A data-driven testing method and system based on intelligent driving

Abstract

The present invention relates to a data-driven testing method and system based on intelligent driving, belonging to the field of intelligent driving technology. The method includes: obtaining point cloud data, speed data, position data, and acceleration data through vehicle-mounted sensors; presetting a vehicle trajectory prediction model and extracting model input data categories, performing balanced processing on the data obtained by the vehicle-mounted sensors based on nonlinear multi-category responses according to the model input data categories and constructing a vehicle power data set, optimizing the vehicle trajectory prediction model through vehicle power data set optimization training; constructing a vehicle motion model, constructing test cases based on the vehicle trajectory prediction model, applying the test cases to the vehicle motion model for test evaluation to obtain test results, and optimizing the vehicle motion model and the vehicle trajectory prediction model according to the test results. The problem of data imbalance in test data collection is solved, and the representativeness and effectiveness of the test cases are improved by dynamically setting the test cases.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent driving, and in particular relates to a data-driven testing method and system based on intelligent driving. Background Art

[0002] With the rapid development of computing power, sensor technology and artificial intelligence, autonomous driving technology has developed rapidly. Intelligent driving cars perceive the surrounding environment through sensors such as lidar, cameras, radars, and use advanced algorithms for environmental perception, path planning and vehicle control to achieve autonomous driving. The simulation environment based on real data can more accurately simulate real road scenes, providing richer possibilities for the testing and evaluation of autonomous driving systems; through the collection, annotation and analysis of a large amount of real data, the performance of intelligent systems can be effectively trained and evaluated. In response to the testing needs of intelligent driving systems, some intelligent testing platforms and tools are also emerging, including software and hardware equipment and platforms for data collection, annotation, simulation, and test analysis. In summary, the data-driven testing method based on intelligent driving mainly involves the development of autonomous driving technology, data-driven methodology, sensor technology, simulation technology, and intelligent testing platforms. The progress and integration of these technologies provide technical support and prospects for the application of data-driven testing methods based on intelligent driving. At present, there are still areas for improvement:

[0003] (1) The data collected by intelligent driving vehicles in real road environments is indeed huge in scale and highly complex in dimension, which brings challenges to data processing and analysis. Traditional statistical methods are difficult to effectively solve problems such as high-dimensional data, the need for real-time performance, and nonlinear relationships. At the same time, data labeling and processing require a lot of manpower and time costs, especially in complex scenarios and details. In addition, the data during the test process may not cover a sufficiently diverse range of scenarios and situations, nor can it overcome the limitations of the data;

[0004] (2) When applying real data to simulation tests, the authenticity of the simulation environment and the reliability of the data, as well as whether the simulation environment can accurately simulate the real road environment, cannot be accurately guaranteed. Comprehensively evaluating the performance and safety of the autonomous driving system based on the collected data only relies on set fixed scenarios and fixed modes. The robustness test of the autonomous driving system requires more accurate simulation data and more complex and extreme test cases for evaluation. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention provides a data-driven testing method and system based on intelligent driving;

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] S1: Setting a vehicle-mounted sensor for intelligent driving data collection, wherein the vehicle-mounted sensor includes a laser radar, a millimeter-wave radar, a positioning unit, and an inertial measurement unit, wherein the laser radar is used to obtain point cloud data of the driving vehicle, the millimeter-wave radar is used to obtain speed data of the driving vehicle, the positioning unit is used to obtain position data of the driving vehicle, and the inertial measurement unit is used to obtain acceleration data of the driving vehicle;

[0008] S2: presetting a vehicle trajectory prediction model and extracting model input data categories, performing equalization processing on the point cloud data, the speed data, the position data, and the acceleration data based on the model input data categories based on a nonlinear multi-category response to obtain a sparse vector based on the input category, constructing a vehicle power data set based on the sparse vector, and optimizing the vehicle trajectory prediction model through optimization training of the vehicle power data set;

[0009] S3: construct a vehicle motion model according to the mechanism state parameters of the intelligent driving vehicle, construct a test case based on the vehicle motion model and the vehicle trajectory prediction model, and apply the test case to the vehicle motion model for test evaluation to obtain a test result.

