Method and system for determining driving learning sample, electronic equipment and medium

Abstract

The invention provides a method and system for determining a driving learning sample, electronic equipment and a medium. The method for determining the driving learning sample comprises the steps of obtaining a driving data set; the driving data set comprises driving data in the driving process of each vehicle; according to the driving data, determining a target vehicle whose driving condition accords with a predetermined condition of a safe driving habit; and determining a driving learning sample according to the target vehicle. The driving data can reflect the driving condition of the vehicle. According to the invention, screening is carried out according to the driving data to determine the target vehicle whose driving condition accords with the predetermined condition of the safe driving habit, and the driving learning sample is determined according to the target vehicle to ensure that the driving condition of the vehicle in the final driving learning sample accords with the safe habit, so that the sample quality of driving learning is improved. And meanwhile, the number of samples for driving learning is reduced through screening, so that the data processing efficiency in the driving learning process is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicles, and particularly to a method, a system, an electronic device, and a medium for determining driving learning samples. Background Art

[0002] At present, vehicles are gradually developing towards the intelligent direction, and the intelligent driving ability of vehicles has received more and more attention. Among many intelligent driving solutions, driving learning is achieved through large models, and the large models are continuously iterated to improve the safety, comfort, and efficiency of the intelligent driving solutions. In the related art, all the vehicle networking data of all users obtained can be input into the large model as training data to learn and determine driving habits for guiding the intelligent driving solution. However, in the actual implementation process, there is a large amount of vehicle networking data, and the quality is uneven, and there are often a large number of invalid data, resulting in a large amount of computing resources being consumed during the training process of the large model. Summary of the Invention

[0003] The present application provides a method, a system, an electronic device, and a medium for determining driving learning samples to improve the sample quality of driving learning, reduce the number of samples, and improve the data processing efficiency during the driving learning process.

[0004] The present application provides a method for determining driving learning samples, including: obtaining a driving data set; the driving data set includes driving data during the driving process of each vehicle; according to the driving data, determining target vehicles whose driving conditions meet the predetermined conditions of safe driving habits; and determining driving learning samples according to the target vehicles.

[0005] Optionally, the driving data includes a driving speed and driving data corresponding to the driving speed; according to the driving data, determining target vehicles whose driving conditions meet the predetermined conditions of safe driving habits includes: determining the driving conditions of each vehicle in each set speed interval according to the corresponding driving data; determining target vehicles whose driving conditions in at least one set speed interval meet the corresponding predetermined conditions according to the driving conditions of each vehicle in each set speed interval; wherein, the set speed interval corresponds to the set condition one by one.

[0006] Optionally, the driving data corresponding to the driving speed includes a driving mileage for indicating the distance change during the driving process of the vehicle; determining the driving conditions of each vehicle in each set speed interval according to the corresponding driving data includes: determining the driving conditions of each vehicle in each set speed interval according to the driving mileage.

[0007] Optionally, the set speed intervals include a first speed interval where the driving speed is greater than or equal to a first speed threshold; the driving data corresponding to the driving speed further includes control data for indicating that the vehicle control system is input with driving operations during the vehicle driving process; according to the driving mileage, determine the driving conditions of each vehicle in each set speed interval, including: according to the driving mileage and the control data, determine the total driving mileage and the average value of the control data of each vehicle in the first speed interval; according to the driving conditions of each vehicle in each set speed interval, determine the target vehicles whose driving conditions in at least one set speed interval meet the corresponding predetermined conditions, including: according to the total driving mileage and the average value of the control data of each vehicle in the first speed interval, determine the target vehicles whose total driving mileage and the average value of the control data in the first speed interval both meet the first predetermined conditions.

[0008] Optionally, according to the total driving mileage and the average value of the control data of each vehicle in the first speed interval, determining the target vehicles whose total driving mileage and the average value of the control data in the first speed interval both meet the first predetermined conditions includes: according to the total driving mileage of each vehicle in the first speed interval, screen out the vehicles whose total driving mileage in the first speed interval is greater than a first mileage threshold as candidate vehicles; according to the average value of the control data of each candidate vehicle in the first speed interval, determine the target vehicles whose average value of the control data meets the predetermined safe operation screening conditions.

