A path and trajectory collaborative optimization method and device based on neural network enhancement, medium and product
By using a neural network collaborative optimization method, a robust Pareto front is generated, which solves the problem of insufficient collaboration in path and trajectory optimization and achieves safe and efficient multi-objective robust optimization.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING FORESTRY UNIV
- Filing Date
- 2024-10-15
- Publication Date
- 2026-07-03
AI Technical Summary
In existing technologies, the optimization process of path and trajectory lacks coordination, making it difficult to achieve global optimum and robustness under multiple optimization objectives and trajectory tracking errors.
A neural network-based collaborative optimization method is adopted, which generates candidate paths and trajectory sets through variational autoencoders and deep neural networks, and uses the concept of robust Pareto front to screen out robust and effective solutions, thereby achieving collaborative optimization of paths and trajectories.
It improves the global optimality and robustness of path and trajectory optimization, ensuring safe and efficient operation under multi-objective conditions, and maintaining stability when faced with trajectory tracking errors.
Smart Images
Figure CN119204374B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method, apparatus, medium, and product for path and trajectory co-optimization based on neural network enhancement, belonging to the interdisciplinary field of transportation and artificial intelligence. Background Technology
[0002] In planned operational spaces such as airports, the coordinated optimization of paths and trajectories has become a key technical requirement for improving operational safety and efficiency. In existing technologies, the optimization process for paths and trajectories is typically divided into two relatively independent steps. First, the shortest path algorithm is used to assign optimal paths to aircraft or ground vehicles. Then, based on the determined paths, detailed trajectory planning is performed to ensure that the operation meets established safety standards and efficiency requirements. This step-by-step optimization method simplifies the computational process to some extent, enabling path and trajectory optimization to be achieved with limited computational resources.
[0003] However, the existing step-by-step optimization method has significant shortcomings. First, the optimization processes for path and trajectory lack sufficient synergy, resulting in optimization results that are not globally optimal. Since path planning and trajectory planning are performed separately, it is difficult to find the optimal balance between the two, thus affecting the overall operational performance. Second, in actual operation, there are often multiple optimization objectives to consider, including time, energy consumption, pollutant emissions, and other aspects. These objectives often conflict, significantly increasing the complexity of the path and trajectory optimization problem. Furthermore, since trajectory tracking errors are unavoidable in actual operation, higher demands are placed on the robustness of path and trajectory optimization. However, existing technologies lack effective multi-objective robust collaborative optimization methods, making it difficult to effectively handle the uncertainties caused by multiple conflicting optimization objectives and trajectory tracking errors, and ensure the robust optimality of paths and trajectories. Therefore, developing an effective multi-objective robust collaborative optimization method for paths and trajectories is particularly important.
[0004] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention
[0005] The technical problem to be solved by this invention is to overcome the shortcomings of current multi-objective robust collaborative optimization of paths and trajectories.
[0006] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution.
[0007] In a first aspect, the present invention provides a path and trajectory co-optimization method based on neural network enhancement, characterized in that it includes:
[0008] Obtain the start and end nodes of the moving target and the graph structure of the problem to be solved;
[0009] Based on the start and end nodes of the moving target and the graph structure of the problem to be solved, optimization is performed using a pre-trained collaborative optimization model to obtain the robust Pareto front of the problem to be solved.
[0010] The training and optimization methods of the collaborative optimization model include:
[0011] The historical start and end nodes of the moving target and the graph structure of the historical problem to be solved are obtained as training samples.
[0012] Based on the training samples, supervised learning methods are used for training to determine the parameters of the variational autoencoder network model and the latent space of the latent space feature vectors generated by the encoder.
[0013] When performing optimization, obtain the start and end nodes of the moving target and the graph structure of the problem to be solved;
[0014] Based on a pre-trained variational autoencoder network model, latent space feature vectors are randomly selected, and a candidate path set is generated according to the start and end nodes of the moving target and the graph structure of the problem to be solved.
