods. This invention forms a distributed architecture scheduling system between zone controllers and AGVs, effectively avoiding collision risks when multiple AGVs are operating and improving task execution efficiency.

Description

Technical Field

[0001] This invention relates to the field of AGV scheduling technology, and more specifically to a multi-AGV partition scheduling system. Background Technology

[0002] Automated Guided Vehicle (AGV) technology is an advanced automated device that is becoming increasingly important with the rapid development of industrial automation. Traditional centralized AGV control methods place the main scheduling system on a single master controller. All AGV path planning commands are issued in real-time by the master controller, which is responsible for data collection, data processing, and path planning algorithm implementation. This places stringent performance requirements on the master controller and demands high real-time performance from the wireless network, making implementation difficult.

[0003] Especially in scenarios where multiple AGVs operate simultaneously, implementing anti-collision strategies is one of the major challenges faced by multi-AGV scheduling systems. Therefore, traditional AGV scheduling systems are not suitable for collision-free path planning in scenarios involving multiple AGVs and multiple task assignments. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-AGV partition scheduling system to solve the following technical problems:

[0005] Traditional AGV scheduling systems are not suitable for collision-free path planning in scenarios involving multiple AGVs and multiple task assignments.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A multi-AGV zone scheduling system includes a central controller, zone controllers, workstations, monitoring terminals, and an AGV integrated control module, wherein:

[0008] The central controller serves as the overall dynamic scheduling and monitoring center for system tasks;

[0009] The area controller is used to receive task requests from the central controller and complete the scheduling and control of AGVs within the area.

[0010] The monitoring terminal is used to monitor the execution and completion of system tasks, as well as to monitor the AGV malfunctions at the execution layer.

[0011] The workstations are for manually controlling AGV task sending terminals, troubleshooting AGV malfunctions, and AGV logistics terminals in each area.

[0012] The AGV integrated control module is the final execution layer of the task, which includes: client, wireless network module, motor drive module, path planning module, and real-time positioning and mapping module;

[0013] The main controller communicates with the monitoring terminal and area controllers via wired connections.

[0014] The area controller communicates with the workstation via wired connection and with each AGV via wireless connection.

[0015] The scheduling method of the scheduling system is as follows:

[0016] S1: The central controller obtains the status information of the partitioned warehouses and issues logistics storage and retrieval commands to the corresponding regional controllers;

[0017] S2: The area controller obtains logistics tasks, assigns task priorities, acquires the current AGV running status information in the area, selects AGVs to execute tasks according to task priorities, and the AGVs autonomously locate themselves, start path planning calculations, and send the spatiotemporal node information of the pre-execution path to the area controller.

[0018] S3: After the main control layer issues tasks according to priority, the area controller obtains the number of AGVs S1 in the task. If S1<1, the AGVs will directly execute the task.

[0019] S4: If S1≥1, the area controller compares the spatiotemporal nodes of the path executed by each running AGV with the spatiotemporal nodes of the pre-execution path of the currently acquired task AGV to see if there is a conflict. If not, the AGV directly executes the task.

[0020] S5: If there is a conflict between the spatiotemporal nodes of the pre-execution path, the area controller issues conflict node constraints and requires the AGV to re-plan the path. The new spatiotemporal nodes of the path are uploaded to the area controller.

[0021] S6: The area controller determines whether there is a conflict between the spatiotemporal nodes of the new path. If there is still a conflict, it returns to step C of S5. If there is no conflict, the AGV starts to execute the task.

[0022] S7: After the AGV completes the logistics task, it updates the status information to the area controller. The area controller updates the task completion status to the main controller. The main controller connects and communicates with the monitoring terminal to synchronously summarize the task status information.

