CN111814904A - A method and system for discriminating driving mode for automatic driving road test - Google Patents

Abstract

The invention relates to an automatic driving road test driving mode distinguishing method, wherein the method comprises the following steps: acquiring initial data; determining a driving duration threshold; obtaining a correct driving data set and a data set to be classified according to the initial data and the driving duration threshold; determining a final classification model based on the correct driving data set; and correcting the driving mode corresponding to each data to be classified in the data set to be classified based on the final classification model. The method can accurately distinguish the driving mode aiming at the data generated by the automatic driving test, and provides a good basis for the subsequent automatic driving capability evaluation.

Description

A method and system for discriminating driving mode for automatic driving road test

technical field

The invention relates to the technical field of pattern recognition, in particular to a method and system for distinguishing driving patterns for automatic driving road tests.

Background technique

Road testing is a key link in the development and application of autonomous driving technology; autonomous driving road testing generates a large amount of test data, recording field information such as test vehicle identification, test time, speed, acceleration, driving mode, etc., but the sensors of autonomous driving vehicles such as Radar, lidar, camera, GPS, etc. will cause performance degradation due to the accumulation of time in bad weather, which may bring about problems such as data detection and transmission of autonomous driving vehicles, affecting the accuracy of road test driving mode data. The accuracy of the recording of automatic or manual driving modes during the test of an autonomous vehicle will directly affect the calculation of indicators such as the manual intervention rate, which in turn affects the accuracy of the assessment of the driving ability of the autonomous vehicle.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a driving mode identification method and system for an automatic driving road test, so as to accurately identify the driving mode in the automatic driving test process.

For achieving the above object, the present invention provides the following scheme:

An automatic driving road test driving mode identification method, comprising:

get initial data;

Determining driving duration thresholds;

Obtain a correct driving data set and a data set to be classified according to the initial data and the driving duration threshold;

determining a final classification model based on the correct driving data set;

The driving mode corresponding to each to-be-classified data in the to-be-classified data set is corrected based on the final classification model.

Preferably, the obtaining initial data includes:

get raw data;

performing data filling and data correction on the original data to obtain repaired data;

The repaired data is segmented according to the driving mode duration to obtain an automatic driving data set and a manual driving data set; the initial data includes the automatic driving data set and the manual driving data set.

Preferably, the determining the driving duration threshold includes:

Select the parameters to be tested;

Determine whether the parameter to be tested satisfies both normal distribution and homogeneous variance;

If yes, the parameter test method is used to obtain the automatic driving duration threshold and the manual driving duration threshold respectively; if not, the non-parametric test method is used to obtain the automatic driving duration threshold and the manual driving duration threshold respectively; the driving duration threshold Including the automatic driving duration threshold and the manual driving duration threshold.

Preferably, the obtaining of the correct driving data set and the data set to be classified according to the initial data and the driving duration threshold includes:

Take the automatic driving data whose duration in the automatic driving data set is greater than or equal to the automatic driving duration threshold as the correct automatic driving data set, and use the automatic driving data set whose duration is greater than or equal to the manual driving duration threshold in the manual driving data set. The manual driving data is used as the correct manual driving data set; the correct driving data set includes the correct automatic driving data set and the correct manual driving data set;

The automatic driving data whose duration in the automatic driving data set is less than the automatic driving duration threshold is regarded as the automatic driving data set to be classified, and the manual driving data whose duration is less than the manual driving duration threshold in the manual driving data set As the to-be-classified manual driving data set; the to-be-classified driving data set includes the to-be-classified automatic driving data set and the to-be-classified manual driving data set.

Preferably, the determining of the final classification model based on the correct driving data set includes:

Perform feature selection in the correct driving data set to obtain a feature data set;

In the feature data set, select a set proportion of data as a training set, and the rest of the data as a test set;

Build each classification model; each described classification model is K nearest neighbor estimation model, support vector machine model, decision tree model, random forest model, BP neural network model respectively;

Train each of the classification models based on the training set; obtain each of the trained classification models;

Each of the classification models is evaluated based on the test set, and the trained classification model corresponding to the highest evaluation value is selected as the final classification model.