[0015] Obtain the ideal speed curve of each candidate path in the candidate path set, and generate a trajectory set using a trajectory generation model based on a deep neural network;
[0016] Repeat the following steps until the optimization calculation time exceeds the threshold:
[0017] By performing mutation and cross-computation on the latent space feature vectors, new latent space feature vectors are obtained.
[0018] Based on the new latent space feature vectors and the graph structure of the starting and ending nodes of the moving target and the problem to be solved, a new set of candidate paths is generated based on a pre-trained variational autoencoder network model.
[0019] Obtain the ideal velocity curve of each candidate path in the new candidate path set, and generate a new trajectory set using a trajectory generation model based on a deep neural network;
[0020] By filtering robust and effective solutions from the existing trajectory set and the new trajectory set, a robust and effective trajectory set is obtained.
[0021] The latent space feature vectors corresponding to the dominant individuals in the robust and effective trajectory set are retained, and the candidate path set is updated to the candidate paths corresponding to the dominant individuals.
[0022] After the optimization calculation is completed, the final set of robust and effective trajectories is used as the robust Pareto front for path and trajectory co-optimization.
[0023] Furthermore, the collaborative optimization model includes:
[0024] Variational autoencoder network model, including encoder and decoder;
[0025] The encoder is used to encode each path of the moving target into a latent space feature vector conditioned on the graph structure and start and end nodes.
[0026] The decoder is used to generate a sequence of nodes for candidate paths step by step using the latent space feature vectors generated by the encoder, which are conditional on the graph structure and start and end nodes.
[0027] A trajectory generation model based on deep neural networks is used to receive the ideal velocity curve of each candidate path and generate a trajectory set that conforms to the motion characteristics of the moving target.
[0028] Furthermore, based on a pre-trained variational autoencoder network model, latent space feature vectors are randomly selected, and a candidate path set is generated according to the start and end nodes of the moving target and the graph structure of the problem to be solved, including:
[0029] The selected latent space feature vectors are input into the decoder, and candidate paths are generated based on the start and end nodes of the moving target and the graph structure of the problem to be solved.
[0030] Multiple latent space feature vectors are used to generate multiple candidate paths, forming a candidate path set.
[0031] Furthermore, mutation and cross-computation are performed on the latent space feature vectors to obtain new latent space feature vectors, including:
[0032] Randomly select three different latent space feature vectors from the current latent space feature vectors as the first, second, and third volumes;
[0033] Calculate the difference between the first individual and the second individual;
[0034] The difference between the first individual and the second individual is added to the third individual to generate a mutated individual;
[0035] By performing a crossover operation between a mutant individual and a third individual, and randomly selecting a crossover point to exchange gene segments in the mutant individual and the third individual, offspring individuals are generated.
[0036] The offspring individuals are used as new latent space feature vectors.
[0037] Furthermore, the calculation expression for the difference between the first individual and the second individual is as follows:
[0038] (1);
[0039] In the formula, This indicates the difference between the first individual and the second individual. Indicates the first entity, It represents the second entity.
[0040] Furthermore, the calculation expression for generating mutated individuals is:
[0041] (2);
[0042] In the formula, Indicates a variant individual. It represents the third entity.
[0043] Furthermore, the definition of the robust and effective solution includes:
[0044] For any trajectory x in the trajectory set:
[0045] If there exists another trajectory y that dominates any trajectory x, then any trajectory x is a non-robust efficient solution when the following condition is met:
[0046] Other trajectories y belong to the set of trajectories;
[0047] The cost of other trajectories y in probability space U is no greater than the cost of any trajectory x in probability space U;
[0048] There are specific variables Belonging to probability space U, such that other trajectories y are in a specific variable The cost of the following path is less than that of any trajectory x under a specific variable. The value of the next generation;
[0049] Any trajectory x is considered a robust and efficient solution when it is not dominated by any other trajectory in the trajectory set and satisfies the time window constraint on the candidate path.
[0050] In a second aspect, the present invention provides a computer device including a memory, a processor, and a computer program stored in the memory, characterized in that the processor executes the computer program to implement the steps of the method described in the first aspect.
