Vehicle control method and related equipment

Abstract

The invention provides a vehicle control method and related equipment. The vehicle control method comprises the steps that driving environment parameters, driving state parameters and historical characteristic parameters of a current vehicle are obtained; performing data processing on the driving environment parameters, the driving state parameters and the historical characteristic parameters to obtain target data; predicting a target distance of the current vehicle according to the target data, wherein the target distance comprises at least one of a vehicle following distance and a braking distance; and controlling the current vehicle to run according to the target distance. According to the invention, the determination precision of the following distance and the braking distance of the vehicle is improved.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and more particularly to a vehicle control method and related equipment. Background Technology

[0002] As people's living standards improve, vehicles have become an indispensable means of transportation in people's daily lives.

[0003] When driving, a vehicle needs to maintain a safe distance from the vehicle in front so that it has enough distance to brake in case of an emergency. In other words, a vehicle needs to maintain an appropriate following distance and braking distance to drive safely.

[0004] In the exemplary technology, the following distance and braking distance are determined by obtaining the speed of the vehicle in front and the speed of the vehicle itself. However, determining the following distance and braking distance solely based on the vehicle speed will result in low accuracy in determining the following distance and braking distance. Summary of the Invention

[0005] Based on the aforementioned technological status, this application provides a vehicle control method and related equipment to address the problem of low accuracy in determining vehicle following distance and braking distance.

[0006] To achieve the above-mentioned technical objectives, this application proposes the following technical solution:

[0007] In a first aspect, this application provides a vehicle control method, including:

[0008] The vehicle acquires current driving environment parameters, driving status parameters, and historical feature parameters. The driving environment parameters include at least one of the speed parameters, lighting parameters, weather parameters, and road parameters of other vehicles around the current vehicle. The driving status parameters include at least one of the vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters of the current vehicle. The historical feature parameters include the current vehicle's historical following distance and braking frequency.

[0009] The driving environment parameters, driving status parameters, and historical feature parameters are processed to obtain target data.

[0010] Predict the target distance of the current vehicle based on the target data, wherein the target distance includes at least one of following distance and braking distance;

[0011] The current vehicle's movement is controlled based on the target distance.

[0012] In some implementations, predicting the target distance of the current vehicle based on the target data includes:

[0013] Spatial and temporal features are extracted from the target data, and an initial distance is determined based on the spatial and temporal features.

[0014] Based on the target data, correction parameters are determined, and the initial distance is corrected according to the correction parameters to obtain the target distance of the current vehicle.

[0015] In some implementations, if the initial distance includes a base following distance, then the correction parameter includes at least one of an environmental correction parameter and a driving state correction parameter.

[0016] If the initial distance includes the base braking distance, then the correction parameters include at least one of the following: friction coefficient decay correction parameters, driver reaction time correction parameters, and obstacle correction parameters.

[0017] In some implementations, predicting the target distance of the current vehicle based on the target data includes:

[0018] The target data is input into the prediction model to obtain the target distance of the current vehicle output by the prediction model. The target distance includes the lower limit of the following distance and the upper limit of the braking distance output by the prediction model.

[0019] In some implementations, the prediction model includes multiple parallel neural networks, and the target distance is the average of the output distances of each of the neural networks.

[0020] In some implementations, predicting the target distance of the current vehicle based on the target data includes:

[0021] Obtain real-time weather data, road construction data, real-time traffic data, and map data corresponding to the area where the vehicle is currently located;

[0022] Based on the real-time weather data, the road construction data, the real-time traffic data, and the map data, the driving environment of the current vehicle in the future time period is predicted, and the driving environment includes the weather environment and the road condition environment.

[0023] The target distance is obtained by predicting the distance to be determined for the current vehicle based on the target data and correcting the distance to be determined based on the driving environment.

[0024] In some implementations, predicting the driving environment of the current vehicle in a future time period based on the real-time weather data, the road construction data, the real-time traffic data, and the map data includes:

[0025] If the current driving environment to be predicted for the vehicle is weather conditions, the area where the current vehicle is located is divided into multiple sub-regions based on the map data.

[0026] The rainfall and wind direction parameters for each sub-region are determined based on the real-time weather data.

[0027] Based on the rainfall and wind direction parameters, the weather conditions for the current vehicle in the future time period are determined.

[0028] In some implementations, predicting the driving environment of the current vehicle in a future time period based on the real-time weather data, the road construction data, the real-time traffic data, and the map data includes:

[0029] If the current driving environment to be predicted for the vehicle is a road condition environment, the probability value of construction at the location of the current vehicle in a future time period is predicted based on road construction data and the map data.

[0030] Based on the construction probability value and the real-time traffic data, the traffic environment in which the current vehicle will be located in the future time period is predicted.

[0031] Thirdly, this application provides a vehicle, including:

[0032] The acquisition module is used to acquire the current vehicle's driving environment parameters, driving status parameters, and historical feature parameters. The driving environment parameters include at least one of the speed parameters, light parameters, weather parameters, and road parameters of other vehicles around the current vehicle. The driving status parameters include at least one of the current vehicle's vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters. The historical feature parameters include the current vehicle's historical following distance and braking frequency.

[0033] The processing module is used to process the driving environment parameters, driving status parameters, and historical feature parameters to obtain target data.

[0034] The prediction module is used to predict the target distance of the current vehicle based on the target data, wherein the target distance includes at least one of following distance and braking distance;

[0035] The control module is used to control the current vehicle's movement based on the target distance.

[0036] Thirdly, this application provides an electronic device, including a memory and a processor, wherein,

[0037] The memory is connected to the processor and is used to store programs;

[0038] The processor is used to implement the vehicle control method as described in the first aspect or any implementation thereof by running a program in the memory.

[0039] Fourthly, this application provides a computer program product, which, when executed by a processor, implements the vehicle control method as described in the first aspect or any implementation thereof.

[0040] Fifthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the vehicle control method as described in the first aspect or any implementation thereof.

[0041] This application provides a vehicle control method and related equipment that acquires current vehicle driving environment parameters, driving state parameters, and historical feature parameters. The method processes these parameters to obtain target data, and predicts the following distance and braking distance of the current vehicle based on this target data. Finally, the method controls the vehicle's movement based on the target distance. In this application, the following distance and braking distance are predicted using driving environment parameters such as speed, lighting, weather, and road parameters of other vehicles around the vehicle; driving state parameters such as vibration, speed, load, braking, and tire wear parameters of the vehicle; and historical following distance, braking frequency, and lighting historical feature parameters. This comprehensive approach, using multiple parameters, improves the accuracy of determining the following distance and braking distance. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0043] Figure 1 A flowchart of a vehicle control method provided in this application embodiment Figure 1 ;

[0044] Figure 2 A flowchart of a vehicle control method provided in this application embodiment Figure 2 ;

[0045] Figure 3 A flowchart of a vehicle control method provided in this application embodiment Figure 3;

[0046] Figure 4 A flowchart of a vehicle control method provided in this application embodiment Figure 4 ;

[0047] Figure 5 A flowchart of a vehicle control method provided in this application embodiment Figure 5 ;

[0048] Figure 6 A flowchart of a vehicle control method provided in this application embodiment Figure 6 ;

[0049] Figure 7 A schematic diagram of the structure of a vehicle provided in an embodiment of this application;

[0050] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0051] The technical solution proposed in this application is applicable to vehicle driving scenarios and aims to improve the accuracy of determining vehicle following distance and braking distance. By employing the technical solution described in this application, the following distance and braking distance of a vehicle are predicted using driving environment parameters such as speed parameters, lighting parameters, weather parameters, and road parameters of other vehicles around the vehicle; driving state parameters such as vehicle vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters; historical following distance and braking frequency and historical lighting characteristic parameters of the vehicle. In other words, the following distance and braking distance of a vehicle are determined by comprehensively considering multiple parameters, thereby improving the accuracy of determining vehicle following distance and braking distance.

[0052] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0053] It should be noted that the user information (including but not limited to electrical equipment information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with relevant laws, regulations and standards, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0054] As people's living standards improve, vehicles have become an indispensable means of transportation in people's daily lives.

[0055] When driving, a vehicle needs to maintain a safe distance from the vehicle in front so that it has enough distance to brake in case of an emergency. In other words, a vehicle needs to maintain an appropriate following distance and braking distance to drive safely.

[0056] In the exemplary technology, the following distance and braking distance are determined by obtaining the speed of the vehicle in front and the speed of the vehicle itself. However, determining the following distance and braking distance solely based on the vehicle speed will result in low accuracy in determining the following distance and braking distance.

[0057] In view of this, embodiments of this application aim to provide a vehicle control method and related equipment, which acquires the current vehicle's driving environment parameters, driving state parameters, and historical characteristic parameters, processes the driving environment parameters, driving state parameters, and historical characteristic parameters to obtain target data, and predicts the current vehicle's following distance and braking distance based on the target data to obtain the target distance, and then controls the current vehicle's movement based on the target distance. In this application, the following distance and braking distance of the vehicle are predicted by using driving environment parameters such as speed parameters, light parameters, weather parameters, and road parameters of other vehicles around the vehicle, driving state parameters such as vehicle vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters, and the vehicle's historical following distance and braking frequency and historical light characteristic parameters. That is, the following distance and braking distance of the vehicle are determined by comprehensively considering multiple parameters, thereby improving the accuracy of the determination of the following distance and braking distance.

[0058] The vehicle control method and related equipment provided in this application can be applied to vehicle driving scenarios.

