Vehicle driving data processing method, device and system, storage medium and equipment

Abstract

The invention discloses a vehicle driving data processing method, device and system, a storage medium and equipment, and the method comprises the steps: obtaining a first driving data set from a preset storage space, the first driving data set comprises at least one piece of driving event data, and each piece of driving event data comprises at least one frame of driving data, the driving event data comprises driving data recorded within a period of time; determining metadata information corresponding to each frame of driving data in the first driving data set according to a preset additional information determination rule, and writing the metadata information corresponding to each frame of driving data into a preset storage space; receiving an SQL (Structured Query Language) statement for each driving scene, and querying metadata information in a preset storage space by utilizing the SQL statements to obtain target metadata information; and determining a second driving data set formed by the target metadata information and the driving data corresponding to the target metadata information as a target data set. According to the invention, high-quality training samples of a large number of planning decision models can be rapidly obtained.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving technology, and more specifically, to a vehicle driving data processing method, device, system, storage medium and equipment. Background Art

[0002] AI (Artificial Intelligence) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. Planning and decision-making, as one of the key technologies in the field of intelligent driving, is currently mainly implemented by rule matching. However, the vehicle driving scenarios are diverse and complex. This decision-making and planning method not only requires a lot of manpower and material resources to collect and set rules, but also has poor flexibility and often causes frustration during vehicle driving, seriously affecting the driving experience and driving safety. In order to solve this technical problem, how to use AI models to realize planning and decision-making has become one of the important research and development directions. The premise of training a high-quality planning and decision-making model is to obtain a high-quality training sample set, and the more data, the better. Therefore, how to quickly obtain a large number of high-quality training samples is an urgent problem to be solved. Summary of the invention

[0003] The present application provides a vehicle driving data processing method, apparatus, system, storage medium and equipment, which can quickly obtain a large number of high-quality training samples for planning decision models.

[0004] The specific technical solutions are as follows:

[0005] In a first aspect, an embodiment of the present application provides a vehicle driving data processing method, the method comprising:

[0006] Acquire a first driving data set from a preset storage space, wherein the first driving data set includes at least one driving event data, each driving event data includes at least one frame of driving data, and the driving event data includes driving data recorded within a period of time;

[0007] Determine metadata information corresponding to each frame of the driving data in the first driving data set according to a preset additional information determination rule, and write the metadata information corresponding to each frame of the driving data into the preset storage space, wherein the metadata information includes additional information for describing the driving data;

[0008] Receiving a structured query language SQL statement for each driving scenario, and using the SQL statement to query the metadata information in the preset storage space to obtain target metadata information, wherein the target metadata information is the metadata information that satisfies the SQL statement, and the SQL statement includes a statement for querying valid metadata information and / or a statement for querying invalid metadata information, the valid metadata information is the metadata information that meets the planning decision model training requirements, and the invalid metadata information is the metadata information that does not meet the planning decision model training requirements;

[0009] The target metadata information and a second driving data set consisting of the driving data corresponding to the target metadata information are determined as the target data set.

[0010] Through the above scheme, it can be seen that the embodiment of the present application can first obtain a first driving data set including at least one driving event data, determine the metadata information corresponding to each frame of driving data in the first driving data set according to the preset additional information determination rule, and write the metadata information corresponding to each frame of driving data into the preset storage space, and then query the required target metadata information in the preset storage space by receiving the SQL statement for each driving scene, and finally determine the second driving data set consisting of the queried target metadata information and the driving data corresponding to the target metadata information as the target data set, so that when the SQL statement is a statement for querying valid metadata information, the obtained target data set can be used as a valid data set to participate in the training of the planning decision model, and when the SQL statement is a statement for querying invalid metadata information, the obtained target data set is used as an invalid data set to avoid affecting the quality of the planning decision model due to the invalid data set participating in the planning decision model training. It can be seen from this that the embodiment of the present application only needs to manually write the SQL statement for each driving scene, and the other steps can be automatically realized, and metadata information is added to assist model training on the basis of the original driving data, and invalid data sets can be queried to avoid affecting the quality of model training, so that a large number of high-quality training samples of planning decision models can be quickly obtained.

[0011] In a first possible implementation of the first aspect, the metadata information includes the curvature of the current road, diverging flow information, the lateral distance of the vehicle deviating from the current lane, the vehicle's intention to change lanes, relevant information of the vehicle in front of the left lane, relevant information of the vehicle in front of the right lane, relevant information of the vehicle behind the left lane, relevant information of the vehicle behind the right lane, whether there is a vehicle in front of the current lane, relevant information of the vehicle in front of the current lane corresponding to the vehicle in front of the current lane when the vehicle in front of the current lane exists, road speed limit information, collision time TTC between the vehicle and the other vehicle during the vehicle's lane change, and at least one of the TTC between the vehicle and the other vehicle during the other vehicle's lane change.

[0012] In a second possible implementation manner of the first aspect, determining metadata information corresponding to each frame of driving data in the first driving data set according to a preset additional information determination rule includes:

[0013] Determining the preset additional information determination rule according to the decision planning rule based on the autonomous driving;

[0014] The metadata information corresponding to each frame of the driving data in the first driving data set is determined according to the preset additional information determination rule.

[0015] It can be seen from the above scheme that the embodiment of the present application can directly determine the preset additional information determination rule through the existing decision planning rules without manual writing, thereby improving the efficiency of obtaining the preset additional information determination rule.

[0016] In a third possible implementation manner of the first aspect, the method further includes:

[0017] When the SQL statement is a statement for querying valid metadata information, inputting the target data set corresponding to the SQL statement into the planning decision model to obtain a loss value of the planning decision model;

[0018] When the loss value is greater than or equal to a preset loss threshold, the SQL statement is updated to be determined as a statement for querying invalid metadata information, or the target data set corresponding to the SQL statement is determined to be an invalid data set.

