Road test data acquisition system for assisting in driving automobile

Abstract

The invention discloses a road test data acquisition system for auxiliary driving of an automobile, relates to the technical field of auxiliary driving of automobiles, and solves the problems that a lot of time and funds need to be spent for re-debugging when the model or performance of a tested automobile is changed in the prior art. The technical problem that the debugging complexity and cost of the auxiliary driving system of the vehicle are increased is solved. According to the method, self-adaptive adjustment can be performed according to vehicle models and performance changes through the constructed extensible model architecture, so that the workload of re-debugging is reduced; the flexibility and the adaptability of a model architecture are jointly guaranteed by adopting a machine learning and self-adaptive algorithm, a modularly-designed vehicle dynamics module, an established vehicle performance index database, an optimized debugging tool and flow, a predefined configuration template, measures such as online real-time calibration and the like; the problem that a large amount of re-debugging work is caused by vehicle model or performance changes is solved, and the complexity and debugging cost of a road test data acquisition system can be reduced.

Description

Technical Field

[0001] The invention belongs to the technical field of automobile assisted driving, and in particular is a road test data acquisition system for an assisted driving automobile. Background Art

[0002] With the development of automobile intelligence, the performance and reliability of assisted driving systems have become the focus of attention. In order to effectively evaluate and optimize assisted driving systems, it is necessary to collect a large amount of accurate and comprehensive road test data.

[0003] However, the existing road test data acquisition system has problems such as incomplete data, low accuracy, and poor real-time performance, and cannot meet the needs of assisted driving system research and development and optimization; therefore, the invention patent application with publication number CN114527674A discloses a model-in-the-loop simulation test method and system based on vehicle road test data, which conducts road tests on a real vehicle equipped with a limited autonomous driving algorithm, and simultaneously collects and stores the vehicle's road test data, conducts vehicle road test data analysis, and explores the key points of the real vehicle road test as the test key points of the model-in-the-loop simulation test, and writes model-in-the-loop simulation test cases, thereby carrying out model-in-the-loop simulation tests, obtaining simulation test results, and comparing the simulation test data with the vehicle road test data, which can accurately verify the effectiveness of the model and the degree of consistency with the real vehicle, and reproduce the vehicle road test effect.

[0004] The vehicle dynamics module in this patent needs to be debugged and calibrated according to the vehicle performance indicators. If the vehicle model or performance changes, a lot of debugging work needs to be done again, which increases the complexity and cost of the system. The models and algorithms relied on for road test data collection of assisted driving vehicles also need to be adjusted according to different vehicle models and performance.

[0005] Therefore, the present invention proposes a road test data collection system for an assisted driving vehicle. Summary of the invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a road test data acquisition system for an assisted driving vehicle, which is used to solve the technical problem that the prior art solution requires a lot of time and money to re-debug when the model or performance of the test vehicle changes, resulting in increased complexity and cost of debugging the vehicle's assisted driving system.

[0007] To achieve the above-mentioned object, a first aspect of the present invention provides a road test data acquisition system for an assisted driving vehicle, comprising: a model architecture module, a sensor module, a data preprocessing module, a data storage module, a positioning and time synchronization module, a communication module, a power management module and a central control module;

[0008] The model architecture module is used to construct an extensible model architecture based on the performance parameters of the vehicle; wherein the performance parameters include vehicle model, configuration and performance index data;

[0009] The sensor module is used to collect performance index data of the vehicle; wherein the performance index data includes conventional performance indexes and special performance indexes; conventional performance indexes include vehicle speed, acceleration and steering angle, etc.; special performance indexes include off-road passability parameters, climbing angle and wading depth, etc.;

[0010] The data preprocessing module is used to filter, remove noise and calibrate the collected performance indicator data;

[0011] The data storage module is used to store the pre-processed performance indicator data;

[0012] The positioning and time synchronization module is used to provide location information and timestamps to ensure the temporal and spatial consistency of data;

