CN115979679A - Method, apparatus and storage medium for testing actual road of automatic driving system - Google Patents

Abstract

The invention relates to the field of automatic driving, and discloses an actual road test method, equipment and storage medium of an automatic driving system. The method includes: acquiring driving environment perception data in real time while the tested vehicle is driving in an automatic driving mode on actual social roads; segmenting the driving period of the tested vehicle to obtain multiple periods, and analyzing the driving environment Segment the perception data, input the driving environment perception data into the driver model respectively, and obtain the anthropomorphic driver control data; input the anthropomorphic driver control data into the vehicle dynamics model respectively, and obtain the first vehicle dynamics model Parameter data: comparing the first vehicle dynamics parameter data corresponding to the same time period with the second vehicle dynamics parameter data to determine the test result of the automatic driving system. This embodiment realizes the actual road test of the automatic driving system, and solves the problem that the intelligent performance of the automatic driving system is difficult to test and evaluate.

Description

Actual road test method, device and storage medium for automatic driving system

technical field

The invention relates to the field of automatic driving, in particular to an actual road test method, device and storage medium of an automatic driving system.

Background technique

Autonomous driving is not only the research frontier in the field of vehicle engineering, but also the development direction of the future automobile industry. It is also an important technical means to solve problems such as traffic safety, energy waste and environmental pollution. As autonomous vehicles enter a new stage of in-depth development, how to comprehensively and accurately test the comprehensive performance of autonomous driving systems is currently a research focus of the entire automotive industry and academia. Compared with traditional vehicle testing, there is no universally recognized solution for testing the functions and performance of autonomous driving products at home and abroad on how to conduct a comprehensive test and evaluation of the unique "intelligence" attributes of autonomous driving vehicles. At present, the most accepted test method for autonomous driving by the international community - the "multi-pillar method" automatic driving test criteria, that is, testing through various methods and methods such as simulation test, field test and actual road test. Among them, the actual road test can be compared It is a good way to verify the intelligent performance of the automatic driving system in various random traffic situations.

The driver model was originally proposed and completed by vehicle dynamics engineers. People call this type of driver model "virtual test driver", which is used for closed-loop testing and simulation, that is, by operating the vehicle at a given or self-set speed Drive along the designated route. Chinese patent 202011312152.7 proposes a test method and device based on a driver model. By triggering and matching the driver model corresponding to other vehicles, the matching other vehicle drives based on its corresponding driver model, so as to assist the automatic driving algorithm of the tested vehicle to achieve accurate testing of the performance of the vehicle under test. However, it does not consider how to test the function and performance of the automatic driving system based on the driver model in the actual road test, and cannot verify the intelligent performance of the automatic driving system in a more complex real environment. Chinese patent 201410055985.8 proposes an auxiliary design system and method for vehicle steering system parameters based on the driver model. The entire test process simulates the human driving process. By changing the steering system parameters, the state response of the car is obtained. Through the state response Analyze and optimize the parameters of the steering system to make the performance of the steering system more in line with human driving characteristics. However, it does not consider the vehicle performance of the automatic driving system, but only focuses on how to test and verify the vehicle steering system based on the driver model. Chinese patent 202210432328.5 proposes a method for generating autonomous driving test scenarios based on personalized driver models, using human driving data as the data source, and building personalized driver models with different styles can effectively improve the authenticity and complexity of test scenarios sex. However, it does not consider how to apply the driver model to the specific link of the actual road test, nor does it give a specific test system and program.

In view of this, the present invention is proposed.

Contents of the invention

In order to solve the above technical problems, the present invention provides an actual road test method, equipment and storage medium of an automatic driving system, which realizes the actual road test of the automatic driving system and solves the problem that the intelligent performance of the automatic driving system is difficult to test and evaluate.