[0019] It can be seen from the above scheme that the embodiment of the present application can determine whether the SQL statement is reasonable by inputting the target data set into the planning decision model, obtaining the loss value of the planning decision model, and comparing the loss value with the preset loss threshold, or whether the target metadata information and corresponding driving data obtained based on the SQL statement query are reasonable. When it is judged to be unreasonable, the SQL statement is updated to be a statement for querying invalid metadata information, or the target data set corresponding to the SQL statement is determined to be an invalid data set, so as to avoid the subsequent use of the invalid data set to train the planning decision model, thereby further improving the quality of the planning decision model, and then improving the accuracy of the planning decision model in making planning decisions.

[0020] In a fourth possible implementation manner of the first aspect, when the SQL statement is a statement for querying valid metadata information, the method further includes:

[0021] According to the driving scenario of the target data set, the target data set is associated with a driving scenario tree, wherein the driving scenario tree includes at least one level of driving scenarios, and each level of driving scenarios includes at least one driving scenario.

[0022] It can be seen from the above scheme that after obtaining a valid target data set (i.e., valid target metadata information and corresponding driving data), the embodiment of the present application associates the target data set to the driving scenario tree according to the driving scenario of the target data set, which not only can realize the classification of the target data set, but also allows the user to intuitively view the number of target data sets for each driving scenario.

[0023] In a second aspect, an embodiment of the present application provides a vehicle driving data processing device, the device comprising:

[0024] an acquisition unit, configured to acquire a first driving data set from a preset storage space, wherein the first driving data set includes at least one driving event data, each driving event data includes at least one frame of driving data, and the driving event data includes driving data recorded within a period of time;

[0025] A first determining unit, configured to determine metadata information corresponding to each frame of the driving data in the first driving data set according to a preset additional information determination rule, wherein the metadata information includes additional information for describing the driving data;

[0026] A writing unit, used for writing the metadata information corresponding to each frame of the driving data into the preset storage space;

[0027] A receiving unit, configured to receive a structured query language SQL statement for each driving scenario;

[0028] A query unit, used to query the metadata information in the preset storage space using the SQL statement to obtain target metadata information, wherein the target metadata information is the metadata information that satisfies the SQL statement, and the SQL statement includes a statement for querying valid metadata information and / or a statement for querying invalid metadata information, wherein the valid metadata information is the metadata information that meets the planning decision model training requirements, and the invalid metadata information is the metadata information that does not meet the planning decision model training requirements;

[0029] The second determining unit is configured to determine a second driving data set consisting of the target metadata information and the driving data corresponding to the target metadata information as a target data set.

[0030] In a first possible implementation manner of the second aspect, the metadata information includes the curvature of the current road, diverging flow information, the lateral distance of the vehicle deviating from the current lane, the vehicle's intention to change lanes, relevant information of the vehicle in front of the left lane, relevant information of the vehicle in front of the right lane, relevant information of the vehicle behind the left lane, relevant information of the vehicle behind the right lane, whether there is a vehicle in front of the current lane, relevant information of the vehicle in front of the current lane corresponding to the vehicle in front of the current lane when there is a vehicle in front of the current lane, road speed limit information, collision time TTC between the vehicle and the other vehicle during the vehicle's lane change, and at least one of the TTC between the vehicle and the other vehicle during the other vehicle's lane change.

[0031] In a second possible implementation manner of the second aspect, the first determining unit includes:

[0032] A first determination module, configured to determine the preset additional information determination rule according to a decision planning rule based on autonomous driving;

[0033] The second determination module is used to determine the metadata information corresponding to each frame of the driving data in the first driving data set according to the preset additional information determination rule.

[0034] In a third possible implementation manner of the second aspect, the device further includes:

[0035] An input unit, used for inputting the target data set corresponding to the SQL statement into the planning decision model to obtain a loss value of the planning decision model when the SQL statement is a statement for querying valid metadata information;

[0036] An updating unit is used to update the SQL statement to a statement for querying invalid metadata information, or to determine the target data set corresponding to the SQL statement as an invalid data set when the loss value is greater than or equal to a preset loss threshold.

[0037] In a fourth possible implementation manner of the second aspect, the device further includes:

[0038] an associating unit, for associating the target data set to a driving scenario tree according to the driving scenario of the target data set when the SQL statement is a statement for querying valid metadata information, wherein the driving scenario tree includes at least one level of driving scenarios, and each level of driving scenarios includes at least one driving scenario.

[0039] Through the above scheme, it can be seen that the embodiment of the present application can first obtain a first driving data set including at least one driving event data, determine the metadata information corresponding to each frame of driving data in the first driving data set according to the preset additional information determination rule, and write the metadata information corresponding to each frame of driving data into the preset storage space, and then query the required target metadata information in the preset storage space by receiving the SQL statement for each driving scene, and finally determine the second driving data set consisting of the queried target metadata information and the driving data corresponding to the target metadata information as the target data set, so that when the SQL statement is a statement for querying valid metadata information, the obtained target data set can be used as a valid data set to participate in the training of the planning decision model, and when the SQL statement is a statement for querying invalid metadata information, the obtained target data set is used as an invalid data set to avoid affecting the quality of the planning decision model due to the invalid data set participating in the planning decision model training. It can be seen from this that the embodiment of the present application only needs to manually write the SQL statement for each driving scene, and the other steps can be automatically realized, and metadata information is added to assist model training on the basis of the original driving data, and invalid data sets can be queried to avoid affecting the quality of model training, so that a large number of high-quality training samples of planning decision models can be quickly obtained.