[0010] Specifically, the equalization processing method is:

[0011] S201: using the input data of the vehicle trajectory prediction model as target data, and using the point cloud data, the speed data, the position data, and the acceleration data as auxiliary data;

[0012] S202: Preset the filtering threshold and the samples of each category in the target data, and calculate the dependency of the auxiliary data corresponding to the target data category. The calculation formula is:

[0013]

[0014] in, is the dependency of auxiliary data on target data in the th target data sample, h1 and h2 represent the h1th and h2th categories of target data, H is the total number of target data categories, is the estimated probability that the auxiliary data corresponds to the target data category h1, is the estimated probability that the auxiliary data corresponds to the target data category h2;

[0015] S203: Calculate the average dependency of the auxiliary data on the target data according to the dependency and samples of each category in the target data, and the calculation formula is:

[0016]

[0017] Among them, l is the sample count, L is the total sample size, and A * (X j , Y) is the auxiliary data X j Corresponding to the dependency of target data Y, is the auxiliary data X in the lth sample j Corresponding dependency on target data Y;

[0018] S204: Arrange the auxiliary data in descending order according to the average dependency to obtain a prediction variable queue, extract components of the prediction variable queue according to the filtering threshold and form a sparse vector corresponding to the target data category.

[0019] Specifically, the vehicle trajectory prediction model predicts the test vehicle trajectory data based on LSTM, and the model calculation formula is:

[0020] i t =σ(W ii x t +b ii +W hi h t-1 +b hi ),

[0021] f t =σ(W if x t +b if +W hf h t-1 +b hf ),

[0022] g t =tanh(W ig x t +b ig +W hg h t-1 +b hg ),

[0023] o t =σ(W io x t +b io +W ho h t-1 +b ho ),

[0024] c t =f t ⊙c t-1 +i t ⊙g t ,

[0025] h t =o t ⊙tanh(ct ),

[0026] Among them, i t is the gate value of the input gate, f t is the gate value of the forget gate, g t To update the gate value of the gate, o t is the gate value of the output gate, h t is the trajectory prediction data, h t-1 is the trajectory prediction data under the t-1 step length, σ is the activation function, tanh is the nonlinear activation function, c t is the cell state in the LSTM network, c t-1 is the cell state at the previous moment, ⊙ represents element-by-element multiplication, W ii is the weight value of different input gate connections, b ii is the bias value between different input gates, W hi is the weight value of the trajectory prediction data and the input gate in the previous calculation step, b hi is the bias value of the trajectory prediction data and the input gate in the previous calculation step, W if is the weight value between the input gate and the forget gate, b if is the bias value between the input gate and the forget gate, W hf is the weight value of the forget gate and the model output, b hf is the bias value between the forget gate and the model output, W ig is the weight value between the input gate and the update gate, b ig is the bias value between the input gate and the update gate, W hg To update the weight value of the gate and model output, b hg To update the bias value of the gate and model output, W io is the weight value between the input gate and the output gate, b io is the bias value between the input gate and the output gate, W ho is the weight value of the output gate and the model output, b ho is the bias value between the output gate and the model output, x t The vehicle location data.

[0027] Specifically, the mechanism state parameters include: speed, acceleration, steering angle, steering rate, vehicle position data, and vehicle wheelbase.

[0028] Specifically, the vehicle motion model describes the motion of the vehicle based on a two-wheel model, and the calculation formula is:

[0029]

[0030] in, Respectively represent the speed change rate of the vehicle on the lateral axis and longitudinal axis, To represent the angular velocity of the vehicle, is the speed change rate of the vehicle, v is the speed of the vehicle, θ is the heading angle of the vehicle, L is the wheelbase of the vehicle, a is the acceleration of the vehicle, and δ is the steering angle of the vehicle.

