Dynamic master-slave system for multi-AGV cooperative scheduling and election method

Abstract

The invention relates to an AGV (Automatic Guided Vehicle) scheduling system and method, in particular to a dynamic master-slave system for multi-AGV cooperative scheduling and an election method, which are used for solving the problems that an existing distributed AGV group lacks a dynamic master node election mechanism, the recovery of a single-point fault of a master node is slow, and AGV roles cannot be dynamically switched. The system comprises a global state synchronization module, a scheduling decision module, a data storage module, a dynamic role management module, a wireless communication module and a heartbeat monitoring and fault processing module which are arranged on each AGV. Each AGV is driven to participate in competitions based on a multivariate evaluation function through a trigger event, a master node is elected through distributed negotiation consensus, and other nodes serve as slave nodes and are registered. The master node periodically broadcasts heartbeat signals, the slave nodes cooperatively monitor, and once it is confirmed that the master node fails, re-election is triggered. According to the invention, decentralized AGV autonomous organization, fault self-recovery and role dynamic switching are realized, and the reliability, the expansibility and the adaptive capability of a multi-AGV system are remarkably improved.

Description

Technical Field

[0001] This invention relates to AGV scheduling systems and methods, specifically to a dynamic master-slave system and dynamic master-slave election method for multi-AGV collaborative scheduling. Background Technology

[0002] With the widespread application of AGVs in flexible production lines, smart warehousing, and other scenarios, multi-AGV collaborative operation systems are evolving from traditional centralized architectures to distributed and decentralized architectures. In centralized architectures, the system relies on a single control server for task scheduling, which presents problems such as high risk of single point of failure, high network deployment and maintenance costs, and limited system scalability.

[0003] In existing technologies, there are solutions for achieving hard synchronization of multiple AGVs through physical connections. For example, Chinese patent CN119937539B discloses an AGV multi-vehicle linkage control system and method. This system consists of multiple independent AGV electrical control systems, each including a controller, remote controller, driver, and motor. The controller connects to the preceding and following AGV systems, drivers, and remote controllers through various interfaces to achieve motion-level linkage. However, such solutions are essentially extensions of motion control and cannot support complex logical task coordination and dynamic system management. Other existing research has proposed multi-AGV scheduling methods, but has not deeply addressed the core issue of how to achieve a highly reliable, self-organizing node election and scheduling mechanism in a distributed AGV fleet without a fixed central server. Specifically, in a fully distributed AGV system, the following key capabilities are still lacking: First, how to dynamically elect a master node to assume centralized scheduling responsibilities. Second, how to achieve rapid and coordinated fault detection and system self-recovery when the master node fails. Third, how to enable ordinary AGVs to dynamically switch roles and support the system to reconstruct its logical architecture during operation. Summary of the Invention

[0004] The purpose of this invention is to solve the technical problems of the lack of a dynamic master node election mechanism and the inability to dynamically switch AGV roles in existing distributed AGV fleets, and to provide a dynamic master-slave system and election method for multi-AGV collaborative scheduling.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A dynamic master-slave system for multi-AGV collaborative scheduling is used to dynamically elect master node AGVs and slave node AGVs in an AGV group composed of multiple AGVs. Its special feature is that it includes a global state synchronization module, a scheduling decision module, a data storage module, a dynamic role management module, a wireless communication module, and a heartbeat monitoring and fault handling module set on each AGV. In each AGV, the dynamic role management module is connected to the wireless communication module, and the heartbeat monitoring and fault handling module is connected to both the wireless communication module and the dynamic role management module. The scheduling decision module is connected to both the data storage module and the global state synchronization module, and the global state synchronization module is connected to the wireless communication module. The wireless communication module in each AGV is connected to the wireless communication modules of other AGVs. When the dynamic role management module responds to an election trigger event, it determines a master node AGV for the AGV group by means of information interaction and negotiation between each AGV in the AGV group, and determines the remaining AGVs as slave node AGVs of the task execution entity. The master node AGV activates its global status synchronization module and scheduling decision module; the global status synchronization module is used to continuously receive real-time status information reported by each slave node AGV and generate a global status view. The scheduling decision module is used to generate scheduling instructions based on the global status view and the scheduling rules pre-stored in the data storage module, and to send the scheduling instructions to the corresponding slave node AGV through the wireless communication module. The heartbeat monitoring and fault handling module in the master node AGV is used to periodically broadcast heartbeat signals to the slave node AGVs; the heartbeat monitoring and fault handling module in each slave node AGV is used to monitor the periodic heartbeat signals broadcast by the master node AGV to determine whether the master node AGV is faulty; and trigger a master-slave election command when a fault is detected in the master node AGV.

