Control methods, devices, vehicles, media, and procedures for vehicle steering correction function

Abstract

This application relates to the field of vehicle technology, and in particular to a control method, device, vehicle, medium, and program for a vehicle's lane correction function. The method includes: acquiring current environmental information and vehicle driving data; inputting the environmental information and vehicle driving data into a pre-trained scene recognition model, which outputs the current scene. The scene includes a one-to-two scenario and non-one-to-two scenarios, where the one-to-two scenario refers to a scenario where the vehicle's driving direction is at a road fork; if the scene is a one-to-two scenario, the vehicle's lane correction function is disabled; otherwise, the vehicle's lane correction function is restored, and the vehicle's driving direction is corrected in real time. This solves the problem in related technologies where inaccurate scene recognition leads to the false triggering of driver assistance functions, affecting the user's driving experience.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to a control method, device, vehicle, medium, and program for a vehicle's steering correction function. Background Technology

[0002] The "one-to-two" scenario refers to a situation in road traffic where a single lane is split into two or more lanes in a specific area due to road design changes, such as road merging, intersections, or road construction.

[0003] In related technologies, ADAS (Advanced Driving Assistance System) may only work based on lane line information and fail to fully recognize changes in lane structure, leading to the erroneous activation of lane departure warning or correction functions in a two-way road section, causing driving discomfort and safety hazards. Summary of the Invention

[0004] This application provides a control method, device, vehicle, medium, and program for a vehicle's correction function, addressing issues in related technologies such as the false triggering of driver assistance functions due to inaccurate scene recognition, which affects the user's driving experience.

[0005] The first aspect of this application provides a method for controlling a vehicle's deviation correction function, comprising the following steps: acquiring current environmental information and vehicle driving data; inputting the environmental information and vehicle driving data into a pre-trained scene recognition model, wherein the scene recognition model outputs the current scene, wherein the scene includes a one-to-two scenario and a non-one-to-two scenario, wherein a one-to-two scenario is a scenario where the vehicle's driving direction is a road fork; if the scenario is a one-to-two scenario, then the vehicle's deviation correction function is turned off; otherwise, the vehicle's deviation correction function is restored, and the vehicle's driving direction is corrected in real time.

[0006] Optionally, the training method for the scene recognition model includes: acquiring historical driving data of the vehicle; extracting multiple vehicle environment data and vehicle data from the historical driving data as positive samples and multiple normal lane data as negative samples; dividing the positive samples and the negative samples into a training dataset and a validation dataset according to a preset ratio; training the scene recognition model using a pattern recognition algorithm and the training dataset, and validating the scene recognition model based on the validation dataset until a preset condition is met to generate a trained scene recognition model.

[0007] Optionally, the pattern recognition algorithm includes a multilayer perceptron model, a Markov chain, and a support vector machine algorithm.

[0008] Optionally, training the scene recognition model using the pattern recognition algorithm and the training dataset includes: when the pattern recognition algorithm is a multilayer perceptron, using real vehicle data labeled with a one-to-two road segment as label data samples, wherein the multilayer perceptron includes an input layer, one or more hidden layers, and an output layer; inputting label data samples from multiple consecutive frames into the input layer of the multilayer perceptron, wherein the hidden layer analyzes the label data samples, identifies scene similarity in the real vehicle data to determine the weights of the training data, and optimizes the network parameters of the scene recognition model based on the weights using a chained back gradient propagation method to obtain the trained scene recognition model.

[0009] Optionally, the step of training the scene recognition model using the pattern recognition algorithm and the training dataset further includes: when the pattern recognition algorithm is a Markov chain, calculating the probability values ​​of various hidden states of the current sample data through a forward and backward algorithm; when the probability of the split-in state is higher than a second threshold, it is determined to be a split-in segment; constructing a scene model based on the split-in segment; and training the scene recognition model based on the constructed scene model.