[0059] Exemplary methods

[0060] Figure 1 A flowchart of a vehicle control method provided in this application embodiment Figure 1 .like Figure 1 As shown, the vehicle control method provided in this embodiment includes:

[0061] Step S101: Obtain the current vehicle's driving environment parameters, driving status parameters, and historical feature parameters. The driving environment parameters include at least one of the speed parameters, lighting parameters, weather parameters, and road parameters of other vehicles around the current vehicle. The driving status parameters include at least one of the current vehicle's vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters. The historical feature parameters include the current vehicle's historical following distance and braking frequency.

[0062] In this embodiment, the vehicle determines the following distance and braking distance during operation. To do this, the vehicle acquires current driving environment parameters, driving status parameters, and historical feature parameters.

[0063] Driving environment parameters include at least one of the following: speed parameters of other vehicles around the vehicle, lighting parameters, weather parameters, and road parameters. Speed ​​parameters are obtained through the vehicle's speed sensor; lighting parameters are obtained through the vehicle's photosensor; weather parameters can be obtained by requesting information from a weather platform; and road parameters can be obtained by capturing images of the road where the vehicle is located through an image acquisition module and identifying the road within the images. Other vehicle speed parameters include vehicle speed and acceleration; lighting parameters include, for example, the on / off status of brake lights; weather parameters are real-time weather parameters, such as visibility and rainfall; and road parameters include, for example, the road surface friction coefficient, curve curvature, gradient, and lane width. Driving environment parameters can be obtained using augmented reality (AR) and virtual reality (VR) technologies. Specifically, AR technology adds the speed, lighting, road, and weather parameters collected by the vehicle's sensors to the electronic map displayed on the vehicle. Furthermore, VR technology can generate multiple virtual vehicles and display them on the vehicle's screen; these virtual vehicles are real vehicles in the same environment as the vehicle in question.

[0064] Vehicle status parameters include vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters. Vibration parameters can be obtained from vibration signals during vehicle operation, such as vibration sensors distributed in the vehicle's chassis, seats, and steering wheel. Speed ​​parameters include the vehicle's speed and acceleration. Load parameters are the vehicle's load weight. Braking parameters include the current state of the braking system. Tire wear parameters are the degree of tire wear.

[0065] Historical characteristic parameters include the vehicle's historical following distance and braking frequency. Historical following distance can be the trend of the vehicle's following distance over a period of time, for example, 5 seconds. Braking frequency refers to the ratio of the number of times the vehicle brakes in the past time period to the duration of the past time period.

[0066] Driving environment parameters, driving status parameters, and historical characteristic parameters can be obtained through the vehicle's sensors.

[0067] In a vehicle, data acquisition modules such as sensors act like the vehicle's "sensory organs," distributed across various key locations. Road condition sensors are installed on the chassis near the tires to detect subtle changes in the road surface in advance; a comprehensive weather monitoring module is placed on the roof to collect real-time external weather information; a tire omnidirectional monitoring system is embedded inside the tires to accurately monitor tire condition; vibration sensors are distributed in the chassis, seats, steering wheel, and other areas to capture vibration signals during vehicle operation; and the vehicle networking module is responsible for communicating with the outside world and obtaining information about the dynamics of surrounding vehicles.

[0068] The data processing module in the vehicle is like the "brain" of the vehicle. After receiving the data from the sensors, the multi-source data fusion algorithm unit integrates and analyzes the data, as shown in the following text on data processing.

[0069] The execution and feedback module in the vehicle acts as the vehicle's "commander of action" and "learning center." The braking and power systems adjust the vehicle's driving state according to the instructions of the intelligent decision-making unit. The human-machine interface provides feedback to the driver and receives feedback from the driver to help the vehicle optimize its algorithms, such as vehicle control based on a determined following distance and braking distance, as described below.

[0070] The following describes the data acquisition modules, such as those for sensors.

[0071] 1. Road Condition Sensor. The electromagnetic wave transmitting unit of the road condition sensor generates an electromagnetic wave signal of a specific frequency through an oscillation circuit. After power amplification, the signal is emitted by the transmitting antenna. Simultaneously, a control signal modulates the transmitted signal through a modulation circuit, enabling it to carry specific information. When the electromagnetic wave is reflected back from the road surface, the receiving antenna receives the weak signal. This signal is first amplified by a low-noise amplifier circuit to increase its strength and reduce noise, then demodulated to reconstruct the original signal, and finally transmitted to the signal processing chip. The signal processing chip analyzes the reflection time, intensity, and other characteristics of the electromagnetic wave to determine the road surface material, the presence of foreign objects, and the degree of dryness or wetness. The processed data is then sent to the vehicle's data processing module via the data output interface.

[0072] In addition, lidar scanning can be used to replace electromagnetic wave detection. By emitting multiple laser beams and receiving the reflected signals, the time-flight principle is used to accurately measure the road surface contour and surface features. It can identify road surface protrusions and depressions at the millimeter level, as well as information such as water depth and ice thickness. At the same time, combined with image recognition algorithms, the road surface image formed by laser point cloud data can be analyzed to determine the road surface material and the presence of foreign objects.

[0073] 2. Weather Comprehensive Monitoring Module. This module includes, for example, a weather radar. Weather radar emits electromagnetic waves and receives reflected waves, analyzing the echo signals to determine the location, direction of movement, and intensity of weather systems. The module also includes a humidity sensor, which utilizes the hygroscopic properties of polymer materials; its resistance or capacitance changes with air humidity, and humidity data is obtained by measuring this change. Additionally, the module includes a light sensor, typically using photodiodes or photoresistors. The resistance changes with varying light intensity, converting light intensity into an electrical signal. Finally, the module includes a raindrop sensor, which measures rainfall and raindrop size by detecting vibrations or capacitance changes caused by raindrop impacts. All sensor data undergoes initial processing, including noise removal and format conversion, before being integrated into complete weather data and sent to the data processing module.

[0074] 3. Tire All-Around Monitoring Module. This module includes an internal tire pressure sensor. Other sensors utilize piezoresistive or capacitive principles; changes in tire pressure cause physical changes in the sensor's internal structure, which are then converted into electrical signals. The pressure data is sent to the central processor every 500 milliseconds. The module also includes a temperature sensor, typically a thermistor, whose resistance changes with temperature. It collects and transmits temperature data every second. Furthermore, it includes a tread wear sensor, which senses changes in tire tread depth and sends wear data every 10 seconds. Finally, it includes a tire deformation sensor, which uses strain gauge technology. When the tire deforms, the strain gauge's resistance changes, and deformation data is sent every 2 seconds, ensuring the vehicle can monitor tire conditions in real time.

[0075] Furthermore, a non-embedded monitoring system based on strain gauge arrays is used. A flexible strain gauge array is installed on the outer side of the tire, and by sensing the deformation stress distribution during tire rolling, tire pressure, wear, and load changes can be indirectly calculated. This solution requires no internal modifications to the tire, is easy to install, and can acquire real-time dynamic mechanical data of the tire.

[0076] 4. Vision Sensors. Vision sensors, such as image acquisition modules, can continuously capture images of surrounding vehicles. These images can be used to determine the speed and headlight parameters of other vehicles. Furthermore, the images acquired by the image acquisition module can be used to identify road parameters, such as lane width.

[0077] Step S102: Perform data processing on driving environment parameters, driving status parameters, and historical feature parameters to obtain target data.

[0078] After obtaining driving environment parameters, driving status parameters, and historical feature parameters, data processing is performed on these parameters to obtain the target data.

[0079] For example, after obtaining driving environment parameters, driving status parameters, and historical feature parameters, the data processing stage will convert parameters of different types and units into a unified format and perform normalization processing to ensure that the parameters after data processing are within the same numerical range, which is convenient for subsequent calculations.

[0080] Driving environment parameters, driving status parameters, and historical feature parameters are first identified and classified. Numerical parameters include:

[0081] Continuous values: vehicle speed (km / h), temperature (°C), road surface friction coefficient (dimensionless);

[0082] Discrete values: gear level (1-5), weather type (0-sunny, 1-rainy, etc.).

[0083] Text parameters include:

[0084] Traffic sign text (e.g., "Construction Ahead") and sensor status descriptions ("Normal", "Abnormal").

[0085] Image and signal type parameters:

[0086] Image from vehicle camera (RGB pixel values), waveform from vibration sensor (time domain signal).

[0087] Standardize the format of driving environment parameters, driving status parameters, and historical status parameters:

[0088] Numerical parameters: converted to floating-point numbers for storage, with units standardized to international standards (e.g., speed in m / s, temperature in K).

[0089] Text parameters:

[0090] Categorical Encoding: Maps weather types to numbers (sunny → 0, rainy → 1, snowy → 2);

[0091] One-hot encoding: For multi-category and unordered data (such as traffic sign types), it generates binary vectors (e.g., "speed limit 60" → [0,1,0,0,...]).

[0092] Image and signal parameters:

[0093] Images: uniform resolution (e.g., 224×224 pixels), pixel values ​​normalized to [0,1];

[0094] Signal: Resampled to a uniform frequency (e.g., vibration signals are uniformly sampled at 100Hz) and converted into a timing array.

[0095] The core methods for parameter normalization include minimization-maximization normalization, with the following formula:

[0096] x_norm = (x - x_min) / (x_max - x_min), where x is the parameter to be normalized, x_norm is the normalization parameter, x_min is the minimum value, and x_max is the maximum value.

[0097] This method is applicable to data with a clear extreme value range, such as vehicle speed (0-120km / h→0-1) and rainfall (0-200mm→0-1). This method preserves the original data distribution and is simple to calculate.

[0098] The formula for the core method of parameter normalization is:

[0099] x_norm = (x - μ) / σ. Where μ is the characteristic mean and σ is the standard deviation. This method is applicable to data that follows a normal distribution, such as tire pressure (which fluctuates significantly due to temperature) and vibration acceleration (which can have positive and negative values). This normalization method eliminates the influence of dimensions and is insensitive to outliers (the standardized mean is 0, and the standard deviation is 1). For example, original tire pressure: 2.5 bar (μ = 2.4 bar, σ = 0.1 bar) → standardized value = (2.5 - 2.4) / 0.1 = 1.0.