[0023] As a further aspect of the present invention: in step S2, the task priority allocation method is as follows:

[0024] A. Poll the current AGV network status and select the connected AGVs;

[0025] B. Obtain the status of connected AGVs, remove AGVs with fault alarms or low battery alarms, and obtain the currently available AGV group for scheduling, denoted as A. S = (a1, a2, ..., a s );

[0026] C. Further classify the schedulable AGV groups to obtain the AGV groups in the task. Idle AGV group Where S1∈(1,S1) represents the total number of AGVs in the task, S2∈(1,S2) represents the total number of idle AGVs, and a selection weight ω is assigned to AGVs in different states. i ω1、

[0027] ω2 represents the selection weights of the AGV and the idle AGV in the task, respectively, and the selection weights satisfy the formula:

[0028]

[0029] D. Determine if the total number of tasks in the current system task queue is greater than S2. If not, directly send tasks to the idle AGVs in sequence. If so, make further judgments based on the weights.

[0030] E. Note The waiting time weight for the nth task is given by N, where N is the total number of tasks in the queue, and t is the weight of the waiting time for the nth task. n Let the waiting time for the nth task be denoted as . The workload weight for the nth task is calculated using the Sth task as an example. i An approximate estimate is made of the Euclidean distance from the endpoint of the last task in the existing task queue of an AGV to the endpoint of the nth task; let be... Let S be the weight of the task to be executed by the S-th AGV in the i-th state (in task or idle); then, assigning priorities at this time is equivalent to finding the minimum value of the objective weight function:

[0031] As a further aspect of the present invention: in step S5, the path planning steps are as follows:

[0032] A. Let the spatiotemporal node information of the starting point of the task currently being performed by the AGV be (M0, t0), where M0 is the starting point coordinate and t0 is the initial time, denoted as M0 = (x0, y0) and t0 = 0;

[0033] B. The path planning algorithm uses the A* algorithm for searching, and adopts the evaluation function: f(i) = g(i) + h(i), where g(i) represents the spatial node M. i The actual cost of the distance from the starting point, h(i) represents the spatial node M. i The estimated cost from the destination is used to select the node with the smallest f(i) value as the next spatial node for the next movement. During the planning process, the AGV can be approximated as moving at a constant speed. Therefore, the time increases by one unit for each layer of nodes searched. Adding a timestamp to the node at that location yields the spatiotemporal node information, denoted as (M). i+1 ,ti+1 ).

[0034] C. Let the set of all spatiotemporal nodes currently recorded in the controller be... s ,M,t>,A s For the Sth AGV, the information of the new planned spatiotemporal nodes can be compared. When the location information and timestamp information in the new planned spatiotemporal nodes are duplicated with the set of recorded spatial nodes, it means that two AGVs will occupy the same node at the same time, resulting in a planning conflict. At this time, the type of conflict can be further determined.

[0035] D. When a conflict occurs, identify the two AGVs involved in the conflict, denoted as A. a A b When its spatiotemporal nodes satisfy the point conflict type, i.e., A a A b At the same time t, they occupy the same node position M. i At that time, the conflict is a node conflict;

[0036] E. When its spatiotemporal nodes satisfy the edge conflict type, that is, between time t and t+1, A a From M i Move to M i+1 Location, A b From M i+1 Move to M i At this point, the conflict is a head-on conflict.

[0037] As a further aspect of the present invention: in step D of step S5, the method further includes:

[0038] D1: At this point, determine the priority weight of the two AGVs.

[0039] D2: Before time t, the area controller issues a stop command to the AGV with lower priority weight. After the high-priority AGV passes through the node, the lower-priority AGV continues to run.

[0040] D3: Upload the updated spatiotemporal node information.

[0041] As a further aspect of the present invention: in step E of step S5, the method further includes:

[0042] E1: The area controller sets the conflict node as an obstacle constraint, and the newly planned AGV replans its path based on the constraint node as an obstacle.

[0043] E2: Upload the updated spatiotemporal node information.

[0044] ​As a further aspect of the present invention: the main controller and the regional controllers are connected and communicate via a wired connection using the TCP / IP protocol, and each regional controller uses the TCP / IP protocol as a server to communicate wirelessly with the AGV client;

[0045] The main controller communicates with the monitoring terminal via industrial Ethernet, while the area controllers communicate with each workstation via industrial Ethernet on the same network segment.

[0046] As a further aspect of the present invention: each AGV in different areas includes an integrated control module, which includes a client, a wireless network module, a motor drive module, a path planning module, and a real-time positioning and mapping module;

[0047] The wireless network module adopts an industrial network data transmission device, which connects to the AGV industrial control computer through an RS232 serial port and uses industrial Ethernet to communicate wirelessly with the area controller.

