Multi-ship collaborative collision avoidance decision-making method and related equipment

Abstract

The embodiment of the invention provides a multi-ship collaborative collision avoidance decision-making method and related equipment, and belongs to the technical field of water traffic safety. The method comprises the steps of collecting current state information of each ship in a water area; performing inter-ship collision risk analysis according to the current state information of each ship to obtain a collision risk identification result of any two ships; a water area traffic map is constructed according to the collision risk identification result, nodes of the water area traffic map represent ship objects, and edges of the water area traffic map represent collision risks of the two nodes; graph structure features of nodes are extracted from the water area traffic graph to obtain multi-dimensional graph structure features of all the nodes; through a node scoring model based on the multilayer graph attention network, performing node importance scoring according to the multidimensional graph structure features of each node; performing key risk ship identification according to the importance score of each ship to obtain a risk ship identification result, and performing multi-ship collaborative collision avoidance decision to obtain a collision avoidance scheme. According to the invention, the reliability of ship anti-collision decision can be improved, and the navigation safety is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water traffic safety, and in particular to a multi-ship cooperative collision avoidance decision method and related equipment. BACKGROUND

[0002] With the rapid development of global maritime trade, the density of maritime traffic continues to increase, and ships frequently intersect in port waters, significantly increasing the risk of collision. Ship collision accidents not only cause casualties and property losses, but also may cause secondary disasters such as marine environmental pollution, which has a serious impact on the marine ecosystem and shipping economy. Currently, the ship traffic service system (VTS) mainly relies on manual monitoring mode, and the operator needs to process a large amount of AIS (Automatic Identification System) data to identify high-risk ships and make collision avoidance decisions within a limited time. However, this manual method cannot accurately and efficiently identify potential collision risks in the entire domain, resulting in low reliability of collision avoidance decisions and high safety risks for ship navigation. SUMMARY

[0003] The main purpose of the embodiments of the present application is to propose a multi-ship cooperative collision avoidance decision method and related equipment, aiming to improve the reliability of ship collision avoidance decisions in water areas and improve the safety of ship navigation in water areas.

[0004] To achieve the above-mentioned purpose, one aspect of the embodiments of the present application proposes a multi-ship cooperative collision avoidance decision method, comprising the following steps: Collecting the current state information of each ship in the water area; Performing collision risk analysis between ships according to the current state information of each ship to obtain collision risk identification results of any two ships; Constructing a water area traffic graph according to the collision risk identification results, wherein the nodes of the water area traffic graph represent ship objects, and the edges represent the existence of collision risk between two nodes; Extracting the graph structure features of the nodes of the water area traffic graph to obtain the multi-dimensional graph structure features of each node; Performing node importance scoring according to the multi-dimensional graph structure features of each node through a node scoring model based on a multi-layer graph attention network to obtain the importance scores of each ship; Performing key risk ship identification according to the importance scores of each ship to obtain risk ship identification results; Performing multi-ship cooperative collision avoidance decision according to the risk ship identification results to obtain a collision avoidance scheme.

[0005] In some embodiments, the collision risk analysis between ships according to the current state information of each ship to obtain collision risk identification results of any two ships comprises the following steps: Calculate the minimum encounter distance and encounter time between any two ships based on the current status information of each ship; Based on the minimum encounter distance and the encounter time, a collision risk analysis is performed on the two corresponding ships to obtain the collision risk values ​​of the two ships; The collision risk identification result is formed based on the collision risk values ​​between all two of the aforementioned vessels.

[0006] In some embodiments, the edge direction of the waterway traffic map represents the collision direction between vessels, and the edge weight represents the collision risk value when there is a collision risk. The multidimensional graph structure features include weighted degree centrality, betweenness centrality, proximity centrality, eigenvector centrality, rate of change of local connectivity coefficients, redundant path count, clustering coefficient, number of cores, and network efficiency-related features.

[0007] In some embodiments, the step of scoring the importance of each ship by using a node scoring model based on a multi-layer graph attention network, according to the multi-dimensional graph structure features of each node, includes the following steps: The multidimensional graph structure features of each node are input into the multi-layer graph attention network of the node scoring model to extract the inter-node association features, and the output feature matrix of the multi-layer graph attention network is obtained. The output feature matrix is ​​input into the multilayer perceptron of the node scoring model to predict the node scores and obtain the importance scores of each ship.

[0008] In some embodiments, the multi-layer graph attention network includes an input layer, a multi-head attention layer, a residual connection module, and a normalization layer; the step of inputting the multi-dimensional graph structure features of each node into the multi-layer graph attention network of the node scoring model to extract inter-node association features and obtain the output feature matrix of the multi-layer graph attention network includes the following steps: The multidimensional graph structure features of all the nodes are input into the input layer for preprocessing to obtain the input features; The input features are input into the multi-head attention layer to perform feature aggregation between nodes, resulting in node-aggregated features. The residual connection module performs a residual connection on the input features and the node aggregation features to obtain residual connection features; The residual connection features are input into the normalization layer for normalization processing to obtain the output feature matrix of the multi-layer graph attention network.

[0009] In some embodiments, the node scoring model is trained through the following steps: A greedy strategy based on network efficiency is used to generate importance labels for nodes, resulting in a training dataset. The initialized node scoring model is trained using the training dataset to obtain a trained node scoring model.