[0006] Furthermore, each of the aforementioned dynamic role management modules includes a parameter calculation unit, a message sending and receiving unit, and a decision logic unit; Each parameter calculation unit is used to calculate the election parameters of this AGV based on a preset multivariate evaluation function when the election is triggered. Each of the aforementioned message transceiver units is connected to a corresponding wireless communication module, used to broadcast an election message containing the election parameters via a wireless network, and to receive election messages from other AGVs; Each decision logic unit is connected to its corresponding parameter calculation unit and message sending and receiving unit, and is used to compare the election parameters in all received election messages and determine the new master node AGV according to the predetermined win and lose determination rules. Each of the heartbeat monitoring and fault handling modules includes a timeout counter, a state co-verifier, and a fault diagnosticator. Each of the timeout counters is used to count when no valid heartbeat signal is received within a preset heartbeat cycle; Each of the aforementioned state collaboration verifiers is connected to a corresponding timeout counter and a wireless communication module, and is used to exchange state confirmation information with other slave node AGVs via a wireless network when the count value of its corresponding timeout counter reaches a threshold. Each of the fault determiners is connected to the state co-verifier and is used to generate the fault trigger signal after receiving information from more than half of the other slave AGVs confirming that the master AGV is suspicious.

[0007] Meanwhile, this invention also provides a dynamic master-slave election method for multi-AGV collaborative scheduling, which is characterized by including the following steps: S1. Build a dynamic master-slave system for collaborative scheduling of the above-mentioned multiple AGVs; S2. Preset the task objectives to be completed and power on all AGVs in the AGV fleet. After AGV initialization, trigger the master-slave election process of the AGV fleet. S3. Each candidate AGV calculates its own election parameters according to the preset multivariate evaluation function through the dynamic role management module, and broadcasts an election message containing the election parameters to the AGV group. S4. Each AGV receives and compares the election messages broadcast by all AGVs through the dynamic role management module. According to the predetermined win / loss determination rules, each AGV autonomously reaches a consensus on the identity of the unique master node AGV. S5. The AGV that is identified as the master node by consensus activates the global state synchronization module and the scheduling decision module, and broadcasts its master node AGV identity through the dynamic role management module; the other AGVs receive the broadcast from the master node AGV, confirm themselves as slave node AGVs, and register with the master node AGV, thus completing the establishment of the AGV group logical architecture. S6. The scheduling decision module of the master node AGV generates scheduling instructions based on the global status view generated by the global status synchronization module and the scheduling rules pre-stored in the data storage module. The scheduling instructions are then sent to the corresponding slave node AGVs through the wireless communication module to schedule the slave node AGVs to perform relevant work tasks. The heartbeat monitoring and fault handling module periodically broadcasts heartbeat signals to the slave node AGVs. S7. The slave node AGV monitors the periodic heartbeat signal broadcast by the master node AGV through the heartbeat monitoring and fault handling module; determines whether the master node AGV is faulty; if the master node AGV is faulty, returns to step S3 to re-elect a new master node AGV until the preset task objective is achieved, realizing dynamic master-slave election for multi-AGV collaborative scheduling.