[0013] The communication module is used to transmit the collected data to a remote server in real time for data analysis and processing;

[0014] The power management module is used to provide a stable power supply for the entire system;

[0015] The central control module is used to coordinate and control the operation of the entire data acquisition system, including setting acquisition parameters, starting and stopping data acquisition, and monitoring system status;

[0016] The data preprocessing module is connected to the model architecture module and the sensor module, and the data preprocessing module communicates and / or is electrically connected to the model architecture module, the sensor module, and the positioning and time synchronization module respectively; the communication module communicates and / or is electrically connected to the positioning and time synchronization module and the power management module respectively.

[0017] Preferably, the constructing of an extensible model architecture based on the performance parameters of the vehicle includes:

[0018] S1: Building a vehicle performance simulation model based on machine learning;

[0019] S2: Decompose the vehicle dynamics module into several submodules;

[0020] S3: Build a vehicle performance index database;

[0021] S4: Develop automated debugging tools, including automated test scripts and visual debugging interfaces.

[0022] S5: predefine the corresponding debugging configuration template based on the performance index data of the vehicle;

[0023] S6: Perform online real-time calibration of the vehicle's assisted driving function;

[0024] S7: Integrates vehicle performance simulation models, vehicle performance indicator databases, and automated debugging tools into a scalable model architecture.

[0025] It should be noted that the vehicle dynamics module mainly analyzes and models the vehicle's motion state and performance, covering longitudinal dynamics (driving force, driving resistance, braking system), lateral dynamics (steering system, lateral force), suspension system, tire model, vehicle mass and inertia, stability and handling. Through comprehensive analysis and modeling, it can more accurately predict and control vehicle motion behavior, provide a theoretical basis for vehicle design, optimization and control strategies, and ensure precise control and safety of the vehicle in assisted driving and autonomous driving.

[0026] Preferably, the vehicle performance simulation model is constructed based on machine learning, including:

[0027] The vehicle's performance index data is extracted as the original input features; the input features are standardized, and the original input features are reduced in dimension based on the principal component analysis algorithm to obtain standard input features; minority class samples are generated based on oversampling technology combined with a generative adversarial network; regularization methods are used to prevent overfitting; in model training, mean square error is selected as the loss function, and stochastic gradient descent is used as the model optimization algorithm, and convergence is accelerated by dynamically adjusting the learning rate; machine learning technology is used to automatically learn and adjust model parameters according to new vehicle performance index data; the optimized model is marked as a vehicle performance simulation model.

[0028] Preferably, the feature dimension reduction of the original input features based on the principal component analysis algorithm includes:

[0029] Extract the original input features; normalize the original input features by the mean and divide them by the standard deviation to obtain standard data, and integrate the standard data into the original feature matrix; calculate the covariance matrix between the standard data, and obtain the eigenvalues ​​and the eigenvectors corresponding to the eigenvalues ​​by eigenvalue decomposition of the covariance matrix; sort the eigenvalues ​​from large to small, select several eigenvalues ​​and corresponding eigenvectors greater than the set threshold, and integrate the corresponding eigenvectors into an eigenvector matrix; multiply the eigenvector matrix by the original data matrix to obtain a reduced-dimensional data matrix; mark the vectors in the reduced-dimensional data matrix as standard input features.

[0030] Preferably, the mean square error function As the loss function; where L is the mean square error, y i is the true value, is the predicted value, and n is the number of samples.

[0031] Preferably, the vehicle dynamics module is decomposed into several submodules, including:

[0032] According to the functional division of each component in the vehicle dynamics module, the vehicle dynamics module is decomposed into a plurality of sub-modules; the plurality of sub-modules are independent of each other and work in coordination.