[0010] In some embodiments, the identification of critical risk vessels based on the importance scores of each vessel to obtain risk vessel identification results includes the following steps: Threshold determination is applied to the edge weights of the waterway traffic map to identify high-risk edges; A ship set is formed based on the ships connected to all the high-risk edges; Based on the importance score of each vessel, the vessels in the vessel set are sorted to obtain the sorting result; Based on the sorting results, several ships are selected as key risk ships to obtain the risk ship identification results.

[0011] In some embodiments, the step of making a multi-vehicle collaborative collision avoidance decision based on the risk vessel identification result to obtain a collision avoidance scheme includes the following steps: When the risk vessel identification results indicate the existence of critical risk vessels, a multi-objective optimization model is constructed with the goal of maximizing the combination of safety and economy; the decision variables of the multi-objective optimization model include the vessel's turning angle and the timing of the vessel's turning. A multi-objective sparrow search algorithm based on a fusion simulated annealing mechanism and an external archiving strategy is used to solve the decision variables of the multi-objective optimization model to obtain a collision avoidance scheme.

[0012] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0013] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0014] The embodiments of this application include at least the following beneficial effects: This application provides a multi-ship collaborative collision avoidance decision-making method, electronic device, and program product. This scheme collects the current state information of each vessel in the waterway, performs collision risk analysis between vessels based on the current state information, obtains the collision risk identification results for any two vessels, and then constructs a waterway traffic map based on the collision risk identification results. Nodes represent vessel objects, and edges represent collision risks between two nodes. The graph structure features of the nodes in the waterway traffic map are extracted to obtain the multi-dimensional graph structure features of each node. Then, using a node scoring model based on a multi-layer graph attention network, the importance of each node is scored based on its multi-dimensional graph structure features to obtain the importance score of each vessel. Based on the importance scores of each vessel, key risk vessels are identified to obtain risk vessel identification results. Based on the risk vessel identification results, multi-ship collaborative collision avoidance decision-making is performed to obtain a collision avoidance scheme. This application constructs a waterway traffic map based on the preliminary collision risk analysis results between two vessels, and then performs node importance analysis on the waterway traffic map based on the node scoring model, thereby accurately and efficiently identifying high-risk vessels throughout the entire area. Based on this, multi-ship collaborative collision avoidance decision-making is performed, improving the reliability of collision avoidance decisions for vessels in the waterway, and thus improving the navigation safety of vessels in the waterway. Attached Figure Description

[0015] Figure 1 This is a flowchart of the multi-ship cooperative collision avoidance decision-making method provided in the embodiments of this application; Figure 2 This is a schematic diagram of the overall concept of multi-ship cooperative collision avoidance decision-making provided in the embodiments of this application; Figure 3 This is a schematic diagram of the multi-layer graph attention mechanism of the multi-head attention layer provided in the embodiments of this application; Figure 4 This is a schematic diagram of key risk vessels in the waterway traffic map provided in the embodiments of this application; Figure 5 This is a schematic diagram of key risk vessels in a waterway traffic map provided in another embodiment of this application; Figure 6 This is a schematic diagram of the four-ship cooperative obstacle avoidance trajectory provided in the embodiments of this application; Figure 7 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0018] Before providing a detailed description of the embodiments of this application, some of the nouns and terms involved in the embodiments of this application will be explained first. The nouns and terms involved in the embodiments of this application are subject to the following interpretations.

[0019] MMSI (Maritime Mobile Service Identity) is a nine-digit code transmitted by a ship's radio communication system on its radio channel to uniquely identify various stations (such as ships, coastal radio stations, and search and rescue units) worldwide. It is equivalent to a ship's "electronic ID card" or "mobile phone number".

[0020] VTS (Vessel Traffic Service), also known as Vessel Traffic Management System, is a comprehensive system that manages vessel traffic and provides advisory services within specific waterways to ensure vessel traffic safety, improve traffic efficiency, and protect the aquatic environment. VTS integrates technologies such as radar, Automatic Identification System (AIS), VHF radio, and Closed-Circuit Television (CCTV) to dynamically monitor vessels in traffic-intensive or environmentally sensitive waterways such as ports and waterways. Its core functions include providing navigational safety information to vessels, providing navigational aids in poor visibility conditions, organizing traffic flow to optimize navigation order, and coordinating joint emergency actions in the event of an emergency.

[0021] AIS (Automatic Identification System) is a new type of digital navigation aid system applied to maritime safety and communication between ships and shore, and between ships themselves. The system consists of shore-based (base station) facilities and shipboard equipment. By integrating technologies such as the Global Positioning System (GPS) and Very High Frequency (VHF) communication, it can automatically and continuously broadcast its name, call sign, position, speed, course, and other dynamic and static information to ships in nearby waters and shore stations, while also receiving similar information from other ships.

[0022] GAT (Graph Attention Networks) is a neural network architecture specifically designed for processing graph-structured data. Its core innovation lies in introducing the attention mechanism into graph neural networks (GNNs) to address the limitations of traditional methods such as graph convolutional networks (GCNs).

[0023] The multi-ship cooperative collision avoidance decision-making method provided in this application relates to the field of maritime traffic safety technology. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or vehicle terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing the multi-ship cooperative collision avoidance decision-making method, but is not limited to the above forms.