The present invention also provides an automatic driving road test driving mode identification system, comprising:

Data acquisition module for acquiring initial data;

a threshold determination module for determining the driving duration threshold;

a data set determination module, configured to obtain a correct driving data set and a data set to be classified according to the initial data and the driving duration threshold;

a model determination module for determining a final classification model based on the correct driving data set;

A mode correction module, configured to correct the driving mode corresponding to each to-be-classified data in the to-be-classified data set based on the final classification model.

Preferably, the data acquisition module includes:

A data acquisition unit for acquiring raw data;

a data processing unit for performing data filling and data correction on the original data to obtain repaired data;

A data segmentation unit, configured to segment the repaired data according to the driving mode duration to obtain an automatic driving data set and a manual driving data set; the initial data includes the automatic driving data set and the manual driving data set.

Preferably, the determination threshold determination module includes:

The parameter unit is used to select the parameters to be tested;

a judging unit for judging whether the parameter to be tested satisfies both normal distribution and homogeneous variance;

The threshold determination unit is used to obtain the automatic driving duration threshold and the manual driving duration threshold respectively by using the parameter test method when it is satisfied; when it is not satisfied, the non-parametric test method is used to obtain the automatic driving duration threshold and the manual driving duration threshold respectively. ; The driving duration threshold includes the automatic driving duration threshold and the manual driving duration threshold.

Preferably, the data set determination module includes:

The correct driving data set determination unit is configured to use the automatic driving data in the automatic driving data set with a duration greater than or equal to the automatic driving duration threshold as the correct automatic driving data set, and use the manual driving data set with a duration greater than or equal to the automatic driving duration. The manual driving data equal to the manual driving duration threshold is used as the correct manual driving data set; the correct driving data set includes the correct automatic driving data set and the correct manual driving data set;

The data to be classified and the determining unit are used to use the automatic driving data in the automatic driving data set whose duration is less than the automatic driving duration threshold as the automatic driving data set to be classified, and use the manual driving data set whose duration is less than the automatic driving data set. The manual driving data of the manual driving duration threshold is used as the manual driving data set to be classified; the driving data set to be classified includes the automatic driving data set to be classified and the manual driving data set to be classified.

Preferably, the model determination module includes:

a feature data set determination module, used for feature selection in the correct driving data set to obtain a feature data set;

A feature data set classification unit, used for selecting a set proportion of data in the feature data set as a training set, and the rest of the data as a test set;

a model construction unit, used for constructing each classification model; each described classification model is respectively a K-nearest neighbor estimation model, a support vector machine model, a decision tree model, a random forest model, and a BP neural network model;

a training unit for training each of the classification models based on the training set; obtaining each of the trained classification models;

A model determining unit, configured to evaluate each of the classification models based on the test set, and select the trained classification model corresponding to the highest evaluation value as the final classification model.

According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

The present invention relates to a driving mode identification method for automatic driving road test, wherein the method includes: acquiring initial data; determining a driving duration threshold; obtaining a correct driving data set and a waiting period according to the initial data and the driving duration threshold classification data set; determining a final classification model based on the correct driving data set; correcting the driving mode corresponding to each to-be-classified data in the to-be-classified data set based on the final classification model. The invention can accurately identify the driving mode according to the data generated by the automatic driving test, and provide a good basis for the subsequent automatic driving capability evaluation.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings required in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some of the present invention. In the embodiments, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative labor.

Fig. 1 is the flow chart of the driving mode identification method for automatic driving road test of the present invention;

FIG. 2 is a schematic structural diagram of the driving mode identification system for automatic driving road test according to the present invention;

Figure 3 is a schematic diagram of the saliency test of automatic driving mode in urban road scenes;

FIG. 4 is a schematic diagram of the saliency test of the artificial driving mode in the urban road scene;

Fig. 5 is a schematic diagram of the saliency test of automatic driving mode in expressway scene;

FIG. 6 is a schematic diagram of the saliency test of the manual driving mode in the expressway scene.

Symbol description: 1-data acquisition module, 2-threshold determination module, 3-data set determination module, 4-model determination module, 5-mode correction module.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The purpose of the present invention is to provide a driving mode identification method and system for an automatic driving road test, so as to accurately identify the driving mode in the automatic driving test process.

In order to make the above objects, features and advantages of the present invention more clearly understood, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

FIG. 1 is a flowchart of the method for discriminating driving modes for automatic driving road tests according to the present invention. As shown in FIG. 1 , the present invention provides a method for discriminating driving modes for automatic driving road tests, including:

Step S1, obtaining initial data.