An embodiment of the present invention provides an actual road test method for an automatic driving system, the method comprising:

Real-time acquisition of driving environment perception data while the tested vehicle is driving in automatic driving mode on actual social roads;

Based on the driving environment perception data, the driving period of the vehicle under test is segmented according to different environmental scene types to obtain multiple time periods, wherein different types of environmental scenes correspond to different time periods;

Segmenting the driving environment perception data according to the plurality of time periods, so as to determine sub-driving environment perception data respectively corresponding to each time period;

Input the sub-driving environment perception data corresponding to each time period respectively into the driver model to obtain anthropomorphic driver control data corresponding to each time period;

Input the anthropomorphic driver control data corresponding to each time period into the vehicle dynamics model to obtain the first vehicle dynamics parameter data corresponding to each time period; the first vehicle dynamics data corresponding to the same time period Compare the data of the physical parameters with the data of the second vehicle dynamics parameters to determine the test result of the automatic driving system, wherein the data of the second vehicle dynamics parameters is read from the data transmitted by the CAN bus in the vehicle under test get.

An embodiment of the present invention provides an electronic device, and the electronic device includes:

processor and memory;

The processor is used to execute the steps of the actual road test method for the automatic driving system described in any embodiment by calling the program or instruction stored in the memory.

An embodiment of the present invention provides a computer-readable storage medium, the computer-readable storage medium stores a program or an instruction, and the program or instruction causes a computer to execute the steps of the actual road test method for an automatic driving system described in any embodiment .

Embodiments of the present invention have the following technical effects:

A practical road test method for automatic driving based on a mature driver model is proposed, which can face the challenge of randomness in the actual road test of automatic driving, and fully utilizes the generalization ability of the mature driver model. The anthropomorphic driving behavior is automatically generated, and the test results of the intelligent performance of the automatic driving system are generated through horizontal comparison with the automatic behavior of the automatic driving system, and a reasonable method for testing and evaluating the intelligent performance of the automatic driving system is scientifically proposed. method.

Description of drawings

In order to more clearly illustrate the specific implementation of the present invention or the technical solutions in the prior art, the following will briefly introduce the accompanying drawings that need to be used in the specific implementation or description of the prior art. Obviously, the accompanying drawings in the following description The drawings show some implementations of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any creative work.

Fig. 1 is a flow chart of an actual road test method for an automatic driving system provided by an embodiment of the present invention;

2 is a schematic structural diagram of an actual road test system for an automatic driving system provided by an embodiment of the present invention;

FIG. 3 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below. Apparently, the described embodiments are only some of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

The actual road test method of the automatic driving system provided by the embodiment of the present invention can face the challenge of randomness in the actual road test of automatic driving, fully utilize the generalization ability of the mature driver model, and can automatically generate the actual and random automatic driving test route Anthropomorphic driving behavior, and through horizontal comparison with the automatic behavior of the automatic driving system, the test results of the intelligent performance of the automatic driving system are generated, and a reasonable method for testing and evaluating the intelligent performance of the automatic driving system is scientifically proposed. The actual road test method for the automatic driving system provided in the embodiment of the present invention can be executed by electronic equipment.

Fig. 1 is a flow chart of an actual road test method for an automatic driving system provided by an embodiment of the present invention. Referring to Figure 1, the actual road test method of the automatic driving system specifically includes:

S110. Acquiring driving environment perception data in real time while the vehicle under test is driving in an automatic driving mode on an actual social road.

Wherein, the driving environment perception data includes obstacle information in front of the vehicle under test, vehicle information in front of the vehicle under test, traffic light information, speed limit sign information, road curvature information, position information of the vehicle under test, and horizontal and longitudinal dynamic parameter information , at least one of the relative speed of the front vehicle compared to the measured vehicle and the relative distance of the front vehicle compared to the measured vehicle.

The driving environment perception data is collected by a camera, a GPS/INS integrated navigation system and a millimeter-wave radar sensor in the vehicle under test, and the camera is installed on the inside of the front windshield of the vehicle under test. Specifically, the camera is calibrated and used to perceive information of obstacles in front of the vehicle, information of vehicles in front, traffic light information, speed limit sign information and road curvature information in real time. The GPS/INS integrated navigation system is used for real-time acquisition of vehicle position information and horizontal and vertical dynamic parameter information. The millimeter-wave radar sensor is used to obtain the relative speed and relative distance information of the front vehicle compared with the tested self-driving vehicle.

S120. Based on the driving environment perception data, segment the driving time period of the vehicle under test according to different environmental scene types to obtain multiple time periods, wherein different time periods correspond to different types of environmental scenes.

S130. Segment the driving environment perception data according to the multiple time periods, so as to determine sub-driving environment perception data corresponding to each time period.