[0040] In a third aspect, an embodiment of the present application provides a vehicle driving data processing system, the system comprising: a storage layer, a computing layer, and an application layer;

[0041] The storage layer is used to store a first driving data set, wherein the first driving data set includes at least one driving event data, each driving event data includes at least one frame of driving data, and the driving event data includes driving data recorded within a period of time;

[0042] The computing layer is used to obtain the first driving data set from the storage layer, determine metadata information corresponding to each frame of the driving data in the first driving data set according to a preset additional information determination rule, and write the metadata information corresponding to each frame of the driving data into the storage layer, wherein the metadata information includes additional information for describing the driving data;

[0043] The application layer is used to receive a structured query language SQL statement for each driving scenario and transmit the SQL statement to the computing layer;

[0044] The computing layer is further used to use the SQL statement to query the metadata information in the storage layer to obtain target metadata information, wherein the target metadata information is the metadata information that satisfies the SQL statement, and the SQL statement includes a statement for querying valid metadata information and / or a statement for querying invalid metadata information, the valid metadata information is the metadata information that meets the planning decision model training requirements, and the invalid metadata information is the metadata information that does not meet the planning decision model training requirements; a second driving data set consisting of the target metadata information and the driving data corresponding to the target metadata information is determined as the target data set.

[0045] Through the above scheme, it can be seen that the embodiment of the present application can first obtain a first driving data set including at least one driving event data from the storage layer by the computing layer, determine the metadata information corresponding to each frame of driving data in the first driving data set according to the preset additional information determination rule, and write the metadata information corresponding to each frame of driving data into the storage layer, and then the application layer receives the SQL statement for each driving scene, and queries the required target metadata information in the storage layer through the computing layer, and finally determines the second driving data set consisting of the queried target metadata information and the driving data corresponding to the target metadata information as the target data set, so that when the SQL statement is a statement for querying valid metadata information, the obtained target data set can be used as a valid data set to participate in the training of the planning decision model, and when the SQL statement is a statement for querying invalid metadata information, the obtained target data set is used as an invalid data set to avoid affecting the quality of the planning decision model due to participating in the planning decision model. It can be seen from this that the embodiment of the present application only needs to manually write the SQL statement for each driving scene, and the other steps can be automatically realized, and metadata information is added to assist model training on the basis of the original driving data, and invalid data sets can be queried to avoid affecting the quality of model training, so that a large number of high-quality training samples of planning decision models can be quickly obtained.

[0046] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any possible implementation manner of the first aspect.

[0047] In a fifth aspect, an embodiment of the present application provides an electronic device, the electronic device comprising:

[0048] one or more processors;

[0049] The processor is coupled to a storage device, and the storage device is used to store one or more programs;

[0050] When one or more programs are executed by one or more processors, the electronic device implements the method described in any possible implementation manner of the first aspect.

[0051] In a sixth aspect, an embodiment of the present application provides a computer program product, which includes instructions. When the instructions are executed on a computer or a processor, the computer or the processor executes the method described in any possible implementation of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art description are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative work.

[0053] Figure 1 A schematic diagram of a process flow of a vehicle driving data processing method provided in an embodiment of the present application;

[0054] Figure 2 An example diagram of a closed-loop vehicle driving data processing method provided in an embodiment of the present application;

[0055] Figure 3 A vehicle driving data processing system architecture diagram provided in an embodiment of the present application;

[0056] Figure 4 Another vehicle driving data processing system architecture diagram provided in an embodiment of the present application;

[0057] Figure 5 A block diagram of a vehicle driving data processing device provided in an embodiment of the present application;

[0058] Figure 6 A schematic diagram of the structure of an electronic device or computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0060] It should be noted that, in the absence of conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The terms "including" and "having" in the embodiments of the present application and the accompanying drawings and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or devices.

[0061] Figure 1 The present invention is a flow chart of a method for processing vehicle driving data. The method can be applied to electronic equipment or computer equipment, and can be applied to a terminal or a server. Specifically, the method can include the following steps:

[0062] S110: Acquire a first driving data set from a preset storage space.

[0063] The first driving data set includes at least one driving event data, each driving event data includes at least one frame of driving data, and the driving event data includes driving data recorded within a period of time. One driving event data may include one driving scene or multiple driving scenes. The "one frame" here may refer to a video frame in the traditional sense, or one driving event data may be divided into multiple data segments according to a preset time interval, and each data segment is called a frame of driving data.

[0064] On urban roads, driving scenarios include at least one of navigation lane change, lane change suppression, detour lane change, overtaking lane change, intersection passage, speed control, lane avoidance, lane centering, and lane avoidance. On highways, driving scenarios include at least one of navigation lane change, lane change suppression, overtaking lane change, speed control, lane centering, and lane avoidance. The preset storage space includes at least one of data lake and database.

[0065] Driving data is objective data obtained by processing the raw data collected by the data collection device on the vehicle. The data collection device includes at least one of an image collector, a laser radar, and a positioning system. Driving data includes at least one of self-vehicle information, other-vehicle information, static landmark information, and traffic light information. Self-vehicle information includes real-time global position, real-time global orientation angle, real-time vehicle speed, real-time following gear, and target vehicle speed of the self-vehicle. Other-vehicle information includes at least one of real-time global position, real-time global orientation angle, real-time vehicle speed, and real-time following gear of other vehicles around the self-vehicle. Static landmark information includes static landmark information on the road surface within a preset distance range around the self-vehicle, such as lane lines, road boundary lines, static obstacles, and at least one of the above. Traffic light information includes the type and status of traffic lights. The types of traffic lights include at least one of motor vehicle traffic lights, non-motor vehicle traffic lights, lane lights, flashing warning lights, pedestrian crossing lights, direction indicators, and road and railway level crossing lights. The status of traffic lights includes the on and off status of traffic lights. The target speed of the self-vehicle is the expected speed set by the user.