[0024] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0025] Figure 1 This is an optional flowchart of the multi-ship cooperative collision avoidance decision-making method provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S101 to S107.

[0026] S101, collect current status information of each vessel in the waterway; S102, Based on the current status information of each vessel, perform collision risk analysis between vessels to obtain the collision risk identification results of any two vessels; S103. Construct a waterway traffic map based on the collision risk identification results. Nodes in the waterway traffic map represent ship objects, and edges indicate that there is a collision risk between two nodes. S104, Extract the graph structure features of the nodes in the waterway traffic map to obtain the multidimensional graph structure features of each node; S105, using a node scoring model based on a multi-layer graph attention network, the importance of each node is scored according to the multi-dimensional graph structure features of each node, thus obtaining the importance score of each ship. S106, Based on the importance score of each vessel, identify key risk vessels and obtain the risk vessel identification results; S107, based on the risk vessel identification results, conduct multi-vessel collaborative collision avoidance decisions to obtain a collision avoidance scheme.

[0027] Steps S101 to S107 as shown in the embodiments of this application construct a waterway traffic map based on the preliminary collision risk analysis results between the two ships, and then perform node importance analysis on the waterway traffic map based on the node scoring model, thereby accurately and efficiently identifying high-risk ships in the entire area, and then making multi-ship collaborative collision avoidance decisions based on this, improving the reliability of collision avoidance decisions for ships in the water, and thus improving the navigation safety of ships in the water.

[0028] In step S101 of some embodiments, the current status information of each vessel in the water can be obtained through AIS. The current status information includes, but is not limited to, the vessel's position, heading, and speed.

[0029] In step S102 of some embodiments, the collision risk analysis between ships can be implemented based on a relevant real-time collision analysis method between two ships, such as a rule analysis method for the distance between ships or a machine learning method, etc., and this application embodiment does not impose specific limitations. The collision risk identification result of any two ships is used to characterize whether there is a collision risk between any two ships, or to represent the collision risk value between any two ships.

[0030] According to some embodiments of this application, step S102 may include, but is not limited to, the following steps: S201, calculate the minimum meeting distance and meeting time between any two ships based on the current status information of each ship; S202, based on the minimum meeting distance and meeting time, a collision risk analysis is performed on the two corresponding ships to obtain the collision risk values ​​of the two ships; S203, based on the collision risk values ​​between all two vessels, forms a collision risk identification result.

[0031] In this embodiment, for any two ships i and j, the minimum encounter distance and the time to reach the encounter point can be calculated based on the positions, headings, and speeds of the two ships. Then, the minimum encounter distance and the time to reach the encounter point are input into a machine learning model (such as a neural network model) used in the ship domain to identify collision risk values ​​to obtain the collision risk value between the two ships. This method integrates the collision risk values ​​between all two vessels to form a collision risk identification result. This embodiment analyzes the collision risk identification result using the minimum encounter distance and encounter time between vessels. The minimum encounter distance represents the probability of a collision between the two vessels, while the encounter time represents the urgency of a future collision, thus improving the reliability of the collision risk identification result.

[0032] In step S103 of some embodiments, a waterway traffic graph is constructed based on the collision risk identification result. Nodes in the waterway traffic graph represent vessel objects, and edges indicate a collision risk between two nodes. In one example, the waterway traffic graph can be an undirected, unweighted graph. The collision risk identification result characterizes whether there is a collision risk between any two vessels. A single vessel is abstracted as a node, and the node ID corresponds to the vessel's MMSI. If there is a collision risk between two vessels, the corresponding two nodes are connected as edges of the graph. In another example, the waterway traffic graph can also be an undirected, weighted graph. The collision risk identification result characterizes the collision risk value between any two vessels. A single vessel is abstracted as a node, and the node ID corresponds to the vessel's MMSI. If there is a collision risk between two vessels, the corresponding edges are connected as edges of the graph. Greater than the expected judgment value (which can be obtained based on statistical analysis methods, such as...). If the collision risk value is zero, then the corresponding two nodes are connected as edges of the graph, and the collision risk value is normalized to obtain the corresponding edge weights; for the collision risk value... Two corresponding nodes with values ​​less than the expected judgment value are not connected. Furthermore, the waterway traffic map can also be a directed weighted graph, where the edge weight represents the collision risk value under the condition of collision risk, the edge direction represents the collision direction between ships, and the collision direction means that node i collides with node j, or node j collides with node i.

[0033] In step S104 of some embodiments, graph structure features of nodes in the waterway traffic map are extracted to obtain multidimensional graph structure features of each node. These multidimensional graph structure features represent the relationships between nodes in the graph. These features may include, but are not limited to, weighted degree centrality, betweenness centrality, proximity centrality, eigenvector centrality, rate of change of local connectivity coefficients, redundant path count, clustering coefficients, core number, and network efficiency-related features, as detailed below: Weighted degree centrality is used to measure the strength of the risk association between a ship and other ships, and is divided into in-degree centrality. out-degree centrality The following are examples: ; ; in, Let i be the set of neighbors pointing to node i. Let i be the set of neighbors that node i points to.

