Path planning method and system for urban roads in congested conditions

Abstract

The present invention discloses a path planning method and system for urban roads in a congested state, which defines corresponding congested sections based on the congestion characteristics of each road, the corresponding position and the distribution map of the urban roads, and is compatible with the overall consideration of the congestion characteristics of each road, the corresponding position and the distribution map of the urban roads, and ensures the accuracy of the congested sections. Further, a corresponding congestion test is performed according to each congested section, and a congestion dynamic map of the urban road is constructed; the user's location and target location are collected, and multiple navigation sub-paths are defined according to the user's location, target location and the congestion dynamic map of the urban road; the user's navigation path is constructed based on the multiple navigation sub-paths, and the user's navigation path and the congestion change amount of the congested section are triggered to autonomously plan the user's navigation path, thereby ensuring the accurate control of the navigation path and achieving the technical effect of autonomous planning of the navigation path.

Description

Technical Field

[0001] The present invention relates to the technical field of urban roads, and in particular to a path planning method and system for urban roads in a congested state. Background Art

[0002] With the development of science and technology, urban roads are used in people's lives and have corresponding congestion forms in different time periods or locations, especially morning rush hour and evening rush hour. In the existing technology, real-time monitoring is performed on urban roads, and multiple road data are collected. The corresponding congested sections are defined according to the multiple road data. However, the accuracy of the congested sections is low, which affects the user's precise control of the navigation path. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art. The present invention provides a path planning method and system for urban roads in a congested state, which collects multiple road data based on various road interaction devices; defines corresponding road congestion characteristics according to the multiple road data; and defines corresponding congested sections based on the congestion characteristics of each road, the corresponding position and the distribution map of the urban roads. It is compatible with the overall consideration of the congestion characteristics of each road, the corresponding position and the distribution map of the urban roads, and ensures the accuracy of the congested sections.

[0004] Furthermore, a corresponding congestion test is performed according to each congested road section, and a dynamic congestion map of urban roads is constructed; the user's location and target location are collected, and multiple navigation sub-paths are defined according to the user's location, target location and the dynamic congestion map of urban roads; the user's navigation path is constructed based on the multiple navigation sub-paths, and the autonomous planning of the user's navigation path is triggered according to the user's navigation path and the congestion change of the congested road section, thereby ensuring the precise control of the navigation path and achieving the technical effect of autonomous planning of the navigation path.

[0005] The embodiment of the present invention provides a path planning method for urban roads in a congested state, which is applied to the path planning scenario of urban roads in a congested state;

[0006] The path planning method for urban roads in a congested state includes:

[0007] Collect multiple road data based on each road interaction device;

[0008] Defining corresponding congestion characteristics of each road according to a plurality of road data;

[0009] Define the corresponding congested road sections based on the congestion characteristics of each road, the corresponding location and the distribution map of urban roads;

[0010] Conduct corresponding congestion tests according to each congested road section and build a dynamic congestion map of urban roads;

[0011] Collect the user's location and target location, and define multiple navigation sub-paths based on the user's location, target location, and a dynamic map of urban road congestion;

[0012] A user's navigation path is constructed based on multiple navigation sub-paths, and autonomous planning of the user's navigation path is triggered according to the user's navigation path and the congestion change amount of the congested road section.

[0013] Optionally, the collecting of multiple road data based on each road interaction device includes:

[0014] Locate each road interaction device;

[0015] Collect the current position of each road interactive device;

[0016] Triggering each road interaction device to collect road data according to the current position of each road interaction device and the interaction dimension corresponding to each road interaction device;

[0017] A plurality of road data are collected based on each road interaction device.

[0018] Optionally, the defining corresponding road congestion features according to the plurality of road data includes:

[0019] Freeze multiple road data;

[0020] Performing data preprocessing on a plurality of road data, and outputting the preprocessed road data;

[0021] Construct a corresponding data set according to the preprocessed road data;

[0022] Match the corresponding data learning model based on each data set;

[0023] Associating various data sets and data learning models;

[0024] The corresponding road congestion characteristics are defined based on each data set and data learning model.

[0025] Optionally, the defining of corresponding congested road sections based on the congestion characteristics of each road, the corresponding location and the distribution map of urban roads includes:

[0026] Determine the congestion characteristics of each road;

[0027] Triggering positioning detection based on various road congestion characteristics;

[0028] Defining corresponding positions according to the location detection of each road congestion feature;

[0029] Correlate the congestion characteristics of each road, the corresponding location and the distribution map of urban roads;

[0030] Multiple interactions are performed on each road congestion feature, its corresponding location, and the distribution map of urban roads;

[0031] The corresponding congested road sections are defined according to the multiple interactions of the congestion characteristics of each road, the corresponding location and the distribution map of the urban roads.