Wherein, the environment scene type includes social vehicles turning right, social vehicles turning left, social vehicles following vehicles, social vehicles cutting in front vehicles, social vehicles cutting ahead vehicles, or social vehicles changing lanes.

In other words, the driving history of the tested vehicle in a specific environmental scene is divided into a time period.

By segmenting the driving period of the vehicle under test according to different environmental scene types based on the driving environment perception data, the test scenes experienced in this actual road test are segmented and segmented to obtain different environmental scenarios. The start and end time corresponding to the scene.

By classifying according to environmental scenarios, the performance of the automatic driving system in specific scenarios can be tested, which is conducive to improving the test accuracy and test coverage, and ensuring the comprehensiveness of the test.

S140. Input the sub-driving environment perception data corresponding to each time period into the driver model to obtain anthropomorphic driver control data corresponding to each time period.

Wherein, the anthropomorphic driver control data includes at least one of the opening degree of the accelerator pedal, the opening degree of the brake pedal and the steering wheel angle. Simulate the driving behavior of real drivers in specific environmental scenarios through the driver model, and compare the driving behavior with the automatic driving behavior of the automatic driving system, so as to determine the intelligent performance of the automatic driving system and realize the test of the automatic driving system .

S150. Input the anthropomorphic driver control data respectively corresponding to each time period into the vehicle dynamics model to obtain first vehicle dynamics parameter data corresponding to each time period.

Further, in order to improve the simulation accuracy of driving behavior, and then improve the test accuracy of the automatic driving system, the driver model is improved so that the driver model can obtain anthropomorphic driving with different driving styles based on the driving environment perception data. staff control the data.

Correspondingly, the anthropomorphic driver control data respectively corresponding to each time period includes: anthropomorphic driver control data of different driving styles respectively corresponding to each time period, the different driving styles include conservative, common type and aggressive type.

Said inputting the anthropomorphic driver control data respectively corresponding to each time period into the vehicle dynamics model to obtain the first vehicle dynamics parameter data respectively corresponding to each time period, including:

Input the anthropomorphic driver control data of different driving styles corresponding to each time period into the vehicle dynamics model, and obtain the sub-vehicle dynamics parameter data of different driving styles corresponding to each time period, the different The sub-vehicle dynamic parameter data of the driving style constitute the first vehicle dynamic parameter data. That is, there are three sub-vehicle dynamics parameter data corresponding to a time period, which are the sub-vehicle dynamics parameter data corresponding to the conservative driver, the sub-vehicle dynamics parameter data corresponding to the ordinary driver, and the sub-vehicle dynamics data corresponding to the aggressive driver. parameter data.

Therefore, it is necessary to screen the above three sub-vehicle dynamics parameter data through a certain strategy, and finally retain one of them. Exemplarily, before comparing the first vehicle dynamics parameter data corresponding to the same time period with the second vehicle dynamics parameter data, and determining the test result of the automatic driving system, the method further includes:

Carry out similarity calculation between the sub-vehicle dynamics parameter data of different driving styles corresponding to a period and the second vehicle dynamics parameter data corresponding to the period, and determine the sub-vehicle dynamics parameter data with the largest similarity as the The first vehicle dynamics parameter data corresponding to a period of time.

S160. Comparing the first vehicle dynamics parameter data corresponding to the same time period with the second vehicle dynamics parameter data, and determining the test result of the automatic driving system.

Wherein, the second vehicle dynamics parameter data is obtained by reading the data transmitted by the CAN bus in the vehicle under test.

Exemplarily, the comparing the first vehicle dynamics parameter data corresponding to the same time period with the second vehicle dynamics parameter data to determine the test result of the automatic driving system includes:

Comparing the first vehicle dynamics parameter data corresponding to the same time period with the second vehicle dynamics parameter data one by one, and determining the similarity of each index based on the root mean square error;

Carrying out weighted summation of the similarities of different indicators to obtain the test result;

Both the first vehicle dynamics parameter data and the second vehicle dynamics parameter data include at least one index of longitudinal acceleration, longitudinal jerk, lateral acceleration, lateral jerk, yaw angle and yaw rate .