[0010] Optionally, training the scene recognition model using the pattern recognition algorithm and the training dataset further includes: when the pattern recognition algorithm is a support vector machine, mapping the input data to a high-dimensional space through a kernel function, dividing different types of data into one-to-two road segments and non-one-to-two road segments through a hyperplane with maximum margin, transforming the one-to-two road segments into a quadratic programming problem through constraints, solving the one-to-two scene using a sequence minimum optimization algorithm, and training the scene recognition model based on the one-to-two scene.

[0011] A second aspect of this application provides a vehicle correction device, comprising: an acquisition module for acquiring current environmental information and vehicle driving data; an input module for inputting the environmental information and vehicle driving data into a pre-trained scene recognition model, wherein the scene recognition model outputs the current scene, wherein the scene includes a one-to-two scenario and a non-one-to-two scenario, wherein a one-to-two scenario is a scenario where the vehicle's driving direction is a road fork; and a processing module for disabling the vehicle's correction function if the scenario is a one-to-two scenario, otherwise restoring the vehicle's correction function and correcting the vehicle's driving direction in real time.

[0012] A third aspect of this application provides a vehicle, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a control method for the vehicle's steering correction function as described in the above embodiments.

[0013] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement a control method for the vehicle's steering correction function as described in the above embodiments.

[0014] The fifth aspect of this application provides a computer program product, which, when executed, is used to implement a control method for the vehicle's deviation correction function as described in the above embodiments.

[0015] Therefore, this application has at least the following beneficial effects:

[0016] In this embodiment, environmental information and vehicle driving data are input into a trained scene recognition model. The scene recognition model outputs the scene at the current moment. If the current scene is a split-scene, the vehicle's correction function is turned off; otherwise, the vehicle's correction function is restored, and the vehicle's driving direction is corrected in real time to improve the accuracy of scene recognition, avoid false triggering of assisted driving functions, and improve the user's driving experience.

[0017] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. Attached Figure Description

[0018] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0019] Figure 1 This is a flowchart of a control method for a vehicle's steering correction function according to an embodiment of this application;

[0020] Figure 2 This is an example diagram of a control method for a vehicle's steering correction function according to an embodiment of this application;

[0021] Figure 3 This is a block diagram of a vehicle steering correction device according to an embodiment of this application;

[0022] Figure 4 This is a structural schematic diagram of a vehicle according to an embodiment of this application. Detailed Implementation

[0023] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0024] The following description, with reference to the accompanying drawings, outlines a method, apparatus, vehicle, storage medium, and program product for controlling a vehicle's steering correction function according to embodiments of this application. Addressing the problem mentioned in the background art where inaccurate scene recognition leads to the false triggering of driver assistance functions, affecting the user's driving experience, this application provides a method for controlling a vehicle's steering correction function. In this method, environmental information and vehicle driving data are input into a trained scene recognition model. The scene recognition model outputs the current scene. If the current scene is a split-scene scenario, the vehicle's steering correction function is disabled; otherwise, the steering correction function is restored, and the vehicle's driving direction is corrected in real time, improving the accuracy of scene recognition, avoiding false triggering of driver assistance functions, and enhancing the user's driving experience. This solves the problem in related technologies where inaccurate scene recognition leads to the false triggering of driver assistance functions, affecting the user's driving experience.

[0025] Specifically, Figure 1 This is a flowchart illustrating a control method for a vehicle's steering correction function provided in an embodiment of this application.

[0026] like Figure 1 As shown, the control method for the vehicle's correction function includes the following steps:

[0027] In step S101, the environmental information and vehicle driving data at the current moment are obtained.

[0028] It is understood that the embodiments of this application can acquire environmental information and vehicle driving data collected at the current moment through cameras or sensors, so as to input the environmental information and vehicle driving data into the trained scene recognition model to recognize the current scene.

[0029] It should be noted that this application can obtain the recharge data collected during vehicle driving, and perceive vehicle operation data such as lane line parameters, vehicle orientation, and vehicle speed.

[0030] In step S102, environmental information and vehicle information are input into the trained scene recognition model, and the scene recognition model outputs the scene at the current moment.

[0031] The scenarios include one-to-two scenarios and non-one-to-two scenarios. The one-to-two scenario refers to a scenario where the vehicle's direction of travel is a road fork.