[0051] Thirdly, the present invention provides a computer-readable storage medium having a computer program / instructions stored thereon, characterized in that the computer program / instructions, when executed by a processor, implement the steps of the method described in the first aspect.
[0052] Fourthly, the present invention provides a computer program product, including a computer program / instructions, characterized in that, when the computer program / instructions are executed by a processor, they implement the steps of the method described in the first aspect.
[0053] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:
[0054] This invention, by acquiring the start and end nodes of the moving target and the graph structure of the problem to be solved, can comprehensively understand the operational requirements and constraints of the moving target. Based on this, using a pre-trained collaborative optimization model, this invention achieves collaborative optimization of path and trajectory, and successfully obtains the robust Pareto front for path and trajectory collaborative optimization. The method provided by this invention is innovative and effectively overcomes the shortcomings of existing technologies where path and trajectory optimization are relatively independent and lack collaboration, significantly improving the global optimality of the optimization results. By comprehensively considering multiple optimization objectives, such as time, energy consumption, and pollutant emissions, the method of this invention can ensure the safe and efficient operation of the moving target while meeting these objectives. Furthermore, due to the introduction of the concept of robust Pareto front, the method of this invention can maintain high robustness and stability when facing uncertainties such as trajectory tracking errors, thereby ensuring the stable operation of the moving target.
[0055] The collaborative optimization model training method provided by this invention is unique and efficient. By acquiring the historical start and end nodes of the moving target and the graph structure of the historical problem to be solved as training samples, it can fully utilize the information in historical data to improve the training effect and generalization ability of the model. This invention uses the latent space feature vectors of a pre-trained variational autoencoder network model to generate a candidate path set, which not only reduces computational complexity but also improves the diversity and accuracy of candidate path generation. This invention also obtains the ideal velocity curves of candidate paths and uses a trajectory generation model based on deep neural networks to generate a trajectory set, further realizing refined trajectory planning. During the iterative optimization process, this invention continuously explores new optimization spaces by performing mutation and cross-calculation on the latent space feature vectors until a time threshold is reached. Finally, by selecting dominant individuals that meet robust effectiveness, a robust Pareto front for collaborative optimization is obtained. The method of this invention can continuously output high-quality, robust path and trajectory optimization schemes and also provides strong support for subsequent dynamic operation decisions. Attached Figure Description
[0056] Figure 1 This is a flowchart illustrating the method for obtaining a robust and effective trajectory set provided in an embodiment of the present invention. Detailed Implementation
[0057] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.
[0058] The term "and / or" simply describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.
[0059] Example 1
[0060] This embodiment introduces a path and trajectory co-optimization method based on neural network enhancement, including the following steps:
[0061] Step 1: Train the collaborative optimization model, including:
[0062] Step 1.1: Obtain the historical start and end nodes of the moving target and the graph structure of the historical problem to be solved as training samples.
[0063] The collaborative optimization model includes:
[0064] Variational autoencoder network model, including encoder and decoder;
[0065] The encoder is used to encode each path of the moving target into a latent space feature vector conditioned on the graph structure and start and end nodes.
[0066] The decoder is used to generate a sequence of nodes for candidate paths step by step using the latent space feature vectors generated by the encoder, which are conditional on the graph structure and start and end nodes.
[0067] A trajectory generation model based on deep neural networks is used to receive the ideal velocity curve of each candidate path and generate a trajectory set that conforms to the motion characteristics of the moving target.
[0068] Step 1.2: Based on the training samples, train the variational autoencoder network model, including:
[0069] Based on the historical start and end position nodes contained in training sample i, the encoder is used to convert the historical start and end nodes... The corresponding optimal path is encoded as a graph structure corresponding to the training samples. and based on historical start and end points The latent space feature vectors are conditional. ,in, N is the number of samples. This represents the starting node in the historical start and end node sequence. This represents the end node in the historical start and end nodes.