[0100] Normalization methods also include logarithmic transformation. The formula is: x_norm = log(x+1) / log(max_x+1); x+1 avoids the logarithm being meaningless when x=0, and max_x is the maximum value of the characteristic. Application scenarios include: data with extremely large ranges such as sensor signal strength (e.g., radar echo intensity, numerical range 1-10000) and traffic flow (0-1000 vehicles / h); the effect of this normalization method is: compressing large numerical ranges and highlighting small numerical differences (e.g., the difference between 10 and 20 is more obvious after logarithmic transformation than between 1000 and 1010).

[0101] Normalization methods also include quantile normalization. Its principle is to map the distribution of all features to the same quantile, so that different features have the same empirical distribution. Application scenarios include multi-sensor data fusion (such as camera pixel values ​​and radar distance values), ensuring the comparability of different distribution features. The steps of quantile normalization are as follows:

[0102] 1. Calculate the quantiles (e.g., 0%, 25%, 50%, 75%, 100%) for each feature;

[0103] 2. Map each value of all features to a uniform quantile (e.g., the 25th quantile of a feature corresponds to the 25th quantile of the target distribution).

[0104] In addition, data processing also includes the detection and handling of outliers.

[0105] 1. Outlier identification based on statistics.

[0106] This method follows the 3σ principle: if a data point deviates from the mean by more than 3 times the standard deviation, it is considered an outlier.

[0107] Specific handling methods: Replace with the mean or median, or mark as missing values. For example, box plot method: Calculate the quartiles Q1, Q3 and the interquartile range IQR = Q3 - Q1; outlier boundaries: Q1 below -1.5IQR or Q3 above +1.5IQR.

[0108] Example: Vehicle speed data Q1 = 40km / h, Q3 = 60km / h, IQR = 20km / h → Outlier values ​​<10km / h or >90km / h (non-stationary state).

[0109] In addition, outlier detection can be based on the model. The model is an isolated forest, which uses binary trees to segment the data. Outliers are isolated earlier due to their unique features (shorter path); it is suitable for high-dimensional data (such as multi-sensor fusion features).

[0110] The model can be an autoencoder. The model is trained to reconstruct normal data; data that cannot be accurately reconstructed is considered anomaly. Example: Tire pressure data reconstructed by an autoencoder has an error > 0.05 bar, which is considered an outlier.

[0111] 2. Fill in missing values ​​after identifying outliers.

[0112] Imputation strategies include handling univariate missing values.

[0113] When the parameter is a numerical parameter, it should be filled in as follows:

[0114] Mean fill: Suitable for data without a clear time trend (such as temperature);

[0115] Forward padding: Fill with the most recent valid value (e.g., tire pressure loss for a short period of time);

[0116] Linear interpolation: Calculated linearly based on the values ​​at different time points (e.g., when vehicle speed is missing, it is interpolated from the speed at different time points).

[0117] When the parameter is a categorical parameter, the imputation strategy is as follows:

[0118] Mode filling: Fill in with the category that appears most frequently (e.g., fill in "sunny" when the weather type is missing);

[0119] Added a "Missing" category: to avoid introducing biases in the filling (such as marking missing traffic signs as "unknown").

[0120] When the parameter is a time series parameter, missing values ​​are handled in the following ways:

[0121] Sliding window interpolation: For consecutive missing values ​​(such as vibration sensor failure for 5 seconds), fill them with the mean / median of the data before and after 10 seconds;

[0122] Time series model prediction imputation: Use ARIMA (autoregressive moving average) model or LSTM (Long Short-Term Memory) model to predict missing values ​​based on historical sequences (e.g., when tire temperature is missing, predict based on historical temperature trends).

[0123] Data processing can be implemented through pipelines. This can be achieved through modular processing flows, where the letters below represent modules. The modular processing flow is as follows:

[0124] A [Driving environment parameters, driving status parameters, and historical feature parameters] --> B1 [Data type identification]; A --> B2 [Outlier detection]; A --> B3 [Missing value labeling]; B1 --> C1 [Numerical parameter normalization]; B1 --> C2 [Text parameter encoding]; B2 --> C3 [Outlier replacement]; B3 --> C4 [Missing value imputation]; C1 --> D [Feature merging]; C2 --> D; C3 --> D; C4 --> D; D --> E [Standardized feature vector].

[0125] The implementation methods include code implementation, that is, by writing code and using the scripts encapsulated in the written code to perform the data processing of the above parameters.

[0126] After processing the data for the above parameters, the data processing effect can be verified and optimized. This specifically includes:

[0127] 1. Evaluation of normalization effect. This evaluation includes visual checks: plotting feature histograms before and after normalization to ensure reasonable distribution (e.g., the value range is strictly within [0,1] after Min-Max normalization); plotting feature scatter plots after PCA (principal component analysis) dimensionality reduction to observe whether the separability of different categories of data has improved.

[0128] The assessment includes statistical indicators: calculating the mean (the mean should be around 0) and standard deviation (the standard deviation should be around 1) of the normalized features; and calculating the correlation coefficient between features to ensure that preprocessing does not damage the original data relationships.

[0129] 2. Dynamic update mechanism. Normalized parameters are recalculated every 1000 frames of data (or daily) to adapt to sensor drift or environmental changes; when the proportion of outliers in the new data is greater than 10%, parameter retraining is triggered to avoid distortion of old parameters.

[0130] The vehicle processes driving environment parameters, driving status parameters, and historical characteristic parameters using the methods described above. The processed parameters are defined as target data, which includes parameter names and corresponding values ​​for each type of parameter. This method converts driving environment parameters, driving status parameters, and historical characteristic parameters into values ​​of the same dimension, facilitating subsequent predictions of following distance and braking distance.

[0131] It should be noted that vehicles can distribute data processing modules across various subsystems. For example, small computing units can be integrated into tire monitoring systems and weather monitoring modules to perform preliminary data processing and feature extraction. Each subsystem transmits the processed data to the central decision-making unit via a high-speed in-vehicle network, reducing the pressure of centralized data transmission and improving system response speed and reliability.

[0132] Step S103: Predict the target distance of the current vehicle based on the target data. The target distance includes at least one of following distance and braking distance.

[0133] After obtaining the target data, the vehicle makes a prediction based on the target data to obtain the target distance, which includes at least one of the following distance and braking distance.

[0134] A prediction model can be set up in the vehicle. By inputting the target data into the prediction model, the target distance output by the prediction model can be obtained. The prediction model can be trained based on training samples, and the prediction model can be an LSTM model.

[0135] The training samples can be processed through the aforementioned data processing steps, converting data of different types and units into a standard format suitable for machine learning model input. This not only eliminates the influence of units of measurement but also improves data quality through outlier handling and missing value imputation, laying the foundation for feature fusion and model training. This process ensures that sensor data (such as temperature in °C and vehicle speed in km / h), image data (pixel values), and text data (weather type) are comparable in the same mathematical space, significantly improving the model's convergence speed and prediction accuracy. The convolutional neural network layer in the prediction model slides convolutional kernels across the data to extract spatial features from data such as road surface texture and weather cloud maps, for example, identifying whether the road surface is dry asphalt or wet cement. The LSTM layer in the prediction model processes time-series features of data on vehicle vibration and tire parameter changes over time to determine whether the vehicle's driving state is stable, thereby predicting following distance and braking distance. The stability of a vehicle can be determined based on data from the following sensors: Accelerometer (x / y / z three-axis, sampling frequency 100Hz): captures vehicle vibration, pitch, and roll; Tire pressure monitoring system: real-time tire pressure and temperature data (sampling frequency 10Hz); Wheel speed sensor: wheel speed difference (sampling frequency 50Hz); Steering wheel angle sensor: steering angle and angular velocity (sampling frequency 20Hz).

[0136] In addition, a fusion method based on fuzzy logic can be used. A fuzzy rule base can be established for data such as road conditions, weather, and tires. For example, parameters such as road slippage, rainfall, and tire wear can be divided into different fuzzy sets (e.g., "dry," "wet," "severe water accumulation," "light wear," "heavy wear," etc.). The range of safe following distance and braking distance can be derived through fuzzy inference rules, and then weighted and adjusted using expert experience to obtain the target data.

[0137] Additionally, edge computing devices can be deployed on the vehicle to process data with high real-time requirements (such as data related to emergency braking decisions), enabling rapid initial judgments and responses. Simultaneously, large amounts of historical data and complex model training tasks can be uploaded to the cloud, leveraging the powerful computing capabilities of cloud computing to optimize the algorithm model. The optimized results can then be distributed back to the vehicle, enabling continuous vehicle evolution.

[0138] Step S104: Control the current vehicle's movement based on the target distance.

[0139] Once the target distance is obtained, the vehicle can drive according to the target distance, that is, ensure that the distance between the vehicle and the vehicle in front is greater than or equal to the following distance included in the target distance, and ensure that the current braking distance of the vehicle is greater than or equal to the braking distance included in the target distance.

[0140] In this embodiment, various parameters of different types and accuracies, such as road conditions, weather, temperature, and tire condition, are effectively fused and processed. These parameters come from a wide range of sources, including vehicle sensors, external weather stations, and road monitoring equipment. Integrating these parameters into a unified system and extracting valuable information is the foundation for accurately judging following distance and braking distance.