[0032] It is understood that, in the embodiments of this application, environmental information and vehicle information can be input into a trained scene recognition model, and the scene recognition model outputs the scene at the current moment, so as to control the vehicle's correction function according to the recognized scene.

[0033] In this embodiment, the training method for the scene recognition model is as follows: acquire historical driving data of the vehicle; extract multiple vehicle environment data and vehicle data of one-to-two road segments from the historical driving data as positive samples and multiple normal lane data as negative samples; divide the positive samples and negative samples into training dataset and validation dataset according to a preset ratio; train the scene recognition model using a pattern recognition algorithm and the training dataset, and validate the scene recognition model based on the validation dataset until the preset conditions are met to generate the trained scene recognition model.

[0034] The preset ratio can be 7:3 or 6:4, which can be set according to actual needs without specific limitations.

[0035] Pattern recognition algorithms include multilayer perceptron models, Markov chains, and support vector machine algorithms.

[0036] It is understood that, in this embodiment of the application, multiple extracted vehicle environment data and vehicle data of one-to-two road segments can be used as positive samples and multiple normal lane data as negative samples; the positive samples and negative samples are divided into training datasets and validation datasets according to a preset ratio; the scene recognition model is trained using a pattern recognition algorithm and the training dataset, and the scene recognition model is validated based on the validation dataset until the preset conditions are met to generate a trained scene recognition model, so as to improve the accuracy of scene recognition.

[0037] In this embodiment, a scene recognition model is trained using a pattern recognition algorithm and a training dataset, including: when the pattern recognition algorithm is a multilayer perceptron, real vehicle data labeled with a one-to-two road segment is used as label data samples, wherein the multilayer perceptron includes an input layer, one or more hidden layers and an output layer; label data samples of multiple consecutive frames are input to the input layer of the multilayer perceptron, the hidden layer analyzes the label data samples, identifies the scene similarity in the real vehicle data to determine the weights of the training data, and optimizes the network parameters of the scene recognition model based on the weights using a chained back gradient propagation method to obtain the trained scene recognition model.

[0038] The first threshold can be set according to actual needs, without specific limitations.

[0039] It is understood that, in the embodiments of this application, when the pattern recognition algorithm is a multilayer perceptron model, the network model parameters can be optimized by using the chained back gradient propagation method to train a scene recognition model, thereby improving the accuracy of the scene recognition model in recognizing scenes.

[0040] In this embodiment of the application, the scene recognition model is trained using a pattern recognition algorithm and a training dataset. The method further includes: when the pattern recognition algorithm is a Markov chain, the probability values ​​of various hidden states of the current sample data are calculated by forward and backward algorithms. When the probability of the split-in state is higher than the second threshold, it is determined to be a split-in segment. A scene model is constructed based on the split-in segment, and the scene recognition model is trained based on the constructed scene model.

[0041] It is understood that, in the embodiments of this application, when the pattern recognition algorithm is a Markov chain, the probability values ​​of various hidden states of the current sample data can be calculated by forward and backward algorithms. When the probability of the split-in state is higher than the second threshold, it is determined to be a split-in road segment. Based on the constructed scene model, the scene recognition model is trained to improve the accuracy of the scene recognition model in recognizing the scene.

[0042] In this embodiment of the application, the scene recognition model is trained using a pattern recognition algorithm and a training dataset. The training process further includes: when the pattern recognition algorithm is a support vector machine, the input data is mapped to a high-dimensional space through a kernel function, different types of data are divided into one-to-two road segments and non-one-to-two road segments through a hyperplane with the maximum margin, the one-to-two road segments are transformed into a quadratic programming problem through constraints, and the one-to-two scene is obtained by solving the sequence minimum optimization algorithm. The scene recognition model is then trained based on the one-to-two scene.

[0043] It is understood that in the embodiments of this application, if the pattern recognition algorithm is a support vector machine, the input data is mapped to a high-dimensional space through the kernel function, and different types of data are divided into two categories through the hyperplane with the maximum margin. The problem is transformed into a quadratic programming problem through constraints, and the SVM scene recognition model is obtained by using the sequence minimum optimization algorithm. The scene recognition model is trained based on the SVM scene recognition model to improve the accuracy of scene recognition model in recognizing scenes.