Step S2, determining the driving duration threshold.

Step S3, obtaining a correct driving data set and a data set to be classified according to the initial data and the driving duration threshold.

Step S4, determining a final classification model based on the correct driving data set.

Step S5, correcting the driving mode corresponding to each to-be-classified data in the to-be-classified data set based on the final classification model.

Wherein, the initial data includes an automatic driving data set and a manual driving data set.

Specifically, the step S1 includes:

Step S11, acquiring original data. The raw data is the test data generated in the road test phase of automatic driving. Preferably, the raw data includes vehicle number, longitude, latitude, positioning time, vehicle speed, and driving mode. Specific examples are shown in Table 1.

Table 1 Raw data example

Step S12, performing data filling and data correction on the original data to obtain repaired data. For example, the abnormal positioning time format data is converted into YYYY-MM-DDHH:MM:SS (year-month-day-hour-minute-second) format.

Step S13, segment the repaired data according to the driving mode duration to obtain an automatic driving data set and a manual driving data set; the initial data includes the automatic driving data set and the manual driving data set.

As an optional implementation manner, the step S2 of the present invention includes:

Step S21, select the parameter to be tested; the parameter to be tested needs to be selected in combination with the field, quality and feasibility and convenience of the significant difference test of the original data.

Step S22, it is judged whether the parameter to be tested satisfies both normal distribution and homogeneous variance.

Preferably, the method for testing whether the parameter to be tested satisfies the normal distribution includes: normal probability paper method, Shapiro-Wilk test method, Kolmogorov test method and skewness-kurtosis test method. Methods for testing whether the parameters to be tested satisfy the homogeneity of variance include: Hartley test, Bartlett test and modified Bartlett test.

Step S23, if yes, then adopt the parameter checking method to obtain the automatic driving duration threshold and the manual driving duration threshold respectively; if not, adopt the non-parametric checking method to obtain the automatic driving duration threshold and the manual driving duration threshold respectively; the driving The duration threshold includes the automatic driving duration threshold and the manual driving duration threshold.

Further, the correct driving data set of the present invention includes a correct automatic driving data set and a correct manual driving data set; the to-be-classified driving data set includes a to-be-classified automatic driving data set and a to-be-classified manual driving data set.

Specifically, the step S3 includes:

Step S31, take the automatic driving data whose duration is greater than or equal to the automatic driving duration threshold in the automatic driving data set as the correct automatic driving data set, and set the duration in the manual driving data set is greater than or equal to the manual driving duration. The manual driving data of the duration threshold is used as the correct manual driving data set.

Step S32, taking the automatic driving data in the automatic driving data set whose duration is less than the automatic driving duration threshold as the automatic driving data set to be classified, and taking the automatic driving data in the manual driving data set whose duration is less than the manual driving duration threshold. The manual driving data is used as the manual driving data set to be classified.

In order to accurately identify and correct the driving mode corresponding to the data in the driving data set to be classified, the present invention needs to select an optimal classification model from the constructed models. Specifically, the step S4 of the present invention includes:

Step S41, perform feature selection in the correct driving data set to obtain a feature data set. The most effective features are selected to achieve the purpose of reducing the dimension of the feature space, and the selected features should be able to better distinguish the automatic driving mode from the manual driving mode.

In step S42, a set proportion of data is selected from the feature data set as a training set, and the rest of the data is used as a test set. In this embodiment, the set ratio is 70%.

Step S43, based on the pattern recognition method of machine learning, construct each classification model according to the selected features; each described classification model is respectively a K-nearest neighbor estimation model, a support vector machine model, a decision tree model, a random forest model, and a BP neural network model.

In step S44, each of the classification models is trained based on the training set to obtain each of the trained classification models.

Step S45: Evaluate each of the classification models based on the test set, and select the trained classification model corresponding to the highest evaluation value as the final classification model. Specifically, three quantities of accuracy, precision, and recall are used for evaluation.

Specifically, the above method will be further described by taking the test data of autonomous driving roads in Shanghai as an example.

The acquired raw data includes vehicle number, longitude, latitude, positioning time, vehicle speed and driving mode; the collection frequency is 1Hz, the driving modes involved include automatic driving and manual driving, and the scenarios involved include urban road scenes and expressway scenes. Among them, the test data for urban road scenes is 33,000, and the expressway scene is 403,000.