[0008] Furthermore, step S7 specifically includes: S7.1 Each slave node AGV monitors the periodic heartbeat signal broadcast from the master node AGV and determines whether it has received the heartbeat signal from the master node AGV through the heartbeat timeout timer set in it. If it does not receive the heartbeat message from the master node AGV within N consecutive heartbeat cycles, the master node AGV is marked as suspicious locally in this AGV, where N≥2. S7.2 Any slave AGV that marks the master node AGV as suspicious shall broadcast a query request message containing the "master node AGV is suspicious" status to the AGV fleet. S7.3. Each slave node AGV responds to the query request message. If it also marks the master node AGV as suspicious, it replies with a confirmation message. S7.4 If the slave AGV that initiates the query receives confirmation messages from more than half of the other slave AGVs within a preset time, it will collaboratively determine that the master AGV is faulty, return to step S3 to re-elect a new master AGV, and continue until the preset task objective is achieved, thus realizing dynamic master-slave election for multi-AGV collaborative scheduling.

[0009] Furthermore, step S4 also includes a verification and anti-collision mechanism for campaign messages, specifically: The campaign message contains the digital signature of the sending node or a message verification code; Upon receiving a message, the receiving node first verifies the validity of its signature and only processes messages that pass the verification. If, during the election process, the same AGV receives conflict declarations from different nodes that claim to be the same node, it will only accept the latest declaration based on the timestamp or sequence number in the message.

[0010] Furthermore, in step S4, the predetermined rules for determining the winner are as follows: Each AGV compares the election parameters in all the election messages it receives, and determines the AGV with the highest election parameters as the master node AGV; If the election parameters are the same, a decision will be made according to a preset priority strategy, which includes: comparing the size of the AGV's physical identifier ID, or comparing the age of its last update timestamp.

[0011] Furthermore, in step S5, the registration information sent by the slave AGV to the master AGV during registration includes at least: This node's unique identifier, current task status, current fault status, current location information, current battery status, current load capacity, and a list of executable task types.

[0012] Furthermore, in step S3, the preset multivariate evaluation function is a weighted function, and its variables include at least: The remaining battery percentage of the AGV, the current load rate of the AGV's central processor, the average communication signal strength with other AGVs in the group, and the historical continuous normal working time.

[0013] Furthermore, in step S3, the weight parameters of the multivariate evaluation function are adjusted offline or online based on the historical operating data of the vehicle group to optimize the impact of the election results on the overall operating efficiency of the system.

[0014] Furthermore, in step S4, the consensus-reaching phase is also set with a first timeout period; if, when the first timeout period expires, there are still AGVs that have not reached a consensus on the identity of the master node AGV, the current election state is cleared and the process returns to step S3 to re-initiate the election. In step S5, a second timeout period is set for the identity verification stage of the master node AGV and slave node AGV. If the master node AGV does not receive the registration information of all slave node AGVs before the second timeout period expires, the master node AGV broadcasts the list of registered slave node AGVs to the AGV fleet. Unregistered slave node AGVs are considered offline and can be added to the system later through the re-registration process.

[0015] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention provides a dynamic master-slave system for multi-AGV collaborative scheduling. Through dynamic role management and heartbeat monitoring mechanisms, it realizes dynamic election and collaborative fault detection of master node AGVs, effectively solving the problems of high single-point failure risk and lack of self-recovery capability in distributed AGV systems, and improving the reliability and robustness of the system. At the same time, through parameter calculation and election comparison mechanisms, it realizes objective election based on multiple indicators; through state collaborative verification, it avoids misjudgment of a single node, making the master node fault judgment more accurate and reliable, and enhancing the overall stability of the system.

[0016] (2) The present invention provides a dynamic master-slave election method for multi-AGV collaborative scheduling, which provides a complete set of dynamic master-slave election and recovery methods, enabling AGV vehicle groups to autonomously establish and maintain a logical architecture without a central server, realizing the self-organization and continuous operation of the system, and fundamentally improving the feasibility of distributed collaboration.

[0017] (3) The present invention provides a dynamic master-slave election method for multi-AGV collaborative scheduling. It adopts a two-stage fault determination mechanism of "suspicious marking-collaborative confirmation". The fault of the master node is confirmed through consensus of the majority of nodes, which effectively prevents false triggering caused by instantaneous network fluctuations and ensures the accuracy of system reconstruction decisions.