[0044] In step S103, if the scenario is a split-in scenario, the vehicle's correction function is turned off; otherwise, the vehicle's correction function is restored, and the vehicle's driving direction is corrected in real time.

[0045] It is understood that, in the embodiments of this application, if the current scenario is a one-to-two scenario, the vehicle's correction function will be turned off; otherwise, the vehicle's correction function will be restored, thereby improving the accuracy of scenario recognition, avoiding false triggering of assisted driving functions, and improving the user's driving experience.

[0046] According to the vehicle correction function control method proposed in the embodiments of this application, environmental information and vehicle driving data are input into a trained scene recognition model. The scene recognition model outputs the scene at the current moment. If the current scene is a split-scene, the vehicle correction function is turned off; otherwise, the vehicle correction function is restored, and the vehicle's driving direction is corrected in real time, thereby improving the accuracy of scene recognition, avoiding false triggering of assisted driving functions, and improving the user's driving experience.

[0047] The following will combine Figure 2 The control method for the vehicle's steering correction function in this application is described in detail below:

[0048] Incorporating pattern recognition into functional decision-making optimization algorithms mainly consists of three parts: data collection, model training, and policy execution.

[0049] Data collection: Using vehicle driving data, a large amount of vehicle environment data and vehicle data from the two-way road section were collected as positive samples, and a large amount of normal lane data were collected as negative samples.

[0050] Model training: Using vehicle environment data, especially lane line data, vehicle orientation, speed and other information, the most representative data features are extracted through correlation verification. The model is then trained using pattern recognition algorithms, such as neural networks, Markov chains, support vector machines and other methods.

[0051] Strategy execution: A one-to-two scene recognition model is added to the vehicle's deviation correction function suppression. When the model determines that the scene is a one-to-two scene for multiple consecutive frames, the vehicle's deviation correction function is set to a suppressed state. When the model determines that the current scene is not a one-to-two scene for multiple consecutive frames, the vehicle's deviation correction function is restored.

[0052] It should be noted that in the suppression state described in this application, the vehicle's lane correction system is activated under automatic vehicle control. That is, if the vehicle deviates from the lane while driving normally, the vehicle's lane correction system will issue an alarm.

[0053] At the same time, the vehicle's steering system controls the steering wheel to ensure that the vehicle returns to the center of the lane when there is a slight turn, thereby correcting the vehicle's driving direction.

[0054] In actual vehicles, the main execution process is as follows: First, the vehicle and environmental data are obtained through perception and the vehicle body. Then, the pattern recognition method is used to determine whether the current vehicle is in a one-way road segment. When the vehicle is identified to be in a one-way road segment, the decision module suppresses the vehicle correction function.

[0055] It should be noted that this application also includes: an offline training part, which collects environment and data from real vehicles, labels the data to determine whether it is a one-to-two road segment, constructs training samples, and trains a scene recognition model that can identify whether a scene is a one-to-two road segment by combining a pattern recognition algorithm.

[0056] In the real-time execution section, the scene recognition model is embedded into the decision-making module of the software. When the scene recognition model identifies the current scene as a one-way road segment, the decision-making module suppresses the vehicle correction function. When the recognized scene enters a normal road segment, the suppression is lifted.

[0057] Specifically, pattern recognition algorithms are used to achieve scene identification. Considering the computational power and recognition accuracy requirements of the algorithm, classic machine learning algorithms are employed for clustering. These include, but are not limited to, methods such as neural networks, Markov chains, and support vector machines.

[0058] Neural Network: A multilayer perceptron model is used to achieve a non-linear mapping between input and output. Real vehicle data labeled as indicating whether it is a one-way / two-way road segment is used as the label data sample. Considering the correlation of data, multiple consecutive frames of data are used as model input. The network model parameters are continuously optimized through a chained backward gradient propagation method to train the scene model. The model is then embedded into the software's decision-making function to determine whether to suppress certain actions.

