Path optimization method and device of unmanned vehicle and electronic equipment

Abstract

The invention discloses a path optimization method and device for an unmanned vehicle and electronic equipment. The method comprises the steps that network flow and traffic information in a factory are acquired, the network flow is used for representing the data transmission quantity between network devices in the factory, and the traffic information is used for representing the operation state of an unmanned vehicle in the factory; predicting the network traffic through the graph neural network model to obtain a traffic prediction result; the traffic cost of the unmanned vehicle is determined according to the flow prediction result and the traffic information, and the traffic cost is used for representing the passing cost of the unmanned vehicle on different paths; and determining the optimal path of the unmanned vehicle according to the flow prediction result and the traffic cost. The AGV scheduling method and device solve the technical problems that in a complex industrial network environment, an AGV scheduling method in the related technology cannot cope with real-time changes of network flow and traffic demands, scheduling delay exists, and the system response speed is low.

Description

Technical Field

[0001] This application relates to the technical field of AGV, and more specifically, to a path optimization method, device, and electronic device for unmanned vehicles. Background Art

[0002] In the current intelligent manufacturing environment, automated guided vehicles have become one of the key elements for improving production efficiency and flexibility. The core lies in how to quickly and effectively allocate AGVs (Automated Guided Vehicles) after network problems occur to achieve the best logistics efficiency. However, this method often faces significant limitations when dealing with actual factory operations, especially in dynamic and complex scenarios.

[0003] Specifically, the AGV scheduling method in the related art starts path planning and task allocation only after actual production requirements or material handling requests occur, which not only increases the response time but also makes it difficult to ensure the efficient operation of AGVs in a dynamically changing production environment. At the same time, problems such as insufficient network coverage and unstable signals further exacerbate the difficulty of resource allocation, especially in the multi-variety and small-batch production mode, where fixed network infrastructure is difficult to adapt to rapidly changing production requirements. In addition, the lack of resource allocation optimization and intelligent prediction capabilities limits the forward-looking and overall efficiency of AGV scheduling. Especially when dealing with complex scenarios of large scale, multiple tasks, and multiple devices, it is difficult to achieve a global optimal solution, resulting in unreasonable AGV operation paths and insufficient instant response capabilities of network traffic and traffic demands.

[0004] To address the above problems, no effective solutions have been proposed yet. Summary of the Invention

[0005] Embodiments of this application provide a path optimization method, device, and electronic device for unmanned vehicles to at least solve the technical problem that in a complex industrial network environment, the AGV scheduling method in the related art cannot cope with the real-time changes of network traffic and traffic demands, resulting in scheduling delays and slow system response speeds.

[0006] According to one aspect of the embodiments of the present application, a path optimization method for an unmanned vehicle is provided, including: obtaining network traffic and traffic information in a factory, where the network traffic is used to represent the data transmission volume between various network devices in the factory, and the traffic information is used to represent the operating state of the unmanned vehicles in the factory; predicting the network traffic through a graph neural network model to obtain a traffic prediction result, where the traffic prediction result is used to represent the predicted data transmission volume between various network devices in the factory in a future time period; determining the traffic cost of the unmanned vehicle based on the traffic prediction result and the traffic information, where the traffic cost is used to represent the passing cost of the unmanned vehicle on different paths; and determining the optimal path of the unmanned vehicle based on the traffic prediction result and the traffic cost.

[0007] Optionally, predicting the network traffic to obtain a traffic prediction result includes: determining the node features of the network devices through the first neural network in the graph neural network model, where the node features are used to represent the multi-dimensional state information of the network devices at different time points; determining the temporal features of the network devices through the second neural network in the graph neural network model, where the temporal features are used to reflect the periodic state changes of the network devices; and fusing the node features and the temporal features through the fully connected layer in the graph neural network model to obtain the traffic prediction result.

[0008] Optionally, determining the node features of the network devices through the first neural network in the graph neural network model includes: determining the network topology graph corresponding to the factory, where the network topology graph includes network device nodes corresponding to the network devices and the communication connection relationships between the various network device nodes; obtaining the node information and neighborhood information of the target network device from the network topology graph, where the target network device is any one of the network devices in the factory, the node information is used to represent the operating state of the target network device, and the neighborhood information is used to represent the communication requirements between the target network device and other network devices; and fusing the node information and the neighborhood information through the first neural network to obtain the target node features of the target network device.

[0009] Optionally, determining the traffic cost of the unmanned vehicle based on the traffic prediction result and the traffic information includes: obtaining the target traffic information of the unmanned vehicle on the first path from the traffic information, where the first path is any one of the initial path sets of the unmanned vehicle; determining the physical distance of the first path and determining the first network load corresponding to the first path based on the traffic prediction result; and determining the first traffic cost of the unmanned vehicle on the first path based on the target traffic information, the physical distance, and the first network load.

[0010] Optionally, the method further includes: determining a first fitness value of all initial paths of the driverless vehicle through a fitness function; determining, as a first optimized path, an initial path corresponding to a fitness value exceeding a preset threshold in the first fitness value, and iteratively optimizing the first optimized path through crossover operation and mutation operation, and stopping the optimization after reaching a preset number of iterations to obtain a second optimized path, where the first optimized path is used to represent a path to be optimized among the initial paths, and the second optimized path is used to represent a path obtained after multiple iterative optimizations of the first optimized path; determining a second fitness value of the second optimized path through the fitness function, and determining, as the optimal path, the optimized path corresponding to the maximum value in the second fitness value.