[0141] In this embodiment, the current vehicle's driving environment parameters, driving state parameters, and historical feature parameters are acquired. These parameters are then processed to obtain target data. Based on this target data, the following distance and braking distance of the current vehicle are predicted to obtain the target distance. Finally, the vehicle's movement is controlled based on the target distance. In this application, the following distance and braking distance of the vehicle are predicted using driving environment parameters such as speed parameters, lighting parameters, weather parameters, and road parameters of other vehicles around the vehicle; driving state parameters such as vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters of the vehicle; and historical following distance and braking frequency and lighting historical feature parameters of the vehicle. This comprehensive approach, using multiple parameters, improves the accuracy of determining the following distance and braking distance.

[0142] Figure 2 A flowchart of a vehicle control method provided in this application embodiment Figure 2 ,based on Figure 1 In the embodiment shown, step S103 includes:

[0143] Step S201: Extract spatial features and temporal features from the target data, and determine the initial distance based on the spatial features and temporal features.

[0144] In this embodiment, after obtaining the target data, the vehicle extracts spatial features and temporal features based on the target data.

[0145] For example, target data can be processed using driving environment parameters, driving state parameters, and historical feature parameters. The resulting fused feature vector can then be slided across the feature vector using a neural network convolution kernel to extract spatial features from data such as road surface texture and weather cloud images. This allows for identification of whether the road surface is dry asphalt or wet concrete. Furthermore, temporal features can be extracted from data such as vehicle vibration and tire parameters that change over time, using LSTM for this purpose.

[0146] After obtaining temporal and spatial features, an initial distance is determined based on these features. For example, the temporal and spatial features are merged in the model's feature fusion layer. The model's neural network output layer then calculates the following distance and braking distance of the vehicle in its current state based on the merged features, using these as the initial distance. This model can be a trained neural network model used to predict vehicle braking and following distances.

[0147] Step S202: Determine the correction parameters based on the target data, and correct the initial distance based on the correction parameters to obtain the target distance of the current vehicle.

[0148] The vehicle can determine correction parameters based on the target data, and then correct the initial distance using these parameters to obtain the target distance for the current vehicle.

[0149] In one example, when the initial distance includes the base distance, the correction parameters include at least one of environmental correction parameters and driving condition correction parameters. The environmental correction parameters are determined by weather or road conditions. For example, in rainy or icy conditions, the environmental correction parameter k_env = 1.5–2.0 (increasing the initial base distance by 50%–100%); for curves or slopes, the environmental correction parameter k_env = 1.2–1.5. The specific value of the environmental correction parameter k_env is dynamically adjusted based on curvature and slope, i.e., determined based on weather or road condition parameters in the target data. The driving condition correction parameters are determined by parameters such as braking performance and load in the target data. If the load exceeds the limit, the driving condition correction parameter k_vehicle = 1.1–1.3; if the braking system is abnormal, the driving condition correction parameter k_vehicle = 1.2–1.5.

[0150] In another example, if the initial distance includes a base braking distance, the correction parameters include at least one of a friction coefficient attenuation correction parameter, a driver reaction time correction parameter, and an obstacle correction parameter. For example, the friction coefficient attenuation correction parameter refers to the decrease in friction coefficient due to tire wear, and is calculated using the road friction coefficient within the road parameters of the target data. The driver reaction time correction parameter can be obtained by learning the reaction time distribution from the vehicle's historical braking data in the target data. The obstacle correction parameter refers to the additional braking demand caused by sudden obstacles, such as a vehicle braking suddenly ahead, triggering a compensation value after visual recognition. This parameter can be obtained by correcting a preset obstacle correction value using the braking frequency in the target data; if the braking frequency is greater than a threshold, the preset obstacle correction value is increased. This part of the correction parameter can be the compensation part of the neural network in the model. That is, after the model determines the initial distance, it determines the correction parameters from the target data based on the compensation mechanism in the neural network, and then corrects the initial distance using these parameters. The correction parameters can be learned by the model.

[0151] Specifically, by inputting data such as tire pressure and brake fluid temperature, the network predicts the actual braking deceleration through a neural network, correcting the 'a' value in the physical model. The 'a' value is the correction parameter. In addition, there is an emergency weighting: when the brake lights of the vehicle ahead are detected (visual input), the network outputs an additional compensation distance (e.g., 5-10 meters), which means that the distance corrected by the correction parameter is compensated again to ensure that the braking distance covers sudden scenarios.

[0152] In this embodiment, the vehicle extracts spatial and temporal features from the target data, determines an initial distance based on the spatial and temporal features, and then corrects the initial distance based on the correction parameters determined by the target data to obtain an accurate target distance.

[0153] Figure 3 A flowchart of a vehicle control method provided in this application embodiment Figure 3 .based on Figure 1 In the embodiment shown, step S103 includes:

[0154] Step S301: Input the target data into the prediction model to obtain the target distance of the current vehicle output by the prediction model. The target distance includes the lower limit of the following distance and the upper limit of the braking distance output by the prediction model.

[0155] In this embodiment, a prediction model is installed in the vehicle. The vehicle inputs target data into the prediction model to obtain the target distance of the current vehicle. The target distance includes the lower limit of the following distance and the upper limit of the braking distance output by the prediction model. It should be noted that if the following distance between vehicles is too small, the driving safety of the vehicle cannot be guaranteed. Therefore, the model needs to predict the lower limit of the following distance to limit the following distance from being too small. On the other hand, if the braking distance of the vehicle is too long, it indicates that the vehicle speed is too high or the vehicle is prone to skidding on the road surface. Therefore, the model predicts the upper limit of the braking distance to limit the braking distance of the vehicle from being too long, thereby limiting the vehicle speed from being too high.

[0156] Specifically, the prediction model uses a two-layer fully connected network (FC) + activation function, with the following structure: input layer (fused feature vector, dimension N) → fully connected layer 1 (number of neurons N / 2, activation function ReLU) → fully connected layer 2 (number of neurons 2, linear activation) → output layer (following distance d_follow, braking distance d_brake).

[0157] Activation function selection in the prediction model: The output layer uses linear activation (no activation function) to ensure that the output value can be positive or negative (in practical applications, it is restricted to non-negative values ​​through post-processing).

[0158] The following logic for calculating following distance in the prediction model is as follows:

[0159] Based on the distance model: the output layer calculation follows the principle of "dynamic safety margin," integrating neural network prediction with traditional safety models (such as the "three-second rule"), as shown in the following formula: d_follow=f_mlp([X])×k_env×k_vehicle; where f_mlp([X]) is the base distance (in meters) predicted by the neural network based on the input features, and k_env is the environmental correction coefficient (determined by weather and road conditions): for example, rainy / icy roads: k_env=1.5~2.0 (distance increases by 50%-100%), curves / slopes: k_env=1.2~1.5 (dynamically adjusted according to curvature / slope). k_vehicle is the vehicle state correction coefficient (determined by braking performance and load): overload: k_vehicle=1.1~1.3; abnormal braking system: k_vehicle=1.2~1.5.

[0160] Dynamic mapping of neural networks in the predictive model: Data-driven safety threshold learning. The network is trained using a large amount of historical driving data (normal following and emergency braking scenarios) to learn the safe distance threshold under different scenarios. For example, when the vehicle speed is 60 km / h and the road surface is dry, the network outputs a base distance of 20 meters. With k_env = 1.0 (good environment) and k_vehicle = 1.0 (standard vehicle condition), the following distance is calculated to be 20 meters. When the vehicle speed is the same but the road surface is slippery, k_env = 1.8, and the final following distance becomes 36 meters.

[0161] The principle of braking distance calculation in predictive models:

[0162] 1. Fusion of physical model and neural network: The calculation of braking distance d_brake combines Newton's second law with neural network prediction, as shown in the following formula: d_brake=(v 2 ) / (2×a)+f_mlp([X]). The physical model part is: (v 2 ) / (2×a) represents the theoretical braking distance, where v is the vehicle speed (m / s); a is the braking acceleration (m / s²). 2 The coefficient of friction μ and the slope θ are determined by: a = ugcosθ + gsinθ (g is the acceleration due to gravity, and u is predicted by a sensor or neural network).

[0163] 2. Neural network compensation part: f_mlp([X]) is used to correct factors not considered by the physical model: friction coefficient decay caused by tire wear (the neural network predicts the actual μ value through sensor data); driver reaction time fluctuations (the reaction time distribution is learned through historical braking data); additional braking demand caused by sudden obstacles (such as sudden braking of the vehicle in front, the compensation value is triggered after visual recognition).

[0164] 3. Neural network compensation mechanism: Braking performance prediction: Input data such as tire pressure and brake fluid temperature, and the network predicts the actual braking deceleration to correct the 'a' value in the physical model; Emergency weighting: When the brake lights of the vehicle in front are detected (visual input), the network outputs an additional compensation distance (e.g., 5-10 meters) to ensure that the braking distance covers sudden scenarios.

[0165] The predictive model performs post-processing and safety constraints on the output data. Specifically, it performs threshold pruning, such as setting a lower limit for following distance: d_follow ≥ 5 meters (to avoid collisions when stationary); and setting an upper limit for braking distance: d_brake ≤ 1.5 × theoretical maximum safe distance (determined by regulations or vehicle performance). By setting the lower and upper limits for following distance and braking distance, the distance between the vehicle and the vehicle in front can be limited to a reasonable range, such as a following distance greater than 5 meters and a braking distance less than or equal to 1.5 × theoretical maximum safe distance.

[0166] Furthermore, the prediction model incorporates uncertainty assessment and redundancy design. Specifically, it outputs prediction confidence (such as the probability distribution of following distance) through additional nodes in the output layer. When the confidence level is <70%, a backup safety model (such as rule-based distance calculation) is triggered.

[0167] Furthermore, the prediction model employs a multi-network voting mechanism: simultaneously running 2-3 independently trained neural networks, and taking the average of their outputs as the final result, thus reducing the risk of misjudgment by a single model. Understandably, the prediction model comprises multiple parallel neural networks, and the target distance is the average of the output distances predicted by each neural network based on the target data. For example, by simultaneously or sequentially inputting the target value into each neural network, the average of the output values ​​of each neural network is the target distance. The structures of the neural networks can be identical or different, and each neural network can be trained using training samples from the same dataset or from different datasets.