[0048] The real-time localization and mapping module achieves real-time localization and map creation of the AGV in a priori map through laser and vision fusion SLAM technology. It employs a laser and vision fusion localization algorithm to extract laser point cloud information and IMU measurements, and uses extended Kalman filtering to fuse them to obtain an optimized posterior data frame. The scanned and fused posterior data frame is interpolated and matched to a sub-map, continuously updated to obtain a global map. During localization, the initial camera pose is interpolated using the LiDAR pose stored during grid map construction. Laser measurements are used to reconstruct the visual scale, achieving alignment between the visual map and the grid map. The module also calls the navigation2 function package to subscribe to known grid map information.

[0049] The prior map set for AGV includes information on warehouse racks, logistics terminal sites, parking areas, and charging points.

[0050] The path planning module uses the A* algorithm to implement global path planning.

[0051] As a further aspect of the present invention: each AGV integrates an A* path planning module, which can perform initial global planning after the task is issued to find the shortest path to the target node. Within the same area, a single AGV does not need to obtain the spatiotemporal node information of the current planned path of other AGVs.

[0052] As a further aspect of the present invention: In step S4, the AGV spatiotemporal node information must be uploaded to the regional scheduler before the AGV actually runs. The regional controller does not need to participate in the specific path planning calculation. The regional scheduler determines whether there is a conflict among all current spatiotemporal nodes. If there is no conflict, the task can be started. If there is a conflict, the regional controller determines the type of conflict.

[0053] As a further aspect of the present invention: In step S5, the spatiotemporal node conflict types are classified into opposing conflicts and node conflicts. When the area controller determines that it is an opposing conflict, node constraints are added to the newly receiving task AGV to avoid collision. If it is a node conflict, the area controller determines the task priority and issues a stop waiting command to the low-priority AGV before the conflict node to avoid collision.

[0054] The beneficial effects of this invention are as follows: Compared with existing technologies, it effectively solves the problems of difficulty in implementing traditional centralized control, high requirements for the performance of the main controller, and poor scalability. By retaining a single AGV autonomous path planning module, it effectively reduces the computational difficulty of path planning for the main controller and improves the efficiency of anti-collision strategies in the context of multiple AGVs and multiple tasks operating in parallel. The layered architecture gives the system excellent overall scalability and reliability. Attached Figure Description

[0055] The invention will now be further described with reference to the accompanying drawings.

[0056] Figure 1 This is a schematic diagram of the overall structure of the AGV partition scheduling system of the present invention;

[0057] Figure 2 This is a block diagram of the area controller and AGV structure of the AGV partition scheduling system of the present invention;

[0058] Figure 3 This is a block diagram of the AGV real-time positioning and path planning module of the AGV partition scheduling system of the present invention;

[0059] Figure 4 This is a flowchart of the AGV partition scheduling system's scheduling tasks and task execution process. Detailed Implementation

[0060] 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.

[0061] like Figure 1 As shown, the multi-AGV zone scheduling system of the present invention includes a central controller, regional controllers, workstations, monitoring terminals, and an AGV integrated control module;

[0062] The central controller serves as the overall dynamic scheduling and monitoring center for system tasks;

[0063] The area controller is used to receive task requests from the central controller and complete the scheduling and control of AGVs within the area.

[0064] The monitoring terminal is used to monitor the execution and completion of system tasks, as well as to monitor the AGV malfunctions at the execution layer.

[0065] The workstations are for manually controlling AGV task sending terminals, troubleshooting AGV malfunctions, and AGV logistics terminals in each area.

[0066] The AGV integrated control module is the final execution layer of the task, which includes: client, wireless network module, motor drive module, path planning module, and real-time positioning and mapping module;

[0067] The main controller communicates with the monitoring terminal and area controllers via wired connections.

[0068] The area controller communicates with the workstation via wired connection and with each AGV via wireless connection.