[0018] (4) The present invention provides a dynamic master-slave election method for multi-AGV collaborative scheduling. By introducing message verification and conflict handling mechanisms, it effectively resists security threats such as unauthorized node access or message tampering, and solves the information conflict problem that may be caused by network latency, ensuring the consistency between the election process and the system state. At the same time, it provides clear and computable rules for determining the winner, ensuring that all nodes can quickly and unambiguously reach a unique consensus in distributed negotiation, improving election efficiency and avoiding election deadlock caused by ambiguous rules.

[0019] (5) The present invention provides a dynamic master-slave election method for multi-AGV collaborative scheduling. Through standardized registration information, the master node can fully grasp the status and capabilities of the slave nodes, laying a data foundation for achieving accurate task scheduling and collaborative optimization based on global resources and load. At the same time, by using an evaluation function that includes multi-dimensional indicators such as power consumption, load, and communication quality, the AGV with the best overall status can be selected as the master node, thereby improving the scheduling decision quality and continuous working potential of the entire vehicle group.

[0020] (6) The present invention provides a dynamic master-slave election method for multi-AGV collaborative scheduling. Through an evaluation function with adjustable weights, the system has the ability to learn and optimize. It can dynamically adjust the election preference according to the actual operation data, so that the election mechanism can continuously adapt to specific scenarios and optimize the long-term operating efficiency of the system.

[0021] (7) The present invention provides a dynamic master-slave election method for multi-AGV collaborative scheduling. By introducing multiple timeout mechanisms, it sets clear fault recovery boundaries for key processes such as election and registration, avoiding the system from falling into infinite waiting due to individual node anomalies, and ensuring the system's rapid self-healing and continuous service capabilities under abnormal conditions. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the structure of a dynamic master-slave system for multi-AGV collaborative scheduling according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating an embodiment of a dynamic master-slave election method for multi-AGV collaborative scheduling according to the present invention. Detailed Implementation

[0023] To better understand the purpose, structure, and function of this invention, the following detailed description, in conjunction with the accompanying drawings, provides a dynamic master-slave system and election method for multi-AGV collaborative scheduling.

[0024] This invention provides a dynamic master-slave system for collaborative scheduling of multiple AGVs, used for dynamically electing master and slave AGVs in an AGV group composed of multiple AGVs; such as Figure 1As shown, it includes a global status synchronization module, a scheduling decision module, a data storage module, a dynamic role management module, a wireless communication module, and a heartbeat monitoring and fault handling module, all set on each AGV.

[0025] In each AGV, the dynamic role management module is connected to the wireless communication module, and the heartbeat monitoring and fault handling module is connected to both the wireless communication module and the dynamic role management module. The scheduling decision module is connected to both the data storage module and the global state synchronization module, and the global state synchronization module is connected to the wireless communication module. The wireless communication module in each AGV is connected to the wireless communication modules of other AGVs. When the dynamic role management module responds to the election trigger event, it determines a master node AGV for the AGV group to assume the responsibility of centralized scheduling through information interaction and negotiation among each AGV in the AGV group, and determines the remaining AGVs as slave node AGVs of the task execution entity.

[0026] Each dynamic role management module includes a parameter calculation unit, a message sending and receiving unit, and a decision logic unit. Each parameter calculation unit is used to calculate the election parameters of the AGV based on a preset multivariate evaluation function when the election is triggered. Each message sending and receiving unit is connected to the corresponding wireless communication module to broadcast election messages containing election parameters via a wireless network and to receive election messages from other AGVs. Each decision logic unit is connected to its corresponding parameter calculation unit and message sending and receiving unit to compare the election parameters in all received election messages and determine the new master node AGV according to the predetermined win / loss determination rules.

[0027] The master node AGV activates its global status synchronization module and scheduling decision module. The global status synchronization module continuously receives real-time status information reported by each slave node AGV and generates a global status view. The scheduling decision module generates scheduling instructions based on the global status view and the scheduling rules pre-stored in the data storage module, and sends the scheduling instructions to the corresponding slave node AGVs via the wireless communication module. The heartbeat monitoring and fault handling module in the master node AGV periodically broadcasts heartbeat signals to the slave node AGVs. The heartbeat monitoring and fault handling module in each slave node AGV monitors the periodic heartbeat signals broadcast by the master node AGV to determine whether the master node AGV is faulty, and triggers a master-slave election instruction when a master node AGV fault is detected.