[0168] It should be noted that the prediction model is trained based on various training samples. The training samples include sample data obtained by processing the driving state parameters of other vehicles, driving state parameters and historical feature parameters. The data processing process is the same as the data processing process in the above embodiment, and will not be repeated here. In addition, the sample data also includes the following distance and braking distance of other vehicles, and the sample data is labeled as driving safe or driving unsafe.

[0169] Furthermore, the construction of training samples can be determined in the following way.

[0170] Training samples were constructed using normal driving data: measured values ​​of following distance and braking distance were collected under different vehicle speeds, weather conditions, and road conditions.

[0171] Emergency scenario data is used to build training samples: simulate scenarios such as emergency braking and obstacle avoidance, and record the actual braking distance;

[0172] Training samples are constructed from edge case data: driving data in extreme weather (heavy rain, heavy snow) and special roads (long downhill, continuous curves).

[0173] The loss function combination of the prediction model is: the mean squared error loss odd function, which measures the deviation between the predicted distance and the actual value. Specifically, the mean squared error loss odd function \(L_{MSE}=\frac{1}{N}\sum(d_{pred}-d_{gt})\) 2 , where \(d_{pred}\) is the predicted distance, \(d_{gt}\) is the actual value, and \(N\) is the number of training samples;

[0174] The quantile loss function is used to optimize the prediction accuracy at different probability quantiles (e.g., the 90% quantile ensures the safety margin). Specifically, the quantile loss function \(L_{Quantile}=\sum(\tau - I(d_{pred}<d_{gt}))(d_{gt}-d_{pred})\), where \(\tau\) is the quantile. For example, \(\tau = 0.9\) means that in 90% of the cases, the predicted distance \(\geq\) the actual distance. \(d_{pred}\) is the predicted distance, and \(d_{gt}\) is the actual value. \(I\) is a conditional function. When the condition is met, \(I = 1\); when the condition is not met, \(I = 0\), and the condition can be set as needed. By training the model in the above way, the neural network output layer in the trained prediction model realizes an accurate mapping from multi-dimensional features to the safe distance, which not only meets the constraints of physical laws but also adapts to the real-time changes in complex dynamic environments, providing a reliable distance control basis for the autonomous driving decision-making of the vehicle.

[0175] In this embodiment, by inputting the target data into the prediction model, the following distance and braking distance of the vehicle can be quickly determined through the prediction model.

[0176] Figure 4 is the flowchart of a vehicle control method provided by an embodiment of the present application Figure 4 . Based on Figures 1 to 3 any of the embodiments shown in

[0177] Step S401, obtain the real-time weather data, road construction data, real-time road condition data, and map data corresponding to the area where the current vehicle is located.

[0178] In this embodiment, the vehicle corrects the predicted distance by predicting its own driving environment in the future time period, that is, predicts the driving environment of the vehicle within 1 to 3 minutes in the future, so as to adjust the following distance and braking distance in advance and ensure the safe driving of the vehicle.

[0179] For this reason, the device obtains the real-time weather data, road construction data, real-time road condition data, and map data corresponding to the area where the current vehicle is located.

[0180] Step S402, predict the driving environment of the current vehicle in the future time period according to the real-time weather data, road construction data, real-time road condition data, and map data. The driving environment includes the weather environment and the road condition environment.

[0181] After obtaining real-time weather data, road construction data, real-time traffic data, and map data, the driving environment of the vehicle in the future time period is predicted based on the real-time weather data, road construction data, real-time traffic data, and map data.

[0182] For example, the device is equipped with an environmental simulation model. By inputting real-time weather data, road construction data, real-time traffic data, and map data into the environmental simulation model, the driving environment of the current vehicle in a future time period can be obtained from the output of the environmental simulation model. The future time period is a relatively short period of time, for example, 1 to 3 minutes in the future. The environmental simulation model is described below.

[0183] The environmental simulation model is built upon a fundamental kinematic model. Specifically, predicting the vehicle's driving environment is essentially predicting its trajectory, which is based on Newton's principles of kinematics, establishing equations of motion in three-dimensional space. In a Cartesian coordinate system, the vehicle's position coordinates are represented as (x(t), y(t), z(t)), where the z-axis corresponds to altitude. The fundamental kinematic equations are as follows:

[0184] 1. Relationship between displacement and velocity:

[0185] x(t) = x0 + ∫0 t v(t)·cos(θ(t))·dt;

[0186] y(t)=y0+∫0 t v(t)·sin(θ(t))·dt;

[0187] z(t) = z0 + ∫0 t v(t)·sin(α(t))·dt;

[0188] Where (x0, y0, z0) are the initial position coordinates of the vehicle, v(t) is the vehicle velocity at time t, θ(t) is the vehicle heading angle at time t (the angle between the vehicle and the x-axis in the horizontal plane); α(t) is the vehicle pitch angle at time t (the angle between the vehicle and the horizontal plane, positive for uphill).

[0189] 2. Relationship between velocity and acceleration: v(t)=v0+∫0 t a(t)·dt. Where v0 is the initial velocity of the vehicle; a(t) is the acceleration of the vehicle at time t, which is affected by factors such as road slope and centrifugal force of curves.

[0190] The environmental simulation model also integrates road geometry features, meaning it includes a curve curvature processing model, a slope impact calculation model, and a three-dimensional road model.

[0191] Specifically, the curve curvature processing model is used to treat road curves as circular arc segments, with the radius of curvature R of the arc segment obtained from a high-precision map. The centripetal acceleration of a vehicle traveling on a curve is: a_c = v 2 / R, where v is the vehicle speed; this acceleration causes the vehicle to deviate laterally, requiring a correction to the equation of motion: θ(t)=θ0+∫0 t (v(t) / R)·dt; where θ0 is the initial heading angle when entering the curve, and v(t) is the vehicle speed at time t.

[0192] The input parameters for the slope impact calculation model include the road surface slope. The impact of the road surface slope i (positive for uphill and negative for downhill) on vehicle motion is reflected in the gravity component: a_g=g·i·sin(α), where g is the gravitational acceleration (9.8m / s²). 2 ); Driving resistance: F_r=m·g·(f·cos(α)+i·sin(α)), where m is the vehicle mass and f is the rolling resistance coefficient; Correction equation for acceleration due to gradient: a(t)=a_d-a_g-a_r; where a_d is the acceleration generated by the driving force and a_r is the resistance acceleration.

[0193] 3D road model parameterization: road information in high-precision maps is converted into parametric equations with arc length s as parameters: x=x(s),y=y(s),z=z(s),R=R(s),i=i(s); where R(s) is the radius of curvature at arc length s, and i(s) is the slope at arc length s.

[0194] The environmental simulation model needs to predict and iterate trajectories, and the process is as follows:

[0195] 1. Discretization time step.

[0196] Numerical integration is performed using the variable step size Euler method, with the time step Δt dynamically adjusted according to the driving environment: Δt=Δt0·(1+k·σ), where Δt0 is the baseline step size (0.1s), k is the environmental complexity coefficient (0.5-2.0), and σ is the road condition change rate (calculated from the curvature and slope change rates).

[0197] 2. Iterate through the above calculation steps.

[0198] 3. During the calculation process, it is necessary to perform detailed calculations of key parameters.

[0199] 3.1 Calculation of velocity correction term: a_c = v 2(t) / R(t), a_c is the centripetal acceleration; a_g=g·i(t)·sin(α(t)); a_g is the gravitational acceleration component; a_r=μ·g·cos(α(t)), a_r is the rolling resistance acceleration, μ is the rolling resistance coefficient; a(t+Δt)=a_d-a_c-a_g-a_r.

[0200] 3.2 Heading Angle Update: R(t)>1000m, applicable to approximately straight-line driving, θ(t+Δt)=θ(t); or curve driving: θ(t+Δt)=θ(t)+(v(t) / R(t))·Δt.

[0201] 3.3. Position coordinate update: x(t+Δt)=x(t)+v(t)·cos(θ(t))·Δt; y(t+Δt)=y(t)+v(t)·sin(θ(t))·Δt; z(t+Δt)=z(t)+v(t)·sin(α(t))·Δt.

[0202] Furthermore, the environmental simulation model can also be compensated based on dynamic characteristics. Specifically:

[0203] Tire lateral slip characteristic correction. The tire lateral slip characteristic is described by the following formula: F_y=D·sin(C·atan(B·α-E·(B·α-atan(B·α)))), where F_y is the tire lateral force, α is the slip angle, and B, C, D, and E are tire characteristic parameters.

[0204] Correction of sideslip angle to heading angle: θ_correction=arctan(F_y / (m·v(t)));θ(t+Δt)=θ(t+Δt)+θ_correction.

[0205] Vehicle center of gravity offset compensation: Considering the vehicle center of gravity offset e (unit: m), the lateral offset of the trajectory is: Δy=e·sin(θ(t)); x(t+Δt)=x(t+Δt)-Δy·sin(θ(t)); y(t+Δt)=y(t+Δt)+Δy·cos(θ(t)).

[0206] In addition, the environmental simulation model will perform real-time corrections and error control. Specifically, this includes Kalman filter optimization: constructing a state-space model for trajectory optimization: state vector: x = [x, y, z, v, θ, α]^T; observation vector: z = [x_sensor, y_sensor, v_sensor]^T.

[0207] State transition equation: x(k+1)=F·x(k)+G·u(k)+w(k);

[0208] The observation equation is: z(k)=H·x(k)+v(k); where F is the state transition matrix, G is the control matrix, H is the observation matrix, and w(k) and v(k) are the process noise and observation noise, respectively.