[0031] A data-driven test system based on intelligent driving, comprising: a data acquisition module, a trajectory prediction module, and a test optimization module;

[0032] The data acquisition module is used to set the vehicle-mounted sensors for intelligent driving data acquisition, and the vehicle-mounted sensors include laser radar, millimeter wave radar, positioning unit, and inertial measurement unit. The laser radar is used to obtain the point cloud data of the driving vehicle, the millimeter wave radar is used to obtain the speed data of the driving vehicle, the positioning unit is used to obtain the position data of the driving vehicle, and the inertial measurement unit is used to obtain the acceleration data of the driving vehicle;

[0033] The trajectory prediction module is used to preset a vehicle trajectory prediction model and extract model input data categories, perform equalization processing on the point cloud data, the speed data, the position data, and the acceleration data based on the model input data categories based on a nonlinear multi-category response to obtain a sparse vector based on the input category, construct a vehicle dynamics data set based on the sparse vector, and optimize the vehicle trajectory prediction model through optimization training of the vehicle dynamics data set;

[0034] The test optimization module is used to build a vehicle motion model according to the mechanism state parameters of the intelligent driving vehicle, build a test case based on the vehicle motion model and the vehicle trajectory prediction model, apply the test case to the vehicle motion model for test evaluation to obtain test results, and optimize and adjust the calculation method of the vehicle motion model and the vehicle trajectory prediction model according to the test results.

[0035] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements a data-driven testing method based on intelligent driving when executing the computer program.

[0036] A storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, a data-driven testing method based on intelligent driving is implemented.

[0037] The beneficial effects of the present invention are:

[0038] (1) By balancing the large amount of collected data and migrating the collected high-dimensional data to a binary response scenario based on the vehicle trajectory prediction model, the correlation between the collected data and the data calculated by the prediction model is ensured. When faced with massive data, there is no need to process the data through manual labeling, which greatly saves computing costs. By integrating target data and auxiliary data, the problem of unbalanced collected data and undersampling is overcome.

[0039] (2) The vehicle trajectory prediction model is optimized through real-time data collection, and the vehicle motion model is combined to obtain the vehicle's predicted trajectory in the medium and long time domains. The vehicle kinematic model is used to obtain the vehicle's short-term predicted trajectory. The predicted trajectory sequence is weighted in different time domain segments to achieve environmental vehicle trajectory prediction with high accuracy and high robustness. Test cases are dynamically set according to the trajectory prediction results. There is no need to set the test mode based on prior knowledge. The test cases can be dynamically adjusted according to the latest trajectory prediction results to timely reflect the performance of the model in the current situation. The test cases are adjusted according to different prediction situations to cover more edge cases and special situations and discover the limitations of the model. It can better adapt to actual road and vehicle driving conditions and improve the representativeness and effectiveness of test cases. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0041] Figure 1 The present invention is a flow chart of a data-driven testing method and system based on intelligent driving. DETAILED DESCRIPTION

[0042] The example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein: on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art. In addition, the described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to give a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, well-known methods, devices, implementations, or operations are not shown or described in detail to avoid blurring various aspects of the present disclosure. The blocks shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices. The flowcharts shown in the accompanying drawings are only exemplary illustrations and do not necessarily include all the content and operations / steps, nor do they necessarily have to be executed in the order described. For example, some operations / steps may be decomposed, while some operations / steps may be combined or partially combined, so the actual execution order may change according to the actual situation.

[0043] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0044] See also Figure 1 , a data-driven testing method and system based on intelligent driving:

[0045] S1: Setting a vehicle-mounted sensor for intelligent driving data collection, wherein the vehicle-mounted sensor includes a laser radar, a millimeter-wave radar, a positioning unit, and an inertial measurement unit, wherein the laser radar is used to obtain point cloud data of the driving vehicle, the millimeter-wave radar is used to obtain speed data of the driving vehicle, the positioning unit is used to obtain position data of the driving vehicle, and the inertial measurement unit is used to obtain acceleration data of the driving vehicle;

[0046] S2: presetting a vehicle trajectory prediction model and extracting model input data categories, performing equalization processing on the point cloud data, the speed data, the position data, and the acceleration data based on the model input data categories based on a nonlinear multi-category response to obtain a sparse vector based on the input category, constructing a vehicle power data set based on the sparse vector, and optimizing the vehicle trajectory prediction model through optimization training of the vehicle power data set;

[0047] S3: construct a vehicle motion model according to the mechanism state parameters of the intelligent driving vehicle, construct a test case based on the vehicle motion model and the vehicle trajectory prediction model, and apply the test case to the vehicle motion model for test evaluation to obtain a test result.