[0066] S120: Determine metadata information corresponding to each frame of driving data in the first driving data set according to a preset additional information determination rule, and write the metadata information corresponding to each frame of driving data into a preset storage space.

[0067] The metadata information includes additional information for describing driving data. The metadata information includes at least one of the following: the curvature of the current road, diverging and merging information, the lateral distance of the vehicle from the current lane, the vehicle's lane change intention, relevant information of the vehicle in front of the left lane, relevant information of the vehicle in front of the right lane, relevant information of the vehicle behind the left lane, relevant information of the vehicle behind the right lane, whether there is a vehicle in front of the current lane, relevant information of the vehicle in front of the current lane when there is a vehicle in front of the current lane, road speed limit information, TTC (Time to Collision) between the vehicle and other vehicles during the vehicle's lane change, and TTC between the vehicle and other vehicles during the vehicle's lane change.

[0068] The front vehicle in the left lane refers to the vehicle that is closest to the own vehicle and relatively located in front of the own vehicle in the left lane adjacent to the current lane where the own vehicle is located; the front vehicle in the right lane refers to the vehicle that is closest to the own vehicle and relatively located in front of the own vehicle in the right lane adjacent to the current lane where the own vehicle is located; the rear vehicle in the left lane refers to the vehicle that is closest to the own vehicle and relatively located behind the own vehicle in the left lane adjacent to the current lane where the own vehicle is located; the rear vehicle in the right lane refers to the vehicle that is closest to the own vehicle and relatively located behind the own vehicle in the right lane adjacent to the current lane where the own vehicle is located; the front vehicle in the current lane refers to the vehicle that is closest to the own vehicle in the current lane where the own vehicle is located.

[0069] The relevant information of the front vehicle in the left lane includes the position of the front vehicle in the left lane relative to the vehicle, the orientation angle of the front vehicle in the left lane relative to the vehicle, the speed of the front vehicle in the left lane relative to the vehicle, etc. The relevant information of the front vehicle in the right lane, the relevant information of the rear vehicle in the left lane, and the relevant information of the rear vehicle in the right lane are the same as the relevant information of the front vehicle in the left lane, and will not be repeated here.

[0070] The implementation method of determining the metadata information corresponding to each frame of driving data in the first driving data set according to the preset additional information determination rule includes: determining the preset additional information determination rule according to the decision planning rule based on autonomous driving; determining the metadata information corresponding to each frame of driving data in the first driving data set according to the preset additional information determination rule.

[0071] In the related art, decision planning can be performed by matching decision planning rules based on automatic driving. "Decision planning rules" refer to traditional planning algorithms based on optimization, which are rules pre-set one by one based on driving experience. Decision planning can be performed by matching decision planning rules during automatic driving, and the pre-set decision planning rules themselves will not change with the automatic learning of different actual driving scenes. Therefore, decision planning rules are different from planning algorithms based on deep learning, that is, different from the planning decision model trained based on driving data in the embodiment of the present application. The embodiment of the present application can determine the preset additional information determination rule based on the decision planning rule, such as extracting key information from the decision planning rule, and using the calculation method of the key information as a preset additional information determination rule. The key information includes some customized data in addition to the objective data in the driving data, such as only the self-vehicle information and other vehicle information in the driving data, but no information associated with the self-vehicle and other vehicles, such as the concepts of the front vehicle in the left lane and the rear vehicle in the left lane. Therefore, when such information exists in the decision planning rule, key information can be extracted from the decision planning rule, and the calculation method of the key information can be used as a preset additional information determination rule. The calculation method of the key information can refer to the prior art, which will not be repeated here.

[0072] In addition to using existing decision planning rules to determine the preset additional information determination rules, users can also customize the preset additional information determination rules.

[0073] The metadata information corresponding to each frame of driving data may include one type of information or multiple types of information, that is, one frame of driving data corresponds to at least one type of metadata information.

[0074] S130: receiving a structured query language SQL statement for each driving scene, and using the SQL statement to query metadata information in a preset storage space to obtain target metadata information.

[0075] Among them, the target metadata information is metadata information that satisfies the SQL statement; the SQL statement includes a statement for querying valid metadata information, and / or a statement for querying invalid metadata information; the valid metadata information is metadata information that meets the planning decision model training requirements; the invalid metadata information is metadata information that does not meet the planning decision model training requirements.

[0076] In an exemplary embodiment, only one planning decision model including all driving scenarios can be trained; in another exemplary embodiment, in order to improve the training efficiency and quality of the planning decision model, corresponding planning decision models can be trained for different driving scenarios, such as for highway scenarios and urban road scenarios. In this case, it is necessary to first filter out the data in each driving scenario from all driving data and metadata information. Therefore, SQL statements can be written in the SQL operation interface for each driving scenario to query, first obtain the target metadata information of each driving scenario, and then obtain the driving data corresponding to the target metadata information.

[0077] S140: Determine a second driving data set consisting of target metadata information and driving data corresponding to the target metadata information as a target data set.

[0078] After obtaining the target metadata information and the driving data corresponding to the target metadata information, the driving event data to which the driving data corresponding to the target metadata information belongs can be called target driving event data, and the target driving event data and the target metadata information corresponding to each driving data in the target driving event data can be combined into a target data set.

[0079] The vehicle driving data processing method provided in the embodiment of the present application can first obtain a first driving data set including at least one driving event data, determine the metadata information corresponding to each frame of driving data in the first driving data set according to the preset additional information determination rule, and write the metadata information corresponding to each frame of driving data into the preset storage space, and then query the required target metadata information in the preset storage space by receiving the SQL statement for each driving scene, and finally determine the second driving data set consisting of the queried target metadata information and the driving data corresponding to the target metadata information as the target data set, so that when the SQL statement is a statement for querying valid metadata information, the obtained target data set can be used as a valid data set to participate in the training of the planning decision model, and when the SQL statement is a statement for querying invalid metadata information, the obtained target data set is used as an invalid data set to avoid affecting the quality of the planning decision model due to the invalid data set participating in the planning decision model training. It can be seen from this that the embodiment of the present application only needs to manually write the SQL statement for each driving scene, and the other steps can be automatically realized, and metadata information is added to assist model training on the basis of the original driving data, and invalid data sets can be queried to avoid affecting the quality of model training, so that a large number of high-quality training samples of planning decision models can be quickly obtained.