[0209] The error control algorithm is as follows: ε = ||x_predicted - x_measured||, if ε > ε_threshold, Δt = Δt·(1-k·ε / ε_threshold), where ε is the error between the predicted and measured positions, ε_threshold is the threshold, and k is the correction coefficient. In essence, the environmental simulation model first predicts the initial driving trajectory, then compensates for the initial trajectory based on dynamic characteristics to obtain an intermediate driving trajectory, and finally corrects the intermediate trajectory using real-time correction and error control methods to serve as the driving environment.

[0210] It's worth noting that vehicles can also build predictive models based on knowledge graphs. Historical weather data, road construction information, traffic accident records, and other data are integrated with real-time collected data to construct a knowledge graph. Graph neural network algorithms are then used to mine the relationships between these data points, such as the probability of accidents on a particular road segment under specific weather and time conditions, thereby predicting future changes in the driving environment and providing a basis for vehicle decision-making. In essence, by inputting target data into a model built from a knowledge graph, the vehicle's driving environment for future time periods can be obtained.

[0211] Step S403: Predict the distance to be determined for the current vehicle based on the target data, and correct the distance to be determined based on the driving environment to obtain the target distance.

[0212] After obtaining the driving environment data, the distance to the current vehicle is predicted based on the target data as the distance to be determined. The distance to be determined can be the following distance and braking distance predicted by the prediction model from the target data. The vehicle then corrects the distance to be determined based on the driving environment to obtain the target distance.

[0213] For example, if the driving environment is predicted to be the slipperiness of the road surface due to rain in the next 5 kilometers, the distance to be determined is increased to obtain the target distance; if the driving environment is predicted to be the peak congestion of the construction section in the next hour, the speed is reduced to increase the distance to be determined to obtain the target distance, and the traffic light timing is dynamically adjusted or detour prompts are issued.

[0214] The following explains the predictions of the environmental drill model.

[0215] Mountain road curve scenario. Input parameters: Initial position: (0,0,100)m, Initial speed: 20m / s, Curve radius: R=100, Slope: i=5°. Tire characteristic parameters: B=10, C=1.5, D=2000, E=0.2.

[0216] Calculation process: Centripetal acceleration: a_c = 20 2 / 100=4m / s 2 The gravitational acceleration component: a_g = 9.8 * 0.05 = 0.49 m / s² 2 Rolling resistance acceleration: a_r = 0.01·9.8 = 0.098 m / s² 2 Corrected acceleration: a = a_d - 4 - 0.49 - 0.098 (assuming a_d = 0 is the deceleration state); Velocity after 1 second: v = 20 - (4 + 0.49 + 0.098)·1 = 15.412 m / s.

[0217] Change in heading angle: θ = 0 + (20 / 100)·1 = 0.2 rad (approximately 11.5°).

[0218] Position update: x = 0 + 20·cos(0)·1 = 20m; y = 0 + 20·sin(0)·1 + e·cos(0) = 0 + e (e is the centroid offset); z = 100 + 20·sin(arctan(0.05))·1 ≈ 100.995m.

[0219] Urban road lane change scenario. Input parameters: Initial position: (0,0,0)m; Initial speed: 15m / s; Lane change curvature radius: R=50m (assuming the lane change trajectory is an arc); Slope: i=0; Centroid offset: e=0.5m. Calculation process: Side slip angle correction: when α=5°, F_y=2000·sin(1.5·atan(10·0.087-0.2·(10·0.087-atan(10·0.087))))≈1500N; Heading angle correction: θ_correction=arctan(1500 / (1500·15))≈0.0067rad; Centroid offset compensation: Δy=0.5·sin(θ), assuming θ=0.1rad, Δy=0.05m. Position after lane change: y = 15·sin(0.1)·3 + 0.05·3 ≈ 4.575m (3-second lane change process).

[0220] The algorithm implementation architecture of the environment simulation model is as follows: hierarchical design of computing modules; parallel computing optimization scheme.

[0221] GPU parallel computing is used to accelerate trajectory prediction, and the computational task is decomposed into:

[0222] Thread block 1: Straight line trajectory calculation; Thread block 2: Curved trajectory calculation; Thread block 3: Slope impact calculation; Thread block 4: Dynamics compensation calculation; Thread block 5: Filtering optimization calculation. Each thread processes calculations for one time step, and data exchange is accelerated through shared memory, which can improve computational efficiency by 10-20 times compared to CPU serial calculation.

[0223] In addition, the prediction principle of the environmental drill model is explained as follows:

[0224] Data input layer: multi-source heterogeneous data acquisition and preprocessing.

[0225] Data types: Real-time sensing data: weather sensors (rainfall, wind speed, temperature), road surface sensors (humidity, friction coefficient), road condition images collected by vehicle cameras / radar (such as construction signs, obstacles). Historical data: historical weather records for the same region and season over the past 3-5 years (such as rainfall distribution, extreme weather periods), and historical records of road construction (construction cycle, frequency).

[0226] Supporting data: High-precision map data (road slope, curvature), traffic flow data (real-time traffic volume, congestion index).

[0227] The preprocessing workflow includes normalization: data at different scales (e.g., rainfall in mm and temperature in °C) are normalized to a unified range using Min-Max normalization or Z-Score. Feature extraction: visual features such as construction signs and waterlogged areas are extracted from image data using CNNs (e.g., ResNet); trend terms and periodic components are extracted from time-series data (e.g., rainfall changes over time). Missing value handling: anomalous sensor data is supplemented using sliding window interpolation or forward imputation. Core model layer: prediction logic of RNN and its variants: taking LSTM (Long Short-Term Memory) network as an example, its prediction principle is as follows:

[0228] Neuron structure: LSTM controls the flow of information through "forget gate", "input gate" and "output gate", which solves the problem of long-sequence gradient vanishing in traditional RNNs. It is suitable for processing time-series data with long-term dependencies, such as weather and road conditions.

[0229] Forget gate f_t = σ(W_f·[h_{t-1},x_t]+b_f);

[0230] Input gate i_t = σ(W_i·[h_{t-1},x_t]+b_i);

[0231] Cell state update c_t = f_t·c_{t-1} + i_t·tanh(W_c·[h_{t-1},x_t] + b_c);

[0232] Output gate o_t = σ(W_o·[h_{t-1},x_t]+b_o);

[0233] The hidden state h_t = o_t·tanh(c_t), where σ is the sigmoid activation function, W is the weight matrix, b is the bias term, h is the hidden state, c is the cell state, and x is the input data.

[0234] Temporal modeling process. Sequence input: Preprocessed time-series data (e.g., rainfall over the past 24 hours, road surface moisture every 10 minutes) are input into the LSTM network at time steps (e.g., 5 minutes / step). State propagation: The hidden state h_t at each time step contains the temporal characteristics of historical data (e.g., the trend of rainfall intensity changes), and the cell state c_t remembers long-term dependencies (e.g., seasonal rainfall patterns). Multi-step prediction: Using "Teacher Forcing" or "Autoregressive" modes, weather / road condition changes at multiple future time steps are predicted based on the current hidden state (e.g., the movement trajectory of rainfall areas in the next hour, the probability of congestion on construction sections in the next 30 minutes).

[0235] The environmental simulation model in this embodiment achieves accurate prediction of vehicle trajectories by integrating kinematic models, road geometry features, and vehicle dynamics characteristics. This solution has been verified through real-vehicle testing; when navigating a curve with a radius of curvature of 100m at 60km / h, the trajectory prediction error is less than 0.5m, meeting the pre-simulation requirements of intelligent driving systems. It also supports flexible configuration based on different road scenarios and vehicle parameters, providing a solid foundation for trajectory calculation in dynamic environment simulations. The machine learning prediction module combines real-time weather and road surface data, using algorithms such as recurrent neural networks (RNNs) to predict weather changes (such as the movement of rainfall areas) and road condition changes (such as road construction ahead). The prediction results are displayed on the vehicle's windshield as virtual images overlaid on the real scene through the vehicle's built-in AR-HUD (Augmented Reality-Head-Up Display system), while the simulation results are sent to the intelligent decision-making unit to help it make more accurate decisions.

[0236] Figure 5 A flowchart of a vehicle control method provided in this application embodiment Figure 5 .based on Figure 4 In the embodiment shown, step S402 includes:

[0237] Step S501: If the current driving environment to be predicted for the vehicle is the weather environment, the area where the current vehicle is located is divided into multiple sub-regions according to the map data.

[0238] In this embodiment, when the driving environment to be predicted for the vehicle is the weather environment, the area where the vehicle is currently located is divided into multiple sub-regions based on map data.

[0239] Step S502: Determine the rainfall parameters and wind direction parameters for each sub-region based on real-time weather data.

[0240] The vehicle determines the rainfall and wind direction parameters for each sub-region based on real-time weather data. Rainfall and wind direction can be determined using an environmental simulation model.

[0241] Step S503: Determine the weather conditions for the current vehicle in the future time period based on rainfall and wind direction parameters.

[0242] After obtaining rainfall and wind direction parameters, the weather conditions for the vehicle in the future can be obtained based on these parameters. For example, the environmental simulation model extracts spatial features by dividing the geographical area into grids (e.g., 1km × 1km), with the rainfall and wind direction of each grid serving as spatial dimension inputs. The environmental simulation model integrates spatiotemporal data; it can be a spatiotemporal LSTM or a convolutional LSTM. The convolutional layers of the environmental simulation model extract the spatial distribution features of the rainfall area (e.g., the shape of the rain band), and the LSTM layers capture its temporal movement patterns.

[0243] It should be noted that in this embodiment, the environmental simulation model uses mean square error to measure the deviation between the predicted and actual rainfall values, and combines binary cross-entropy to assess the accuracy of the probability of rainfall occurring.