[0011] Optionally, the fitness function is determined in the following manner: determining a predicted traffic volume value of the driverless vehicle in a preset network area, and obtaining a network capacity threshold of the preset network area; determining an adjustment strength for the driverless vehicle based on the predicted traffic volume value and the network capacity threshold; and determining the fitness function based on the adjustment strength and the traffic cost.

[0012] Optionally, the method further includes: determining a second network load of the driverless vehicle on a second path according to a traffic prediction result, where the second path is any path in the second optimized path; determining a second traffic cost of the second path; and updating the optimal path based on the second traffic cost and the second network load.

[0013] According to another aspect of the embodiments of the present application, there is also provided a path optimization device for a driverless vehicle, including: an acquisition module, configured to acquire network traffic and traffic information in a factory, where the network traffic is used to represent the data transmission volume between each network device in the factory, and the traffic information is used to represent the running state of the driverless vehicle in the factory; a prediction module, configured to predict the network traffic through a graph neural network model to obtain a traffic prediction result, where the traffic prediction result is used to represent the predicted data transmission volume between each network device in the factory in a future time period; a first determination module, configured to determine the traffic cost of the driverless vehicle based on the traffic prediction result and the traffic information, where the traffic cost is used to represent the passing cost of the driverless vehicle on different paths; and a second determination module, configured to determine the optimal path of the driverless vehicle based on the traffic prediction result and the traffic cost.

[0014] According to yet another aspect of the embodiments of the present application, there is also provided an electronic device, including: a memory and a processor, where the memory is configured to store program instructions; and the processor is connected to the memory and configured to execute to implement the above-mentioned path optimization method for a driverless vehicle.

[0015] According to another aspect of the embodiments of the present application, a non-volatile storage medium is further provided. The non-volatile storage medium includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the above-mentioned path optimization method for the unmanned vehicle by running the computer program.

[0016] According to another aspect of the embodiments of the present application, a computer program product is further provided, including computer instructions that implement the above-mentioned path optimization method for the unmanned vehicle when executed by a processor.

[0017] In the embodiments of the present application, by obtaining the network traffic and traffic information in the factory, where the network traffic is used to represent the data transmission volume between various network devices in the factory, and the traffic information is used to represent the running state of the unmanned vehicles in the factory; predicting the network traffic through a graph neural network model to obtain a traffic prediction result, where the traffic prediction result is used to represent the predicted data transmission volume between various network devices in the factory in a future time period; determining the traffic cost of the unmanned vehicle based on the traffic prediction result and the traffic information, where the traffic cost is used to represent the passing cost of the unmanned vehicle on different paths; and determining the optimal path of the unmanned vehicle based on the traffic prediction result and the traffic cost, the purpose of intelligent prediction and dynamic optimization of the unmanned vehicle scheduling and path selection is achieved, thereby realizing the technical effects of improving the response speed, resource utilization efficiency and overall operation efficiency of the factory logistics system, and further solving the technical problem that in a complex industrial network environment, the AGV scheduling method in the related art cannot cope with the real-time changes of network traffic and traffic demand, resulting in scheduling delay and slow system response speed. Description of the Drawings

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings:

[0019] Figure 1 is a hardware structure diagram of a computer terminal for implementing the path optimization method for an unmanned vehicle according to an embodiment of the present application;

[0020] Figure 2 is a flowchart of a path optimization method for an unmanned vehicle according to an embodiment of the present application;

[0021] Figure 3 is a structure diagram of a path optimization device for an unmanned vehicle according to an embodiment of the present application. Detailed Embodiments

[0022] To enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

[0023] It should be noted that the terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of this application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0024] First, some nouns or terms that appear during the process of explaining and illustrating the embodiments of this application are applicable to the following explanations:

[0025] Automated Guided Vehicle (AGV): An automated vehicle that can navigate along a predetermined path or through wireless signals without human intervention, widely used in industries such as warehousing, manufacturing, and logistics for tasks such as material handling and cargo transportation. AGVs are usually equipped with sensors, control systems, and software, and can achieve automated logistics handling in a production environment, and are an important part of realizing intelligent manufacturing and automated warehousing systems.

[0026] Mobile Ad Hoc Network (MANET): A self-organizing wireless network in which nodes (such as mobile devices like AGVs) can dynamically form a network without relying on a pre-existing network infrastructure (such as base stations, routers, etc.). In a MANET, each node is both a user and a forwarder of the network, and can dynamically adjust the network topology according to network requirements and environmental changes to achieve communication. Mobile ad hoc networking has a wide range of applications in fields such as military, emergency rescue, and automated logistics.

[0027] Genetic Algorithm (GA): A search optimization technique inspired by Darwin's natural selection and genetic principles. In a genetic algorithm, the solutions to a problem are encoded as chromosomes, and a population contains multiple chromosomes. The genetic algorithm generates new solution sets by simulating the selection, crossover (gene recombination), and mutation operations in the biological evolution process, and finally finds the optimal solution or approximate optimal solution to the problem. Genetic algorithms are applicable to solving optimization problems with high complexity and large solution spaces, such as path planning and scheduling tasks.