[0034] Betweenness centrality measures a vessel's control over the shortest path (path length measured by edge weights) on a water traffic chart. Calculated using the following formula: ; in, Let be the number of paths that pass through i in the shortest path from s to e. Let be the total number of shortest paths from s to e.

[0035] Proximity centrality Calculated using the following formula: ; in, Let i be the shortest path length from i to j; Eigenvector centrality is calculated based on the principal eigenvectors of the adjacency matrix. Calculated using the formula: ; Where v is the eigenvector corresponding to the largest eigenvalue of the adjacency matrix.

[0036] The rate of change of LCC (Local Connectivity Coefficient) is calculated as follows: ; Wherein, LCC(G) is the original network local connectivity coefficient; Redundant path count is calculated using the following formula: ; in, Let u be the set of paths from u to v that do not pass through i. Clustering coefficients are calculated using the following formula: ; in, Let i be the degree of node i; Core count: determined using the k-core decomposition algorithm. Let k be the value of the largest k-core to which node i belongs; Network efficiency correlation characteristics: Calculated based on the change in global network efficiency before and after node removal, network efficiency correlation characteristics. Calculated using the following formula: ; in, .

[0037] In step S105 of some embodiments, a node scoring model based on a multi-layer graph attention network is used to score the importance of each node according to its multi-dimensional graph structure features, thereby obtaining the importance score of each ship. The node scoring model is a deep learning model that includes a multi-layer graph attention network. This model is used to classify the importance of nodes and obtain the importance score of each node (i.e., ship). The node scoring model includes a multi-layer graph attention network and a classification network connected in sequence. The multi-layer graph attention network is mainly used to aggregate features between nodes based on their multi-dimensional graph structure features. Through the multi-head attention mechanism of the multi-layer graph attention network, the comprehensiveness of feature aggregation can be improved. The classification network is used to classify and predict the output of the multi-layer graph attention network to obtain the importance score of each ship. It is understood that the classification network can be a machine learning module such as a feedforward neural network; this embodiment does not impose specific limitations.

[0038] According to some embodiments of this application, step S105 may include, but is not limited to, the following steps: S301, input the multidimensional graph structure features of each node into the multi-layer graph attention network of the node scoring model to extract the inter-node association features, and obtain the output feature matrix of the multi-layer graph attention network. S302, the output feature matrix is ​​input into the multilayer perceptron of the node scoring model to predict the node scores and obtain the importance scores of each ship.

[0039] In this embodiment, the classification network of the node scoring model can use a multilayer perceptron for classification prediction. The multilayer perceptron in this embodiment is also called an MLP scorer. The multilayer perceptron has powerful nonlinear modeling and representation learning capabilities, which can improve the accuracy of ship importance scoring. The multilayer graph attention network adopts a multi-head attention mechanism, which includes an input layer, a multi-head attention layer, a residual connection module, and a normalization layer.

[0040] In step S301 of some embodiments, the multi-layer graph attention network includes an input layer, a multi-head attention layer, a residual connection module, and a normalization layer; step S301 may include, but is not limited to, the following steps: S401: Input the multidimensional graph structure features of all nodes into the input layer for preprocessing to obtain the input features; S402, input the input features into the multi-head attention layer to perform feature aggregation between nodes, and obtain node aggregated features; S403, The input features and node aggregation features are residually connected through the residual connection module to obtain residual connected features; S404 inputs the residual connection features into the normalization layer for normalization processing, and obtains the output feature matrix of the multi-layer graph attention network.

[0041] In this embodiment, please refer to Figure 2 A multi-layer graph attention network (MLN) may include an input layer, three multi-head attention layers, a residual connection module, and a layer normalization layer. The output of the MLP quantifies node importance using an MLP scorer. By using a multi-layer graph attention network based on a fusion of multi-head attention, residual connections, and layer normalization, the features of the ship nodes are learned to obtain a ship node importance score. The layers of the multi-layer graph attention network are described below: The input layer is used to initialize the input feature matrix of the graph attention network. The input feature matrix consists of 9-dimensional structural feature vectors of ship nodes, where N is the number of ship nodes.

[0042] Multi-head attention layers are used to weighted aggregate the neighbor features of ship nodes. Each attention layer can contain 8 attention heads, capturing multi-dimensional risk associations between nodes through different attention heads. Specifically, the multi-layer graph attention mechanism of multi-head attention layers is as follows: Figure 3 As shown, the calculation process for each attention head is as follows: ; ; ; in, Let be the attention weight between node i and node j in the k-th attention head. For attention parameter vectors, The characteristic transformation matrix, and These are the input feature vectors of the node and its neighboring nodes, respectively. This represents the normalized attention weights. Indicates a node All neighbors Summation of exponential fractions, Let the neighbor aggregation features of node i be . and All are activation functions. The outputs of multiple attention heads are concatenated to obtain the output of the multi-head attention layer.

[0043] The residual connection module is used to perform residual connections between the inputs and outputs of a multi-head attention layer, such as overlaying. Specifically, when the input and output dimensions of the attention layer are the same, residuals are added directly; when the dimensions are different, residuals are added through a projection matrix. Align the dimensions before performing residual joins.

[0044] The normalization layer is used to normalize the output of the multi-head attention layer after it has passed through the residual connection module. The formula is as follows: , The input feature vector; For feature vectors The mean; For feature vectors The variance; It is a very small constant to prevent the denominator from being zero; For learnable scaling parameters; The offset parameter is learnable.