[0080] In one implementation, in order to improve the accuracy of planning decision-making by the planning decision model, another embodiment of the present application further provides the following method:

[0081] When the SQL statement is a statement for querying valid metadata information, the target data set corresponding to the SQL statement is input into the planning decision model to obtain the loss value of the planning decision model; when the loss value is greater than or equal to the preset loss threshold, the SQL statement is updated to be determined as a statement for querying invalid metadata information, or the target data set corresponding to the SQL statement is determined as an invalid data set; when the loss value is less than the preset loss threshold, the purpose of the SQL statement is not updated, or the attribute of the metadata information obtained by the SQL statement query is not updated, or the attribute of the target data set corresponding to the SQL statement is not updated. Among them, the attribute includes validity or invalidity.

[0082] When the SQL statement is a statement for querying valid metadata information, the target data set corresponding to the SQL statement is input into the planning decision model to obtain predicted data, and the loss value is calculated based on the predicted data and the true value data. When the loss value is greater than or equal to the preset loss threshold, it means that the target data set is not suitable for training the planning decision model. Of course, in order to prevent misjudgment, when the loss value is greater than or equal to the preset loss threshold, the model parameters of the planning decision model can be adjusted first, and the target data set is input into the planning decision model again to calculate the loss value. If the loss value is still greater than or equal to the preset loss threshold, the SQL statement is updated to be determined as a statement for querying invalid metadata information, or the target data set corresponding to the SQL statement is directly determined as an invalid data set. When the loss value is less than the preset loss threshold, it means that the target data set is suitable for training the planning decision model, and there is no need to update the SQL statement or the attributes of the data corresponding to the SQL statement.

[0083] The embodiment of the present application can determine whether the SQL statement is reasonable by inputting the target data set into the planning decision model, obtaining the loss value of the planning decision model, and comparing the loss value with a preset loss threshold, or whether the target metadata information and corresponding driving data obtained based on the SQL statement query are reasonable. When it is determined to be unreasonable, the SQL statement is updated to be a statement for querying invalid metadata information, or the target data set corresponding to the SQL statement is determined to be an invalid data set, so as to avoid the subsequent use of the invalid data set to train the planning decision model, thereby further improving the quality of the planning decision model, and then improving the accuracy of the planning decision model in making planning decisions.

[0084] In one implementation, in order to intuitively and quickly obtain the data collection status of each driving scenario and the data participation of each driving scenario in the planning decision-making model, another embodiment of the present application also provides the following method: when the SQL statement is a statement for querying valid metadata information, the target data set is associated with the driving scenario tree according to the driving scenario of the target data set. The driving scenario tree includes at least one level of driving scenarios, and each level of driving scenarios includes at least one driving scenario.

[0085] The first-level driving scenarios include expressways and / or urban roads. When the first-level driving scenario is an expressway, the second-level driving scenarios include at least one of navigation lane change, lane change suppression, lane change by overtaking, vehicle speed control, lane centering, lane avoidance, etc.; when the first-level driving scenario is an urban road, the second-level driving scenarios include at least one of navigation lane change, lane change suppression, lane change by detour, lane change by overtaking, intersection passage, vehicle speed control, lane avoidance, lane centering, and lane avoidance.

[0086] In an embodiment of the present application, the user can determine the driving scene of the target data set, and associate the target data set to the corresponding driving scene in the driving scene tree through human-computer interaction, or an automatic association method can be used. The automatic association method includes: when the driving event data in the target data set to be associated contains the driving data recorded in a preset time period before and after being taken over during the automatic driving process of the self-vehicle based on the planning decision model, the driving scene is determined according to the taken over scene corresponding to the target data set to be associated; according to the determined driving scene, the target data set to be associated is associated with the driving scene tree. For example, when the self-vehicle is performing automatic driving based on the planning decision model, when overtaking and changing lanes on a highway, a dangerous behavior is about to occur, and the driver takes over, then it can be determined that the driving scene is overtaking and changing lanes on the highway, and the target data set is associated with the overtaking and changing lanes on the highway.

[0087] After different target data sets are associated with the driving scenario tree respectively, the amount of data in each driving scenario in the driving scenario tree can be counted regularly or irregularly, and the statistical results can be output and displayed in the form of charts and the like.

[0088] In one embodiment, if Figure 2 As shown, the embodiment of the present application can provide a closed-loop vehicle driving data processing method, firstly collect data to obtain a first driving data set, pre-process the first driving data set (including the processing method for obtaining a target data set from the first driving data set as described in any of the above embodiments), obtain the target data set, and then associate the target data set with the driving scene at the last level of the driving scene tree to implement the process of subdividing the target data set, and perform model iterative training on the target data set to obtain a calculated loss value. When the loss value is greater than or equal to a preset loss threshold, it indicates that the target data set is not suitable for training a planning decision model. In subsequent data collection, data of the same type as the target data set may no longer be included in the model training.

[0089] Corresponding to the above method embodiment, another embodiment of the present application provides a vehicle driving data processing system, such as Figure 3 As shown, the system includes: a storage layer 210, a computing layer 220 and an application layer 230;

[0090] The storage layer 210 is used to store a first driving data set, wherein the first driving data set includes at least one driving event data, each driving event data includes at least one frame of driving data, and the driving event data includes driving data recorded within a period of time;

[0091] The computing layer 220 is configured to obtain the first driving data set from the storage layer, determine metadata information corresponding to each frame of the driving data in the first driving data set according to a preset additional information determination rule, and write the metadata information corresponding to each frame of the driving data into the storage layer, wherein the metadata information includes additional information for describing the driving data.