[0069] Its scheduling method is as follows:

[0070] S1: The central controller obtains the status information of the partitioned warehouses and issues logistics storage and retrieval commands to the corresponding regional controllers;

[0071] S2: The area controller obtains logistics tasks, assigns task priorities, acquires the current AGV running status information in the area, selects AGVs to execute tasks according to task priorities, and the AGVs autonomously locate themselves, start path planning calculations, and send the spatiotemporal node information of the pre-execution path to the area controller.

[0072] S3: The area controller determines whether multiple AGVs are running simultaneously. If not, the AGVs directly execute the task.

[0073] S4: If multiple AGVs are running simultaneously, the area controller compares the spatiotemporal nodes of the path executed by each running AGV with the spatiotemporal nodes of the pre-execution path of the AGV currently acquiring the task to see if there is a conflict. If not, the AGV directly executes the task.

[0074] S5: If there is a conflict between the spatiotemporal nodes of the pre-execution path, the area controller issues conflict node constraints and requires the AGV to re-plan the path. The new spatiotemporal nodes of the path are uploaded to the area controller.

[0075] S6: The area controller determines whether there is a conflict between the spatiotemporal nodes of the new path. If not, the AGV starts to execute the task.

[0076] S7: After the AGV completes the logistics task, it updates the status information to the area controller. The area controller updates the task completion status to the main controller. The main controller connects and communicates with the monitoring terminal to synchronously summarize the task status information.

[0077] like Figure 2 As shown, the regional controller control system includes a shunting algorithm module, a spatiotemporal node comparison algorithm module, and a task priority algorithm module.

[0078] This shunting algorithm module monitors the uploaded status information of AGVs within the receiving area and handles the transitions between AGVs' task execution status, idle status, and queuing status. This facilitates the selection of suitable AGVs to perform tasks within the area.

[0079] The spatiotemporal node comparison algorithm module is used to receive the spatiotemporal node information of the pre-execution path uploaded by AGVs in the region, and compare it with the received historical spatiotemporal nodes to determine whether there are conflicting nodes in the current new path. If there are conflicting nodes, the conflict type is determined and constraint information is added and sent to the AGV to realize a collision-free strategy for multiple AGVs and multiple tasks.

[0080] This fault monitoring module is used to detect whether there are faults in the AGVs and workstations within the area. If a fault occurs in the AGVs and workstations within the area, the area status will be reported to the monitoring terminal, and the faulty vehicle will be locked and added as a constraint node to prevent collisions.

[0081] like Figure 2 As shown, each AGV includes a wireless network module, a motor drive module, a path planning module, and a real-time positioning and mapping module.

[0082] This AGV wireless network module uses industrial network data transmission equipment, connects to the AGV industrial control computer via RS232 serial port, and communicates wirelessly with the area controller via industrial Ethernet. The module will upload AGV status and path spatiotemporal node information to the area controller, and synchronously receive task commands and constraint node information from the area controller.

[0083] This AGV real-time positioning and mapping module uses laser and vision fusion SLAM technology to realize the real-time positioning and map creation of the AGV in the prior map, and calls the navi gat ion2 function package to subscribe to known grid map information.

[0084] This AGV path planning module uses the A* algorithm to achieve global path planning.

[0085] The AGV motor drive module, after global path planning by the path planning module, converts the path information into angular velocity information. After the central processing unit industrial control computer subscribes to the information, it calculates the differential speed of the AGV drive wheels and sends commands to the motor drive module. The data obtained by the motor encoder is synchronously fed back to the central processing unit industrial control computer.

[0086] like Figure 3 As shown, the implementation method of AGV real-time positioning and path planning is as follows:

[0087] The SLAM drive module integrates a laser vision information fusion algorithm. After data fusion and optimization from the IMU sensor front end, the precise pose information of the AGV in the grid map is obtained. This pose information is uploaded to the AGV industrial control computer, and the path planning module calls the A* path planning algorithm to complete the initial global planning. The industrial control computer then sends predetermined angular velocity and linear velocity control commands to the motor drive module, which controls the differential speed of the two wheels.

[0088] In scenarios where multiple AGVs operate simultaneously, the spatiotemporal nodes of the newly planned path may conflict with those of the historically planned path. If a new constraint node is received after the initial planning, the path planning module will add the constraint and then call the A* algorithm again to perform path planning.