This embodiment has the following technical effects: a method for actual road testing of automatic driving based on a mature driver model is proposed, which can face the challenge of randomness in actual road testing of automatic driving, fully utilizes the generalization ability of a mature driver model, and can be aimed at actual , random automatic driving test route, automatically generate anthropomorphic driving behavior, and generate test results for the intelligent performance of the automatic driving system through horizontal comparison with the automatic behavior of the automatic driving system, and scientifically propose a method for testing A reasonable way to evaluate the intelligent performance of autonomous driving.

On the basis of the technical solutions of the above embodiments, the embodiments of the present invention also provide a solution for obtaining the driver model, with the purpose of obtaining the driving behavior of drivers who can accurately simulate different driving styles in different environmental scenarios.

Specifically, obtain P groups of driver model training data sets; each set of driver model training data sets includes driving environment perception training data, driver control training data, and driving style evaluation information; at least part of the driver model training The driving style evaluation information in the data set is different;

Based on the P group of driver model training data sets, a generalized regression neural network (Generalized regression neural network, GRNN) is trained to obtain the driver model.

After the acquisition of the P group of driver model training data sets, before training the generalized regression neural network (Generalized regression neural network, GRNN) based on the P group of driver model training data sets, it also includes:

Based on the driving environment perception training data and/or driver control training data in each set of driver model training data, determine the values of N characteristic parameters corresponding to each set of driver model training data;

Construct N-dimensional space, take the values of the N characteristic parameters as coordinate values, and determine the corresponding points of each group of driver model training initial data in N-dimensional space;

Use the K-means clustering analysis algorithm to cluster P corresponding points in the N-dimensional space, and obtain M clustering results;

Determining a representative point for each clustering result, the representative point is a corresponding point in the clustering result to which it belongs, and the distance between each representative point and the cluster center of the clustering result to which it belongs is greater than the first set Set a threshold, and the distance between each representative point and the cluster center of other clustering results is greater than the second set threshold;

Correcting the driving style evaluation information in each set of driver model training data, so that the driving style evaluation information in each set of driver model training data after correction is the same as that of representatives belonging to the same clustering result. The driving style evaluation information of the points is consistent.

Before determining the values of N characteristic parameters corresponding to each set of driver model training data based on the driving environment awareness training data and/or driver control training data in each set of driver model training data, Also includes:

In the alternative parameters, principal component analysis is carried out to determine N characteristic parameters; the alternative parameters include average vehicle speed, standard deviation of vehicle speed, maximum value of vehicle speed, average value of longitudinal acceleration, standard deviation of longitudinal acceleration, maximum value of longitudinal acceleration, Average value of lateral acceleration, standard deviation of lateral acceleration, maximum value of lateral acceleration, average value of steering wheel angle, standard deviation of steering wheel angle, and maximum value of steering wheel angle.

Further, refer to the schematic structural diagram of an actual road test system for an automatic driving system as shown in FIG. Module 240, automatic driving test scene segmentation processing module 250, mature driver model 260, vehicle dynamics model 270, vehicle dynamics parameter similarity comparison module 280, automatic driving system intelligent performance test result display module 290, driving style classification A discrimination module 300 and a mature driver's natural driving behavior database 310 .

Wherein, the forward-facing test camera 210 is used for real-time perception of obstacle information in front of the vehicle, vehicle information in front, traffic light information, speed limit sign information and road curvature information. The GPS/INS integrated navigation system 220 is used to obtain the position information of the vehicle and the information of the horizontal and vertical dynamic parameters in real time. The millimeter-wave radar sensor 230 is used to obtain information about the relative speed and distance of the vehicle in front compared to the self-driving vehicle under test. The vehicle CAN bus data reading module 240 is used to read the second vehicle dynamics parameter data of the automatic driving system.

The automatic driving test scene segmentation processing module 250 is used to segment the test scene experienced in this actual road test, that is, based on the driving environment perception data, according to different environmental scene types, the Segment the driving time period to obtain multiple time periods, wherein different types of environmental scenes correspond to different time periods, and divide the driving environment perception data according to the multiple time periods to determine the sub-driving environment corresponding to each time period sense data.

The mature driver model 260 is used to obtain anthropomorphic driver control data corresponding to each time period based on the sub-driving environment perception data corresponding to each time period.

The vehicle dynamics model 270 is used to obtain the first vehicle dynamics parameter data respectively corresponding to each time period based on the anthropomorphic driver control data respectively corresponding to each time period.