[0048] Specifically, the equalization processing method is:

[0049] S201: using the input data of the vehicle trajectory prediction model as target data, and using the point cloud data, the speed data, the position data, and the acceleration data as auxiliary data;

[0050] S202: Preset the filtering threshold and the samples of each category in the target data, and calculate the dependency of the auxiliary data corresponding to the target data category. The calculation formula is:

[0051]

[0052] in, is the dependency of the auxiliary data on the target data in the lth target data sample, h1 and h2 represent the h1th and h2th categories of the target data, and H is the total number of categories of the target data. is the estimated probability that the auxiliary data corresponds to the target data category h1, is the estimated probability that the auxiliary data corresponds to the target data category h2;

[0053] S203: Calculate the average dependency of the auxiliary data on the target data according to the dependency and samples of each category in the target data, and the calculation formula is:

[0054]

[0055] Among them, l is the sample count, L is the total sample size, and A * (X j , Y) is the auxiliary data X j Corresponding to the dependency of target data Y, is the auxiliary data X in the lth sample j Corresponding dependency on target data Y;

[0056] S204: Arrange the auxiliary data in descending order according to the average dependency to obtain a prediction variable queue, extract components of the prediction variable queue according to the filtering threshold and form a sparse vector corresponding to the target data category.

[0057] In this embodiment, the category labels are firstly one-hot encoded, and then balanced processing is performed for each category so that the number of samples in each category reaches the number of the maximum category. Finally, the processed data is converted into a sparse vector.

[0058] Specifically, the vehicle trajectory prediction model predicts the test vehicle trajectory data based on LSTM, and the model calculation formula is:

[0059] i t =σ(W ii x t +b ii +W hi h t-1 +b hi ),

[0060] f t =σ(W if x t +b if +W hf h t-1 +b hf ),

[0061] g t =tanh(W ig x t +b ig +W hg h t-1 +b hg ),

[0062] o t =σ(W io x t +b io +W ho h t-1 +b ho ),

[0063] c t =f t ⊙c t-1 +i t ⊙g t ,

[0064] h t =o t ⊙tanh(c t ),

[0065] Among them, i t is the gate value of the input gate, f t is the gate value of the forget gate, g t To update the gate value, o t is the gate value of the output gate, h t is the trajectory prediction data, ht-1 is the trajectory prediction data under the t-1 step length, σ is the activation function, tanh is the nonlinear activation function, c t is the cell state in the LSTM network, c t-1 is the cell state at the previous moment, ⊙ represents element-by-element multiplication, W ii is the weight value of different input gate connections, b ii is the bias value between different input gates, W hi is the weight value of the trajectory prediction data and the input gate in the previous calculation step, b hi is the bias value of the trajectory prediction data and the input gate in the previous calculation step, W if is the weight value between the input gate and the forget gate, b if is the bias value between the input gate and the forget gate, W hf is the weight value of the forget gate and the model output, b hf is the bias value between the forget gate and the model output, W ig is the weight value between the input gate and the update gate, b ig is the bias value between the input gate and the update gate, W hg To update the weight value of the gate and model output, b hg To update the bias value of the gate and model output, W io is the weight value between the input gate and the output gate, b io is the bias value between the input gate and the output gate, W ho is the weight value of the output gate and the model output, b ho is the bias value between the output gate and the model output, x t The vehicle location data.

[0066] In this embodiment, by using historical vehicle trajectory data as input, the Keras interface of TensorFlow is used to build an LSTM model, create a Sequential model, and add an LSTM layer and a fully connected layer to it; the model is configured by adjusting the number of neurons, activation function and other parameters of the LSTM layer, and the mean square error is used as the loss function for training; finally, the model predicts the new vehicle trajectory data and obtains the prediction result. The trajectory prediction model initially set uses the trajectory data of the public data set as the training set required for the training model, and then randomly selects continuous trajectory sequences under a variety of different working conditions as the verification set, and sets different prediction time domain ranges. There are many measurable states in the actual driving process of the vehicle; in terms of trajectory prediction, it is only necessary to consider the vehicle as a rigid body based on the kinematic model, so the longitudinal displacement, lateral displacement and speed are predicted, and the root mean square error and mean absolute error of the three predicted output values ​​are statistically compared with the true value.

[0067] Specifically, the mechanism state parameters include: speed, acceleration, steering angle, steering rate, vehicle position data, and vehicle wheelbase.