[0089] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A multi-AGV zone scheduling system, characterized in that, Includes a central controller, area controllers, workstations, monitoring terminals, and an AGV integrated control module, among which: The central controller serves as the overall dynamic scheduling and monitoring center for system tasks; The area controller is used to receive task requests from the central controller and complete the scheduling and control of AGVs within the area. The monitoring terminal is used to monitor the execution and completion of system tasks, as well as to monitor the AGV malfunctions at the execution layer. The workstations are for manually controlling AGV task sending terminals, troubleshooting AGV malfunctions, and AGV logistics terminals in each area. The AGV integrated control module is the final execution layer of the task, which includes: client, wireless network module, motor drive module, path planning module, and real-time positioning and mapping module; The main controller communicates with the monitoring terminal and area controllers via wired connections. The area controller communicates with the workstation via wired connection and with each AGV via wireless connection. The scheduling method of the scheduling system is as follows: S1: The central controller obtains the status information of the partitioned warehouses and issues logistics storage and retrieval commands to the corresponding regional controllers; S2: The area controller obtains logistics tasks, assigns task priorities, acquires the current AGV running status information in the area, selects AGVs to execute tasks according to task priorities, and the AGVs autonomously locate themselves, start path planning calculations, and send the spatiotemporal node information of the pre-execution path to the area controller. S3: After the central control layer issues tasks according to priority, the regional controller obtains the number of AGVs in the task. ,like Then the AGV will directly execute the task; S4: If The area controller then compares the spatiotemporal nodes of the path executed by each running AGV with the spatiotemporal nodes of the pre-execution path of the currently acquired task AGV to see if there is a conflict. If not, the AGV directly executes the task. S5: If there is a conflict between the spatiotemporal nodes of the pre-execution path, the area controller issues conflict node constraints and requires the AGV to re-plan the path. The new spatiotemporal nodes of the path are uploaded to the area controller. S6: The area controller determines whether there is a conflict between the spatiotemporal nodes of the new path. If there is still a conflict, it returns to step C of S5. If there is no conflict, the AGV starts to execute the task. S7: After the AGV completes the logistics task, it updates the status information to the area controller. The area controller updates the task completion status to the main controller. The main controller connects and communicates with the monitoring terminal to synchronously summarize the task status information. In step S2, the task priority allocation method is as follows: A. Poll the current AGV network status and select the connected AGVs; B. Obtain the status of connected AGVs, remove AGVs with fault alarms or low battery alarms, obtain the current group of AGVs available for scheduling, and record... ; C. Further classify the schedulable AGV groups to obtain the AGV groups in the task. Idle AGV group ,in This indicates the total number of AGVs in the task. This represents the total number of idle AGVs, and assigns selection weights to AGVs in different states. These represent the selection weights for AGVs in the task and idle AGVs, respectively, and the selection weights satisfy the formula: ; D. Determine if the total number of tasks in the current system task queue is greater than [the specified value]. If not, tasks are directly assigned to idle AGVs in sequence; if yes, further judgment is made based on weight. E. Note Let N be the waiting time weight for the nth task, where N is the total number of tasks in the queue. Let the waiting time for the nth task be denoted as . The workload weight for the nth task is determined by the weight of the nth task. An approximate estimate is made of the Euclidean distance from the endpoint of the last task in the existing task queue of an AGV to the endpoint of the nth task; let be... Let be the weight of the task to be executed by the S-th AGV in the i-th state; then, assigning priorities at this point is equivalent to finding the minimum value of the objective weight function: .