[0045] In this embodiment, the outputs of multiple attention heads are concatenated, and residual connections and layer normalization are combined to obtain the output feature matrix of the multi-layer graph attention network. In step S302 of some embodiments, the MLP scorer may include two hidden layers (with LeakyReLU activation function) and one output layer, outputting a single-valued node importance score. .

[0046] According to some embodiments of this application, the node scoring model in step S105 can be obtained by, but is not limited to, the following steps: S501 uses a greedy strategy based on network efficiency to generate importance labels for nodes, thus obtaining the training dataset; S502, use the training dataset to train the initialized node scoring model to obtain the trained node scoring model.

[0047] Specifically, the process of generating the importance labels for nodes is as follows: Calculate the network global efficiency of the original waterway traffic map ; After removing each node i in the graph sequentially, calculate the global network efficiency of the waterway traffic graph after the removal.

[0048] The true importance label of node i is calculated using the following formula: ; The importance label for each node can be calculated using the process described above. This forms a training dataset. The embodiments of this application, based on a greedy strategy, automatically generate node importance labels, which improves model training efficiency. Considering network efficiency during the node importance label generation process allows for consideration of the impact of node collision risks on the overall graph structure, thus improving the accuracy of importance label marking.

[0049] Please continue to refer to Figure 2 The initialized node scoring model was trained using the training dataset, with mean squared error (MSE) as the loss function during training. The model is trained using the Adam optimizer and iterated until the loss converges.

[0050] In step S106 of some embodiments, the importance score of a ship is obtained using the above-described node scoring model. Based on the importance scores of each ship, critical risk ships are identified to obtain the risk ship identification result. In one example, a threshold determination can be made based on the ship's importance score to identify critical risk ships; alternatively, several ships with higher importance scores can be selected to identify critical risk ships. In another example, critical risk ships can be identified by combining the collision risk values ​​between the ships with the ship's importance score, improving the reliability of critical risk ships. Specifically, high-risk associated edges are filtered based on a preset edge weight threshold, and critical risk ships are determined by combining the ship node importance score.

[0051] According to some embodiments of this application, step S106 may include, but is not limited to, the following steps: S601, threshold determination is performed on the edge weights of the waterway traffic map to identify high-risk edges; S602, forming a ship ensemble based on all ships connected to high-risk edges; S603, based on the importance score of each vessel, sort the vessels in the vessel set to obtain the sorting result; S604: Based on the ranking results, select several vessels as key risk vessels to obtain the risk vessel identification results.

[0052] In this embodiment, an edge weight threshold can be set. Filter out all High-risk associated edges. Ships corresponding to these edges have a higher collision risk and require priority attention. Extract the endpoints of these high-risk associated edges to form a high-priority ship set. Then, score the ships in this set according to node importance. By sorting in descending order, the top 20% of vessels can be selected as critical risk vessels. Furthermore, if a vessel is simultaneously the endpoint of multiple high-risk associated edges and ranks highly in the score, it is identified as a core critical risk vessel, and collision avoidance plans can be developed specifically for it. For example, the critical risk vessels identified in the waterway traffic map in this application embodiment... Figure 4 and Figure 5 As shown in the figure, the nodes represent ships, and the thickness of the edges represents the weight of the risk edges; the red nodes are critical risk ships, and the green nodes are non-critical ships.

[0053] In step S107 of some embodiments, a multi-vessel cooperative collision avoidance decision is made based on the risk vessel identification result to obtain a collision avoidance scheme. In one example, if the risk vessel identification result indicates the existence of a critical risk vessel, then a multi-vessel cooperative collision avoidance decision is executed to obtain a collision avoidance scheme; in another example, if the risk vessel identification result indicates the existence of a critical risk vessel, then a decision object is determined based on all critical risk vessels, and then a multi-vessel cooperative collision avoidance decision is executed based on the decision object to obtain a collision avoidance scheme; determining the decision object based on all critical risk vessels can mean using vessels located within a certain range of the critical risk vessel as the decision object, or it can mean using vessels connected to the critical risk vessel on the waterway traffic chart as the decision object. The multi-vessel cooperative collision avoidance decision can be implemented based on a multi-objective optimization algorithm or a deep learning algorithm. For example, please continue to refer to... Figure 2 A multi-objective optimization model for multi-ship cooperative collision avoidance can be constructed. This model uses turning angle and turning timing as decision variables, and minimizes the safety objective of collision risk and the economic objective of minimizing operating costs as optimization objectives. An improved multi-objective sparrow search algorithm that integrates simulated annealing mechanism and external archiving strategy is used to solve the multi-objective optimization model for multi-ship cooperative collision avoidance and obtain the Pareto optimal collision avoidance scheme.

[0054] According to some embodiments of this application, step S107 may include, but is not limited to, the following steps: S701, when the risk vessel identification results indicate the existence of critical risk vessels, a multi-objective optimization model is constructed with the goal of maximizing the combination of safety and economy; the decision variables of the multi-objective optimization model include the vessel's turning angle and the timing of the vessel's turning. S702 employs a multi-objective sparrow search algorithm based on a fusion simulated annealing mechanism and an external archiving strategy to solve the decision variables of the multi-objective optimization model and obtain a collision avoidance scheme.