[0028] GNN (Graph Neural Networks): A machine learning model for processing graph-structured data. GNN can capture the complex dependencies between nodes in a graph and update node representations through an information propagation mechanism, and is applicable to tasks such as graph classification, node classification, and link prediction.

[0029] GCN (Graph Convolutional Networks): A specific implementation of graph neural networks (GNN). GCN aggregates the neighborhood information of nodes through graph convolutional layers to update the feature representations of nodes. Different from the application of convolutional operations in traditional convolutional neural networks on regular grids such as images, GCN can achieve similar functions on irregular graph structures and capture the information of local structures in the graph.

[0030] ST-GNN (Spatial-Temporal Graph Neural Network): An extended version of graph neural networks (GNN) specifically designed to process data with spatio-temporal attributes. In many real-world scenarios, data not only has a graph structure of nodes and edges but also has a time dimension, such as traffic flow, meteorological data, video frames, etc. ST-GNN can analyze both the static graph structure and dynamic time series characteristics of the data to make more accurate predictions or decisions.

[0031] MSE (Mean Squared Error): A commonly used metric for evaluating the performance of a prediction model. MSE is the average of the sum of the squares of the differences between the predicted values and the true values. During the model training process, MSE is often used as a loss function, and the algorithm tries to minimize MSE to optimize the model parameters to improve the prediction accuracy.

[0032] To solve the problem of poor AGV scheduling efficiency in the related art, the embodiments of the present application provide a path optimization method for unmanned vehicles, which can run on Figure 1 the computer terminal shown below, and the computer terminal will be described as follows.

[0033] The embodiment of the path optimization method for an unmanned vehicle provided by the embodiments of the present application can be executed on a mobile terminal, a computer terminal, or a similar computing device. Figure 1 The following shows a hardware structure block diagram of a computer terminal for implementing the path optimization method of an unmanned vehicle. As Figure 1 shown, the computer terminal 10 may include one or more processors (illustrated as 102a, 102b,..., 102n in the figure) (the processor may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission module 106 for communication functions connected by wired and / or wireless networks. In addition, it may further include: a display, a keyboard, a cursor control device, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, and a BUS bus. Those of ordinary skill in the art can understand that Figure 1 the structure shown is only schematic and does not limit the structure of the above-mentioned electronic device. For example, the computer terminal 10 may further include more or fewer components than those Figure 1 shown, or have a different configuration from that Figure 1 shown.

[0034] It should be noted that the above one or more processors and / or other data processing circuits are generally referred to as "data processing circuits" in this article. The data processing circuit may be embodied in whole or in part as software, hardware, firmware, or any combination thereof. In addition, the data processing circuit may be a single independent processing module, or be incorporated in whole or in part into any one of the other elements in the computer terminal 10. As involved in the embodiments of the present application, the data processing circuit is a kind of processor control (such as the selection of a variable resistor terminal path connected to an interface).

[0035] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the path optimization method of the unmanned vehicle in the embodiments of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, implements the above-mentioned path optimization method of the unmanned vehicle. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, a flash memory, or other non-volatile solid-state memories. In some instances, the memory 104 may further include a memory remotely disposed relative to the processor, and these remote memories may be connected to the computer terminal 10 through a network. Examples of the above networks include, but are not limited to, the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0036] The transmission module 106 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission module 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission module 106 can be a Radio Frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0037] The display can be, for example, a touch-screen liquid crystal display (LCD), which enables the user to interact with the user interface of the computer terminal 10.

[0038] It should be noted here that in some alternative embodiments, the above Figure 1 illustrated computer terminal may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of both hardware elements and software elements. It should be pointed out that Figure 1 is only an example of a specific specific instance and is intended to show the types of components that may exist in the above computer terminal.

[0039] Under the above operating environment, an embodiment of a path optimization method for an unmanned vehicle is provided in an embodiment of the present application. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0040] Figure 2 is a flowchart of a path optimization method for an unmanned vehicle according to an embodiment of the present application. As Figure 2 shown, the method includes the following steps:

[0041] Step S202, obtain the network traffic and traffic information in the factory, where the network traffic is used to represent the data transmission volume between various network devices in the factory, and the traffic information is used to represent the operating status of the unmanned vehicles in the factory.

[0042] In the above step S202, a sensor network, a communication system, or real-time monitoring technology can be used to collect the network traffic between various network devices in the factory and the traffic information of unmanned vehicles (taking AGV as an example in the present application). Among them, the network traffic reflects the activity degree and data demand of internal communication in the factory, while the traffic information provides the operating status data such as the current distribution, operating speed, direction, and task status of the AGV.

[0043] Step S204, predicting the network traffic through a graph neural network model to obtain a traffic prediction result, where the traffic prediction result is used to represent the predicted data transmission volume between each network device in the factory within a future time period.

[0044] In the above step S204, by using a graph neural network (GNN) model to perform in-depth learning analysis on the collected network traffic data, the expected data transmission volume between each network device in the factory within a future time period can be accurately predicted. Among them, the GNN model can capture the complex relationships and spatio-temporal dependencies between network devices, and generate a high-precision traffic prediction result by aggregating node features and propagating information.

[0045] Step S206, determining the traffic cost of the unmanned vehicle according to the traffic prediction result and traffic information, where the traffic cost is used to represent the passing cost of the unmanned vehicle on different paths.