The vehicle dynamics parameter similarity comparison module 280 is used to compare the first vehicle dynamics parameter data corresponding to the same period of time with the second vehicle dynamics parameter data to determine the test result of the automatic driving system.

The automatic driving system intelligent performance test result display module 290 is used to display the test result.

The driving style classification and discrimination module 300 is used to cluster and analyze a large number of mature drivers' driving behavior data, and form driving style classifications for different scenarios, namely conservative, common and aggressive, and label them at the same time . Specifically, Principal Component Analysis (PCA) is used to reduce the dimensionality of high-dimensional data features. Since the dimensions of each feature parameter are different, the magnitude of data varies greatly, and data with large magnitudes will cover up the data reflected by small magnitudes. Therefore, the Z-score standardization method is firstly used to process the data, so that the data of each dimension can be scaled to the same area in proportion to eliminate the dimensional influence, and then PCA processing is performed on the standardized data set. When the cumulative variance contribution rate of the principal components is close to 1 (usually 85%), the first m feature variables are used instead of the original p variables for comprehensive analysis, which not only simplifies the calculation steps, but also retains the information of the original features. Subsequently, the K-means clustering analysis algorithm is used to identify the specific driving style of mature drivers. The K-means clustering analysis algorithm selects the input quantity k and divides n data objects into k clusters according to the principle of similarity. After clustering Satisfied: The similarity of data objects in the same cluster is high, while the similarity of data objects in different clusters is low.

The mature driver's natural driving behavior database 310 stores a large amount of mature driver's natural driving behavior data. Specifically, the age of the driver is between 20 and 50 years old, and the data collection frequency is 10 Hz. There are 170 male drivers and 30 female drivers. In order to meet the diversity of road conditions, the selected routes involve highways, urban expressways, national highways and country roads, including bends, straights and ramps. In addition, the data also includes the evaluation results of each driver's overall driving style by 3 professional driving evaluators. The driving evaluators are all professional drivers with more than 20 years of driving experience. During the evaluation process, each driver drives the vehicle on a given route according to his usual driving habits. Three professional driving evaluators sit in the car at the same time and through intensive observation The driving style is evaluated by the driver's operation (such as the use of the throttle pedal and the brake pedal) and the subjective feelings of the driving evaluator (driver emotion, ride comfort). The evaluation results are divided into three categories: conservative, ordinary and aggressive. The driving evaluator evaluates the classification labels on the driver's driving style.

At present, there is no uniform standard for the characteristic parameters that represent driving style attributes. Many scholars at home and abroad have selected speed, acceleration, throttle pedal position, and throttle pedal pressure as representative characteristic parameters to evaluate driving style, and the effect is good. This patent selects the vehicle speed, longitudinal acceleration, lateral acceleration and steering wheel angle of 200 mature drivers on the basis of previous research. This paper uses the original data with a frequency of 10Hz to calculate the average value (Mean) of 4 vehicle parameters per second. , standard deviation (Std), and maximum value (Max) are three types of statistics, forming a total of 12-dimensional feature parameters, which provide data samples for the subsequent driving style classification and discrimination module.

FIG. 3 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. As shown in FIG. 3 , an electronic device 400 includes one or more processors 401 and a memory 402 .

The processor 401 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 400 to perform desired functions.

Memory 402 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and/or a cache memory (cache). The non-volatile memory may include, for example, a read-only memory (ROM), a hard disk, a flash memory, and the like. One or more computer program instructions can be stored on the computer-readable storage medium, and the processor 401 can run the program instructions to realize the actual road test method of the automatic driving system in any embodiment of the present invention described above and / or other desired functionality. Various contents such as initial external parameters and thresholds may also be stored in the computer-readable storage medium.

In an example, the electronic device 400 may further include: an input device 403 and an output device 404, and these components are interconnected through a bus system and/or other forms of connection mechanisms (not shown). The input device 403 may include, for example, a keyboard, a mouse, and the like. The output device 404 can output various information to the outside, including early warning prompt information, braking force and so on. The output device 404 may include, for example, a display, a speaker, a printer, a communication network and remote output devices connected thereto, and the like.

Of course, for simplicity, only some components related to the present invention in the electronic device 400 are shown in FIG. 3 , and components such as bus, input/output interface, etc. are omitted. In addition, according to specific application conditions, the electronic device 400 may also include any other appropriate components.