[0055] In step S701 of some embodiments, in multi-vessel encounter scenarios, traditional collision avoidance methods struggle to balance safety and economic objectives and do not fully consider the constraints of maritime collision avoidance rules, resulting in poor practicality of the decision-making scheme. The multi-objective optimization model of this application's embodiments determines collision avoidance rule constraints based on maritime collision avoidance rules. These constraints include the responsibility allocation rules between vessels traveling in the straight course and those giving way, and the turning and avoidance direction rules, as detailed below: Responsibility division constraints: Based on the encounter situation of the vessels (such as head-on encounter, overtaking, and cross encounter), the vessel traveling in the straight and the vessel giving way are clearly defined. The vessel traveling in the straight has no obligation to take collision avoidance maneuvers, while the vessel giving way must take collision avoidance measures proactively. Turning direction constraints: In a cross-encounter scenario, the giving way vessel should turn to the right and avoid turning to the left; in an overtaking scenario, the overtaking vessel should overtake from the starboard side of the vessel being overtaken.

[0056] The decision variables in the multi-objective optimization model for multi-ship cooperative collision avoidance are defined as follows: Ship turning angle The steering angle of a ship when avoiding a collision can be in the range of [-30°, 30°] (this range can be determined by taking into account the ship's maneuverability, such as the maximum steering angle not exceeding 30°). Ship turning timing The time at which the ship begins its turning maneuver can be within a range of values. , The time when a ship reaches the critical state of collision risk (such as the time of encounter being 0).

[0057] The objective functions of the multi-objective optimization model include a safety objective function and an economic objective function, as detailed below: The safety objective function aims to minimize collision risk and is measured by the deviation between the minimum encounter distance (DCPA) and the safe distance. The safety objective function is expressed as follows: ; in, For security weights, For a safe distance, This represents the minimum meeting distance between ships. For example, (Safety is given higher weight than economy). (as set according to the port waterway navigation rules) This is the minimum DCPA value among all ship pairs. The smaller the size, the lower the risk of a collision.

[0058] The economic objective function aims to minimize the ship's operating costs, which include steering angle costs and steering time costs. The economic objective function is expressed as follows: ; in, Indicates economic weight, Represents the time cost coefficient (unit: ), Indicates the time cost of turning around. This represents the cost of steering angle. For example, , The larger the steering angle and the later the steering timing, the higher the operating cost. The larger.

[0059] The constraint settings for the multi-objective optimization model are explained below: Steering angle constraints: To avoid excessive steering that could affect the ship's maneuverability; Turning timing constraints: Ensure that collision avoidance maneuvers are completed before the risk of collision becomes critical; Collision avoidance rules constraints: Straight-going vessels , (No collision avoidance maneuver will be performed), the yielding vessel must meet the turning direction constraints (such as when meeting at an intersection). (That is, turn right).

[0060] In step S702 of some embodiments, the optimization algorithms of related technologies are prone to getting trapped in local optima when dealing with multi-objective collision avoidance problems, and the solution set diversity is insufficient, failing to provide VTS operators with flexible decision-making options. This application embodiment employs an improved multi-objective sparrow search algorithm that integrates simulated annealing and external archiving strategies, which can improve the convergence speed and solution set diversity of the solution process, thereby improving richer and more reliable multi-ship cooperative collision avoidance schemes. Specifically, the solution process of the multi-objective optimization model for multi-ship cooperative collision avoidance in this application embodiment is as follows: Initialize a sparrow population, with each individual in the population corresponding to a set of decision variables. ; Calculate the safety objective function value and the economic objective function value for each individual in the initial population; Initialize the external archive to store non-dominated solutions, storing the objective function values ​​corresponding to the individuals in the external archive; Iterative optimization, as follows: Select the discoverer and update the position based on the following formula: ; in, is a random number in the range [0,1], ST is the safety threshold, Q is a normally distributed random number, L is a vector of all 1s, and T is the maximum number of iterations.

[0061] Select followers and update positions based on the following formula: ; in, For the worst individual position, To be the optimal discoverer position, Let be a random vector with elements ±1. for The false rebellion;

[0062] Select the vigilant and update their position based on the following formula: ; in, For the optimal individual position, These are step size control parameters. A random number in the range [-1, 1] It is a very small constant. Let i be the objective function value for individual i. This represents the worst objective function value.

[0063] To improve the algorithm's ability to escape local optima, a simulated annealing acceptance criterion is adopted for newly generated individuals. Specifically, the following rules are used to select newly generated individuals: ; The above screening rule means that if the objective function value of the new individual is better than that of the current individual ( If the objective function value of the new individual is poor, then accept it directly; ), then by probability accept; T(t) is the current temperature. , The coefficient of performance is the cooling factor; for example, the initial temperature. Cooling coefficient .

[0064] To maintain solution set diversity, an external archive is set up to store non-dominated solutions, as follows: After each iteration, merge the parent and child populations; Perform non-dominated sorting on the merged population and add the first frontier solution to the external archive; If the external archive capacity exceeds a preset value (e.g., 200), the congestion of the solution is calculated as follows: ; in, Let i be the neighboring solutions of the i-th solution on the m-th objective. Let these be the maximum and minimum values ​​of the m-th target; Delete the solution with the least congestion until the archive capacity meets the requirements.