[0244] In addition, the model receives new sensor data at set intervals and fine-tunes its parameters through online learning to adapt to sudden weather events (such as severe convective rainfall). The set interval is, for example, 10 minutes.

[0245] In this embodiment, the vehicle can accurately predict the weather conditions for the future time period by using the rainfall and wind direction of the sub-region where it is currently located.

[0246] Figure 6 A flowchart of a vehicle control method provided in this application embodiment Figure 6 Based on the embodiment shown in 4, step S402 includes:

[0247] Step S601: Given that the current driving environment to be predicted for the vehicle is the road condition environment, predict the probability value of construction at the current vehicle's location in the future time period based on road construction data and map data.

[0248] In this embodiment, when the driving environment that the vehicle needs to predict is the road condition environment, the probability value of construction at the vehicle's location in the future time period is predicted based on road construction data and map data.

[0249] Step S602: Based on the construction probability value and real-time road condition data, predict the road condition environment in which the current vehicle will be located in the future time period.

[0250] The vehicle predicts the road conditions it will encounter in the future based on construction probability values ​​and real-time road condition data.

[0251] For example, the device can use an environmental simulation model to predict the path environment. Specifically, it first performs multi-source data fusion: text data: crawling construction plan texts from traffic department announcements and using NLP technology to extract key information such as construction time and road sections; visual data: recognizing construction signs (such as warning signs and fences) through vehicle-mounted cameras and locating the construction area using object detection algorithms such as YOLO; time-series data: combining historical construction data (such as regular maintenance of a road section every June) with real-time traffic flow data (such as a sudden increase in congestion on a road section).

[0252] Model prediction logic: Construction probability prediction: Taking the construction plan extracted from the text, the visually recognized site signs, and historical construction patterns as input, the model outputs the construction probability of each road segment in the next 24 hours through LSTM. Impact range extrapolation: If construction is predicted on a certain road segment, the model combines high-precision maps (such as the number of lanes and slope) and traffic flow models to predict the duration of congestion caused by construction and suggest detour routes.

[0253] It should be noted that in this embodiment, the model uses the F1 score to evaluate the accuracy of construction area identification and uses the mean absolute error (MAE) to measure the error in congestion duration prediction.

[0254] In addition, the model receives new sensor data at set intervals and fine-tunes its parameters through online learning to adapt to temporary construction scenarios. The set interval is, for example, 10 minutes.

[0255] In this embodiment, the vehicle can accurately predict the road conditions in the future by using road construction data, map data, and real-time traffic data.

[0256] In one embodiment, after determining the target distance for the vehicle, it is determined whether the target distance matches the current driving state of the vehicle. If the current vehicle speed is too high and the actual following distance is too close, resulting in insufficient braking distance, then the target distance does not match the current driving state, and the vehicle determines that an adjustment strategy is needed. For example, if the following distance is too close or the current vehicle speed is too high, resulting in insufficient braking distance, a command is sent to the braking system to increase the pressure of the brake hydraulic system or electric braking device to increase braking force; at the same time, a command is sent to the power system to reduce the engine throttle opening or the transmission gear to reduce the vehicle speed. In addition, the driver is reminded of the current vehicle status and adjustment status through the human-machine interface using sound prompts, image displays, etc. When the target distance matches the current driving state of the vehicle, no adjustment is required, and the vehicle maintains its current driving state.

[0257] In addition, the vehicle displays the target distance. If the driver deems the target distance unreasonable, pressing the feedback button will immediately record all data collected by the sensors, vehicle driving status data, and the aforementioned adjustment strategies. This data is used as training samples and input into the multi-source data fusion algorithm and dynamic environment simulation model. Through machine learning, the algorithm parameters and model structure are adjusted, enabling the system to more accurately determine following distance and braking distance in subsequent operations, continuously improving vehicle performance.

[0258] Compared with the above embodiments, this application breaks through the limitations of traditional single or few-factor judgment, constructs a brand-new algorithm, integrates road conditions, weather, temperature, and tire data, and adds vehicle vibration data and surrounding vehicle dynamic distribution data. It adopts a neural network architecture combining CNN (Convolutional Neural Network) and LSTM to achieve deep fusion of multi-source data. Specifically, it uses high-precision vibration sensors and vehicle networking modules to collect new data, enriching the data dimensions; CNN extracts spatial features, and LSTM processes time-series data; the combination of the two fully explores the complex relationships between data, solving the problem of multi-source data fusion and processing, and providing a foundation for accurate distance calculation.

[0259] Furthermore, this application provides a dynamic environment simulation and adaptive decision-making model. The innovation lies in creating a dynamic environment simulation model based on real-time data, combining physical modeling and machine learning, and utilizing augmented reality (where driving status parameters on the electronic map in the vehicle are displayed using augmented reality technology) and virtual reality visualization technology (such as displaying various 3D virtual vehicles in the current environment and the distances between them) to predict changes in the driving environment 1-3 minutes in advance, and adjust following and braking strategies accordingly. Specifically, this is manifested in predicting vehicle trajectories based on maps and real-time data, and using machine learning models such as RNNs (Recurrent Neural Networks) to predict weather and road condition changes. This advance strategy adjustment addresses the lack of dynamic environment prediction, avoids delays in vehicle response to sudden road conditions, and improves real-time performance and reliability.

[0260] Furthermore, the vehicle in this application features a multi-dimensional perception and intelligent control system. The vehicle's innovations include: the development of a novel multi-functional road condition sensor, an integrated weather monitoring module incorporating multiple sensors, and a tire omnidirectional monitoring system, constructing a multi-dimensional perception system and implementing a dual mechanism of driver feedback and automatic adjustment. Specifically, the new sensor addresses the challenge of accurate data collection in complex environments; the driver feedback mechanism accumulates data to optimize algorithms, while the automatic adjustment mechanism responds to unforeseen circumstances, resolving algorithm optimization and system reliability issues to ensure stable system operation.

[0261] Exemplary device

[0262] Corresponding to the vehicle control method described above, this application also provides a vehicle. Figure 7 This is a schematic diagram of a vehicle module provided in an embodiment of this application. The vehicle provided in this embodiment includes:

[0263] The acquisition module 710 is used to acquire the current vehicle's driving environment parameters, driving status parameters, and historical feature parameters. The driving environment parameters include at least one of the speed parameters, light parameters, weather parameters, and road parameters of other vehicles around the current vehicle. The driving status parameters include at least one of the current vehicle's vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters. The historical feature parameters include the current vehicle's historical following distance and braking frequency.

[0264] The processing module 720 is used to process driving environment parameters, driving status parameters, and historical feature parameters to obtain target data.

[0265] Prediction module 730 is used to predict the target distance of the current vehicle based on target data, wherein the target distance includes at least one of following distance and braking distance;

[0266] The control module 740 is used to control the current vehicle's movement based on the target distance.

[0267] In some implementations, vehicle 700 is also used for:

[0268] Extract spatial and temporal features from the target data, and determine the initial distance based on the spatial and temporal features;

[0269] The correction parameters are determined based on the target data, and the initial distance is corrected based on the correction parameters to obtain the target distance of the current vehicle.

[0270] In some implementations, if the initial distance includes the base following distance, the correction parameters include at least one of the environmental correction parameters and the driving state correction parameters.

[0271] If the initial distance includes the base braking distance, the correction parameters include at least one of the following: friction coefficient decay correction parameters, driver reaction time correction parameters, and obstacle correction parameters.

[0272] In some implementations, vehicle 700 is also used for:

[0273] The target data is input into the prediction model to obtain the target distance of the current vehicle output by the prediction model. The target distance includes the lower limit of the following distance and the upper limit of the braking distance output by the prediction model.

[0274] In some implementations, the prediction model comprises multiple parallel neural networks, and the target distance is the average of the output distances of each neural network.

[0275] In some implementations, vehicle 700 is also used for:

[0276] Obtain real-time weather data, road construction data, real-time traffic data, and corresponding map data for the area where the vehicle is currently located;

[0277] Based on real-time weather data, road construction data, real-time traffic data, and map data, the driving environment of the current vehicle in the future time period is predicted. The driving environment includes weather environment and road condition environment.

[0278] The target distance is obtained by predicting the distance to be determined for the current vehicle based on the target data and correcting the distance to be determined based on the driving environment.

[0279] In some implementations, vehicle 700 is also used for:

[0280] Given that the current driving environment to be predicted for the vehicle is the weather environment, the area where the vehicle is currently located is divided into multiple sub-regions based on map data.

[0281] The rainfall and wind direction parameters for each sub-region are determined based on real-time weather data.

[0282] Based on rainfall and wind direction parameters, determine the weather conditions for the current vehicle in the future time period.

[0283] In some implementations, vehicle 700 is also used for:

[0284] Given that the current driving environment to be predicted for the vehicle is the road condition environment, based on road construction data and map data, predict the probability value of construction at the current vehicle's location in the future time period.

[0285] Based on the construction probability value and real-time traffic data, predict the traffic environment in which the current vehicle will be in the future time period.

[0286] The vehicle provided in this embodiment belongs to the same concept as the vehicle control method provided in the above embodiments of this application. It can execute the vehicle control method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects for executing the vehicle control method. Technical details not described in detail in this embodiment can be found in the specific processing content of the vehicle control method provided in the above embodiments of this application, and will not be repeated here.

[0287] The functions implemented by the various modules in the vehicle can be implemented by the same or different processors, and this application embodiment does not limit this.