[0028] The slave node AGV runs a lightweight client, responsible for receiving instructions and reporting status (including location information read via RFID). All nodes communicate with each other via a wireless local area network (such as Wi-Fi). The master node AGV makes scheduling decisions based on pre-set collaborative task rules (e.g., a task flow sequence table loaded from a database) and the global status reported by the slave node AGVs, and issues task instructions (such as "move to location X") to specific slave node AGVs.

[0029] Each heartbeat monitoring and fault handling module includes a timeout counter, a state coordination verifier, and a fault determiner. Each timeout counter counts when no valid heartbeat signal is received within a preset heartbeat cycle. Each state coordination verifier is connected to its corresponding timeout counter and wireless communication module, and is used to exchange status confirmation information with other slave AGVs via wireless network when the count value of its corresponding timeout counter reaches a threshold. Each fault determiner is connected to the state coordination verifier and generates a fault trigger signal after receiving information from more than half of the other slave AGVs confirming that the master AGV is suspicious.

[0030] Based on the aforementioned dynamic master-slave system for multi-AGV collaborative scheduling, the present invention provides a dynamic master-slave election method for multi-AGV collaborative scheduling, as follows: Figure 2 As shown, it includes the following steps: S1. Preset the task objectives to be completed and power on all AGVs in the AGV fleet. After the AGVs are initialized, the master-slave election process of the AGV fleet is triggered.

[0031] S2. Each candidate AGV calculates its own election parameters according to the preset multivariate evaluation function through the dynamic role management module, and broadcasts an election message containing the election parameters to the AGV group.

[0032] The preset multivariate evaluation function is a weighted function, whose variables include at least: the remaining battery percentage of the AGV, the current load rate of the AGV's central processor, the average communication signal strength with other AGVs in the group, and the historical continuous normal working time. For example, AGV2 calculates: remaining battery 85% (weight 0.4), CPU load 20% (weight 0.3), average signal strength -60dBm (weight 0.2), historical continuous normal working time 100h (weight 0.1), its weighted election parameter = 85*0.5 +(100-20)*0.3 + (100-60)*0.2+100*0.1 = 42.5 + 24 + 8+10 = 84.5. AGV2 encapsulates this value into an election message and broadcasts it.

[0033] The weight parameters of the multivariate evaluation function are adjusted offline or online based on the historical operation data of the vehicle group to optimize the impact of the election results on the overall operating efficiency of the system.

[0034] S3. Each AGV receives and compares the election messages broadcast by all AGVs through the dynamic role management module. Based on predetermined win / loss rules, each AGV autonomously reaches a consensus regarding the identity of the unique master node AGV. This also includes a verification and anti-collision mechanism for the election messages: the election message contains the digital signature or message verification code of the sending node; upon receiving the message, the receiving node first verifies the validity of its signature and only processes messages that pass verification; if, during the election process, the same AGV receives conflict declarations from different nodes claiming to be from the same node, it accepts only the latest declaration based on the timestamp or sequence number in the message. Assume AGV1's value is 90 (90% battery), AGV2's value is 84.5, AGV3's value is 65, AGV4's value is 90 (50% battery), and AGV5's value is 80. After independent comparison, each AGV identifies AGV1 as having the highest election parameters. According to the predetermined "highest value wins" rule, all AGVs reach a consensus: AGV1 is the new generation master node.

[0035] The predetermined rules for determining the winner are as follows: Each AGV will compare the election parameters in all the election messages it receives, and the AGV with the highest election parameters will be determined as the master node AGV; if the election parameters are the same, the decision will be made according to the preset priority strategy, which includes: comparing the size of the AGV's physical identifier ID, or comparing the age of its last update timestamp.