Perform data filling and data correction on the obtained original data to obtain repaired data.

The patched data is segmented to obtain an automatic driving urban road dataset, an automatic driving expressway dataset, a manual driving urban road dataset and a manual driving expressway dataset.

The vehicle speed is selected as the parameter to be tested. After calculation, the speed parameters in Shanghai do not satisfy the normal distribution and homogeneous variance.

The rank sum test in the non-parametric test is used to perform the significance test of the vehicle speed before and after a specific duration of 5-20 seconds. Among them, the amount of data under the specific duration of 5 seconds and 6 seconds in the expressway scene is too small and does not meet the minimum sample size. Therefore, the vehicle speed before and after a specific period of 7-20 seconds is selected for significance test. In order to ensure that the sample size of the vehicle speed before and after a certain period of time is the same, the vehicle speed data of the same sample size are randomly selected for significance testing. The specific significance test results are shown in Figure 3, Figure 4, Figure 5 and Figure 6, wherein Figure 3 is a schematic diagram of the significance test of the automatic driving mode in the urban road scene; Figure 4 is a schematic diagram of the significance test of the manual driving mode in the urban road scene ; Figure 5 is a schematic diagram of the saliency test of the automatic driving mode in the expressway scene; Figure 6 is a schematic diagram of the saliency test of the manual driving mode in the expressway scene.

According to the principle of rank sum test, the smaller the P value is, the greater the significant difference is. Compare all the significance test results. Therefore, the thresholds for the duration of automatic driving mode and manual driving mode in urban road scenarios and expressway scenarios are finally determined as shown in Table 2. .

Table 2 Duration thresholds

Correct driving is obtained by classifying the data in the automatic driving urban road dataset, the automatic driving expressway dataset, the manual driving urban road dataset and the manual driving expressway dataset according to the threshold information obtained in Table 2 datasets and datasets to be classified.

Select the average speed, the standard deviation of the speed, the average value of the absolute value of the speed difference, the 85th percentile value of the absolute value of the speed difference, and the 15th percentile value of the absolute value of the speed difference as the feature values, and the correct driving data set is extracted based on the selected feature values. Feature dataset.

Select 70% from the feature dataset as the training set and 30% as the test set.

Build K-nearest neighbor estimation, support vector machines, decision trees, random forests and BP neural network 5 classification models. Use the training set to train each classification model, and use the test set to evaluate each classification model. The evaluation results are shown in Table 3 and Table 4.

Table 3 Evaluation results of urban road scene model

Table 4 Evaluation results of expressway scene model

In the urban road scene and expressway scene, the recognition accuracy and precision rate of the random forest classification model are the highest among the five classification models, and the recall rate of the support vector machine is higher. Considering three indicators, the random forest classification model The performance is the best, so the random forest classification model is selected as the final classification model to identify and correct the posture pattern corresponding to the data in the data set to be classified.

See Table 5, Table 6 and Table 7 for the specific correction data.

Table 5 Identification results of original automatic driving mode data

Table 6 Discrimination results of original manual driving mode data

Table 7 Summary of driving mode discrimination results in test data

It can be seen from the data in Table 7 that 40% of the driving modes corresponding to the data to be classified are wrong. Therefore, it is necessary to correct the posture mode in the road test of automatic driving.

FIG. 2 is a schematic structural diagram of the driving mode identification system for an automatic driving road test according to the present invention. As shown in FIG. 2 , the present invention also provides a driving mode identification system for an automatic driving road test, including: a data acquisition module 1, a threshold determination module 2, Data set determination module 3 , model determination module 4 and mode correction module 5 .

The data acquisition module 1 is used to acquire initial data.

The threshold determination module 2 is used to determine the driving duration threshold.

The data set determination module 3 is configured to obtain a correct driving data set and a data set to be classified according to the initial data and the driving duration threshold.

The model determination module 4 is used to determine a final classification model based on the correct driving data set.

The mode correction module 5 is configured to correct the driving mode corresponding to each to-be-classified data in the to-be-classified data set based on the final classification model.

As an optional implementation manner, the data acquisition module 1 of the present invention includes: a data acquisition unit, a data processing unit, and a data segmentation unit.

The data acquisition unit is used to acquire original data.

The data processing unit is configured to perform data filling and data correction on the original data to obtain repaired data.