[0059] A Markov chain consists of three main modules: model parameter learning, evaluation of observed sequence probability prediction, and model parameter training using the Baum-Welch algorithm based on the EM algorithm. The module constructs a scene model and calculates the probability values ​​of various hidden states in the current observed sequence using a forward and backward algorithm. When the probability of a split-in state is higher than a threshold, it is judged as a split-in segment. The model is then embedded into the software's decision-making function to determine whether to suppress it.

[0060] Support Vector Machines (SVMs) map input data to a higher-dimensional space using kernel functions, making it easier to classify different types of data into two classes using a maximum-margin hyperplane. The problem is transformed into a quadratic programming problem using the KKT conditions and solved using the SMO algorithm. This yields an SVM scene recognition model; the model is then embedded into the software's decision-making function to determine whether to suppress certain elements.

[0061] In summary, this application uses a pattern recognition method to identify lane line scenes as split-in scenarios, suppresses functions, improves the accuracy of scene recognition, and enhances the user's driving experience.

[0062] Next, the vehicle correction device according to the embodiments of this application is described with reference to the accompanying drawings.

[0063] Figure 3 This is a block diagram of a vehicle correction device according to an embodiment of this application.

[0064] like Figure 3 As shown, the vehicle's correction device 10 includes: an acquisition module 100, an input module 200, and a processing module 300.

[0065] The acquisition module 100 is used to acquire the environmental information and vehicle driving data at the current moment; the input module 200 is used to input the environmental information and vehicle information into the trained scene recognition model, and the scene recognition model is used to input the environmental information and vehicle driving data into the pre-trained scene recognition model, and the scene recognition model outputs the scene at the current moment. The scene includes a one-to-two scene and a non-one-to-two scene. The one-to-two scene is a scene where the vehicle's driving direction is a road fork; the processing module 300 is used to turn off the vehicle's correction function if the scene is a one-to-two scene, otherwise restore the vehicle's correction function and correct the vehicle's driving direction in real time.

[0066] It should be noted that the explanation of the control method embodiment for the vehicle's correction function described above also applies to the vehicle's correction device in this embodiment, and will not be repeated here.

[0067] According to the vehicle correction device proposed in the embodiments of this application, environmental information and vehicle driving data are input into a trained scene recognition model. The scene recognition model outputs the scene at the current moment. If the current scene is a split-scene, the vehicle correction function is turned off; otherwise, the vehicle correction function is restored, and the vehicle's driving direction is corrected in real time, thereby improving the accuracy of scene recognition, avoiding false triggering of assisted driving functions, and improving the user's driving experience.

[0068] Figure 4 A schematic diagram of the structure of a vehicle provided in an embodiment of this application. The vehicle may include:

[0069] The memory 401, the processor 402, and the computer program stored on the memory 401 and capable of running on the processor 402.

[0070] When the processor 402 executes the program, it implements the control method for the vehicle's deviation correction function provided in the above embodiments.

[0071] Furthermore, the vehicle also includes:

[0072] Communication interface 403 is used for communication between memory 401 and processor 402.

[0073] The memory 401 is used to store computer programs that can run on the processor 402.

[0074] The memory 401 may include high-speed RAM (Random Access Memory) memory, and may also include non-volatile memory, such as at least one disk storage.

[0075] If the memory 401, processor 402, and communication interface 403 are implemented independently, then the communication interface 403, memory 401, and processor 402 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 4 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0076] Optionally, in a specific implementation, if the memory 401, processor 402, and communication interface 403 are integrated on a single chip, then the memory 401, processor 402, and communication interface 403 can communicate with each other through an internal interface.

[0077] Processor 402 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement embodiments of this application.

[0078] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described vehicle correction function control method.

[0079] This application also provides a computer program product, including a computer program or instructions, which, when executed, is used to implement a control method for the vehicle's deviation correction function as described in the above embodiments.

[0080] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0081] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of this application, "N" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0082] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing a custom logical function or process step, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may be performed in a different order than shown or discussed, including performing functions in a substantially simultaneous manner or in a reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present application pertain.