[0036] The consensus-reaching phase also includes a first timeout period. If, when the first timeout period expires, there are still AGVs that have not reached a consensus on the identity of the master node AGV, the current election state is cleared, and the process returns to step S2 to re-initiate the election.

[0037] S4. The AGV identified as the master node by consensus activates the global state synchronization module and scheduling decision module, and broadcasts its master node AGV identity through the dynamic role management module. The remaining AGVs receive the broadcast from the master node AGV, confirm themselves as slave node AGVs, and register with the master node AGV, completing the establishment of the AGV group logical architecture. A second timeout is set during the identity confirmation phase for both master and slave node AGVs. If the master node AGV does not receive registration information from all slave node AGVs before the second timeout expires, it broadcasts a list of registered slave node AGVs to the AGV group. Unregistered slave node AGVs are considered offline and can be re-registered later. For example, AGV1 automatically switches to master node mode and broadcasts an "election announcement." AGV2, AGV3, AGV4, and AGV5 receive this, switch to slave node mode, and send a "registration message" to AGV1, attaching their initial state. After AGV1 receives all registrations, the global state view is established, and the system enters a stable operation phase.

[0038] The registration information sent by the slave AGV to the master AGV when registering includes at least: the unique identifier of this node, the current task status (idle, busy, paused), the current fault status (fault present or absent), the current location information, the current power status, the current load capacity, and a list of executable task types.

[0039] S5. The scheduling decision module of the master node AGV generates scheduling instructions based on the global status view generated by the global status synchronization module and the scheduling rules pre-stored in the data storage module. The scheduling instructions are then sent to the corresponding slave node AGVs through the wireless communication module to schedule the slave node AGVs to perform relevant work tasks. The heartbeat monitoring and fault handling module periodically broadcasts heartbeat signals to the slave node AGVs.

[0040] S6. The slave node AGV monitors the periodic heartbeat signal broadcast by the master node AGV through the heartbeat monitoring and fault handling module; determines whether the master node AGV is faulty; if the master node AGV is faulty, returns to step S2 to re-elect a new master node AGV until the preset task objective is achieved, realizing dynamic master-slave election for multi-AGV collaborative scheduling.

[0041] S6.1 Each slave AGV monitors the periodic heartbeat signal broadcast from the master AGV and determines whether it has received the heartbeat signal from the master AGV through its internal heartbeat timeout timer. If it does not receive the heartbeat message from the master AGV within N consecutive heartbeat cycles, the master AGV is marked as suspicious locally on the AGV, where N≥2.

[0042] S6.2 Any slave AGV that marks the master node AGV as suspicious shall broadcast a query request message containing the "master node AGV is suspicious" status to the AGV fleet.

[0043] S6.3 Each slave node AGV responds to the query request message. If it also marks the master node AGV as suspicious, it replies with a confirmation message.

[0044] S6.4 If the slave AGV that initiates the query receives confirmation messages from more than half of the other slave AGVs within a preset time, it will collaboratively determine that the master AGV is faulty, return to step S3 to re-elect a new master AGV, and continue until the preset task objective is achieved, thus realizing dynamic master-slave election for multi-AGV collaborative scheduling.

[0045] For example, during stable operation, the master node AGV1 broadcasts a heartbeat message every T seconds (e.g., 1 second). The slave node AGV2 has an internal timer that resets each time it receives a heartbeat. If AGV2 does not receive a heartbeat after T seconds for three consecutive times (N=3), it will not act immediately but will first broadcast a "Master node suspicious - query" message to inquire with its companions. If AGV3 and AGV4 reply with "also suspicious" confirmation, then more than half (3 out of 4) of the slave nodes confirm the anomaly. At this point, AGV2, AGV3, and AGV4 collaboratively determine that the master node AGV1 is faulty and jointly trigger an election event.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present invention.