[0032] Optionally, performing a corresponding congestion test according to each congested road section and constructing a congestion dynamic map of urban roads includes:

[0033] Freeze each congested road section;

[0034] Collect the location of each congested road section, and match the corresponding congestion test according to the location of each congested road section;

[0035] Carry out corresponding congestion tests according to each congested road section;

[0036] The corresponding congestion level is defined based on the congestion test of each congested road section, and a congestion dynamic map of urban roads is constructed according to the congestion level, each congested road section, the distribution map of urban roads and the time model.

[0037] Optionally, the collecting of the user's location and target location, and defining a plurality of navigation sub-paths according to the user's location, the target location and a congestion dynamic map of urban roads, includes:

[0038] Collect the user's location and target location;

[0039] Define relative space based on the user's location and the target location;

[0040] Multiple movement paths are defined based on the relative space and the movement pattern of the user.

[0041] Optionally, the collecting of the user's location and target location, and defining a plurality of navigation sub-paths according to the user's location, the target location and a congestion dynamic map of urban roads, further includes:

[0042] Collect dynamic images of urban road congestion;

[0043] Linking multiple mobility paths and congestion dynamic graphs of urban roads;

[0044] A plurality of navigation sub-paths are defined according to a plurality of moving paths and a dynamic map of congestion of urban roads.

[0045] Optionally, constructing the user's navigation path based on the multiple navigation sub-paths, and triggering autonomous planning of the user's navigation path according to the user's navigation path and the congestion change amount of the congested road section, includes:

[0046] Freeze multiple navigation sub-paths;

[0047] Constructing a user's navigation path based on multiple navigation sub-paths, the user's movement pattern, and movement time period;

[0048] Freeze the user's navigation path;

[0049] Carry out congestion control on congested sections and collect the congestion changes of congested sections within the preset time.

[0050] Optionally, the constructing of the user's navigation path based on the multiple navigation sub-paths, and triggering the autonomous planning of the user's navigation path according to the user's navigation path and the congestion change amount of the congested road section, further includes:

[0051] Associating the user's navigation path and the congestion change of the congested road section;

[0052] The user's navigation path is autonomously planned based on the user's navigation path and the congestion change of the congested road section, so that the user's navigation path is dynamically updated with the change of time or the change of the congested road section.

[0053] In addition, an embodiment of the present invention further provides a path planning system for urban roads in a congested state, wherein the path planning system for urban roads in a congested state comprises:

[0054] A road data module, used for collecting a plurality of road data based on various road interaction devices;

[0055] A road congestion module, used for defining corresponding road congestion features according to a plurality of road data;

[0056] A congested road section module, used to define corresponding congested road sections based on the congestion characteristics of each road, the corresponding location and the distribution map of urban roads;

[0057] The congestion dynamic map module is used to perform corresponding congestion tests according to each congested road section and construct a congestion dynamic map of urban roads;

[0058] A navigation sub-path module is used to collect the user's location and target location, and define multiple navigation sub-paths according to the user's location, target location and the congestion dynamic map of urban roads;

[0059] The navigation path module is used to construct a user's navigation path based on multiple navigation sub-paths, and trigger autonomous planning of the user's navigation path according to the user's navigation path and the congestion change amount of the congested road section.

[0060] In an embodiment of the present invention, through the method in the embodiment of the present invention, multiple road data are collected based on each road interaction device; corresponding road congestion characteristics are defined according to the multiple road data; corresponding congested sections are defined based on the congestion characteristics of each road, the corresponding positions and the distribution map of urban roads, which is compatible with the overall consideration of the congestion characteristics of each road, the corresponding positions and the distribution map of urban roads, and ensures the accuracy of the congested sections.

[0061] Furthermore, a corresponding congestion test is performed according to each congested road section, and a dynamic congestion map of urban roads is constructed; the user's location and target location are collected, and multiple navigation sub-paths are defined according to the user's location, target location and the dynamic congestion map of urban roads; the user's navigation path is constructed based on the multiple navigation sub-paths, and the autonomous planning of the user's navigation path is triggered according to the user's navigation path and the congestion change of the congested road section, thereby ensuring the precise control of the navigation path and achieving the technical effect of autonomous planning of the navigation path. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0063] Figure 1 is a flow chart of a method for planning a path in a congested urban road according to an embodiment of the present invention;

[0064] Figure 2 is a flow chart of S11 in the path planning method for urban roads in a congested state in an embodiment of the present invention;

[0065] Figure 3 is a flow chart of S12 in the path planning method for urban roads in a congested state in an embodiment of the present invention;

[0066] Figure 4 is a flow chart of S13 in the path planning method for urban roads in a congested state in an embodiment of the present invention;

[0067] Figure 5 is a flow chart of S14 in the path planning method for urban roads in a congested state in an embodiment of the present invention;

[0068] Figure 6 is a flow chart of S15 in the path planning method for urban roads in a congested state in an embodiment of the present invention;

[0069] Figure 7 is a flow chart of S16 in the path planning method for urban roads in a congested state in an embodiment of the present invention;

[0070] Figure 8 is a schematic diagram of the structure of a path planning system for urban roads in a congested state according to an embodiment of the present invention;