[0083] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiment, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, it can be implemented using any one or a combination of the following technologies known in the art: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array, a field programmable gate array, etc.

[0084] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0085] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A control method for the vehicle's deviation correction function, characterized in that, Includes the following steps: Obtain current environmental information and vehicle driving data; The environmental information and vehicle driving data are input into a pre-trained scene recognition model. The scene recognition model outputs the scene at the current moment. The scene includes a one-to-two road segment and non-one-to-two road segments. The one-to-two road segment is a scene where the vehicle's driving direction is a road fork. The training method of the scene recognition model is as follows: acquire historical driving data of the vehicle; extract multiple one-to-two road segment vehicle environmental data and vehicle data as positive samples and multiple normal lane data as negative samples from the historical driving data; divide the positive samples and the negative samples into a training dataset and a validation dataset according to a preset ratio; train the scene recognition model using a pattern recognition algorithm and the training dataset, and validate the scene recognition model based on the validation dataset until the preset conditions are met to generate a trained scene recognition model. If the scenario is a one-to-two scenario, the vehicle's correction function will be turned off; otherwise, the vehicle's correction function will be restored, and the vehicle's driving direction will be corrected in real time.

2. The control method for the vehicle's correction function according to claim 1, characterized in that, The pattern recognition algorithms include multilayer perceptron models, Markov chains, and support vector machine algorithms.

3. The control method for the vehicle's correction function according to claim 2, characterized in that, The process of training the scene recognition model using the pattern recognition algorithm and the training dataset includes: When the pattern recognition algorithm is a multilayer perceptron, the real vehicle data of the one-to-two road segment is used as the label data sample. The multilayer perceptron includes an input layer, one or more hidden layers and an output layer. Label data samples from multiple consecutive frames are input into the input layer of a multilayer perceptron. The hidden layer analyzes the label data samples, identifies the scene similarity in the real vehicle data, determines the weights of the training data, and optimizes the network parameters of the scene recognition model based on the weights using a chained back gradient propagation method to obtain a trained scene recognition model.

4. The control method for the vehicle's correction function according to claim 2, characterized in that, The step of training the scene recognition model using the pattern recognition algorithm and the training dataset further includes: When the pattern recognition algorithm is a Markov chain, the probability values ​​of various hidden states of the current sample data are calculated by forward and backward algorithms. When the probability of the split-in state is higher than the second threshold, it is determined to be a split-in segment. A scene model is constructed based on the split-in segment, and a scene recognition model is trained based on the constructed scene model.

5. The control method for the vehicle's correction function according to claim 2, characterized in that, The step of training the scene recognition model using the pattern recognition algorithm and the training dataset further includes: When the pattern recognition algorithm is a support vector machine, the input data is mapped to a high-dimensional space through a kernel function. Different types of data are divided into one-to-two road segments and non-one-to-two road segments through a hyperplane with the maximum margin. The one-to-two road segments are transformed into a quadratic programming problem through constraints, and the one-to-two scene is obtained by solving the sequence minimum optimization algorithm. The scene recognition model is trained based on the one-to-two scene.

6. A vehicle steering correction device, characterized in that, A control method for implementing the vehicle's steering correction function as described in any one of claims 1-5, comprising: The acquisition module is used to acquire environmental information and vehicle driving data at the current moment; The input module is used to input the environmental information and vehicle driving data into a pre-trained scene recognition model. The scene recognition model outputs the scene at the current moment. The scene includes a one-to-two scene and a non-one-to-two scene. The one-to-two scene is a scene where the vehicle's driving direction is a road fork segment. The processing module is used to disable the vehicle's correction function if the scenario is a one-to-two scenario, otherwise restore the vehicle's correction function and correct the vehicle's driving direction in real time.

7. A vehicle, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement a control method for the vehicle's steering correction function as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the control method for the vehicle's steering correction function as described in any one of claims 1-5.

9. A computer program product, comprising: A computer program or instruction, characterized in that, when executed, the computer program or instruction implements a control method for the vehicle's deviation correction function as described in any one of claims 1-5.

Images