[0288] It should be understood that the modules in the above-described vehicle can be implemented by a processor calling software. For example, the system includes a processor connected to a memory containing instructions. The processor calls the instructions stored in the memory to implement any of the above methods or to implement the functions of each module of the device. The processor can be a general-purpose processor, such as a CPU or microprocessor, and the memory can be internal to the device or external to the system. Alternatively, the modules in the system can be implemented as hardware circuits. By designing the hardware circuits, some or all of the module functions can be implemented. The hardware circuit can be understood as one or more processors. For example, in one implementation, the hardware circuit is an ASIC, and by designing the logical relationships between the components within the circuit, some or all of the above module functions are implemented. In another implementation, the hardware circuit can be implemented using a PLD, such as an FPGA, which can include a large number of logic gates. The connection relationships between the logic gates are configured through configuration files, thereby implementing some or all of the above module functions. All modules of the above-described vehicle can be implemented entirely by a processor calling software, entirely by hardware circuits, or partially by a processor calling software with the remaining parts implemented by hardware circuits.

[0289] In this application embodiment, a processor is a circuit with signal processing capabilities. In one implementation, the processor can be a circuit with instruction reading and execution capabilities, such as a CPU, microprocessor, GPU, or DSP. In another implementation, the processor can implement certain functions through the logical relationships of hardware circuits. These logical relationships are fixed or reconfigurable. For example, the processor may be a hardware circuit implemented as an ASIC or PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document and configuring the hardware circuit can be understood as the processor loading instructions to implement the functions of some or all of the above modules. Furthermore, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as an NPU, TPU, or DPU.

[0290] As can be seen, each module in the above vehicle can be one or more processors (or processing circuits) configured to implement the above methods, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0291] Furthermore, the modules in the above-mentioned vehicle can be integrated in whole or in part, or they can be implemented independently. In one implementation, these modules are integrated together and implemented in the form of a System-on-Chip (SoC). The SoC may include at least one processor for implementing any of the above methods or implementing the functions of the modules of the device. The at least one processor may be of different types, such as CPU and FPGA, CPU and artificial intelligence processor, CPU and GPU, etc.

[0292] Exemplary electronic devices

[0293] This application provides another structural schematic diagram of an electronic device, see [link to schematic diagram]. Figure 8 As shown, the electronic device includes a memory 800 and a processor 810; wherein the memory 800 is connected to the processor 810 and is used to store programs; the processor 810 is used to implement the vehicle control method disclosed in any of the above embodiments by running the programs stored in the memory 800.

[0294] Specifically, the aforementioned electronic device may further include: a bus, a communication interface 820, an input device 830, and an output device 840. The electronic device may also include a data transceiver module, an image monitoring module, and a signal monitoring module.

[0295] The processor 810, memory 800, communication interface 820, input device 830, and output device 840 are interconnected via a bus. Among them:

[0296] A bus can include a pathway for transmitting information between various components in an electronic device.

[0297] The processor 810 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present invention. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0298] The processor 810 may include a main processor, as well as a baseband chip, modem, etc.

[0299] The memory 800 stores a program that executes the technical solution of this invention, and may also store an operating system and other key business functions. Specifically, the program may include program code, which includes computer operation instructions. More specifically, the memory 800 may include read-only memory (ROM), other types of static storage devices capable of storing static information and instructions, random access memory (RAM), other types of dynamic storage devices capable of storing information and instructions, disk storage, flash memory, etc.

[0300] Input device 830 may include a device for receiving user input data and information, such as a keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor.

[0301] Output device 840 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0302] The communication interface 820 may include a device that uses any transceiver to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0303] The processor 810 executes the program stored in the memory 800 and calls other devices, which can be used to implement the various steps of any of the vehicle control methods provided in the above embodiments of this application.

[0304] This application also proposes a chip, which includes a processor and a data interface. The processor reads and runs a program stored in the memory through the data interface to execute the vehicle control method described in any of the above embodiments. For details of the processing and its beneficial effects, please refer to the above-described embodiments of the vehicle control method.

[0305] Exemplary computer program products and storage media

[0306] In addition to the methods and devices described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in the vehicle control methods according to various embodiments of this application as described in any of the above embodiments of this specification.

[0307] Computer program products can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the power device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0308] Furthermore, embodiments of this application may also be storage media storing computer programs, which are executed by a processor to perform the steps of the vehicle control method according to various embodiments of this application described in any of the above embodiments of this specification, specifically implementing the steps of the above vehicle control method.

[0309] For the foregoing method embodiments, in order to simplify the description, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0310] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For apparatus embodiments, since they are basically similar to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0311] The steps in the methods of the various embodiments of this application can be adjusted, merged, or deleted in order according to actual needs, and the technical features described in each embodiment can be replaced or combined.

[0312] The units of the apparatus in the various embodiments of this application can be merged, divided, and deleted according to actual needs.

[0313] It should be understood that the disclosed terminals, devices, and methods can be implemented in other ways, given the several embodiments provided in this application. For example, the terminal embodiments described above are merely illustrative. For instance, the division of modules or sub-modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple sub-modules or modules may be combined or integrated into another module, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or modules, and may be electrical, mechanical, or other forms.

[0314] The modules or submodules described as separate components may or may not be physically separate. The components that constitute a module or submodule may or may not be physical modules or submodules; that is, they may be located in one place or distributed across multiple network modules or submodules. Some or all of the modules or submodules can be selected to achieve the purpose of this embodiment's solution, depending on actual needs.

[0315] Furthermore, the functional modules or sub-modules in the various embodiments of this application can be integrated into one processing module, or each module or sub-module can exist physically separately, or two or more modules or sub-modules can be integrated into one module. The integrated modules or sub-modules described above can be implemented in hardware or in the form of software functional modules or sub-modules.

[0316] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0317] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software unit executed by a processor, or a combination of both. The software unit can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0318] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0319] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A vehicle control method, characterized in that, include: The vehicle acquires current driving environment parameters, driving status parameters, and historical feature parameters. The driving environment parameters include at least one of the speed parameters, lighting parameters, weather parameters, and road parameters of other vehicles around the current vehicle. The driving status parameters include at least one of the vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters of the current vehicle. The historical feature parameters include the current vehicle's historical following distance and braking frequency. The driving environment parameters, driving status parameters, and historical feature parameters are processed to obtain target data. Predict the target distance of the current vehicle based on the target data, wherein the target distance includes at least one of following distance and braking distance; The current vehicle's movement is controlled based on the target distance.

2. The vehicle control method according to claim 1, characterized in that, The step of predicting the target distance of the current vehicle based on the target data includes: Spatial and temporal features are extracted from the target data, and an initial distance is determined based on the spatial and temporal features. Based on the target data, correction parameters are determined, and the initial distance is corrected according to the correction parameters to obtain the target distance of the current vehicle.

3. The vehicle control method according to claim 2, characterized in that, If the initial distance includes the base following distance, then the correction parameters include at least one of environmental correction parameters and driving state correction parameters; If the initial distance includes the base braking distance, then the correction parameters include at least one of the following: friction coefficient decay correction parameters, driver reaction time correction parameters, and obstacle correction parameters.

4. The vehicle control method according to claim 1, characterized in that, The step of predicting the target distance of the current vehicle based on the target data includes: The target data is input into the prediction model to obtain the target distance of the current vehicle output by the prediction model. The target distance includes the lower limit of the following distance and the upper limit of the braking distance output by the prediction model.

5. The vehicle control method according to claim 4, characterized in that, The prediction model comprises multiple parallel neural networks, and the target distance is the average of the output distances of each of the neural networks.

6. The vehicle control method according to claim 1, characterized in that, The step of predicting the target distance of the current vehicle based on the target data includes: Obtain real-time weather data, road construction data, real-time traffic data, and map data corresponding to the area where the vehicle is currently located; Based on the real-time weather data, the road construction data, the real-time traffic data, and the map data, the driving environment of the current vehicle in the future time period is predicted, and the driving environment includes the weather environment and the road condition environment. The target distance is obtained by predicting the distance to be determined for the current vehicle based on the target data and correcting the distance to be determined based on the driving environment.

7. The vehicle control method according to claim 6, characterized in that, The step of predicting the driving environment of the current vehicle in a future time period based on the real-time weather data, road construction data, real-time traffic data, and map data includes: If the current driving environment to be predicted for the vehicle is weather conditions, the area where the current vehicle is located is divided into multiple sub-regions based on the map data. The rainfall and wind direction parameters for each sub-region are determined based on the real-time weather data. Based on the rainfall and wind direction parameters, the weather conditions for the current vehicle in the future time period are determined.

8. The vehicle control method according to claim 6, characterized in that, The step of predicting the driving environment of the current vehicle in a future time period based on the real-time weather data, road construction data, real-time traffic data, and map data includes: If the current driving environment to be predicted for the vehicle is a road condition environment, the probability value of construction at the current vehicle's location in a future time period is predicted based on road construction data and the map data. Based on the construction probability value and the real-time traffic data, the traffic environment in which the current vehicle will be located in the future time period is predicted.

9. A vehicle, characterized in that, include: The acquisition module is used to acquire the current vehicle's driving environment parameters, driving status parameters, and historical feature parameters. The driving environment parameters include at least one of the speed parameters, light parameters, weather parameters, and road parameters of other vehicles around the current vehicle. The driving status parameters include at least one of the current vehicle's vibration parameters, speed parameters, load parameters, braking parameters, and tire wear parameters. The historical feature parameters include the current vehicle's historical following distance and braking frequency. The processing module is used to process the driving environment parameters, driving status parameters, and historical feature parameters to obtain target data. The prediction module is used to predict the target distance of the current vehicle based on the target data, wherein the target distance includes at least one of following distance and braking distance; The control module is used to control the current vehicle's movement based on the target distance.

10. An electronic device, characterized in that, Including memory and processor, among which, The memory is connected to the processor and is used to store programs; The processor is used to implement the vehicle control method as described in any one of claims 1 to 8 by running a program in the memory.

Images