[0065] During the iteration process, the external archive is updated through non-dominated sorting and crowding calculation. After the iteration is completed, the Pareto optimal collision avoidance scheme in the external archive can be output.

[0066] In one example, the process of solving a multi-objective optimization model using the multi-objective sparrow search algorithm is as follows: Initialization: Set population size Maximum number of iterations External archive capacity Simulated annealing initial temperature ; Population initialization: Randomly generate 50 individuals, each corresponding to a set of decision variables. Ensure that the constraints are met; Objective function calculation: Calculate the safety and economic objective values ​​for each individual; External archive initialization: Add non-dominated solutions from the initial population to the external archive; Iterative optimization: a. Select 30% of individuals as discoverers and update the formula according to the discoverer's location; b. The remaining 70% of individuals are treated as followers, and the update formula is updated according to the follower's position; c. Select 10% of individuals as vigilants and update the formula according to the vigilant's position; d. Apply the simulated annealing acceptance criterion to the newly generated individuals to select valid individuals; e. Merge the parent and child populations, sort them in a non-dominated order, and then update the external archive; f. Lower the simulated annealing temperature ; Iteration Termination: When the number of iterations reaches 100, output the Pareto optimal collision avoidance scheme from the external archive. For example, based on the above algorithm, the cooperative obstacle avoidance trajectory of the four ships guided by the optimal collision avoidance scheme in a four-ship encounter scenario is as follows: Figure 6 As shown.

[0067] According to some embodiments of this application, real AIS data from a port during a certain time period is used, collected once per second, for a total of 659 time nodes and 1762 waterway traffic maps. The dataset is divided into a training set (209 maps), a validation set (90 maps), and a test set (1463 maps) in chronological order.

[0068] The effectiveness of identifying critical risk vessels was evaluated using the rate of change in network efficiency: the decrease in global network efficiency was compared between removing critical risk vessels identified by the GAT method and those identified by the traditional centrality method. Experimental results showed that removing critical risk vessels identified by the GAT method resulted in an average decrease in global network efficiency of 38.2%, significantly higher than the traditional method (average decrease of 15.7%), demonstrating that the GAT method can more accurately identify critical risk vessels that significantly impact network functionality.

[0069] The effectiveness of multi-ship cooperative collision avoidance was verified by constructing encounter scenarios of 4 ships (overtaking + cross encounter) and 6 ships, and comparing the solution performance of the IMOSSA algorithm and the NSGA-II algorithm: Convergence: The IMOSSA algorithm converges in 50 iterations, which is 37.5% faster than NSGA-II (convergence in 80 iterations). Solution set diversity: The Pareto front solution distribution uniformity index (Spacing) of the IMOSSA algorithm is 0.12, which is better than NSGA-II (0.21). Practicality: In the collision avoidance schemes generated by IMOSSA, the minimum distance between ships is greater than 300m, the operating cost is reduced by an average of 18.3%, and it complies with the COLREGs rules.

[0070] This application constructs a directed weighted complex network for maritime traffic, determining ship node characteristics, risk edge weights between nodes, and collision avoidance rule constraints. Based on a graph attention network, it integrates multi-head attention mechanisms, residual connections, and layer normalization to learn the importance of ship nodes and uses edge weights to screen key risk ships. A multi-ship cooperative collision avoidance multi-objective optimization model is constructed, and an improved multi-objective sparrow search algorithm is used to solve for the Pareto optimal collision avoidance scheme. By dynamically learning the risk correlation characteristics between ships, the accuracy of identifying key risk ships is improved, while simultaneously balancing navigation safety and economy, providing flexible and efficient decision support for intelligent maritime traffic supervision.

[0071] This application also provides a multi-ship cooperative collision avoidance decision-making system, including: The first module is used to collect the current status information of each vessel in the waterway; The second module is used to perform collision risk analysis between ships based on the current status information of each ship, and to obtain the collision risk identification results of any two ships. The third module is used to construct a waterway traffic map based on the collision risk identification results. The nodes of the waterway traffic map represent ship objects, and the edges represent the collision risk between two nodes. The fourth module is used to extract the graph structure features of nodes in the waterway traffic map, and obtain the multi-dimensional graph structure features of each node. The fifth module is used to score the importance of each ship by using a node scoring model based on a multi-layer graph attention network, according to the multi-dimensional graph structure features of each node. The sixth module is used to identify key risk vessels based on the importance score of each vessel, and to obtain the risk vessel identification results; The seventh module is used to make multi-vehicle collaborative collision avoidance decisions based on the risk vessel identification results, and obtain a collision avoidance scheme.

[0072] It is understood that the methods described in the above method embodiments are applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0073] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including a tablet computer, a shipboard computer, or similar device.