[0046] In the above step S206, based on the traffic prediction result obtained in step S204 and the traffic information in step S202, the traffic cost of the AGV on different paths can be calculated through a traffic cost function. Among them, the traffic cost comprehensively considers factors such as network load, physical distance, and traffic congestion degree, and is a quantitative index that can reflect the passing efficiency and cost of the AGV on a specific path. By calculating the traffic cost, the system can evaluate the advantages and disadvantages of each path and provide support for the subsequent optimal path selection.

[0047] Step S208, determining the optimal path of the unmanned vehicle according to the traffic prediction result and traffic cost.

[0048] In the above step S208, after fully analyzing the network traffic prediction result and traffic cost, a genetic algorithm (GA) can be used to screen out the optimal path from all possible paths. Specifically, by simulating the natural selection process through the genetic algorithm, integrating the traffic prediction result and traffic cost into the fitness function, evaluating the fitness of each path, and gradually optimizing the population through operations such as selection, crossover, and mutation, the path combination with the lowest comprehensive cost and highest efficiency can be finally found. This step realizes the transformation of path planning from passive response to active optimization, ensures that the AGV can run on the best path, thereby reducing network latency, avoiding traffic congestion, and improving the operation efficiency of the entire logistics system.

[0049] Through the above steps S202 to S208, the purpose of intelligent prediction and dynamic optimization of unmanned vehicle scheduling and path selection is achieved, thus realizing the technical effects of improving the response speed, resource utilization efficiency and overall operation efficiency of the factory logistics system, and further solving the technical problem that in a complex industrial network environment, the AGV scheduling method in the related technology cannot cope with the real-time changes of network traffic and traffic demand, resulting in scheduling delay and slow system response speed. The following is a detailed description.

[0050] In the above step S204, the network traffic is predicted to obtain a traffic prediction result, including: determining the node features of network devices through the first neural network in the graph neural network model, where the node features are used to represent the multi-dimensional state information of network devices at different time points; determining the temporal features of network devices through the second neural network in the graph neural network model, where the temporal features are used to reflect the periodic state changes of network devices; and fusing the node features and temporal features through the fully connected layer in the graph neural network model to obtain the traffic prediction result.

[0051] Optionally, determining the node features of network devices through the first neural network in the graph neural network model includes: determining a network topology graph corresponding to the factory, where the network topology graph includes network device nodes corresponding to network devices and the communication connection relationships between each network device node; obtaining the node information and neighborhood information of the target network device from the network topology graph, where the target network device is any network device in the factory, the node information is used to represent the operating state of the target network device, and the neighborhood information is used to represent the communication requirements between the target network device and other network devices; and fusing the node information and neighborhood information through the first neural network to obtain the target node features of the target network device.

[0052] In the embodiment of the present application, network traffic prediction is a multi-stage and multi-perspective deep learning process. It comprehensively considers the static multi-dimensional state information and dynamic time series changes of network devices, and captures the spatio-temporal characteristics of network traffic through the structured learning ability of the graph neural network to generate accurate traffic prediction results. The specific process can be as follows:

[0053] S1. Data preprocessing.

[0054] In order to achieve accurate prediction of network traffic in the factory, it is first necessary to perform data preprocessing on the network traffic of factory network devices. For example: cleaning the collected data to remove outliers or missing data to ensure the quality of the data; then performing feature selection and feature engineering to retain the features that contribute to network traffic prediction; finally, performing standardization or normalization processing on the data to ensure the numerical comparability of features and the stability of model training. Subsequently, the preprocessed network traffic is input into the GNN model.

[0055] S2. Determine the node features of the network device.

[0056] First, construct a network topology diagram that reflects the internal network devices of the factory and their communication connection relationships. Specifically, network devices can be mapped to nodes in the diagram, where each node carries a feature vector that contains multi-dimensional information of the device, such as traffic, bandwidth, load, etc. Further, define the edges between nodes to reflect the connection relationships between network devices, where the edges can be based on the actual physical connections between devices (such as Ethernet cable connections) or logical communication requirements (such as device A transmitting data to device B).

[0057] Second, for any network device in the factory (i.e., the above-mentioned target network device), obtain its own operating status information (such as traffic, bandwidth occupancy, device load, etc.), that is, node information; and obtain the communication requirement information between the target network device and other network devices, that is, neighborhood information. Among them, the node information reflects the static attributes of the target network device, while the neighborhood information reveals the dynamic interaction relationships between network devices.

[0058] Finally, process the node information and neighborhood information through the first neural network (such as GCN or a similar structure) in the GNN model. After feature fusion and update, output the target node features of each network device, where the target node features synthesize the device's own state and its interaction with other devices in the network environment. This operation captures the spatio-temporal features of the network device status and communication relationships through the local aggregation and information transfer mechanism of the graph convolutional layer, providing more comprehensive input information for traffic prediction.

[0059] S3. Determine the temporal features of the network device.

[0060] Model the periodic changes in the network device status through the second neural network (such as TGNN or a similar structure) in the GNN model to capture the fluctuation law of network traffic over time. That is, by analyzing the patterns and trends in historical data, predict the periodic changes in future network traffic, thereby obtaining temporal features. These temporal features are crucial for understanding the workload and traffic demand of network devices at different time periods and can effectively improve the accuracy of traffic prediction.

[0061] S4. Fuse the node features and temporal features.