[0071] Fig. 9 The figure is a hardware diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0073] See also Figures 1 to 9 , a path planning method for urban roads in a congested state, applied to a path planning scenario for urban roads in a congested state; the path planning method for urban roads in a congested state includes:

[0074] Step S11: collecting multiple road data based on each road interaction device;

[0075] Step S12: defining corresponding congestion characteristics of each road according to the plurality of road data;

[0076] Step S13: defining corresponding congested road sections based on the congestion characteristics of each road, the corresponding location and the distribution map of urban roads;

[0077] Step S14: performing corresponding congestion tests according to each congested road section, and constructing a congestion dynamic map of urban roads;

[0078] Step S15: collecting the user's location and target location, and defining multiple navigation sub-paths according to the user's location, target location, and a congestion dynamic map of urban roads;

[0079] Step S16: constructing a user's navigation path based on the multiple navigation sub-paths, and triggering autonomous planning of the user's navigation path according to the user's navigation path and the congestion change amount of the congested road section.

[0080] In an embodiment of the present invention, through the method in the embodiment of the present invention, multiple road data are collected based on each road interaction device; corresponding road congestion characteristics are defined according to the multiple road data; corresponding congested sections are defined based on the congestion characteristics of each road, the corresponding positions and the distribution map of urban roads, which is compatible with the overall consideration of the congestion characteristics of each road, the corresponding positions and the distribution map of urban roads, and ensures the accuracy of the congested sections.

[0081] Furthermore, a corresponding congestion test is performed according to each congested road section, and a dynamic congestion map of urban roads is constructed; the user's location and target location are collected, and multiple navigation sub-paths are defined according to the user's location, target location and the dynamic congestion map of urban roads; the user's navigation path is constructed based on the multiple navigation sub-paths, and the autonomous planning of the user's navigation path is triggered according to the user's navigation path and the congestion change of the congested road section, thereby ensuring the precise control of the navigation path and achieving the technical effect of autonomous planning of the navigation path.

[0082] refer to Figure 2 , in step S11, a plurality of road data are collected based on each road interaction device;

[0083] In the specific implementation process of the present invention, the specific steps may be:

[0084] S111: locating each road interaction device;

[0085] S112: Collecting the current position of each road interaction device;

[0086] S113: triggering each road interaction device to collect road data according to the current position of each road interaction device and the interaction dimension corresponding to each road interaction device;

[0087] S114: Collecting multiple road data based on various road interaction devices.

[0088] In an embodiment of the present application, each road interaction device is located, each road interaction device is managed and controlled, and the current position of each road interaction device is collected, and the current position of each road interaction device is introduced.

[0089] Furthermore, the collection of road data by each road interaction device is triggered according to the current position of each road interaction device and the interaction dimension corresponding to each road interaction device, and the current position of each road interaction device and the interaction dimension corresponding to each road interaction device are introduced. The current position of each road interaction device and the interaction dimension corresponding to each road interaction device are controlled as a whole, and the multi-dimensional control of the current position of each road interaction device and the interaction dimension corresponding to each road interaction device is compatible, so that multiple road data are collected based on each road interaction device, and the synchronous collection of multiple road data is realized, thereby ensuring the synchronous operation of each road interaction device. Optionally, each road interaction device includes traffic cameras, vehicle-mounted sensors, navigation vehicles, and mobile applications.

[0090] refer to Figure 3 , in step S12, defining corresponding congestion features of each road according to the plurality of road data;

[0091] In the specific implementation process of the present invention, the specific steps may be:

[0092] S121: freeze multiple road data;

[0093] S122: performing data preprocessing on a plurality of road data, and outputting the preprocessed road data;

[0094] S123: constructing a corresponding data set according to the preprocessed road data;

[0095] S124: Matching corresponding data learning models based on each data set;

[0096] S125: Associating various data sets and data learning models;

[0097] S126: Define corresponding road congestion features based on each data set and data learning model.

[0098] In an embodiment of the present application, multiple road data are frozen, and the multiple road data are controlled. At the same time, data preprocessing is performed on the multiple road data, and the preprocessed road data is output. At this time, for the data preprocessing of the multiple road data, the collected road data is cleaned, processed and summarized for further data analysis and model training.

[0099] The data preprocessing steps are as follows:

[0100] Data cleaning

[0101] 1. Missing value processing: Delete records with missing values. Fill missing values, such as using the mean, median, mode, or a specific estimated value.

[0102] 2. Outlier detection and processing: Use statistical methods (such as box plots, standard deviation, Z-score, etc.) to identify outliers. Remove or replace outliers.

[0103] 3. Data consistency check: Ensure that the data follows the same format and units. Standardize the data to eliminate inconsistencies.

[0104] 4. Format error correction: Correct data entry errors such as date format, phone number, etc.

[0105] 5. Duplicate data processing: Identify and delete or merge duplicate records.