[0033] It is worth noting that the vehicle dynamics module can be decomposed into a powertrain submodule, a suspension submodule, a steering submodule and a braking submodule. The powertrain submodule provides torque output information to the suspension submodule, and the suspension submodule provides body posture data to the braking submodule. The collaborative working mechanism is based on event-driven or timed synchronization. When the key parameters of a submodule change, the relevant submodules are notified through message queues or publish / subscribe modes to make corresponding adjustments and calculations. When the vehicle model or performance changes, only the affected submodules need to be debugged instead of the entire module.

[0034] Preferably, the building of the vehicle performance index database includes:

[0035] Collect and organize the performance parameters of various types of vehicles; store the performance parameters in a database and record the performance parameters in a table; replace the corresponding performance parameters recorded in the table according to the performance parameters updated by the vehicle manufacturer and the actual test data; integrate several tables storing the performance parameters into a vehicle performance index database; wherein the database is a relational database and a non-relational database.

[0036] It should be noted that when debugging a new vehicle, the corresponding performance index data can be obtained by comparing and referring to the new vehicle with similar models in the vehicle performance index database, thereby reducing the blindness of debugging the new vehicle.

[0037] Preferably, the development automated debugging tool includes:

[0038] Use programming language to write automated test scripts, and simulate user operations and driving conditions based on the Selenium framework. Use Web architecture to develop a visual debugging interface to achieve real-time performance curve and parameter comparison functions. The visual debugging interface supports multi-window layout and custom views.

[0039] Preferably, the online real-time calibration of the vehicle's assisted driving function includes:

[0040] Real-time data is transmitted using a custom communication protocol. The bandwidth requirements are evaluated based on the amount of data transmitted simultaneously and the sampling frequency to ensure stable and reliable data transmission. Edge computing devices are used for processing, and parallel computing and caching technologies are used to optimize the computing process. A monitoring mechanism is established to monitor and adjust the usage of computing resources in real time.

[0041] Preferably, the data preprocessing module uses a Kalman filter algorithm to filter the performance index data to remove noise interference, and calibrates each sensor in the sensor module through a linear regression algorithm.

[0042] Preferably, the data storage module adopts a solid state hard disk, and the capacity of the solid state hard disk is not less than 1T, and the reading and writing speed is not less than 500MB / s.

[0043] Preferably, the positioning and time synchronization module uses a satellite positioning system to provide a position and uses an atomic clock to provide a timestamp.

[0044] Preferably, the data transmission network of the communication module includes 5G network and satellite communication. The 5G network is preferentially used for data transmission in the 5G signal coverage area, and it automatically switches to satellite communication when the 5G signal is interrupted.

[0045] Preferably, the power management module has overvoltage protection, undervoltage protection and intelligent power distribution functions to adapt to the complex working conditions of the vehicle power supply and ensure stable operation of the system under various conditions.

[0046] Preferably, the central control module adopts a microcontroller, runs a real-time operating system, responds to various control instructions and processes data.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] The present invention constructs an extensible model architecture that can adaptively adjust according to vehicle model and performance changes, thereby reducing the workload of re-debugging; the adopted machine learning and adaptive algorithms, modularly designed vehicle dynamics modules, established vehicle performance index database, optimized debugging tools and processes, predefined configuration templates, and online real-time calibration and other measures jointly ensure the flexibility and adaptability of the model architecture, overcome the problem of a large amount of re-debugging work caused by changes in vehicle model or performance, and are conducive to reducing the complexity and debugging cost of the road test data acquisition system, as well as improving the work efficiency of the road test data acquisition system. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 It is a schematic diagram of the principle of the road test data acquisition system of the present invention;

[0051] Figure 2 The figure is an overall flow chart of constructing the model architecture module in the present invention. DETAILED DESCRIPTION

[0052] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] See also Figure 1-Figure 2 , the first aspect of the present invention provides a road test data acquisition system for an assisted driving vehicle, including: a model architecture module, a sensor module, a data preprocessing module, a data storage module, a positioning and time synchronization module, a communication module, a power management module and a central control module;