[0062] Fuse the node features and temporal features through the fully connected layer in the GNN to obtain the final traffic prediction result. Among them, the fully connected layer can process information from different sources, integrate and transform these features by learning the weight matrix, and generate a prediction of the future trend of network traffic.

[0063] Generally speaking, the above traffic prediction process makes full use of the advantages of GNN in processing graph-structured data, combines the static operating status of devices and dynamic time series information, enabling the model to comprehensively understand and predict changes in network traffic. Among them, the first neural network is responsible for learning the static and neighborhood features of devices, which helps to understand the direct relationships and communication requirements between devices; while the second neural network focuses on the dynamic changes in device status and can capture periodic patterns. Finally, the fully connected layer serves as a "bridge" to fuse static and dynamic features and generate predictions for the future state of network traffic.

[0064] In step S206 above, determining the traffic cost of the unmanned vehicle based on the traffic prediction result and traffic information includes: obtaining the target traffic information of the unmanned vehicle on the first path from the traffic information, where the first path is any path in the initial path set of the unmanned vehicle; determining the physical distance of the first path, and determining the first network load corresponding to the first path according to the traffic prediction result; determining the first traffic cost of the unmanned vehicle on the first path based on the target traffic information, physical distance, and first network load.

[0065] In the embodiments of the present application, by quantifying the traffic cost of the path, the optimal path can be identified, that is, the path that achieves the best balance in terms of network resource utilization, traffic efficiency, and task completion time.

[0066] Specifically, for the path planning of the AGV, the traffic cost of each path can be calculated through a traffic control algorithm or a traffic cost function according to the current network traffic, traffic information, and path congestion situation. Suppose there are N paths, and the traffic cost calculation formula for the path can be as follows:

[0067] C i = w1·Distance i + w2·Traffic i + w3·Network load i

[0068] In the formula, C i represents the traffic cost of the AGV on the i-th path (such as the first path), that is, the first traffic cost; Distance i represents the physical distance of the first path; Traffic i represents the target traffic information of the AGV on the first path (including the traffic density, traffic capacity, etc. of the first path); Network load i represents the first network load corresponding to the first path obtained based on the traffic prediction result; w1, w2, and w3 are weight coefficients.

[0069] In the above step S208, determining the optimal path of the driverless vehicle according to the traffic flow prediction result and the traffic cost includes: determining the first fitness value of all initial paths of the driverless vehicle through a fitness function; determining the initial paths corresponding to the fitness values exceeding a preset threshold in the first fitness values as the first optimized paths, and iteratively optimizing the first optimized paths through crossover operations and mutation operations, and stopping the optimization after reaching the preset number of iterations to obtain the second optimized paths, where the first optimized paths are used to represent the paths to be optimized in the initial paths, and the second optimized paths are used to represent the paths obtained after multiple iterative optimizations of the first optimized paths; determining the second fitness value of the second optimized paths through the fitness function, and determining the optimized path corresponding to the maximum value in the second fitness values as the optimal path.

[0070] In the embodiments of the present application, a genetic algorithm can be used to determine the optimal path of the AGV according to the traffic flow prediction result and the traffic cost to achieve dynamic optimization of path selection. The specific process can be as follows:

[0071] First, generate a population containing a set of multiple initial paths, where each path corresponds to an individual, representing a driving route that the AGV may select.

[0072] Secondly, apply the fitness function to each individual (i.e., each initial path) in the population to calculate the first fitness value of each path to evaluate its ability to complete the task under network load and traffic conditions. Among them, the fitness function comprehensively considers the physical distance, network traffic and traffic cost to quantify the selection efficiency and cost of the path.

[0073] According to the calculated first fitness value, select the individuals with fitness values exceeding the preset threshold through a selection mechanism (such as roulette wheel selection) and determine them as the first optimized paths. The selection operation ensures that the paths with higher fitness in the population have the opportunity to be inherited to the next generation, thus gradually approaching the optimal solution. Then, perform crossover and mutation operations on the first optimized paths to generate offspring individuals. Among them, the crossover operation enables information exchange between paths, which may lead to new path combinations, while the mutation operation explores a new solution space by randomly changing some features in the path to increase the diversity of the population.

[0074] Repeat the selection, crossover and mutation operations until the preset number of iterations is reached to obtain the second optimized paths. In each iteration process, the system updates the fitness value, selects better paths for genetic operations, and gradually optimizes the path set to improve the efficiency and effect of AGV scheduling.

[0075] Finally, in the second optimization path, the fitness function is used again to calculate the fitness value of each path, that is, the second fitness value, and the path with the largest fitness value is selected as the optimal path. After considering the physical distance, network load, and traffic cost, this optimal path can complete the task with the lowest cost and the highest efficiency.

[0076] In the above process, not only the iterative optimization characteristics of the genetic algorithm are reflected, but also the real-time consideration of network traffic and traffic conditions is incorporated. By dynamically adjusting and optimizing the path of the AGV, it is ensured that in a complex and changeable production environment, the AGV can select the most suitable driving route according to the predicted network demand and real-time traffic conditions, thereby improving the foresight and flexibility of logistics scheduling and further enhancing the overall response speed and resource utilization efficiency of the intelligent manufacturing system.