[0106] 6. Data type conversion: Convert data into a format suitable for analysis, such as converting a string into a numeric type.

[0107] Furthermore, corresponding data sets are constructed based on the preprocessed road data; based on matching the corresponding data learning models to each data set, a data learning model is introduced to associate each data set and the data learning model; then, based on each data set and the data learning model, corresponding road congestion characteristics are defined, achieving deep interaction between each data set and the data learning model, thereby ensuring the accuracy of each road congestion characteristic.

[0108] refer to Figure 4 , in step S13, in the floor cleaning space, the corresponding congested road sections are defined based on the congestion characteristics of each road, the corresponding location and the distribution map of the urban roads;

[0109] In the specific implementation process of the present invention, the specific steps may be:

[0110] S131: freeze the congestion characteristics of each road;

[0111] S132: triggering positioning detection based on various road congestion characteristics;

[0112] S133: defining corresponding positions according to the positioning detection of each road congestion feature;

[0113] S134: Associating the congestion characteristics of each road, the corresponding location, and the distribution map of the urban roads;

[0114] S135: performing multiple interactions on the congestion characteristics of each road, the corresponding location, and the distribution map of urban roads;

[0115] S136: Define corresponding congested road sections according to multiple interactions of congestion characteristics of each road, corresponding locations, and a distribution map of urban roads.

[0116] In an embodiment of the present application, multiple road data are collected based on various road interaction devices; corresponding road congestion characteristics are defined based on the multiple road data; corresponding congested road sections are defined based on the congestion characteristics of each road, the corresponding positions, and the distribution map of urban roads, which is compatible with the overall consideration of the congestion characteristics of each road, the corresponding positions, and the distribution map of urban roads, and ensures the accuracy of the congested road sections.

[0117] At this time, each road congestion feature is frozen; positioning detection is triggered based on each road congestion feature, so that the corresponding position is defined according to the positioning detection of each road congestion feature, thereby clarifying the location of each road congestion feature.

[0118] Therefore, the congestion characteristics of each road, the corresponding location and the distribution map of urban roads are associated; multiple interactions are performed on the congestion characteristics of each road, the corresponding location and the distribution map of urban roads; and the corresponding congested sections are defined according to the multiple interactions of the congestion characteristics of each road, the corresponding location and the distribution map of urban roads, which is compatible with the overall consideration of the congestion characteristics of each road, the corresponding location and the distribution map of urban roads and ensures the accuracy of the congested sections.

[0119] refer to Figure 5 , S14: performing corresponding congestion tests according to each congested road section, and constructing a congestion dynamic map of urban roads;

[0120] In the specific implementation process of the present invention, the specific steps may be:

[0121] S141: freeze each congested road section;

[0122] S142: collecting the locations of each congested road section, and matching the corresponding congestion test according to the locations of each congested road section;

[0123] S143: performing a corresponding congestion test according to each congested road section;

[0124] S144: defining corresponding congestion levels based on congestion tests of various congested road sections, and constructing a congestion dynamic map of urban roads according to the congestion levels, various congested road sections, a distribution map of urban roads, and a time model.

[0125] In an embodiment of the present application, each congested road section is frozen, and management and control are performed on each congested road section. At this time, the location of each congested road section is collected, and a corresponding congestion test is matched according to the location of each congested road section; a corresponding congestion test is performed according to each congested road section, congestion tests of each congested road section are implemented, and overall management and control are performed on the congestion tests of each congested road section.

[0126] Therefore, the corresponding congestion level is defined based on the congestion test of each congested road section, and the congestion dynamic map of urban roads is constructed according to the congestion level, each congested road section, the distribution map of urban roads and the time model. It is compatible with the overall consideration of the congestion level, each congested road section, the distribution map of urban roads and the time model, and realizes multi-dimensional control of the congestion level, each congested road section, the distribution map of urban roads and the time model, ensuring the accuracy of the congestion dynamic map of urban roads. The changes of congested sections are presented based on the congestion dynamic map of urban roads, which facilitates the autonomous planning of navigation trajectories.

[0127] refer to Figure 6 , S15: collecting the user's location and target location, and defining multiple navigation sub-paths according to the user's location, target location and the congestion dynamic map of the urban road;

[0128] In the specific implementation process of the present invention, the specific steps may be:

[0129] S151: Collecting the user's location and target location;

[0130] S152: defining a relative space according to the user's location and the target location;

[0131] S153: defining a plurality of movement paths based on the relative space and the movement pattern of the user;

[0132] S154: Collecting a dynamic map of urban road congestion;

[0133] S155: associating multiple moving paths and a congestion dynamic graph of urban roads;

[0134] S156: Defining a plurality of navigation sub-paths according to the plurality of moving paths and the congestion dynamic graph of the urban roads.

[0135] In an embodiment of the present application, the user's location and target location are collected, and the user's location and target location are controlled, so that a relative space is defined according to the user's location and target location, and a relative space is introduced to facilitate path planning for the relative space.