[0092] The application layer 230 is used to receive a structured query language SQL statement for each driving scenario and transmit the SQL statement to the computing layer;

[0093] The computing layer 220 is also used to use the SQL statement to query the metadata information in the storage layer to obtain target metadata information, wherein the target metadata information is the metadata information that satisfies the SQL statement, and the SQL statement includes a statement for querying valid metadata information and / or a statement for querying invalid metadata information, the valid metadata information is the metadata information that meets the planning decision model training requirements, and the invalid metadata information is the metadata information that does not meet the planning decision model training requirements; a second driving data set consisting of the target metadata information and the driving data corresponding to the target metadata information is determined as the target data set.

[0094] In one embodiment, the metadata information includes the curvature of the current road, diverging flow information, the lateral distance of the vehicle deviating from the current lane, the vehicle's intention to change lanes, relevant information of the vehicle in front of the left lane, relevant information of the vehicle in front of the right lane, relevant information of the vehicle behind the left lane, relevant information of the vehicle behind the right lane, whether there is a vehicle in front of the current lane, relevant information of the vehicle in front of the current lane corresponding to the vehicle in front of the current lane when the vehicle in front of the current lane exists, road speed limit information, collision time TTC between the vehicle and the other vehicle during the vehicle's lane change, and at least one of the TTC between the vehicle and the other vehicle during the other vehicle's lane change.

[0095] In one embodiment, the computing layer 220 is used to determine the preset additional information determination rule according to the decision planning rule based on autonomous driving; and determine the metadata information corresponding to each frame of the driving data in the first driving data set according to the preset additional information determination rule.

[0096] In one embodiment, the application layer 230 is further used to associate the target data set with a driving scenario tree according to the driving scenario of the target data set when the SQL statement is a statement for querying valid metadata information, wherein the driving scenario tree includes at least one level of driving scenarios, and each level of driving scenarios includes at least one driving scenario.

[0097] like Figure 4 As shown, the computing layer reads at least one driving event data from the storage layer as the first driving data set, and then at least one computing module including data mining logic for calculating metadata information based on the driving data calculates the first driving data set, wherein each computing module can be located in the same computer device or in different computer devices, each computing module reads different driving event data in the first driving data set respectively, processes different driving event data in parallel, writes each metadata information obtained into the storage layer, and associates the metadata information with the corresponding driving data. The user inputs an SQL statement based on the SQL user interface of the application layer, or inputs an SQL search term, and the computing layer automatically generates an SQL statement based on the SQL search term, and the computing layer generates a query logic code based on the SQL statement, loads multiple metadata information from the storage layer, and queries target metadata information from the multiple metadata information, reads the driving data corresponding to the target metadata information from the storage layer, and transmits the target metadata information and the corresponding driving data to the application layer to form the relevant information of the target data set (such as the statistical data volume), so that the user can query the relevant information of the target data set through the data service interface of the application layer. Similar to data mining logic, the query logic is located in the query module. Each query module can be located on the same computer device or on different computer devices. Each query module can query in parallel to improve query efficiency.

[0098] The vehicle driving data processing system provided by the embodiment of the present application can first obtain a first driving data set including at least one driving event data from the storage layer by the computing layer, determine the metadata information corresponding to each frame of driving data in the first driving data set according to the preset additional information determination rule, and write the metadata information corresponding to each frame of driving data into the storage layer, and then receive the SQL statement for each driving scene by the application layer, query the required target metadata information in the storage layer through the computing layer, and finally determine the second driving data set consisting of the queried target metadata information and the driving data corresponding to the target metadata information as the target data set, so that when the SQL statement is a statement for querying valid metadata information, the obtained target data set can be used as a valid data set to participate in the training of the planning decision model, and when the SQL statement is a statement for querying invalid metadata information, the obtained target data set is used as an invalid data set to avoid affecting the quality of the planning decision model due to the invalid data set participating in the planning decision model training. It can be seen from this that the embodiment of the present application only needs to manually write the SQL statement for each driving scene, and the other steps can be automatically realized, and metadata information is added to assist model training on the basis of the original driving data, and invalid data sets can be queried to avoid affecting the quality of model training, so that a large number of high-quality training samples of planning decision models can be quickly obtained.

[0099] Based on the above method embodiment, another embodiment of the present application provides a vehicle driving data processing device, such as Figure 5 As shown, the device comprises:

[0100] An acquiring unit 310 is configured to acquire a first driving data set from a preset storage space, wherein the first driving data set includes at least one driving event data, each driving event data includes at least one frame of driving data, and the driving event data includes driving data recorded within a period of time;

[0101] A first determining unit 320, configured to determine metadata information corresponding to each frame of the driving data in the first driving data set according to a preset additional information determination rule, wherein the metadata information includes additional information for describing the driving data;

[0102] A writing unit 330, configured to write the metadata information corresponding to each frame of the driving data into the preset storage space;

[0103] A receiving unit 340 is used to receive a structured query language SQL statement for each driving scenario;

[0104] A query unit 350 is used to query the metadata information in the preset storage space using the SQL statement to obtain target metadata information, wherein the target metadata information is the metadata information that satisfies the SQL statement, and the SQL statement includes a statement for querying valid metadata information and / or a statement for querying invalid metadata information, wherein the valid metadata information is the metadata information that meets the planning decision model training requirements, and the invalid metadata information is the metadata information that does not meet the planning decision model training requirements;

[0105] The second determining unit 360 is configured to determine a second driving data set consisting of the target metadata information and the driving data corresponding to the target metadata information as a target data set.