[0077] Optionally, the fitness function is determined in the following manner: determining the predicted traffic value of the unmanned vehicle in the preset network area, and obtaining the network capacity threshold of the preset network area; determining the adjustment strength for the unmanned vehicle based on the predicted traffic value and the network capacity threshold; and determining the fitness function based on the adjustment strength and the traffic cost.

[0078] In the embodiments of the present application, the design strategy of the fitness function ensures that the AGV scheduling not only focuses on the minimum traffic cost on the physical route, but also takes into account the dynamic changes in network traffic and the limitations of network capacity. Specifically, the fitness function F(X) can be defined by the following formula:

[0079]

[0080] In the formula, c(p ij ) is the traffic cost on path p ij , representing the cost for the AGV to travel from location i to location j. x ij represents the selection of the AGV on path p ij in individual X, usually a binary variable indicating whether to select this path p ij (for example, 1 means selection and 0 means non-selection). γ represents the penalty factor, used to control the penalty strength for non-compliance with the constraint conditions. Penalty(X) represents the penalty function, used to penalize the parts that do not meet the constraint conditions, such as violations of network traffic or traffic restrictions, as follows:

[0081]

[0082] In the formula, M represents the number of network areas; represents the predicted traffic demand of the preset network area (such as the kth network area), and Threshold k represents the network capacity threshold of this preset network area.

[0083] Optionally, the above method further includes: determining a second network load of the unmanned vehicle on a second path according to the traffic prediction result, where the second path is any one of the second optimized paths; determining a second traffic cost of the second path; and updating the optimal path according to the second traffic cost and the second network load.

[0084] In the embodiments of the present application, the genetic algorithm generates a series of optimized paths in the offline or preprocessing stage, and the path with the highest fitness is regarded as the optimal response to the preset network traffic prediction and traffic information. However, the factory environment is dynamic, and the network traffic and traffic conditions change over time. Therefore, the output of the genetic algorithm can be regarded as the optimal solution at a certain point in time or based on a certain prediction situation.

[0085] Nomadic networking refers to dynamically adjusting the path selection of the AGV according to the real-time updated network traffic prediction and traffic information during the actual operation of the AGV. This process is a real-time decision based on the set of optimal paths output by the genetic algorithm. The "nomadic" attribute of nomadic networking is reflected in that the AGV can flexibly adjust its travel route according to the latest network and traffic conditions, just like a nomadic people migrating with the environment.

[0086] Therefore, in the actual factory environment, even if the path of the AGV is pre-optimized using the genetic algorithm, when unexpected traffic peaks, network congestion or traffic obstacles occur in the actual operation environment, it is still necessary to make immediate adjustments in combination with nomadic networking to avoid delays or interruptions.

[0087] Suppose that at time t, the AGV needs to select a path to complete a task. Through the calculation of network traffic prediction and traffic cost, the AGV can select the optimal path according to the following decision rule:

[0088]

[0089] In the formula, P * (t) represents the optimal path selected by the AGV at time t; C i represents the traffic cost of the AGV on the second path, that is, the second traffic cost; represents the second network load corresponding to the second path obtained based on the traffic prediction result; γ represents a weight coefficient used to control the influence of the traffic prediction result on the path selection.

[0090] Generally speaking, the genetic algorithm provides an initial set of optimal paths for nomadic networking, and these paths are determined as the best choices suitable for the current prediction situation at the end of the algorithm. In nomadic networking, during the actual operation of AGVs, the path set provided by the genetic algorithm and real-time traffic prediction and traffic information are used to perform dynamic path selection and optimization, ensuring the immediacy and flexibility of AGV scheduling to cope with unpredictable changes in the factory environment.

[0091] In this application, the traffic prediction ability of GNN and the optimization advantages of the genetic algorithm are combined to propose a new solution idea for the nomadic networking and scheduling problems of AGVs in the intelligent manufacturing environment. Through the deep learning ability of the graph neural network, the system can accurately predict the future traffic changes in the factory network, capture complex spatio-temporal dependencies, and provide forward-looking network traffic information for AGV scheduling. On this basis, the genetic algorithm uses these prediction results and real-time traffic data to dynamically optimize the AGV path and task allocation by intelligently adjusting the fitness function, crossover rate, and mutation rate, ensuring that AGVs can respond efficiently when facing dynamic changes in the production environment, avoiding the waste of network resources and the negative impacts brought by traffic congestion. This innovative method not only improves the efficiency and intelligence level of AGV scheduling but also optimizes the configuration of network resources, demonstrating the powerful potential of data-driven and intelligent decision-making in intelligent manufacturing systems and providing strong technical support for realizing highly automated logistics management and production scheduling.

[0092] According to an embodiment of the present application, a path optimization device for an unmanned vehicle is provided. It should be noted that the path optimization device for an unmanned vehicle in the embodiment of the present application can be used to execute the path optimization method for an unmanned vehicle provided in the embodiment of the present application. The following introduces the path optimization device for an unmanned vehicle provided in the embodiment of the present application.

[0093] Figure 3 is a structural diagram of a path optimization device for an unmanned vehicle provided according to an embodiment of the present application. As Figure 3 shown, the device includes:

[0094] An acquisition module 30, configured to acquire network traffic and traffic information in the factory, where the network traffic is used to represent the data transmission volume between various network devices in the factory, and the traffic information is used to represent the running state of unmanned vehicles in the factory;

[0095] A prediction module 32, configured to predict the network traffic through a graph neural network model to obtain a traffic prediction result, where the traffic prediction result is used to represent the predicted data transmission volume between various network devices in the factory in a future time period;

[0096] The first determination module 34 is configured to determine the traffic cost of the driverless vehicle according to the traffic flow prediction result and traffic information, where the traffic cost is used to represent the passing cost of the driverless vehicle on different paths;

[0097] The second determination module 36 is configured to determine the optimal path of the driverless vehicle according to the traffic flow prediction result and the traffic cost.