[0136] Furthermore, multiple movement paths are defined based on the relative space and the user's movement pattern, which is compatible with the overall consideration of the relative space and the user's movement pattern, and realizes multi-dimensional control of the relative space and the user's movement pattern, ensuring the rationality and integrity of the multiple movement paths.

[0137] Therefore, the congestion dynamic map of urban roads is collected; multiple moving paths and the congestion dynamic map of urban roads are associated; and multiple navigation sub-paths are defined according to the multiple moving paths and the congestion dynamic map of urban roads. This is compatible with the overall consideration of the congestion dynamic map of multiple moving paths and urban roads, realizes multi-dimensional control of the congestion dynamic map of multiple moving paths and urban roads, and ensures the accuracy of multiple navigation sub-paths.

[0138] refer to Figure 7 , S16: constructing a user's navigation path based on multiple navigation sub-paths, and triggering autonomous planning of the user's navigation path according to the user's navigation path and the congestion change amount of the congested road section;

[0139] In the specific implementation process of the present invention, the specific steps may be:

[0140] S161: freeze multiple navigation sub-paths;

[0141] S162: constructing a navigation path for the user based on the multiple navigation sub-paths, the user's movement pattern, and the movement time period;

[0142] S163: freeze the user's navigation path;

[0143] S164: performing congestion control on the congested road section, and collecting the congestion change amount of the congested road section within a preset time;

[0144] S165: Associating the navigation path of the user with the congestion change amount of the congested road section;

[0145] S166: triggering autonomous planning of the user's navigation path based on the user's navigation path and the congestion change amount of the congested road section, so that the user's navigation path is dynamically updated with changes in time or changes in the congested road section.

[0146] In the specific implementation process of the present invention, a corresponding congestion test is performed according to each congested road section, and a congestion dynamic map of urban roads is constructed; the user's location and target location are collected, and multiple navigation sub-paths are defined according to the user's location, target location and the congestion dynamic map of urban roads; the user's navigation path is constructed based on the multiple navigation sub-paths, and the autonomous planning of the user's navigation path is triggered according to the user's navigation path and the congestion change of the congested road section, thereby ensuring the precise control of the navigation path and achieving the technical effect of autonomous planning of the navigation path.

[0147] At this time, multiple navigation sub-paths are frozen and managed. At the same time, the user's navigation path is constructed based on multiple navigation sub-paths, the user's movement pattern and movement time period. Multiple navigation sub-paths, the user's movement pattern and movement time period are introduced, and multiple navigation sub-paths, the user's movement pattern and movement time period are controlled in multiple dimensions to ensure the accuracy of the user's navigation path.

[0148] Furthermore, the user's navigation path is frozen; congestion control is performed on congested sections, and the congestion change of the congested sections is collected within a preset time. The congestion change is introduced to associate the user's navigation path and the congestion change of the congested sections, and then based on the user's navigation path and the congestion change of the congested sections, autonomous planning of the user's navigation path is triggered, so that the user's navigation path is dynamically updated with changes in time or changes in congested sections.

[0149] Therefore, corresponding congestion tests are performed according to each congested road section, and a dynamic congestion map of urban roads is constructed; the user's location and target location are collected, and multiple navigation sub-paths are defined according to the user's location, target location and the dynamic congestion map of urban roads; the user's navigation path is constructed based on the multiple navigation sub-paths, and the autonomous planning of the user's navigation path is triggered according to the user's navigation path and the congestion change of the congested road section, thereby ensuring the precise control of the navigation path and achieving the technical effect of autonomous planning of the navigation path.

[0150] In addition, for the management of road data, this embodiment performs annealing by applying modified likelihood regularization to general GAN ​​training (annealing is used as an algorithm strategy in the field of computing and optimization, especially in simulated annealing algorithms. Simulated annealing is a probabilistic algorithm used to find the global minimum of a function. It simulates the cooling of metal in the physical annealing process, gradually reduces the energy state of the system, thereby helping the algorithm to jump out of the local minimum and finally find the global minimum). As shown below:

[0151]

[0152] GAN is the abbreviation of Generative Adversarial Networks, which is a deep learning model. GAN contains two competing neural networks:

[0153] 1) Generator: Its goal is to generate data samples that are as realistic as possible. These samples are similar to the real dataset in statistical properties, so that it is difficult for the discriminator to distinguish which are real samples and which are fake samples generated by the generator.

[0154] 2) Discriminator: Its goal is to distinguish between real data samples and fake data samples generated by the generator. The better the discriminator is trained, the more realistic the samples generated by the generator need to be in order to fool the discriminator.

[0155] The training process of GAN can be seen as a zero-sum game in which the generator and the discriminator compete with each other. The generator tries to generate more and more realistic samples, while the discriminator tries to improve its ability to distinguish between real and fake samples. This process is optimized through backpropagation and gradient descent until a balance or Nash equilibrium is reached, at which point the quality of the samples generated by the generator is optimal.