[0054] Model architecture module: used to build a scalable model architecture based on the vehicle's performance parameters; the performance parameters include vehicle model, configuration and performance indicator data;

[0055] Sensor module: used to collect vehicle performance index data; sensor modules include but are not limited to vehicle speed sensors, acceleration sensors, gyroscopes, lidars, cameras, steering wheel angle sensors and pedal travel sensors; performance index data include conventional performance indexes and special performance indexes; conventional performance indexes include vehicle speed, acceleration and steering angle, etc.; special performance indexes include off-road passability parameters, climbing angle and wading depth, etc.;

[0056] Data preprocessing module: used to filter, denoise and calibrate the collected performance indicator data;

[0057] A data storage module, used for storing preprocessed performance indicator data;

[0058] Positioning and time synchronization module: used to provide location information and timestamps to ensure the temporal and spatial consistency of data;

[0059] Communication module: used to transmit the collected data to the remote server in real time for data analysis and processing;

[0060] Power management module: used to provide stable power supply for the entire system;

[0061] Central control module: used to coordinate and control the operation of the entire data acquisition system, including setting acquisition parameters, starting and stopping data acquisition, and monitoring system status;

[0062] The data preprocessing module is connected to the model architecture module and the sensor module. The data preprocessing module communicates and / or is electrically connected to the model architecture module, the sensor module, and the positioning and time synchronization module respectively; the communication module communicates and / or is electrically connected to the positioning and time synchronization module and the power management module respectively.

[0063] Example 1

[0064] For example, the application of the assisted driving car road test data acquisition system on cars; a new electric car is selected as a test vehicle, and the road test data acquisition system proposed by the present invention is installed;

[0065] In the model architecture module, the corresponding model architecture is constructed according to the performance characteristics of the car, such as acceleration performance and cruising range. Machine learning algorithms, such as random forest or support vector machine, are used to conduct in-depth analysis and accurate prediction of performance index data such as vehicle speed and power consumption.

[0066] In the sensor module, the speed sensor measures the vehicle speed with an error of no more than ±0.1km / h; the acceleration sensor obtains acceleration and deceleration conditions with a measurement accuracy of ±0.05m / s 2 The steering wheel angle sensor records steering operations with an accuracy of ±1°. The lidar and camera can sense the surrounding environment. The camera resolution is at least 1080p, and the lidar scan frequency is not less than 10Hz.

[0067] The data preprocessing module uses Kalman filtering algorithm and linear regression algorithm to filter, denoise and calibrate the data collected by the sensor; the Kalman filtering algorithm can remove high-frequency noise, and the linear regression algorithm can accurately calibrate the sensor according to the known standard value to ensure that the data accuracy is above 99%;

[0068] The data storage module uses a high-speed solid-state hard drive with a storage capacity of 2T and a read and write speed of more than 500MB / s. It can quickly and stably store pre-processed performance indicator data to ensure the integrity and traceability of data storage;

[0069] The positioning and time synchronization module uses satellite positioning systems (such as Beidou or GPS) and atomic clocks to provide data with nanosecond-level timestamps and centimeter-level position accuracy, thereby ensuring the temporal and spatial consistency of the data;

[0070] In areas with good 5G signals, the communication module will use the 5G network to transmit data to the remote server in real time. When the signal strength of the 5G network is lower than the threshold (e.g. -100dBm), the system will automatically switch to satellite communication to ensure the continuity of data transmission, and the packet loss rate of data transmission will be controlled within 1%.

[0071] The power management module supplies power to the system and can adapt to the complex changes in the car's power supply. For example, it provides the system with stable voltage (within a fluctuation range of ±5%) and current under different operating conditions such as vehicle starting, acceleration, and charging.

[0072] The central control module coordinates the work of each module and sets appropriate acquisition parameters; for example, according to different test scenarios and requirements, the sampling frequency of the sensor, the interval of data transmission and other parameters can be flexibly adjusted.