In addition to the above-mentioned methods and devices, embodiments of the present invention may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the functions provided by any of the embodiments of the present invention. Steps in the method of practical road testing of automated driving systems.

The computer program product can be written in any combination of one or more programming languages for executing the program codes for the operations of the embodiments of the present invention, and the programming languages include object-oriented programming languages, such as Java, C++, etc. , also includes conventional procedural programming languages, such as the "C" language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server to execute.

In addition, the embodiments of the present invention may also be a computer-readable storage medium, on which computer program instructions are stored, and the computer program instructions, when executed by a processor, cause the processor to perform the automatic operation provided by any embodiment of the present invention. Steps in the actual road test method of the driving system.

The computer readable storage medium may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, but not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or devices, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: electrical connection with one or more conductors, portable disk, hard disk, random access memory (RAM), read only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

It should be noted that the terms used in the present invention are only used to describe specific embodiments, but not to limit the scope of the application. As shown in the specification and claims of the present invention, words such as "a", "an", "an" and/or "the" do not specifically refer to the singular, and may also include the plural, unless the context clearly indicates an exception. The term "comprising", "comprising" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method or apparatus comprising a set of elements includes not only those elements but also other elements not expressly listed, Alternatively, elements inherent in such a process, method, or apparatus are also included. Without further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method or apparatus comprising said element.

It should also be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. indicate the orientation or positional relationship Based on the orientation or positional relationship shown in the drawings, it is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be construed as a limitation of the invention. Unless otherwise clearly specified and limited, the terms "mounted", "connected", "connected" and so on should be interpreted in a broad sense, for example, it can be fixed connection, detachable connection, or integral connection; it can be mechanical connection , can also be an electrical connection; it can be a direct connection, or an indirect connection through an intermediary, or an internal connection between two components. Those of ordinary skill in the art can understand the specific meanings of the above terms in the present invention in specific situations.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: It is still possible to modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present invention.

Claims (10)

1. An actual road testing method for an automatic driving system, which is integrated in a vehicle to be tested, comprises the following steps:

the method comprises the steps that driving environment perception data are obtained in real time in the process that a detected vehicle runs in an automatic driving mode on an actual social road;

based on the driving environment perception data, segmenting the running time period of the detected vehicle according to different environment scene types to obtain a plurality of time periods, wherein the environment scene types corresponding to the different time periods are different;

segmenting the driving environment perception data according to the plurality of time periods to determine sub-driving environment perception data respectively corresponding to each time period;

respectively inputting the sub-driving environment perception data respectively corresponding to each time interval into a driver model to obtain anthropomorphic driver control data respectively corresponding to each time interval;

inputting the anthropomorphic driver control data respectively corresponding to each time interval into a vehicle dynamics model respectively to obtain first vehicle dynamics parameter data respectively corresponding to each time interval;

and comparing the first vehicle dynamics parameter data corresponding to the same time period with the second vehicle dynamics parameter data to determine the test result of the automatic driving system, wherein the second vehicle dynamics parameter data is obtained by reading data transmitted by a CAN bus in the tested vehicle.

2. The method according to claim 1, wherein the driving environment perception data is acquired by a camera, a GPS/INS integrated navigation system and a millimeter wave radar sensor in the tested vehicle, and the camera is installed on the inner side of a front windshield of the tested vehicle;

the driving environment perception data comprise at least one of obstacle information in front of the detected vehicle, vehicle information in front of the detected vehicle, traffic light information, speed limit sign information, road curvature information, position information of the detected vehicle, transverse and longitudinal dynamic parameter information, relative speed of the front vehicle compared with the detected vehicle and relative distance of the front vehicle compared with the detected vehicle;

the anthropomorphic driver control data comprises at least one of an accelerator pedal opening, a brake pedal opening, and a steering wheel angle;

the environmental scene types comprise a right turn of the social vehicle, a left turn of the social vehicle, following driving of the social vehicle, front vehicle cut-in of the social vehicle, front vehicle cut-out of the social vehicle or lane change of the social vehicle.