[0156] In addition, this embodiment assumes that training is performed in the continuous time limit, that is, the maximization step is solved exactly and instantly, and the evolution of ρ follows the functional gradient of the regularization loss. t To approach μ, we must let λ t tends to 0, otherwise it will cause the sampler to be over-dispersed (λ<0) or over-concentrated (λ>0). However, in some cases, one may want to keep λ ∞ Non-zero, for example, to enhance the robustness of the model (λ<0) or to exclude outliers (λ>0).

[0157] This example assumes two conditions: 1) the availability of empirical data samples for estimating V(μ, ρ; D); and 2) a tractable (non-normalized) distribution for computing logμ(X'). In practice, however, only one of these two conditions holds, that is, either a set of data samples or a model likelihood. The following describes how to construct an alternative implementation.

[0158] 3.1) Replace the likelihood function with an empirical score function estimator

[0159] In general, density estimation is a challenging task in itself. However, in practice, the requirement for a tractable μ function can be relaxed to its score function, given as To understand this, let G(z;θ) be the parameterized generator and the gradient from the regularization term is obtained by the chain rule as For modern automatic differentiation learning platforms, we can use R μ (z) = G(z;θ) T StopGrad{S μ (G(z; θ))} to replace the term logμ(G(z; θ)) in the objective, where StopGrad(·) is a platform-specific API used to stop gradient computation during backpropagation.

[0160] There are many ways to estimate the score function from the data. This embodiment uses a denoising autoencoder estimator because it is simple in structure and robust to complex distributions. If a mapping φ θ (x):χ→χ minimizes the L2 reconstruction loss of independent and identically distributed Gaussian noise-contaminated samples with strength σ, then it is called a denoising autoencoder estimator.

[0161]

[0162] When σ approaches 0, where s σ (x)=(φ θ (x)-x) / σ 2 , and its dependence on μ is omitted to simplify the formula. This embodiment trains a denoising autoencoder estimator modeled by a neural network, optimizes the above formula with a sufficiently small σ, and then inserts it into the score function estimator s σ (x). Since the update of the generator does not involve back-propagation through the denoising autoencoder estimator, this scheme is very efficient.

[0163] 3.2) Solving the problem of non-normalized distribution by generating flows

[0164] When learning based on the deregularized distribution u(x), the D-maximization step can no longer be performed due to the lack of real samples (D-step: train the discriminator D to distinguish between real data and fake data generated by the generator G. This step is achieved by maximizing the probability that the discriminator correctly classifies real and fake data.). However, for f-GANs, it can still be achieved through a bridge measure v that is both easy to sample and easy to evaluate (Bridge measure v (usually denoted by the Greek letter "v" and pronounced "nu" in English) is an auxiliary distribution in statistics and machine learning that is used to connect or bridge two different distributions: one is the target distribution ρ (usually a complex or deregularized distribution) that we want to learn or sample, and the other is an easy-to-sample base distribution, such as a simple noise distribution q(z).) Recall that for f-GANs, D-maximization is equivalent to the estimation of the density ratio r(x). We can estimate using the regular f-GAN D-maximization step using samples from ρ and v. And directly assume v(x)∝exp(-γ(x)) to evaluate Since r(x) = r1(x) + r2(x), D(x) can now be estimated in f-GANs and subsequently updated. In order to distinguish it from conventional GANs that learn from empirical samples, this embodiment is called Unregularized GAN (U-GAN).

[0165] In implementation, this embodiment approaches the bridge metric v by generating a flow in the following form:

[0166] x=f K of K-1 ...f2of1(z),z~q(z)

[0167] Among them, {f K} is a reversible mapping with an easily computable Jacobian matrix, and q(z) is some simple noise distribution. k =f K (z k -1), x=z k +1 and z0=z, we can then trace back the likelihood function by Generally speaking, v should be kept close to ρ in order to make a more accurate estimate of r1(x). Therefore, when optimizing the maximum likelihood estimation objective for v To track the development of the model distribution ρ t .

[0168] This embodiment selects a masked autoregressive flow to construct the bridge measure v. The masked autoregressive flow is a special case of a backward shift-scale flow.

[0169] z k+1 =t(z k+1 )+s(z k+1 )⊙z k

[0170] where t(z),s(z) respect the causal dependencies of the autoregressive flow, so the Jacobian matrix is ​​triangular by design (the Jacobian matrix is ​​a mathematical concept used to describe the local linear approximation of a differentiable function near a point.) and can be computed in parallel. The inverse design makes the masked autoregressive flow fast in evaluation but slow in generation, because the forward pass (e.g., z→x) is similar to the Gaussian elimination process. This feature is particularly suitable for the framework of this embodiment because only the (simpler) evaluation step needs to be backpropagated: included in the G maximization step (G step: train the generator G so that the fake data it generates is as "deceiving" as possible to the discriminator D, so that the discriminator misjudges it as real data. This step is achieved by minimizing the probability that the generated data is correctly identified as fake data by the discriminator, but in fact it maximizes the probability that the discriminator misjudges the fake data as real data.) and the maximum likelihood estimation step. In practice, masked autoregressive flow blocks are stacked and interleaved with permutation layers to build flexible flows.