[0073] During the model architecture module construction process, follow the steps below:

[0074] S1: Using machine learning and adaptive algorithms, the original data is first standardized to make its mean 0 and standard deviation 1; at the same time, feature scaling is performed, such as mapping the data to the interval [0,1]; by using oversampling techniques, such as the SMOTE algorithm, the minority class samples are increased to achieve a balanced ratio; L2 regularization is used to prevent overfitting, and the regularization parameters are selected based on the cross-validation results; the mean square error is selected as the loss function, and the stochastic gradient descent algorithm is used for optimization. The learning rate is initially set to 0.01, and the learning rate is dynamically adjusted according to the validation set error during training;

[0075] S2: The vehicle dynamics module is carefully decomposed into sub-modules such as powertrain, suspension, steering and braking. Each sub-module interacts through a well-defined and easy-to-understand API, and the data format adopts a concise and efficient JSON format. For example, the API provided by the powertrain sub-module can obtain the current torque output and power distribution. The collaborative work is event-driven. When the key parameters of the braking sub-module (such as braking torque, brake disc temperature, etc.) change significantly (exceed the preset threshold), the relevant sub-modules are notified in time through an efficient message queue (such as RabbitMQ), triggering them to make corresponding adjustments and calculations.

[0076] S3: Establish a vehicle performance index database to comprehensively store the performance data of the car and similar vehicles, including but not limited to special performance indicators such as vehicle power performance, energy consumption index, comfort parameters, etc.; the data is stored in a MySQL relational database with a reasonable table structure design; the database is updated and improved monthly based on new model information, performance improvement data and actual test feedback provided by the manufacturer; during the data update process, transaction processing is used to ensure data consistency and integrity;

[0077] S4: Develop automated test scripts written in Python and Selenium to simulate various complex and extreme driving conditions, such as emergency lane changes on highways and driving in bad weather conditions. Use JavaScript and D3.js to create a visual debugging interface that not only supports multi-window layout, making it convenient to view the change curves of multiple parameters at the same time, but also supports custom views. Users can select specific time periods and parameter combinations for detailed analysis according to their needs, and can zoom and pan the data.

[0078] S5: Prepare comprehensive and detailed debugging configuration templates for common electric car performance categories, such as economy, sports, and comfort. When testing the new car, apply the closest template through preliminary evaluation of vehicle performance, and make small but precise modifications based on actual test data and the personalized characteristics of the vehicle. For example, for a sports car, the template may focus on optimizing power output and suspension hardness, while for an economy car, the template may focus more on energy recovery efficiency and the setting of low-power driving mode.

[0079] S6: When the vehicle is running, the system is calibrated and adjusted in real time; the accuracy of sensors such as vehicle speed and acceleration needs to reach the industry-leading level. For example, the accuracy of the vehicle speed sensor is within ±0.5km / h, and the accuracy of the acceleration sensor is within ±0.1m / s 2 Within; data is transmitted at a frequency of at least 1000 times per second through the CAN bus, and the communication protocol is used for data transmission. The communication protocol contains fields such as data packet header, data type, data length, check bit, etc. to ensure the accuracy and integrity of the data; edge computing devices (such as NVIDIA Jetson series) are used to process data, establish a real-time monitoring mechanism, and dynamically adjust the priority of computing tasks and allocate resources according to the usage of computing resources (such as CPU utilization, memory occupancy, etc.);

[0080] S7: Integrates vehicle performance simulation models, vehicle performance indicator databases, and automated debugging tools into a scalable model architecture.

[0081] Through this embodiment, the road test data of the electric car was successfully collected, providing strong support for the optimization of the assisted driving system.