Claims

1. A dynamic master-slave system for collaborative scheduling of multiple AGVs, used for dynamically electing master node AGVs and slave node AGVs in an AGV fleet composed of multiple AGVs; characterized in that: It includes a global status synchronization module, a scheduling decision module, a data storage module, a dynamic role management module, a wireless communication module, and a heartbeat monitoring and fault handling module, all set on each AGV; In each AGV, the dynamic role management module is connected to the wireless communication module, and the heartbeat monitoring and fault handling module is connected to both the wireless communication module and the dynamic role management module. The scheduling decision module is connected to both the data storage module and the global state synchronization module, and the global state synchronization module is connected to the wireless communication module. The wireless communication module in each AGV is connected to the wireless communication modules of other AGVs. When the dynamic role management module responds to an election trigger event, it determines a master node AGV for the AGV group by means of information interaction and negotiation between each AGV in the AGV group, and determines the remaining AGVs as slave node AGVs of the task execution entity. The master node AGV activates its global status synchronization module and scheduling decision module; the global status synchronization module is used to continuously receive real-time status information reported by each slave node AGV and generate a global status view. The scheduling decision module is used to generate scheduling instructions based on the global status view and the scheduling rules pre-stored in the data storage module, and to send the scheduling instructions to the corresponding slave node AGV through the wireless communication module. The heartbeat monitoring and fault handling module in the master node AGV is used to periodically broadcast heartbeat signals to the slave node AGVs; the heartbeat monitoring and fault handling module in each slave node AGV is used to monitor the periodic heartbeat signals broadcast by the master node AGV to determine whether the master node AGV is faulty; and trigger a master-slave election command when a fault is detected in the master node AGV.

2. The dynamic master-slave system for multi-AGV collaborative scheduling according to claim 1, characterized in that: Each of the aforementioned dynamic role management modules includes a parameter calculation unit, a message sending and receiving unit, and a decision logic unit; Each parameter calculation unit is used to calculate the election parameters of this AGV based on a preset multivariate evaluation function when the election is triggered. Each of the aforementioned message transceiver units is connected to a corresponding wireless communication module, used to broadcast an election message containing the election parameters via a wireless network, and to receive election messages from other AGVs; Each decision logic unit is connected to its corresponding parameter calculation unit and message sending and receiving unit, and is used to compare the election parameters in all received election messages and determine the new master node AGV according to the predetermined win and lose determination rules. Each of the heartbeat monitoring and fault handling modules includes a timeout counter, a state co-verifier, and a fault diagnosticator. Each of the timeout counters is used to count when no valid heartbeat signal is received within a preset heartbeat cycle; Each of the aforementioned state collaboration verifiers is connected to a corresponding timeout counter and a wireless communication module, and is used to exchange state confirmation information with other slave node AGVs via a wireless network when the count value of its corresponding timeout counter reaches a threshold. Each of the fault determiners is connected to the state co-verifier and is used to generate the fault trigger signal after receiving information from more than half of the other slave AGVs confirming that the master AGV is suspicious.

3. A dynamic master-slave election method for multi-AGV collaborative scheduling, characterized in that, Includes the following steps: S1. Construct a dynamic master-slave system for multi-AGV collaborative scheduling as described in claim 1 or 2; S2. Preset the task objectives to be completed and power on all AGVs in the AGV fleet. After AGV initialization, trigger the master-slave election process of the AGV fleet. S3. Each candidate AGV calculates its own election parameters according to the preset multivariate evaluation function through the dynamic role management module, and broadcasts an election message containing the election parameters to the AGV group. S4. Each AGV receives and compares the election messages broadcast by all AGVs through the dynamic role management module. According to the predetermined win / loss determination rules, each AGV autonomously reaches a consensus on the identity of the unique master node AGV. S5. The AGV that is identified as the master node by consensus activates the global state synchronization module and the scheduling decision module, and broadcasts its master node AGV identity through the dynamic role management module; the other AGVs receive the broadcast from the master node AGV, confirm themselves as slave node AGVs, and register with the master node AGV, thus completing the establishment of the AGV group logical architecture. S6. The scheduling decision module of the master node AGV generates scheduling instructions based on the global status view generated by the global status synchronization module and the scheduling rules pre-stored in the data storage module. The scheduling instructions are then sent to the corresponding slave node AGVs through the wireless communication module to schedule the slave node AGVs to perform relevant work tasks. The heartbeat monitoring and fault handling module periodically broadcasts heartbeat signals to the slave node AGVs. S7. The slave node AGV monitors the periodic heartbeat signal broadcast by the master node AGV through the heartbeat monitoring and fault handling module; determines whether the master node AGV is faulty; if the master node AGV is faulty, returns to step S3 to re-elect a new master node AGV until the preset task objective is achieved, realizing dynamic master-slave election for multi-AGV collaborative scheduling.