[0171] 4) Route planning: Based on the predicted traffic congestion, an optimization algorithm is used to calculate the best driving path.

[0172] The optimization algorithm of this embodiment adopts a greedy algorithm:

[0173] The principle of the greedy algorithm is to achieve global optimality through local optimality, and it uses the method of gradually constructing the optimal solution. At each stage, a seemingly optimal solution is made, and once a decision is made, it will not be changed.

[0174] In an embodiment of the present invention, through the method in the embodiment of the present invention, multiple road data are collected based on each road interaction device; corresponding road congestion characteristics are defined according to the multiple road data; corresponding congested sections are defined based on the congestion characteristics of each road, the corresponding positions and the distribution map of urban roads, which is compatible with the overall consideration of the congestion characteristics of each road, the corresponding positions and the distribution map of urban roads, and ensures the accuracy of the congested sections.

[0175] Furthermore, a corresponding congestion test is performed according to each congested road section, and a dynamic congestion map of urban roads is constructed; the user's location and target location are collected, and multiple navigation sub-paths are defined according to the user's location, target location and the dynamic congestion map of urban roads; the user's navigation path is constructed based on the multiple navigation sub-paths, and the autonomous planning of the user's navigation path is triggered according to the user's navigation path and the congestion change of the congested road section, thereby ensuring the precise control of the navigation path and achieving the technical effect of autonomous planning of the navigation path.

[0176] See also Figure 8 , Figure 8 It is a schematic diagram of the structural composition of a path planning system for urban roads in a congested state in an embodiment of the present invention.

[0177] like Figure 8 As shown, a path planning system for a city road in a congested state, the path planning system for a city road in a congested state comprises:

[0178] A road data module 21, used for collecting a plurality of road data based on various road interaction devices;

[0179] A road congestion module 22, configured to define corresponding road congestion features according to a plurality of road data;

[0180] A congested road section module 23, used to define corresponding congested road sections based on the congestion characteristics of each road, the corresponding location and the distribution map of urban roads;

[0181] The congestion dynamic map module 24 is used to perform corresponding congestion tests according to each congested road section and construct a congestion dynamic map of the urban road;

[0182] A navigation sub-path module 25 is used to collect the user's location and target location, and define multiple navigation sub-paths according to the user's location, target location and the congestion dynamic map of the urban road;

[0183] The navigation path module 26 is used to construct a user's navigation path based on multiple navigation sub-paths, and trigger autonomous planning of the user's navigation path according to the user's navigation path and the congestion change amount of the congested road section.

[0184] See also Fig. 9 , refer to the following Fig. 9 The electronic device 40 according to this embodiment of the present invention will be described. Fig. 9 The electronic device 40 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0185] like Fig. 9 As shown, the electronic device 40 is in the form of a general computing device. The components of the electronic device 40 may include but are not limited to: at least one processing unit 41, at least one storage unit 42, and a bus 43 connecting different system components (including the storage unit 42 and the processing unit 41).

[0186] The storage unit stores program codes, which can be executed by the processing unit 41, so that the processing unit 41 executes the steps according to various exemplary embodiments of the present invention described in the above “Embodiment Method” section of this specification.

[0187] The storage unit 42 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 421 and / or a cache memory unit 422 , and may further include a read-only memory unit (ROM) 423 .

[0188] The storage unit 42 may also include a program / utility 424 having a set (at least one) of program modules 425, such program modules 425 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of network electromagnetics.

[0189] Bus 43 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0190] The electronic device 40 may also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device 40, and / or any device that enables the electronic device 40 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 44. Furthermore, the electronic device 40 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 45. Fig. 9 As shown, the network adapter 45 communicates with other modules of the electronic device 40 via the bus 43. It should be understood that although Fig. 9 Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 40, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup planning systems.

[0191] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a multi-parameter sensor device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0192] Those skilled in the art can understand that all or some of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium can include: a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc. In addition, it stores computer program instructions, and when the computer program instructions are executed by a computer, the computer executes the above method.

[0193] In addition, the path planning method and system for urban roads in congested conditions provided by the embodiments of the present invention are introduced in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. For general technical personnel in this field, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A path planning method for urban roads in a congested state, characterized in that: Applied to path planning scenarios in congested urban roads; The path planning method for urban roads in a congested state includes: Collect multiple road data based on each road interaction device; Defining corresponding congestion characteristics of each road according to a plurality of road data; Define the corresponding congested road sections based on the congestion characteristics of each road, the corresponding location and the distribution map of urban roads; Conduct corresponding congestion tests according to each congested road section and build a dynamic congestion map of urban roads; Collect the user's location and target location, and define multiple navigation sub-paths based on the user's location, target location, and a dynamic map of urban road congestion; A user's navigation path is constructed based on multiple navigation sub-paths, and autonomous planning of the user's navigation path is triggered according to the user's navigation path and the congestion change amount of the congested road section.