[0068] Specifically, the vehicle motion model describes the motion of the vehicle based on a two-wheel model, and the calculation formula is:

[0069]

[0070] in, Respectively represent the speed change rate of the vehicle on the lateral axis and longitudinal axis, To represent the angular velocity of the vehicle, is the speed change rate of the vehicle, v is the speed of the vehicle, θ is the heading angle of the vehicle, L is the wheelbase of the vehicle, a is the acceleration of the vehicle, and δ is the steering angle of the vehicle.

[0071] A data-driven test system based on intelligent driving, comprising: a data acquisition module, a trajectory prediction module, and a test optimization module;

[0072] The data acquisition module is used to set the vehicle-mounted sensors for intelligent driving data acquisition, and the vehicle-mounted sensors include laser radar, millimeter wave radar, positioning unit, and inertial measurement unit. The laser radar is used to obtain the point cloud data of the driving vehicle, the millimeter wave radar is used to obtain the speed data of the driving vehicle, the positioning unit is used to obtain the position data of the driving vehicle, and the inertial measurement unit is used to obtain the acceleration data of the driving vehicle;

[0073] The trajectory prediction module is used to preset a vehicle trajectory prediction model and extract model input data categories, perform equalization processing on the point cloud data, the speed data, the position data, and the acceleration data based on the model input data categories based on a nonlinear multi-category response to obtain a sparse vector based on the input category, construct a vehicle dynamics data set based on the sparse vector, and optimize the vehicle trajectory prediction model through optimization training of the vehicle dynamics data set;

[0074] The test optimization module is used to build a vehicle motion model according to the mechanism state parameters of the intelligent driving vehicle, build a test case based on the vehicle motion model and the vehicle trajectory prediction model, apply the test case to the vehicle motion model for test evaluation to obtain test results, and optimize and adjust the calculation method of the vehicle motion model and the vehicle trajectory prediction model according to the test results.

[0075] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements a data-driven testing method based on intelligent driving when executing the computer program.

[0076] A storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, a data-driven testing method based on intelligent driving is implemented.

[0077] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device.

[0078] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0079] The program code included on the computer readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or their combinations, and the programming language includes object-oriented programming languages-such as Java, Smalltalk, C++, and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0080] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A data-driven testing method based on intelligent driving, characterized in that: include: S1: Setting a vehicle-mounted sensor for intelligent driving data collection, wherein the vehicle-mounted sensor includes a laser radar, a millimeter-wave radar, a positioning unit, and an inertial measurement unit, wherein the laser radar is used to obtain point cloud data of the driving vehicle, the millimeter-wave radar is used to obtain speed data of the driving vehicle, the positioning unit is used to obtain position data of the driving vehicle, and the inertial measurement unit is used to obtain acceleration data of the driving vehicle; S2: Preset the vehicle trajectory prediction model and extract the model input data category; S201: using the input data of the vehicle trajectory prediction model as target data, and using the point cloud data, the speed data, the position data, and the acceleration data as auxiliary data; S202: Preset the filtering threshold and the samples of each category in the target data, and calculate the dependency of the auxiliary data corresponding to the target data category. The calculation formula is: in, is the dependency of the auxiliary data on the target data in the lth target data sample, h1 and h2 represent the h1th and h2th categories of the target data, and H is the total number of categories of the target data. is the estimated probability that the auxiliary data corresponds to the target data category h1, is the estimated probability that the auxiliary data corresponds to the target data category h2; S203: Calculate the average dependency of the auxiliary data on the target data according to the dependency and samples of each category in the target data, and the calculation formula is: Among them, l is the sample count, L is the total sample size, and A * (X j , Y) is the auxiliary data X j Corresponding to the dependency of target data Y, is the auxiliary data X in the lth sample j Corresponding dependency on target data Y; S204: Arranging the auxiliary data in descending order according to the average dependency to obtain a prediction variable queue, extracting components of the prediction variable queue according to the filtering threshold and forming a sparse vector corresponding to the target data category; Constructing a vehicle dynamics dataset according to the sparse vector, and optimizing the vehicle trajectory prediction model through optimization training of the vehicle dynamics dataset; S3: constructing a vehicle motion model according to the mechanism state parameters of the intelligent driving vehicle, wherein the mechanism state parameters include: speed, acceleration, steering angle, steering rate, vehicle position data, and vehicle wheelbase; constructing test cases based on the vehicle motion model and the vehicle trajectory prediction model, applying the test cases to the vehicle motion model for test evaluation to obtain test results, and optimizing and adjusting the calculation methods of the vehicle motion model and the vehicle trajectory prediction model according to the test results.