[0074] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0075] Please see Figure 7 , Figure 7 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 902 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901. The input / output interface 903 is used to implement information input and output; The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904); The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0076] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0077] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0078] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0079] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0080] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0081] The multi-ship cooperative collision avoidance decision-making method and related equipment provided in this application have at least one of the following beneficial effects: High recognition accuracy: By fusing 9-dimensional structural features through a multi-layer graph attention network and dynamically learning the risk association weights between ships, the accuracy of identifying key risk ships is improved by 22.5% compared with traditional methods; The decision-making is highly scientific: the multi-objective optimization model takes into account both safety and economy, combines standard collision avoidance rules, the collision avoidance scheme is highly practical, and the risk of ship collision is reduced by more than 40%. Superior algorithm performance: The improved multi-objective sparrow search algorithm introduces simulated annealing and external archiving, which improves the convergence speed by 37.5% and the solution set diversity is significantly better than that of traditional algorithms; Wide applicability: Based on real AIS data verification, it can be applied to complex maritime traffic scenarios such as ports and narrow waterways, providing technical support for the intelligent upgrade of VTS systems.

[0082] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0083] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0084] The system embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0085] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0086] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or modules is not necessarily limited to those steps or modules explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0087] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0088] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0089] The modules described above as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0090] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0091] If the integrated module is implemented as a software functional module 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 the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0092] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A multi-ship cooperative collision avoidance decision-making method, characterized in that, Includes the following steps: Collect current status information of each vessel in the waterway; Based on the current status information of each vessel, a collision risk analysis between vessels is performed to obtain the collision risk identification results for any two vessels. A waterway traffic map is constructed based on the collision risk identification results. Nodes in the waterway traffic map represent ship objects, and edges indicate that there is a collision risk between two nodes. The graph structure features of the nodes in the waterway traffic map are extracted to obtain the multidimensional graph structure features of each node; By using a node scoring model based on a multi-layer graph attention network, the importance score of each ship is obtained by scoring the importance of each node according to the multi-dimensional graph structure features of each node. Based on the importance scores of each vessel, key risk vessels are identified, and risk vessel identification results are obtained; Based on the risk vessel identification results, a multi-vessel collaborative collision avoidance decision is made to obtain a collision avoidance scheme.

2. The method according to claim 1, characterized in that, The step of performing collision risk analysis between ships based on the current status information of each ship to obtain the collision risk identification result of any two ships includes the following steps: Calculate the minimum encounter distance and encounter time between any two ships based on the current status information of each ship; Based on the minimum encounter distance and the encounter time, a collision risk analysis is performed on the two corresponding ships to obtain the collision risk values ​​of the two ships; The collision risk identification result is formed based on the collision risk values ​​between all two of the aforementioned vessels.

3. The method according to claim 2, characterized in that, The edge direction of the waterway traffic map represents the collision direction between ships, and the edge weight represents the collision risk value when there is a collision risk. The multidimensional graph structure features include weighted degree centrality, betweenness centrality, proximity centrality, eigenvector centrality, rate of change of local connectivity coefficients, redundant path count, clustering coefficient, number of cores, and network efficiency-related features.

4. The method according to claim 2, characterized in that, The method of using a node scoring model based on a multi-layer graph attention network to score the importance of each node according to the multi-dimensional graph structure features, and obtaining the importance score of each ship, includes the following steps: The multidimensional graph structure features of each node are input into the multi-layer graph attention network of the node scoring model to extract the inter-node association features, and the output feature matrix of the multi-layer graph attention network is obtained. The output feature matrix is ​​input into the multilayer perceptron of the node scoring model to predict the node scores and obtain the importance scores of each ship.

5. The method according to claim 4, characterized in that, The multi-layer graph attention network includes an input layer, a multi-head attention layer, a residual connection module, and a normalization layer; the step of inputting the multi-dimensional graph structure features of each node into the multi-layer graph attention network of the node scoring model to extract the inter-node association features and obtain the output feature matrix of the multi-layer graph attention network includes the following steps: The multidimensional graph structure features of all the nodes are input into the input layer for preprocessing to obtain the input features; The input features are input into the multi-head attention layer to perform feature aggregation between nodes, resulting in node-aggregated features. The residual connection module performs a residual connection on the input features and the node aggregation features to obtain residual connection features; The residual connection features are input into the normalization layer for normalization processing to obtain the output feature matrix of the multi-layer graph attention network.

6. The method according to any one of claims 1 to 5, characterized in that, The node scoring model is trained through the following steps: A greedy strategy based on network efficiency is used to generate importance labels for nodes, resulting in a training dataset. The initialized node scoring model is trained using the training dataset to obtain a trained node scoring model.

7. The method according to claim 3, characterized in that, The process of identifying key risk vessels based on the importance scores of each vessel, and obtaining the risk vessel identification results, includes the following steps: Threshold determination is applied to the edge weights of the waterway traffic map to identify high-risk edges; A ship set is formed based on the ships connected to all the high-risk edges; Based on the importance score of each vessel, the vessels in the vessel set are sorted to obtain the sorting result; Based on the sorting results, several ships are selected as key risk ships to obtain the risk ship identification results.

8. The method according to claim 1, characterized in that, The step of making a multi-vessel collaborative collision avoidance decision based on the risk vessel identification results to obtain a collision avoidance scheme includes the following steps: When the risk vessel identification results indicate the existence of critical risk vessels, a multi-objective optimization model is constructed with the goal of maximizing the combination of safety and economy; the decision variables of the multi-objective optimization model include the vessel's turning angle and the timing of the vessel's turning. A multi-objective sparrow search algorithm based on a fusion simulated annealing mechanism and an external archiving strategy is used to solve the decision variables of the multi-objective optimization model to obtain a collision avoidance scheme.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 8.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 8.

Images