[0106] In a possible implementation, the metadata information includes the curvature of the current road, diverging flow information, the lateral distance of the vehicle deviating from the current lane, the vehicle's intention to change lanes, relevant information of the vehicle in front of the left lane, relevant information of the vehicle in front of the right lane, relevant information of the vehicle behind the left lane, relevant information of the vehicle behind the right lane, whether there is a vehicle in front of the current lane, relevant information of the vehicle in front of the current lane corresponding to the vehicle in front of the current lane when there is a vehicle in front of the current lane, road speed limit information, collision time TTC between the vehicle and the other vehicle during the vehicle's lane change, and at least one of the TTC between the vehicle and the other vehicle during the other vehicle's lane change.

[0107] In a possible implementation, the first determining unit 320 includes:

[0108] A first determination module, configured to determine the preset additional information determination rule according to a decision planning rule based on autonomous driving;

[0109] The second determination module is used to determine the metadata information corresponding to each frame of the driving data in the first driving data set according to the preset additional information determination rule.

[0110] In a possible implementation manner, the device further includes:

[0111] An input unit, used for inputting the target data set corresponding to the SQL statement into the planning decision model to obtain a loss value of the planning decision model when the SQL statement is a statement for querying valid metadata information;

[0112] An updating unit is used to update the SQL statement to a statement for querying invalid metadata information, or to determine the target data set corresponding to the SQL statement as an invalid data set when the loss value is greater than or equal to a preset loss threshold.

[0113] In a possible implementation manner, the device further includes:

[0114] an associating unit, for associating the target data set to a driving scenario tree according to the driving scenario of the target data set when the SQL statement is a statement for querying valid metadata information, wherein the driving scenario tree includes at least one level of driving scenarios, and each level of driving scenarios includes at least one driving scenario.

[0115] The vehicle driving data processing device provided in the embodiment of the present application can first obtain a first driving data set including at least one driving event data, determine the metadata information corresponding to each frame of driving data in the first driving data set according to the preset additional information determination rule, and write the metadata information corresponding to each frame of driving data into the preset storage space, and then query the required target metadata information in the preset storage space by receiving the SQL statement for each driving scene, and finally determine the second driving data set consisting of the queried target metadata information and the driving data corresponding to the target metadata information as the target data set, so that when the SQL statement is a statement for querying valid metadata information, the obtained target data set can be used as a valid data set to participate in the training of the planning decision model, and when the SQL statement is a statement for querying invalid metadata information, the obtained target data set is used as an invalid data set to avoid affecting the quality of the planning decision model due to the invalid data set participating in the planning decision model training. It can be seen from this that the embodiment of the present application only needs to manually write the SQL statement for each driving scene, and the other steps can be automatically realized, and metadata information is added to assist model training on the basis of the original driving data, and invalid data sets can be queried to avoid affecting the quality of model training, so that a large number of high-quality training samples of planning decision models can be quickly obtained.

[0116] Based on the above method embodiments, another embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method described in any of the above embodiments is implemented.

[0117] Based on the above method embodiment, another embodiment of the present application provides an electronic device or a computer device, such as Figure 6 As shown, including:

[0118] one or more processors 410;

[0119] The processor 410 is coupled to a storage device 420, and the storage device 420 is used to store one or more programs;

[0120] When the one or more programs are executed by the one or more processors 410, the electronic device or computer device implements the method described in any of the above embodiments.

[0121] Based on the above method embodiments, another embodiment of the present application provides a vehicle, which includes the device as described in any of the above embodiments, or includes the electronic device as described above.

[0122] Based on the above embodiments, another embodiment of the present application provides a computer program product, which includes instructions. When the instructions are executed on a computer or a processor, the computer or the processor executes the method described in any of the above embodiments.

[0123] The above device embodiment corresponds to the method embodiment and has the same technical effect as the method embodiment. For specific description, please refer to the method embodiment. The device embodiment is obtained based on the method embodiment. For specific description, please refer to the method embodiment part, which will not be repeated here. It can be understood by those skilled in the art that the accompanying drawings are only schematic diagrams of one embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present application.

[0124] Those skilled in the art can understand that the modules in the device in the embodiment can be distributed in the device in the embodiment according to the description of the embodiment, or can be changed accordingly and located in one or more devices different from the embodiment. The modules in the above embodiment can be combined into one module, or can be further divided into multiple sub-modules.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A vehicle driving data processing method, characterized in that: The method comprises: Acquire a first driving data set from a preset storage space, wherein the first driving data set includes at least one driving event data, each driving event data includes at least one frame of driving data, and the driving event data includes driving data recorded within a period of time; Determine metadata information corresponding to each frame of the driving data in the first driving data set according to a preset additional information determination rule, and write the metadata information corresponding to each frame of the driving data into the preset storage space, wherein the metadata information includes additional information for describing the driving data; Receiving a structured query language SQL statement for each driving scenario, and using the SQL statement to query the metadata information in the preset storage space to obtain target metadata information, wherein the target metadata information is the metadata information that satisfies the SQL statement, and the SQL statement includes a statement for querying valid metadata information and / or a statement for querying invalid metadata information, the valid metadata information is the metadata information that meets the planning decision model training requirements, and the invalid metadata information is the metadata information that does not meet the planning decision model training requirements; The target metadata information and a second driving data set consisting of the driving data corresponding to the target metadata information are determined as the target data set.