The data segmentation unit is configured to segment the repaired data according to the driving mode duration to obtain an automatic driving data set and a manual driving data set; the initial data includes the automatic driving data set and the manual driving data set.

As an optional implementation manner, the threshold determination module 2 of the present invention includes: a parameter unit, a judgment unit, and a threshold determination unit.

The parameter unit is used for selecting parameters to be checked.

The judging unit is used for judging whether the parameter to be tested satisfies both normal distribution and homogeneous variance.

The threshold determination unit is used to obtain the automatic driving duration threshold and the manual driving duration threshold respectively by adopting a parameter inspection method when it is satisfied; when not satisfied, adopting a non-parametric inspection method to obtain the automatic driving duration threshold and the manual driving duration respectively. Threshold; the driving duration threshold includes the automatic driving duration threshold and the manual driving duration threshold.

As an optional implementation manner, the data set determination module 3 of the present invention includes: a correct driving data set determination unit and a data to be classified and determination unit.

The correct driving data set determination unit is configured to use the automatic driving data in the automatic driving data set with a duration greater than or equal to the automatic driving duration threshold as a correct automatic driving data set, and set the manual driving data set with a duration greater than or equal to the automatic driving duration threshold. Or the manual driving data equal to the manual driving duration threshold is used as the correct manual driving data set; the correct driving data set includes the correct automatic driving data set and the correct manual driving data set.

The data to be classified and the determination unit are used to use the automatic driving data in the automatic driving data set whose duration is less than the automatic driving duration threshold as the automatic driving data set to be classified, and the manual driving data set whose duration is less than all the automatic driving data sets. The manual driving data of the manual driving duration threshold is used as the manual driving data set to be classified; the driving data set to be classified includes the automatic driving data set to be classified and the manual driving data set to be classified.

As an optional implementation manner, the model determination module 4 of the present invention includes: a feature data set determination module, a feature data set classification unit, a model construction unit, a training unit, and a model determination unit.

The feature data set determination module is configured to perform feature selection in the correct driving data set to obtain a feature data set.

The feature data set classification unit is configured to select a set proportion of data in the feature data set as a training set, and the rest of the data as a test set.

The model construction unit is used to construct each of the classification models; each of the classification models is respectively a K-nearest neighbor estimation model, a support vector machine model, a decision tree model, a random forest model, and a BP neural network model.

The training unit is configured to train each of the classification models based on the training set, and obtain each of the trained classification models.

The model determining unit is configured to evaluate each of the classification models based on the test set, and select the trained classification model corresponding to the highest evaluation value as the final classification model.

The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part can be referred to the description of the method.

In this paper, specific examples are used to illustrate the principles and implementations of the present invention. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the present invention; meanwhile, for those skilled in the art, according to the present invention There will be changes in the specific implementation and application scope. In conclusion, the contents of this specification should not be construed as limiting the present invention.

Claims (10)

1. An automated driving road test driving pattern recognition method, comprising:

acquiring initial data;

determining a driving duration threshold;

obtaining a correct driving data set and a data set to be classified according to the initial data and the driving duration threshold;

determining a final classification model based on the correct driving data set;

and correcting the driving mode corresponding to each data to be classified in the data set to be classified based on the final classification model.

2. The automated driven road test driving pattern recognition method of claim 1, wherein the obtaining initial data comprises:

acquiring original data;

performing data filling and data correction on the original data to obtain repair data;

segmenting the patching data according to the duration of the driving mode to obtain an automatic driving data set and a manual driving data set; the initial data includes the autonomous driving data set and the manual driving data set.

3. The automated driven road test driving pattern recognition method of claim 1, wherein the determining a driving duration threshold comprises:

selecting parameters to be detected;

judging whether the parameters to be detected simultaneously meet normal distribution and variance uniformity;

if so, respectively obtaining an automatic driving duration threshold value and a manual driving duration threshold value by adopting a parameter inspection method; if not, respectively obtaining an automatic driving duration threshold and a manual driving duration threshold by adopting a nonparametric inspection method; the driving duration threshold includes the automated driving duration threshold and the manual driving duration threshold.