2. The multi-AGV partition scheduling system according to claim 1, characterized in that, In step S5, the path planning steps are as follows: A. Record the spatiotemporal node information of the starting point of the task currently being performed by the AGV as follows: ,in As the starting coordinates, Let the initial time be denoted as . ; B. Path planning algorithm usage The algorithm performs the search using an evaluation function: Representing spatial nodes The actual value of the distance from the starting point Representing spatial nodes The estimated cost from the destination is selected each time, with the minimum cost. The node with the value is used as the spatial node for the next move in the search; During the planning process, the AGV can be approximated as moving at a constant speed. Therefore, the time increases by one unit for each layer of nodes searched. Adding a timestamp to the node at that location yields the spatiotemporal node information, denoted as […]. ; C. Let the set of all spatiotemporal nodes currently recorded in the controller be... , For the Sth AGV, the information of the new planned spatiotemporal nodes can be compared. When the location information and timestamp information in the new planned spatiotemporal nodes are duplicated with the set of recorded spatial nodes, it means that two AGVs will occupy the same node at the same time, resulting in a planning conflict. At this time, the type of conflict can be further determined. D. When a conflict occurs, identify the two AGVs involved in the conflict, denoted as... , When its spatiotemporal nodes satisfy the point conflict type, that is... At the same time t, they occupy the same node position. At that time, the conflict is a node conflict; E. When its spatiotemporal nodes satisfy the edge conflict type, that is, between time t and t+1, From Move to place, From Move to At this point, the conflict is a head-on conflict.

3. A multi-AGV zone scheduling system according to claim 2, characterized in that, Step D of step S5 further includes: D1: At this point, determine the priority weight of the two AGVs. ; D2: Before time t, the area controller issues a stop command to the AGV with lower priority weight. After the high-priority AGV passes through the node, the lower-priority AGV continues to run. D3: Upload the updated spatiotemporal node information.

4. A multi-AGV partition scheduling system according to claim 2, characterized in that, Step E of step S5 further includes: E1: The area controller sets the conflict node as an obstacle constraint, and the newly planned AGV replans its path based on the constraint node as an obstacle. E2: Upload the updated spatiotemporal node information.

5. A multi-AGV zone scheduling system according to claim 1, characterized in that, The main controller and the area controllers communicate via a wired connection using the TCP / IP protocol, while each area controller uses the TCP / IP protocol as a server to communicate wirelessly with the AGV client. The main controller communicates with the monitoring terminal via industrial Ethernet, while the area controllers communicate with each workstation via industrial Ethernet on the same network segment.

6. A multi-AGV zone scheduling system according to claim 1, characterized in that, Each AGV in different areas contains an integrated control module, which includes a client, a wireless network module, a motor drive module, a path planning module, and a real-time positioning and mapping module. The wireless network module adopts an industrial network data transmission device, which connects to the AGV industrial control computer through an RS232 serial port and uses industrial Ethernet to communicate wirelessly with the area controller. The real-time localization and mapping module achieves real-time localization and map creation of the AGV in a priori map through laser and vision fusion SLAM technology. It employs a laser and vision fusion localization algorithm to extract laser point cloud information and IMU measurements, and uses extended Kalman filtering to fuse them to obtain an optimized posterior data frame. The scanned and fused posterior data frame is interpolated and matched to a sub-map, continuously updated to obtain a global map. During localization, the initial camera pose is interpolated using the LiDAR pose stored during grid map construction. Laser measurements are used to reconstruct the visual scale, achieving alignment between the visual map and the grid map. The module also calls the navigation2 package to subscribe to known grid map information. The prior map set for AGV includes information on warehouse racks, logistics terminal sites, parking areas, and charging points. The path planning module utilizes The algorithm implements global path planning.

7. A multi-AGV zone scheduling system according to claim 1, characterized in that, Each AGV integrates The path planning module can perform initial global planning after the task is issued to find the shortest path to the target node. Individual AGVs within the same area do not need to obtain the spatiotemporal node information of other AGVs' currently planned paths.

8. A multi-AGV zone scheduling system according to claim 1, characterized in that, In step S4, the AGV spatiotemporal node information must be uploaded to the regional scheduler before the AGV actually runs. The regional controller does not need to participate in the specific path planning calculation. The regional scheduler determines whether there is a conflict among all current spatiotemporal nodes. If there is no conflict, the task can be started. If there is a conflict, the regional controller determines the type of conflict.

9. A multi-AGV partition scheduling system according to claim 1, characterized in that, In step S5, the spatiotemporal node conflict types are classified into head-on conflicts and node conflicts. When the area controller determines that it is a head-on conflict, it adds node constraints to the newly receiving AGV to avoid collisions. If it is a node conflict, the area controller determines the task priority and issues a stop waiting command to the low-priority AGV before the conflict node to avoid collisions.

Images