2. The method according to claim 1, characterized in that The metadata information includes the curvature of the current road, diverging flow information, the lateral distance of the vehicle deviating from the current lane, the vehicle's intention to change lanes, relevant information of the vehicle in front of the left lane, relevant information of the vehicle in front of the right lane, relevant information of the vehicle behind the left lane, relevant information of the vehicle behind the right lane, whether there is a vehicle in front of the current lane, relevant information of the vehicle in front of the current lane corresponding to the vehicle in front of the current lane when there is the vehicle in front of the current lane, road speed limit information, collision time TTC between the vehicle and the other vehicle during the vehicle's lane change, and at least one of the TTC between the vehicle and the other vehicle during the other vehicle's lane change.

3. The method according to claim 1, characterized in that The determining, according to a preset additional information determination rule, metadata information corresponding to each frame of the driving data in the first driving data set includes: Determining the preset additional information determination rule according to the decision planning rule based on the autonomous driving; The metadata information corresponding to each frame of the driving data in the first driving data set is determined according to the preset additional information determination rule.

4. The method according to claim 1, characterized in that: The method further comprises: When the SQL statement is a statement for querying valid metadata information, inputting the target data set corresponding to the SQL statement into the planning decision model to obtain a loss value of the planning decision model; When the loss value is greater than or equal to a preset loss threshold, the SQL statement is updated to be determined as a statement for querying invalid metadata information, or the target data set corresponding to the SQL statement is determined to be an invalid data set.

5. The method according to any one of claims 1 to 4, characterized in that When the SQL statement is a statement for querying valid metadata information, the method further includes: According to the driving scenario of the target data set, the target data set is associated with a driving scenario tree, wherein the driving scenario tree includes at least one level of driving scenarios, and each level of driving scenarios includes at least one driving scenario.

6. A vehicle driving data processing device, characterized in that: The device comprises: an acquisition unit, configured to acquire a first driving data set from a preset storage space, wherein the first driving data set includes at least one driving event data, each driving event data includes at least one frame of driving data, and the driving event data includes driving data recorded within a period of time; A first determining unit, configured to determine metadata information corresponding to each frame of the driving data in the first driving data set according to a preset additional information determination rule, wherein the metadata information includes additional information for describing the driving data; A writing unit, used for writing the metadata information corresponding to each frame of the driving data into the preset storage space; A receiving unit, configured to receive a structured query language SQL statement for each driving scenario; A query unit, configured to query the metadata information in the preset storage space using the SQL statement to obtain target metadata information, wherein the target metadata information is the metadata information satisfying the SQL statement, the SQL statement includes a statement for querying valid metadata information and / or a statement for querying invalid metadata information, the valid metadata information is the metadata information satisfying the planning decision model training requirements, and the invalid metadata information is the metadata information not satisfying the planning decision model training requirements; The second determining unit is configured to determine a second driving data set consisting of the target metadata information and the driving data corresponding to the target metadata information as a target data set.

7. The device according to claim 6, characterized in that The metadata information includes the curvature of the current road, diverging flow information, the lateral distance of the vehicle deviating from the current lane, the vehicle's intention to change lanes, relevant information of the vehicle in front of the left lane, relevant information of the vehicle in front of the right lane, relevant information of the vehicle behind the left lane, relevant information of the vehicle behind the right lane, whether there is a vehicle in front of the current lane, relevant information of the vehicle in front of the current lane corresponding to the vehicle in front of the current lane when there is the vehicle in front of the current lane, road speed limit information, collision time TTC between the vehicle and the other vehicle during the vehicle's lane change, and at least one of the TTC between the vehicle and the other vehicle during the other vehicle's lane change.

8. The device according to claim 6, characterized in that The first determining unit includes: A first determination module, configured to determine the preset additional information determination rule according to a decision planning rule based on autonomous driving; The second determination module is used to determine the metadata information corresponding to each frame of the driving data in the first driving data set according to the preset additional information determination rule.

9. The device according to claim 6, characterized in that The device also includes: An input unit, used for inputting the target data set corresponding to the SQL statement into the planning decision model to obtain a loss value of the planning decision model when the SQL statement is a statement for querying valid metadata information; An updating unit is used to update the SQL statement to a statement for querying invalid metadata information, or to determine the target data set corresponding to the SQL statement as an invalid data set when the loss value is greater than or equal to a preset loss threshold.

10. The device according to any one of claims 6 to 9, characterized in that: The device also includes: an associating unit, for associating the target data set to a driving scenario tree according to the driving scenario of the target data set when the SQL statement is a statement for querying valid metadata information, wherein the driving scenario tree includes at least one level of driving scenarios, and each level of driving scenarios includes at least one driving scenario.

11. A vehicle driving data processing system, characterized in that: The system comprises: a storage layer, a computing layer and an application layer; The storage layer is used to store a first driving data set, wherein the first driving data set includes at least one driving event data, each driving event data includes at least one frame of driving data, and the driving event data includes driving data recorded within a period of time; The computing layer is used to obtain the first driving data set from the storage layer, determine metadata information corresponding to each frame of the driving data in the first driving data set according to a preset additional information determination rule, and write the metadata information corresponding to each frame of the driving data into the storage layer, wherein the metadata information includes additional information for describing the driving data; The application layer is used to receive a structured query language SQL statement for each driving scenario and transmit the SQL statement to the computing layer; The computing layer is further used to use the SQL statement to query the metadata information in the storage layer to obtain target metadata information, wherein the target metadata information is the metadata information that satisfies the SQL statement, and the SQL statement includes a statement for querying valid metadata information and / or a statement for querying invalid metadata information, the valid metadata information is the metadata information that meets the planning decision model training requirements, and the invalid metadata information is the metadata information that does not meet the planning decision model training requirements; a second driving data set consisting of the target metadata information and the driving data corresponding to the target metadata information is determined as the target data set.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

13. An electronic device, characterized in that: The electronic device comprises: one or more processors; The processor is coupled to a storage device, and the storage device is used to store one or more programs; When the one or more programs are executed by the one or more processors, the electronic device implements the method according to any one of claims 1 to 5.

Images