[0082] Example 2

[0083] For example, the application of the assisted driving car road test data collection system on SUV; select a four-wheel drive SUV as the test vehicle and install this road test data collection system;

[0084] The model architecture module builds the model based on the off-road performance and climbing ability of the SUV, taking into account the special factors such as the high ground clearance of the SUV model, the working characteristics of the four-wheel drive system, and the travel and hardness of the suspension;

[0085] The sensor module includes a high-precision wheel speed sensor (with a measurement accuracy within ±1rpm) for measuring wheel speed and a high-sensitivity gyroscope (with a measurement accuracy within ±0.5°) for detecting the vehicle body tilt angle, which can accurately obtain the vehicle's operating status under complex road conditions.

[0086] The data preprocessing module, data storage module, positioning and time synchronization module, communication module, power management module and central control module work normally according to system requirements;

[0087] In the construction of the model architecture module, the specific steps are as follows:

[0088] S1: Generate minority class samples using a generative adversarial network (GAN), such as performance indicator data under simulated extreme off-road conditions; use L1 regularization to prevent the model from being too complex; select mean square error as the loss function and use the stochastic gradient descent algorithm for optimization; during training, dynamically adjust the learning rate and regularization parameters according to the performance of the validation set to obtain the best model performance;

[0089] S2: The vehicle dynamics module is subdivided into multiple submodules, such as the four-wheel drive system control module, the suspension adjustment module, and the differential lock control module. The modules interact with each other in XML data format, which is clear and standardized. The collaborative work is based on timed synchronization, such as data exchange and status update every 10 milliseconds to ensure coordinated operation between the submodules.

[0090] S3: Establish a vehicle performance index database for SUVs and store it in MongoDB non-relational database; the vehicle performance index database contains special performance indicators such as off-road passability parameters, climbing angle, wading depth, etc.; update the data monthly, and use data backup and recovery mechanism during the update process to prevent data loss;

[0091] S4: Develop efficient automated debugging tools and processes, such as using automated scripts to batch test and evaluate models, and using performance analysis tools to monitor and optimize system resource consumption in real time;

[0092] S5: Prepare a debugging configuration template suitable for SUVs, for example, formulate specific parameter configurations for different off-road scenarios (such as deserts, mountains, muddy roads, etc.);

[0093] S6: Real-time calibration and adjustment, sensor data is transmitted via Bluetooth 5.0 in a low-power, high-stability manner; the transmission rate is no less than 2Mbps, ensuring the real-time and accuracy of the data;

[0094] S7: Integrates vehicle performance simulation models, vehicle performance indicator databases, and automated debugging tools into a scalable model architecture.

[0095] Through this embodiment, road test data of a four-wheel drive SUV model was successfully collected, providing strong support for the optimization of the assisted driving system.

[0096] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A road test data acquisition system for an assisted driving vehicle, comprising: Model architecture module, sensor module, data preprocessing module, data storage module, positioning and time synchronization module, communication module, power management module and central control module; characterized in that the model architecture module: is used to build an extensible model architecture based on the performance parameters of the vehicle; wherein the performance parameters include vehicle model, configuration and performance indicator data; The sensor module is used to collect performance index data of the vehicle; wherein the performance index data includes conventional performance indexes and special performance indexes; The data preprocessing module is used to filter, remove noise and calibrate the collected performance indicator data; The positioning and time synchronization module is used to provide location information and timestamps to ensure the temporal and spatial consistency of data; The communication module is used to transmit the collected data to a remote server in real time for data analysis and processing; The central control module is used to coordinate and control the operation of the entire data acquisition system, including setting acquisition parameters, starting and stopping data acquisition, and monitoring system status.

2. The road test data acquisition system for an assisted driving vehicle according to claim 1, characterized in that: The scalable model architecture is constructed based on the performance parameters of the vehicle, including: S1: Building a vehicle performance simulation model based on machine learning; S2: Decompose the vehicle dynamics module into several submodules; S3: Build a vehicle performance index database; S4: Develop automated debugging tools, including automated test scripts and visual debugging interfaces. S5: predefine the corresponding debugging configuration template based on the performance index data of the vehicle; S6: Perform online real-time calibration of the vehicle's assisted driving function; S7: Integrates vehicle performance simulation models, vehicle performance indicator databases, and automated debugging tools into a scalable model architecture.