2. The path planning method for urban roads in congested conditions according to claim 1, characterized in that: The collecting of multiple road data based on each road interaction device includes: Locate each road interaction device; Collect the current position of each road interactive device; Triggering each road interaction device to collect road data according to the current position of each road interaction device and the interaction dimension corresponding to each road interaction device; A plurality of road data are collected based on each road interaction device.

3. The path planning method for urban roads in congested conditions according to claim 2, characterized in that: Defining the corresponding road congestion features according to the plurality of road data includes: Freeze multiple road data; Performing data preprocessing on a plurality of road data, and outputting the preprocessed road data; Construct a corresponding data set according to the preprocessed road data; Match the corresponding data learning model based on each data set; Associating various data sets and data learning models; The corresponding road congestion characteristics are defined based on each data set and data learning model.

4. The path planning method for urban roads in congested conditions according to claim 3, characterized in that: The method of defining corresponding congested road sections based on the congestion characteristics of each road, the corresponding location and the distribution map of urban roads includes: Determine the congestion characteristics of each road; Triggering positioning detection based on various road congestion characteristics; Defining corresponding positions according to the location detection of each road congestion feature; Correlate the congestion characteristics of each road, the corresponding location and the distribution map of urban roads; Multiple interactions are performed on each road congestion feature, its corresponding location, and the distribution map of urban roads; The corresponding congested road sections are defined according to the multiple interactions of the congestion characteristics of each road, the corresponding location and the distribution map of the urban roads.

5. The path planning method for urban roads in congested conditions according to claim 4, characterized in that: The method of performing a corresponding congestion test according to each congested road section and constructing a congestion dynamic map of urban roads includes: Freeze each congested road section; Collect the location of each congested road section, and match the corresponding congestion test according to the location of each congested road section; Carry out corresponding congestion tests according to each congested road section; The corresponding congestion level is defined based on the congestion test of each congested road section, and a congestion dynamic map of urban roads is constructed according to the congestion level, each congested road section, the distribution map of urban roads and the time model.

6. The path planning method for urban roads in congested conditions according to claim 5, characterized in that: The collecting of the user's location and target location, and defining a plurality of navigation sub-paths according to the user's location, the target location and the congestion dynamic map of the urban road, include: Collect the user's location and target location; Define relative space based on the user's location and the target location; Multiple movement paths are defined based on the relative space and the movement pattern of the user.

7. The path planning method for urban roads in congested conditions according to claim 6, characterized in that: The collecting of the user's location and target location, and defining a plurality of navigation sub-paths according to the user's location, the target location and the congestion dynamic map of the urban road, further includes: Collect dynamic images of urban road congestion; Linking multiple mobility paths and congestion dynamic graphs of urban roads; A plurality of navigation sub-paths are defined according to a plurality of moving paths and a dynamic map of congestion of urban roads.

8. The path planning method for urban roads in a congested state according to claim 7, characterized in that: The method of constructing a user's navigation path based on a plurality of navigation sub-paths and triggering autonomous planning of the user's navigation path according to the user's navigation path and the congestion change amount of the congested road section includes: Freeze multiple navigation sub-paths; Constructing a user's navigation path based on multiple navigation sub-paths, the user's movement pattern, and movement time period; Freeze the user's navigation path; Carry out congestion control on congested sections and collect the congestion changes of congested sections within the preset time.

9. The path planning method for urban roads in congested conditions according to claim 8, characterized in that: The method of constructing a user's navigation path based on a plurality of navigation sub-paths and triggering autonomous planning of the user's navigation path according to the user's navigation path and the congestion change amount of the congested road section also includes: Associating the user's navigation path and the congestion change of the congested road section; The user's navigation path is autonomously planned based on the user's navigation path and the congestion change of the congested road section, so that the user's navigation path is dynamically updated with the change of time or the change of the congested road section.

10. A path planning system for urban roads in congested conditions, characterized in that: The path planning system for urban roads in a congested state is applied to the path planning method for urban roads in a congested state as claimed in any one of claims 1 to 9, and the path planning system for urban roads in a congested state comprises: A road data module, used for collecting a plurality of road data based on various road interaction devices; A road congestion module, used for defining corresponding road congestion features according to a plurality of road data; A congested road section module, used to define corresponding congested road sections based on the congestion characteristics of each road, the corresponding location and the distribution map of urban roads; The congestion dynamic map module is used to perform corresponding congestion tests according to each congested road section and construct a congestion dynamic map of urban roads; A navigation sub-path module is used to collect the user's location and target location, and define multiple navigation sub-paths according to the user's location, target location and the congestion dynamic map of urban roads; The navigation path module is used to construct a user's navigation path based on multiple navigation sub-paths, and trigger autonomous planning of the user's navigation path according to the user's navigation path and the congestion change amount of the congested road section.

Images