4. The automated driven road test driving pattern recognition method of claim 3, wherein obtaining the correct driving data set and the data set to be classified according to the initial data and the driving duration threshold comprises:

taking the automatic driving data with the duration longer than or equal to the automatic driving duration threshold value in the automatic driving data set as a correct automatic driving data set, and taking the manual driving data with the duration longer than or equal to the manual driving duration threshold value in the manual driving data set as a correct manual driving data set; the correct driving data set comprises the correct automated driving data set and the correct manual driving data set;

taking the automatic driving data with the duration less than the automatic driving duration threshold value in the automatic driving data set as an automatic driving data set to be classified, and taking the manual driving data with the duration less than the manual driving duration threshold value in the manual driving data set as a manual driving data set to be classified; the driving data set to be classified comprises the automatic driving data set to be classified and the manual driving data set to be classified.

5. The automated driven road test driving pattern recognition method of claim 1, wherein the determining a final classification model based on the correct driving data set comprises:

performing feature selection in the correct driving data set to obtain a feature data set;

selecting data with a set proportion from the characteristic data set as a training set, and taking the rest data as a test set;

constructing each classification model; each classification model is a K neighbor estimation model, a support vector machine model, a decision tree model, a random forest model and a BP neural network model;

training each of the classification models based on the training set; obtaining each trained classification model;

and evaluating each classification model based on the test set, and selecting the trained classification model corresponding to the highest evaluation value as a final classification model.

6. An automated driving road test driving pattern recognition system, comprising:

the data acquisition module is used for acquiring initial data;

the threshold value determining module is used for determining a driving duration threshold value;

the data set determining module is used for obtaining a correct driving data set and a data set to be classified according to the initial data and the driving duration threshold;

a model determination module for determining a final classification model based on the correct driving data set;

and the mode correction module is used for correcting the driving mode corresponding to each data to be classified in the data set to be classified based on the final classification model.

7. The automated driving road test driving pattern recognition system of claim 6, wherein the data acquisition module comprises:

a data acquisition unit for acquiring original data;

the data processing unit is used for carrying out data filling and data correction on the original data to obtain repair data;

the data segmentation unit is used for segmenting the repair data according to the duration of the driving mode to obtain an automatic driving data set and a manual driving data set; the initial data includes the autonomous driving data set and the manual driving data set.

8. The automated driven road test driving pattern recognition system of claim 6, wherein the determination threshold determination module comprises:

the parameter unit is used for selecting parameters to be detected;

the judging unit is used for judging whether the parameters to be detected simultaneously meet normal distribution and uniform variance;

the threshold value determining unit is used for respectively obtaining an automatic driving duration threshold value and a manual driving duration threshold value by adopting a parameter checking method when the automatic driving duration threshold value and the manual driving duration threshold value are met; if the automatic driving duration threshold value and the manual driving duration threshold value are not met, respectively obtaining an automatic driving duration threshold value and a manual driving duration threshold value by adopting a nonparametric inspection method; the driving duration threshold includes the automated driving duration threshold and the manual driving duration threshold.

9. The automated driving road test driving pattern recognition system of claim 8, wherein the data set determination module comprises:

a correct driving data set determining unit, configured to use automatic driving data in the automatic driving data set, of which duration is greater than or equal to the automatic driving duration threshold, as a correct automatic driving data set, and use manual driving data in the manual driving data set, of which duration is greater than or equal to the manual driving duration threshold, as a correct manual driving data set; the correct driving data set comprises the correct automated driving data set and the correct manual driving data set;

the data to be classified and determining unit is used for taking the automatic driving data with the duration less than the automatic driving duration threshold value in the automatic driving data set as the automatic driving data set to be classified, and taking the manual driving data with the duration less than the manual driving duration threshold value in the manual driving data set as the manual driving data set to be classified; the driving data set to be classified comprises the automatic driving data set to be classified and the manual driving data set to be classified.

10. The automated driven road test driving pattern recognition system of claim 6, wherein the model determination module comprises:

the characteristic data set determining module is used for performing characteristic selection in the correct driving data set to obtain a characteristic data set;

the characteristic data set classification unit is used for selecting data with a set proportion from the characteristic data set as a training set, and taking the rest data as a test set;

the model construction unit is used for constructing each classification model; each classification model is a K neighbor estimation model, a support vector machine model, a decision tree model, a random forest model and a BP neural network model;

a training unit for training each of the classification models based on the training set; obtaining each trained classification model;

and the model determining unit is used for evaluating each classification model based on the test set and selecting the trained classification model corresponding to the highest evaluation value as the final classification model.

Images