[0070] Using the decoder based on the latent space feature vector Historical start and end points and graph structure Generate a sequence containing nodes Candidate paths ,in, This indicates the starting node included in the candidate path, which is the starting position node of the problem to be solved. This indicates the second node contained in the candidate path. This indicates the termination node included in the candidate path, which is the destination node of the problem to be solved.
[0071] Using the optimal path computed offline as the truth value, a variational autoencoder network model is trained using supervised learning methods to determine the parameters of the variational autoencoder network model and the latent space Z to which the latent space feature vectors generated by the encoder belong.
[0072] Step 2: Obtain the start and end nodes of the moving target and the graph structure of the problem to be solved. Based on the start and end nodes of the moving target and the graph structure of the problem to be solved, perform optimization based on a pre-trained collaborative optimization model to obtain the robust Pareto front of the problem to be solved, including:
[0073] Step 2.1: Based on the pre-trained variational autoencoder network model, randomly select latent space feature vectors from the latent space Z, and generate a candidate path set according to the start and end nodes of the moving target and the graph structure of the problem to be solved, including:
[0074] Randomly selected latent space feature vectors Input decoder to move the start and end nodes of the target. and the graph structure of the problem to be solved Generate candidate paths based on the given conditions. ,in , where n is the number of randomly selected latent space features;
[0075] Using n latent space feature vectors Generate n candidate paths to form a candidate path set. .
[0076] Step 2.2: Obtain the candidate path set Each candidate path The ideal velocity curve, in which, , Given the total number of paths in the candidate path set, a trajectory set is generated using a trajectory generation model based on a deep neural network. ,in, This represents the first trajectory in the trajectory set. This represents the last trajectory in the trajectory set, and K represents the total number of ideal velocity curves.
[0077] The probability distribution of the position of each trajectory in the trajectory set at time t is expressed as: .
[0078] Step 2.3: Repeat the following steps until the latent space feature vectors are obtained. The time required for mutation and crossover calculations exceeds the threshold, such as... Figure 1 As shown:
[0079] Step 2.3.1: Process the latent space feature vectors Mutation and crossover calculations are performed to obtain n new feature vectors. ,include:
[0080] From the current latent space feature vector Three different latent space feature vectors are randomly selected as the first volume. The second body and the third body ;
[0081] Calculate the first body Second individual Differences between ;
[0082] The first body Second individual Differences between Add a third body Generate mutated individuals ;
[0083] Utilizing mutated individuals With the third body Perform a crossover operation, randomly selecting a crossover point to exchange mutated individuals. and the third body Gene fragments in the gene pool generate offspring individuals ;
[0084] The expression for calculating the difference between the first individual and the second individual is as follows:
[0085] (1);
[0086] In the formula, This indicates the difference between the first individual and the second individual. Indicates the first entity, Indicates the second entity;
[0087] The calculation expression for generating the mutant individuals is:
[0088] (2);
[0089] In the formula, Indicates a variant individual. It represents the third entity.
[0090] The offspring individuals As a new latent space feature vector .
[0091] Step 2.3.2: Based on the new latent space feature vector Given the graph structure of the problem to be solved, a new set of candidate paths is generated based on a pre-trained variational autoencoder network model. .
[0092] Step 2.3.3: Obtain a new set of candidate paths The ideal velocity curve of each candidate path is used to generate a new trajectory set using a trajectory generation model based on a deep neural network.
[0093] Step 2.3.4: According to the definition of robust effective solution, filter the existing trajectory set and the new trajectory set to obtain the robust effective trajectory set. .
[0094] The definition of the robust solution includes:
[0095] For any trajectory x in the trajectory set:
[0096] If there exists another trajectory y that dominates any trajectory x, then any trajectory x is a non-robust efficient solution when the following condition is met:
[0097] Other trajectories y belong to the set of trajectories;
[0098] The cost of other trajectories y in probability space U is no greater than the cost of any trajectory x in probability space U;
[0099] There are specific variables Belonging to probability space U, such that other trajectories y are in a specific variable The cost of the following path is less than that of any trajectory x under a specific variable. The value of the next generation;
[0100] Any trajectory x is considered a robust and efficient solution when it is not dominated by any other trajectory in the trajectory set and satisfies the time window constraint on the candidate path.