2. The method according to claim 1, characterized in that The vehicle trajectory prediction model predicts the test vehicle trajectory data based on LSTM, and the model calculation formula is: i t =σ(W ii x t +b ii +W hi h t-1 +b hi ), f t =σ(W if x t +b if +W hf h t-1 +b hf ), g t =tanh(W ig x t +b ig +W hg h t-1 +b hg ), o t =σ(W io x t +b io +W ho h t-1 +b ho ), c t =f t ☉c t-1 +i t ☉9 t , h t =o t ☉tanh(c t ), Among them, i t is the gate value of the input gate, f t is the gate value of the forget gate, g t To update the gate value of the gate, o t is the gate value of the output gate, h t is the trajectory prediction data, h t-1 is the trajectory prediction data under the t-1 step length, σ is the activation function, tanh is the nonlinear activation function, c t is the cell state in the LSTM network, c t-1 is the cell state at the previous moment, ⊙ represents element-by-element multiplication, W ii is the weight value of different input gate connections, b ii is the bias value between different input gates, W hi is the weight value of the trajectory prediction data and the input gate in the previous calculation step, b hi is the bias value of the trajectory prediction data and the input gate in the previous calculation step, W if is the weight value between the input gate and the forget gate, b if is the bias value between the input gate and the forget gate, W hf is the weight value of the forget gate and the model output, b hf is the bias value between the forget gate and the model output, W ig is the weight value between the input gate and the update gate, b ig is the bias value between the input gate and the update gate, W hg To update the weight value of the gate and model output, b hg To update the bias value of the gate and model output, W io is the weight value between the input gate and the output gate, b io is the bias value between the input gate and the output gate, W ho is the weight value of the output gate and the model output, b ho is the bias value between the output gate and the model output, x t The vehicle location data.

3. The method according to claim 1, characterized in that The vehicle motion model describes the motion of the vehicle based on a two-wheel model, and the calculation formula is: in, Respectively represent the speed change rate of the vehicle on the lateral axis and longitudinal axis, To represent the angular velocity of the vehicle, is the rate of change of the vehicle's speed, v is the vehicle's speed, θ is the vehicle's heading angle, L is the vehicle's wheelbase, a is the vehicle's acceleration, and δ is the vehicle's steering angle.

4. A data-driven testing system based on intelligent driving, used to execute the data-driven testing method based on intelligent driving according to any one of claims 1 to 3, characterized in that: include: Data acquisition module, trajectory prediction module, test optimization module; The data acquisition module is used to set the vehicle-mounted sensors for intelligent driving data acquisition, and the vehicle-mounted sensors include laser radar, millimeter wave radar, positioning unit, and inertial measurement unit. The laser radar is used to obtain the point cloud data of the driving vehicle, the millimeter wave radar is used to obtain the speed data of the driving vehicle, the positioning unit is used to obtain the position data of the driving vehicle, and the inertial measurement unit is used to obtain the acceleration data of the driving vehicle; The trajectory prediction module is used to preset a vehicle trajectory prediction model and extract model input data categories, perform equalization processing on the point cloud data, the speed data, the position data, and the acceleration data based on the model input data categories based on a nonlinear multi-category response to obtain a sparse vector based on the input category, construct a vehicle dynamics data set based on the sparse vector, and optimize the vehicle trajectory prediction model through optimization training of the vehicle dynamics data set; The test optimization module is used to build a vehicle motion model according to the mechanism state parameters of the intelligent driving vehicle, build a test case based on the vehicle motion model and the vehicle trajectory prediction model, apply the test case to the vehicle motion model for test evaluation to obtain test results, and optimize and adjust the calculation method of the vehicle motion model and the vehicle trajectory prediction model according to the test results.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that When the processor executes the computer program, the data-driven testing method based on intelligent driving as described in any one of claims 1-3 is implemented.

6. A storage medium having a computer program stored thereon, characterized in that When the program is executed by a processor, the data-driven testing method based on intelligent driving as described in any one of claims 1-3 is implemented.

Images