3. The road test data acquisition system for an assisted driving vehicle according to claim 2, characterized in that: The vehicle performance simulation model is constructed based on machine learning, including: The vehicle's performance index data is extracted as the original input features; the original input features are reduced in dimension based on the principal component analysis algorithm to obtain standard input features; minority class samples are generated based on oversampling technology combined with a generative adversarial network; regularization methods are used to prevent overfitting; in model training, mean square error is selected as the loss function, and stochastic gradient descent is used as the model optimization algorithm, and convergence is accelerated by dynamically adjusting the learning rate; machine learning technology is used to automatically learn and adjust model parameters according to new vehicle performance index data; the optimized model is marked as a vehicle performance simulation model.

4. The road test data acquisition system for an assisted driving vehicle according to claim 3, characterized in that: The feature dimension reduction of the original input features based on the principal component analysis algorithm includes: Extract the original input features; normalize the original input features by the mean and divide them by the standard deviation to obtain standard data, and integrate the standard data into the original feature matrix; calculate the covariance matrix between the standard data, and obtain the eigenvalues ​​and the eigenvectors corresponding to the eigenvalues ​​by eigenvalue decomposition of the covariance matrix; sort the eigenvalues ​​from large to small, select several eigenvalues ​​and corresponding eigenvectors greater than the set threshold, and integrate the corresponding eigenvectors into an eigenvector matrix; multiply the eigenvector matrix by the original data matrix to obtain a reduced-dimensional data matrix; mark the vectors in the reduced-dimensional data matrix as standard input features.

5. The road test data acquisition system for an assisted driving vehicle according to claim 2, characterized in that: The vehicle dynamics module is decomposed into several sub-modules, including: According to the functional division of each component in the vehicle dynamics module, the vehicle dynamics module is decomposed into a plurality of sub-modules; the plurality of sub-modules are independent of each other and work in coordination.

6. The road test data acquisition system for an assisted driving vehicle according to claim 2, characterized in that: The step of constructing a vehicle performance index database includes: Collect and organize the performance parameters of various types of vehicles; store the performance parameters in a database and record the performance parameters in a table; replace the corresponding performance parameters recorded in the table according to the performance parameters updated by the vehicle manufacturer and the actual test data; integrate several tables storing the performance parameters into a vehicle performance index database; wherein the database is a relational database and a non-relational database.

7. The road test data acquisition system for an assisted driving vehicle according to claim 2, characterized in that: The development of the automated debugging tool comprises: Use programming language to write automated test scripts, and simulate user operations and driving conditions based on the Selenium framework. Use Web architecture to develop a visual debugging interface to achieve real-time performance curve and parameter comparison functions. The visual debugging interface supports multi-window layout and custom views.

8. The road test data acquisition system for an assisted driving vehicle according to claim 2, characterized in that: The online real-time calibration of the vehicle's auxiliary driving function includes: Real-time data is transmitted using a custom communication protocol. The bandwidth requirements are evaluated based on the amount of data transmitted simultaneously and the sampling frequency to ensure stable and reliable data transmission. Edge computing devices are used for processing, and parallel computing and caching technologies are used to optimize the computing process. A monitoring mechanism is established to monitor and adjust the usage of computing resources in real time.

9. The road test data acquisition system for an assisted driving vehicle according to claim 1, characterized in that: The data preprocessing module uses a Kalman filter algorithm to filter the performance index data to remove noise interference, and calibrates each sensor in the sensor module through a linear regression algorithm.

10. The road test data acquisition system for an assisted driving vehicle according to claim 1, characterized in that: The data transmission network of the communication module includes 5G network and satellite communication. The 5G network is preferentially used for data transmission in the 5G signal coverage area, and it automatically switches to satellite communication when the 5G signal is interrupted.

Images