[0098] Through the acquisition module, prediction module, first determination module and second determination module in the above path optimization device for driverless vehicles, the purpose of intelligent prediction and dynamic optimization of driverless vehicle scheduling and path selection is achieved, thereby realizing the technical effects of improving the response speed, resource utilization efficiency and overall operation efficiency of the factory logistics system, and further solving the technical problem that in a complex industrial network environment, the AGV scheduling method in the related technology cannot cope with the real-time changes of network traffic and traffic demand, and there are problems such as scheduling delay and slow system response speed.

[0099] In the path optimization device for driverless vehicles provided in the embodiment of the present application, the prediction module is further configured to determine the node features of the network device through the first neural network in the graph neural network model, where the node features are used to represent the multi-dimensional state information of the network device at different time points; determine the temporal features of the network device through the second neural network in the graph neural network model, where the temporal features are used to reflect the periodic state changes of the network device; fuse the node features and the temporal features through the fully connected layer in the graph neural network model to obtain the traffic flow prediction result.

[0100] In the path optimization device for driverless vehicles provided in the embodiment of the present application, the prediction module is further configured to determine the network topology graph corresponding to the factory, where the network topology graph includes network device nodes corresponding to network devices and the communication connection relationships between each network device node; obtain the node information and neighborhood information of the target network device from the network topology graph, where the target network device is any network device in the factory, the node information is used to represent the operating state of the target network device, and the neighborhood information is used to represent the communication requirements between the target network device and other network devices; fuse the node information and the neighborhood information through the first neural network to obtain the target node features of the target network device.

[0101] In the path optimization device for driverless vehicles provided in the embodiment of the present application, the first determination module is further configured to obtain the target traffic information of the driverless vehicle on the first path from the traffic information, where the first path is any path in the initial path set of the driverless vehicle; determine the physical distance of the first path and determine the first network load corresponding to the first path according to the traffic flow prediction result; determine the first traffic cost of the driverless vehicle on the first path according to the target traffic information, physical distance and first network load.

[0102] In the path optimization device for an unmanned vehicle provided in the embodiment of the present application, the second determination module is further configured to determine the first fitness value of all initial paths of the unmanned vehicle through a fitness function; determine the initial paths corresponding to the fitness values exceeding a preset threshold in the first fitness values as the first optimized paths, and iteratively optimize the first optimized paths through crossover operations and mutation operations, and stop the optimization after reaching a preset number of iterations to obtain second optimized paths, where the first optimized paths are used to represent the paths to be optimized in the initial paths, and the second optimized paths are used to represent the paths obtained after multiple iterative optimizations of the first optimized paths; determine the second fitness value of the second optimized paths through the fitness function, and determine the optimized path corresponding to the maximum value in the second fitness values as the optimal path.

[0103] In the path optimization device for an unmanned vehicle provided in the embodiment of the present application, the second determination module is further configured to determine the predicted traffic value of the unmanned vehicle in a preset network area, and obtain the network capacity threshold of the preset network area; determine the adjustment strength for the unmanned vehicle according to the predicted traffic value and the network capacity threshold; determine the fitness function according to the adjustment strength and the traffic cost.

[0104] In the path optimization device for an unmanned vehicle provided in the embodiment of the present application, it further includes an update module 38, and the update module is configured to determine the second network load of the unmanned vehicle on a second path according to the traffic prediction result, where the second path is any one of the second optimized paths; determine the second traffic cost of the second path; update the optimal path according to the second traffic cost and the second network load.

[0105] The embodiment of the present application further provides an electronic device, including: a memory and a processor, where the memory is used to store program instructions; the processor is connected to the memory and is configured to execute to implement the above-mentioned path optimization method for an unmanned vehicle.

[0106] It should be noted that the above-mentioned electronic device is used to execute Figure 2 the path optimization method for an unmanned vehicle shown, so the relevant explanations in the above-mentioned path optimization method for an unmanned vehicle also apply to this electronic device, and will not be elaborated here.

[0107] The embodiment of the present application further provides a non-volatile storage medium, and the non-volatile storage medium includes a stored computer program, where the device where the non-volatile storage medium is located executes the above-mentioned path optimization method for an unmanned vehicle by running the computer program.

[0108] It should be noted that the above-mentioned non-volatile storage medium is used to execute Figure 2 the path optimization method for an unmanned vehicle shown, so the relevant explanations in the above-mentioned path optimization method for an unmanned vehicle also apply to this non-volatile storage medium, and will not be elaborated here.

[0109] An embodiment of the present application also provides a computer program product, including computer instructions, which implement the above-mentioned path optimization method for the unmanned vehicle when executed by a processor.

[0110] It should be noted that the above computer program product is used to execute Figure 2 the path optimization method for the unmanned vehicle shown above. Therefore, the relevant explanations in the above path optimization method for the unmanned vehicle also apply to this computer program product, and will not be elaborated here.