[0101] Step 2.3.5: Retain the latent space feature vectors corresponding to the dominant individuals in the robust and effective trajectory set, and update the candidate path set to the candidate paths corresponding to the dominant individuals.
[0102] Step 2.4: After the optimization calculation is completed, the final set of robust and effective trajectories is used as the robust Pareto front for path and trajectory co-optimization.
[0103] Example 2
[0104] Based on the same inventive concept as Embodiment 1, this embodiment introduces a computer device, including a memory, a processor, and a computer program stored in the memory, characterized in that the processor executes the computer program to implement the steps of the method described in Embodiment 1.
[0105] Example 3
[0106] Based on the same inventive concept as other embodiments, this embodiment introduces a computer-readable storage medium storing a computer program / instructions thereon, characterized in that the computer program / instructions, when executed by a processor, implement the steps of the method described in Embodiment 1.
[0107] Example 4
[0108] Based on the same inventive concept as other embodiments, this embodiment introduces a computer program product, including a computer program / instructions, characterized in that the computer program / instructions, when executed by a processor, implement the steps of the method described in Embodiment 1.
[0109] In summary, by acquiring the start and end nodes of the moving target and the problem to be solved, this invention can comprehensively understand the operational requirements and constraints of the moving target. Based on this, using a pre-trained collaborative optimization model, this invention achieves collaborative optimization of path and trajectory, and successfully obtains the robust Pareto front for path and trajectory collaborative optimization. The method provided by this invention is innovative and effectively overcomes the shortcomings of existing technologies where path and trajectory optimization are relatively independent and lack collaboration, significantly improving the global optimality of the optimization results. By comprehensively considering multiple optimization objectives, such as time, energy consumption, and pollutant emissions, the method of this invention can ensure the safe and efficient operation of the moving target while meeting these objectives. Furthermore, due to the introduction of the concept of robust Pareto front, the method of this invention can maintain high robustness and stability when facing uncertainties such as trajectory tracking errors, thereby ensuring the stable operation of the moving target.
[0110] The collaborative optimization model training method provided by this invention is unique and efficient. By acquiring the historical start and end nodes of the moving target and the graph structure of the historical problem to be solved as training samples, it can fully utilize the information in historical data to improve the training effect and generalization ability of the model. This invention uses the latent space feature vectors of a pre-trained variational autoencoder network model to generate a candidate path set, which not only reduces computational complexity but also improves the diversity and accuracy of candidate path generation. This invention also obtains the ideal velocity curves of candidate paths and uses a trajectory generation model based on deep neural networks to generate a trajectory set, further realizing refined trajectory planning. During the iterative optimization process, this invention continuously explores new optimization spaces by performing mutation and cross-calculation on the latent space feature vectors until a time threshold is reached. Finally, by selecting dominant individuals that meet robust effectiveness, a robust Pareto front for collaborative optimization is obtained. The method of this invention can continuously output high-quality, robust path and trajectory optimization schemes and also provides strong support for subsequent dynamic operation decisions.
[0111] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0112] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0113] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0114] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0115] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.