3. The method according to claim 1, characterized in that the driver model is capable of deriving anthropomorphic driver control data of different driving styles based on driving environment perception data;

the personified driver control data corresponding to each time period respectively includes: anthropomorphic driver control data of different driving styles respectively corresponding to each time period, the different driving styles including conservative, ordinary and aggressive;

the step of inputting the anthropomorphic driver control data corresponding to each time interval into the vehicle dynamics model respectively to obtain first vehicle dynamics parameter data corresponding to each time interval respectively comprises the following steps:

inputting the anthropomorphic driver control data of different driving styles corresponding to each time interval into a vehicle dynamics model respectively to obtain sub-vehicle dynamics parameter data of different driving styles corresponding to each time interval respectively, wherein the sub-vehicle dynamics parameter data of different driving styles form the first vehicle dynamics parameter data.

4. The method of claim 3, wherein prior to comparing the first vehicle dynamics parameter data and the second vehicle dynamics parameter data corresponding to the same time period to determine the test result for the autopilot system, the method further comprises:

and respectively carrying out similarity calculation on the sub-vehicle dynamics parameter data of different driving styles corresponding to a period of time and the second vehicle dynamics parameter data corresponding to the period of time, and determining the sub-vehicle dynamics parameter data with the maximum similarity as the first vehicle dynamics parameter data corresponding to the period of time.

5. The method of claim 1, wherein comparing the first vehicle dynamics parameter data and the second vehicle dynamics parameter data corresponding to the same time period to determine the test result of the autopilot system comprises:

comparing the first vehicle dynamics parameter data and the second vehicle dynamics parameter data corresponding to the same time period one by one, and determining the similarity of each index based on the root mean square error;

carrying out weighted summation on the similarity of different indexes to obtain the test result;

the first vehicle dynamics parameter data and the second vehicle dynamics parameter data each include at least one indicator of longitudinal acceleration, longitudinal jerk, lateral acceleration, lateral jerk, yaw angle, and yaw rate.

6. The method of claim 1, further comprising:

acquiring a P group of driver model training data sets; each set of the driver model training data set comprises driving environment perception training data, driver control training data and driving style evaluation information; driving style evaluation information in at least part of the driver model training data sets is different;

and training a Generalized Regression Neural Network (GRNN) based on the P groups of driver model training data groups to obtain the driver model.

7. The method of claim 6, wherein after the obtaining the P sets of driver model training data sets and before training a Generalized Recurrent Neural Network (GRNN) based on the P sets of driver model training data sets, further comprising:

determining values of N characteristic parameters corresponding to each group of driver model training data based on driving environment perception training data and/or driver control training data in each group of driver model training data;

constructing an N-dimensional space, and determining corresponding points of each group of driver model training initial data in the N-dimensional space by taking the values of the N characteristic parameters as coordinate values;

clustering P corresponding points in the N-dimensional space by using a K-means clustering analysis algorithm to obtain M clustering results;

determining a representative point for each clustering result, wherein the representative point is a corresponding point in the clustering result to which the representative point belongs, the distance between each representative point and the clustering center of the clustering result to which the representative point belongs is greater than a first set threshold, and the distance between each representative point and the clustering centers of other clustering results is greater than a second set threshold;

and correcting the driving style evaluation information in each group of the driver model training data so that the driving style evaluation information in each group of the corrected driver model training data is consistent with the driving style evaluation information of the representative points belonging to the same clustering result.

8. The method according to claim 7, wherein before determining values of the N characteristic parameters corresponding to each set of driver model training data based on the driving environment perception training data and/or the driver control training data in each set of driver model training data, further comprising:

performing principal component analysis in the alternative parameters to determine N characteristic parameters; the alternative parameters comprise a vehicle speed average value, a vehicle speed standard deviation, a vehicle speed maximum value, a longitudinal acceleration average value, a longitudinal acceleration standard deviation, a longitudinal acceleration maximum value, a transverse acceleration average value, a transverse acceleration standard deviation, a transverse acceleration maximum value, a steering wheel corner average value, a steering wheel corner standard deviation and a steering wheel corner maximum value.

9. An electronic device, characterized in that the electronic device comprises:

a processor and a memory;

the processor is adapted to perform the steps of the autopilot system actual road testing method of any one of claims 1 to 8 by invoking programs or instructions stored by the memory.

10. A computer-readable storage medium characterized in that it stores a program or instructions for causing a computer to execute the steps of the automatic driving system actual road test method according to any one of claims 1 to 8.

Images