4. The dynamic master-slave election method for multi-AGV collaborative scheduling according to claim 3, characterized in that, Step S7 is as follows: S7.1 Each slave AGV monitors the periodic heartbeat signal broadcast from the master AGV and determines whether it has received the heartbeat signal from the master AGV through the heartbeat timeout timer set within it. If it does not receive the heartbeat message from the master AGV within N consecutive heartbeat cycles, the master AGV is marked as suspicious locally in that AGV, where N≥2. S7.2 Any slave AGV that marks the master node AGV as suspicious shall broadcast a query request message containing the "master node AGV is suspicious" status to the AGV fleet. S7.

3. Each slave node AGV responds to the query request message. If it also marks the master node AGV as suspicious, it replies with a confirmation message. S7.4 If the slave AGV that initiates the query receives confirmation messages from more than half of the other slave AGVs within a preset time, it will collaboratively determine that the master AGV is faulty, return to step S3 to re-elect a new master AGV, and continue until the preset task objective is achieved, thus realizing dynamic master-slave election for multi-AGV collaborative scheduling.

5. The dynamic master-slave election method for multi-AGV collaborative scheduling according to claim 3, characterized in that, Step S4 also includes a verification and anti-collision mechanism for campaign messages, specifically: The campaign message contains the digital signature of the sending node or a message verification code; Upon receiving a message, the receiving node first verifies the validity of its signature and only processes messages that pass the verification. If, during the election process, the same AGV receives conflict declarations from different nodes that claim to be the same node, it will only accept the latest declaration based on the timestamp or sequence number in the message.

6. The dynamic master-slave election method for multi-AGV collaborative scheduling according to claim 5, characterized in that, In step S4, the predetermined rules for determining the winner are as follows: Each AGV compares the election parameters in all the election messages it receives, and determines the AGV with the highest election parameters as the master node AGV; If the election parameters are the same, a decision will be made according to a preset priority strategy, which includes: comparing the size of the AGV's physical identifier ID, or comparing the age of its last update timestamp.

7. The dynamic master-slave election method for multi-AGV collaborative scheduling according to claim 3, characterized in that, In step S5, the registration information sent by the slave AGV to the master AGV during registration includes at least the following: This node's unique identifier, current task status, current fault status, current location information, current battery status, current load capacity, and a list of executable task types.

8. The dynamic master-slave election method for multi-AGV collaborative scheduling according to claim 3, characterized in that, In step S3, the preset multivariate evaluation function is a weighted function, and its variables include at least: The remaining battery percentage of the AGV, the current load rate of the AGV's central processor, the average communication signal strength with other AGVs in the group, and the historical continuous normal working time.

9. The dynamic master-slave election method for multi-AGV collaborative scheduling according to claim 8, characterized in that: In step S3, the weight parameters of the multivariate evaluation function are adjusted offline or online based on the historical operation data of the vehicle group to optimize the impact of the election results on the overall operating efficiency of the system.

10. The dynamic master-slave election method for multi-AGV collaborative scheduling according to any one of claims 3-8, characterized in that: In step S4, the consensus-reaching phase also has a first timeout period; if the first timeout period expires and there are still AGVs that have not reached a consensus on the identity of the master node AGV, the current election state is cleared and the process returns to step S3 to re-initiate the election. In step S5, a second timeout period is set for the identity verification stage of the master node AGV and slave node AGV. If the master node AGV does not receive the registration information of all slave node AGVs before the second timeout period expires, the master node AGV broadcasts the list of registered slave node AGVs to the AGV fleet. Unregistered slave node AGVs are considered offline and can be added to the system later through the re-registration process.

Images