Claims
1. A path and trajectory co-optimization method based on neural network enhancement, characterized in that, include: Obtain the start and end nodes of the moving target and the graph structure of the problem to be solved; Based on the start and end nodes of the moving target and the graph structure of the problem to be solved, optimization is performed using a pre-trained collaborative optimization model to obtain the robust Pareto front of the problem to be solved. The training and optimization methods of the collaborative optimization model include: The historical start and end nodes of the moving target and the graph structure of the historical problem to be solved are obtained as training samples. Based on the training samples, supervised learning methods are used for training to determine the parameters of the variational autoencoder network model and the latent space of the latent space feature vectors generated by the encoder. When performing optimization, obtain the start and end nodes of the moving target and the graph structure of the problem to be solved; Based on a pre-trained variational autoencoder network model, latent space feature vectors are randomly selected, and a candidate path set is generated according to the start and end nodes of the moving target and the graph structure of the problem to be solved. Obtain the ideal speed curve of each candidate path in the candidate path set, and generate a trajectory set using a trajectory generation model based on a deep neural network; Repeat the following steps until the optimization calculation time exceeds the threshold: Mutation and crossover calculations are performed on the latent space feature vectors to obtain new latent space feature vectors, including: Randomly select three different latent space feature vectors from the current latent space feature vectors as the first, second, and third volumes; Calculate the difference between the first individual and the second individual; The difference between the first individual and the second individual is added to the third individual to generate a mutated individual; By performing a crossover operation between a mutant individual and a third individual, and randomly selecting a crossover point to exchange gene segments in the mutant individual and the third individual, offspring individuals are generated. The offspring individuals are used as new latent space feature vectors; Based on the new latent space feature vectors and the graph structure of the starting and ending nodes of the moving target and the problem to be solved, a new set of candidate paths is generated based on a pre-trained variational autoencoder network model. Obtain the ideal velocity curve of each candidate path in the new candidate path set, and generate a new trajectory set using a trajectory generation model based on a deep neural network; By filtering robust and effective solutions from the existing trajectory set and the new trajectory set, a robust and effective trajectory set is obtained. The latent space feature vectors corresponding to the dominant individuals in the robust and effective trajectory set are retained, and the candidate path set is updated to the candidate paths corresponding to the dominant individuals. After the optimization calculation is completed, the final set of robust and effective trajectories is used as the robust Pareto front for path and trajectory co-optimization. The definition of the robust and effective solution includes: For any trajectory x in the trajectory set: Other trajectories y dominate any trajectory x, and any trajectory x is a non-robust efficient solution, when the following conditions are met: other trajectories y belong to the set of trajectories; the cost of other trajectories y in the probability space U is not greater than the cost of any trajectory x in the probability space U; and there exists a specific variable. Belonging to probability space U, such that other trajectories y are in a specific variable The cost of the following path is less than that of any trajectory x under a specific variable. The value of the next generation; Any trajectory x is considered a robust and efficient solution when it is not dominated by any other trajectory in the trajectory set and satisfies the time window constraint on the candidate path.
2. The path and trajectory co-optimization method based on neural network enhancement according to claim 1, characterized in that, The collaborative optimization model includes: Variational autoencoder network model, including encoder and decoder; The encoder is used to encode each path of the moving target into a latent space feature vector conditioned on the graph structure and start and end nodes. The decoder is used to generate a sequence of nodes for candidate paths step by step using the latent space feature vectors generated by the encoder, which are conditional on the graph structure and start and end nodes. A trajectory generation model based on deep neural networks is used to receive the ideal velocity curve of each candidate path and generate a trajectory set that conforms to the motion characteristics of the moving target.
3. The path and trajectory co-optimization method based on neural network enhancement according to claim 1, characterized in that, Based on a pre-trained variational autoencoder network model, latent space feature vectors are randomly selected, and a candidate path set is generated according to the start and end nodes of the moving target and the graph structure of the problem to be solved, including: The selected latent space feature vectors are input into the decoder, and candidate paths are generated based on the start and end nodes of the moving target and the graph structure of the problem to be solved. Multiple latent space feature vectors are used to generate multiple candidate paths, forming a candidate path set.
4. The path and trajectory co-optimization method based on neural network enhancement according to claim 1, characterized in that, The expression for calculating the difference between the first individual and the second individual is as follows: （1）； In the formula, This indicates the difference between the first individual and the second individual. Indicates the first entity, It represents the second entity.
5. The path and trajectory co-optimization method based on neural network enhancement according to claim 1, characterized in that, The calculation expression for generating the mutant individuals is as follows: （2）； In the formula, Indicates a variant individual. It represents the third entity.
6. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1-5.
7. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-5.
8. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-5.