[0111] The serial numbers of the above embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0112] In the above embodiments of the present application, the descriptions of the respective embodiments have their own emphases. For parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0113] In several embodiments provided by the present application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units can be a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of units or modules can be in an electrical or other form.

[0114] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0115] In addition, the functional units in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0116] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), mobile hard disks, magnetic disks, or optical discs.

[0117] The above are only the preferred embodiments of this application. It should be noted that for those of ordinary skill in the art, without departing from the principle of this application, several improvements and refinements can still be made, and these improvements and refinements should also be regarded as the protection scope of this application.

Claims

1. A path optimization method for an unmanned vehicle, characterized in that: include: Acquire network traffic and traffic information within the factory, wherein the network traffic is used to indicate the amount of data transmission between various network devices within the factory, and the traffic information is used to indicate the operating status of unmanned vehicles within the factory; Predicting the network traffic through a graph neural network model to obtain a traffic prediction result, wherein the traffic prediction result is used to represent the predicted data transmission volume between various network devices in the factory in a future time period; Determining the traffic cost of the unmanned vehicle according to the traffic prediction result and the traffic information, wherein the traffic cost is used to represent the travel cost of the unmanned vehicle on different paths; An optimal path for the unmanned vehicle is determined based on the traffic prediction result and the traffic cost.

2. The method according to claim 1, characterized in that Predicting the network traffic to obtain traffic prediction results includes: Determining node features of the network device through a first neural network in the graph neural network model, wherein the node features are used to represent multidimensional state information of the network device at different time points; Determining a timing feature of the network device through a second neural network in the graph neural network model, wherein the timing feature is used to reflect a periodic state change of the network device; The traffic prediction result is obtained by fusing the node features and the time series features through the fully connected layer in the graph neural network model.

3. The method according to claim 2, characterized in that Determining the node characteristics of the network device by a first neural network in the graph neural network model includes: Determine a network topology diagram corresponding to the factory, wherein the network topology diagram includes network device nodes corresponding to the network devices, and communication connection relationships between the network device nodes; Acquire node information and neighborhood information of a target network device from the network topology diagram, wherein the target network device is any network device in the factory, the node information is used to indicate the operation status of the target network device, and the neighborhood information is used to indicate the communication requirements between the target network device and other network devices; The node information and the neighborhood information are fused through the first neural network to obtain the target node characteristics of the target network device.

4. The method according to claim 1, determining the traffic cost of the unmanned vehicle based on the traffic prediction result and the traffic information, comprising: Acquire target traffic information of the unmanned vehicle on a first path from the traffic information, wherein the first path is any one path in an initial path set of the unmanned vehicle; Determining a physical distance of the first path, and determining a first network load corresponding to the first path according to the traffic prediction result; A first traffic cost of the unmanned vehicle on the first path is determined according to the target traffic information, the physical distance, and the first network load.

5. The method according to claim 1, further comprising: Determining a first fitness value of all initial paths of the unmanned vehicle by means of a fitness function; Determine an initial path corresponding to a fitness value exceeding a preset threshold value in the first fitness value as a first optimized path, and iteratively optimize the first optimized path through a crossover operation and a mutation operation, stop the optimization after reaching a preset number of iterations, and obtain a second optimized path, wherein the first optimized path is used to represent a path to be optimized in the initial path, and the second optimized path is used to represent a path obtained after multiple iterative optimizations of the first optimized path; The second fitness value of the second optimization path is determined by the fitness function, and the optimization path corresponding to the maximum value of the second fitness value is determined as the optimal path.

6. The method according to claim 5, wherein the fitness function is determined by: Determining a predicted flow value of the unmanned vehicle in a preset network area, and obtaining a network capacity threshold of the preset network area; Determining the adjustment strength of the unmanned vehicle according to the predicted traffic value and the network capacity threshold; The fitness function is determined according to the adjustment strength and the traffic cost.

7. The method according to claim 5, further comprising: Determining a second network load of the unmanned vehicle on a second path according to the traffic prediction result, wherein the second path is any one of the second optimized paths; determining a second traffic cost of the second path; The optimal path is updated according to the second traffic cost and the second network load.

8. A path optimization device for an unmanned vehicle, characterized in that: include: An acquisition module, used to acquire network traffic and traffic information in the factory, wherein the network traffic is used to represent the amount of data transmission between various network devices in the factory, and the traffic information is used to represent the operating status of unmanned vehicles in the factory; A prediction module, used to predict the network traffic through a graph neural network model to obtain a traffic prediction result, wherein the traffic prediction result is used to represent the predicted data transmission volume between various network devices in the factory in a future time period; A first determination module, configured to determine a traffic cost of the unmanned vehicle according to the traffic prediction result and the traffic information, wherein the traffic cost is used to represent the travel cost of the unmanned vehicle on different paths; The second determination module is used to determine the optimal path of the unmanned vehicle according to the traffic prediction result and the traffic cost.

9. An electronic device, characterized in that: include: A memory and a processor, wherein the memory is used to store program instructions; The processor is connected to the memory and is used to execute the path optimization method for the unmanned vehicle as described in any one of claims 1 to 7.

10. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the path optimization method for the unmanned vehicle as described in any one of claims 1 to 7 by running the computer program.

11. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by the processor, the path optimization method for the unmanned vehicle described in any one of claims 1 to 7 is implemented.

Images