[0009] Optionally, the average value of the control data includes the average deceleration used to represent the deceleration condition during the vehicle driving; determining the target vehicle whose average value of the control data meets the predetermined safe operation screening condition according to the average value of the control data of each candidate vehicle in the first speed range includes: determining the N candidate vehicles with the smallest average deceleration as the target vehicles according to the average deceleration of each candidate vehicle in the first speed range; or, the average value of the control data includes the average steering angle used to represent the steering condition of the steering wheel; determining the target vehicle whose average value of the control data meets the set safe operation screening condition according to the average value of the control data of each candidate vehicle in the first speed range includes: determining the M candidate vehicles with the smallest average steering angle as the target vehicles according to the average steering angle of each candidate vehicle in the first speed range; or, the average value of the control data includes the average deceleration used to represent the deceleration condition during the vehicle driving and the average steering angle used to represent the steering condition of the steering wheel; determining the target vehicle whose average value of the control data meets the set safe operation screening condition according to the average value of the control data of each candidate vehicle in the first speed range includes: determining the N candidate vehicles with the smallest average deceleration according to the average deceleration of each candidate vehicle in the first speed range; determining the M candidate vehicles with the smallest average steering angle according to the average steering angle of each candidate vehicle in the first speed range; determining the L candidate vehicles with the smallest comprehensive average deceleration and average steering angle as the target vehicles according to the average deceleration and average steering angle of the N candidate vehicles and the M candidate vehicles in the first speed range; where N, M, and L are all set positive integer values.

[0010] Optionally, the set speed range includes a second speed range where the driving speed is less than the first speed threshold; determining the driving condition of the vehicle in each set speed range according to the driving mileage includes: determining the total driving mileage of the vehicle in the second speed range according to the driving mileage; determining the target vehicle whose driving condition in at least one set speed range meets the corresponding predetermined condition according to the driving condition of each vehicle in each set speed range includes: determining the target vehicle whose total driving mileage in the second speed range meets the second predetermined condition according to the total driving mileage of each vehicle in the second speed range.

[0011] Optionally, determining the target vehicle whose total driving mileage in the second speed range meets the second predetermined condition according to the total driving mileage of each vehicle in the second speed range includes: screening out the vehicles whose total driving mileage in the first speed range is greater than the second mileage threshold according to the total driving mileage of each vehicle in the second speed range as the target vehicles.

[0012] Optionally, driving learning samples are determined according to target vehicles, including: labeling the driving styles of target vehicles according to the driving data of each target vehicle, and determining driving learning samples according to the labeled target vehicles; wherein, the driving data includes the control data of driving operations during vehicle driving.

[0013] Optionally, the driving data includes the number of lane changes and / or deceleration values during vehicle driving; labeling the driving styles of target vehicles according to the driving data of each target vehicle includes: labeling the driving styles of target vehicles according to the number of lane changes and / or deceleration values of each target vehicle.

[0014] Optionally, determining target vehicles whose driving conditions meet set conditions according to the driving data includes: determining the vehicle identification numbers of target vehicles whose driving conditions meet set conditions according to the driving data; determining driving learning samples according to the target vehicles includes: issuing a driving video data acquisition instruction to each target vehicle according to the vehicle identification number of the target vehicle; and determining driving learning samples according to the acquired driving video data of the target vehicle.

[0015] Optionally, each vehicle includes an in-vehicle information acquisition device, the in-vehicle information acquisition device is communicatively connected to a cloud server, and the in-vehicle acquisition device is used to upload the original driving data of each vehicle during driving to the cloud server, so that the cloud server processes the original driving data and forwards the processed driving data to a set columnar database management system; obtaining a driving data set includes: regularly querying the processed driving data in the columnar database management system to obtain the driving data of each vehicle during driving.

[0016] Optionally, it includes: a data acquisition module for obtaining a driving data set; the driving data set includes the driving data of each vehicle during driving; a target vehicle determination module for determining target vehicles whose driving conditions meet a predetermined condition of safe driving habits according to the driving data; and a sample determination module for determining driving learning samples according to the target vehicles.

[0017] This application provides an electronic device, including one or more processors for implementing the foregoing method for determining driving learning samples.

[0018] This application provides a computer-readable storage medium, on which a program is stored, and when the program is executed by a processor, the foregoing method for determining driving learning samples is implemented.

[0019] The method, system, electronic device, and medium for determining driving learning samples provided by this application obtain a driving data set including driving data of each vehicle during driving, and the driving data can reflect the driving conditions of the vehicle. Screening is performed based on the driving data to determine target vehicles whose driving conditions meet the predetermined conditions of safe driving habits, and driving learning samples are determined based on the target vehicles. In this way, it can be ensured that the driving conditions of the vehicles in the final driving learning samples all meet safe habits, which is beneficial to improving the quality of the driving learning samples. At the same time, screening is beneficial to reducing the number of driving learning samples, thereby improving the data processing efficiency during the driving learning process. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flowchart of a method for determining driving learning samples provided by an embodiment of this application;

[0021] Figure 2 is a flowchart of a method for determining driving learning samples provided by another embodiment of this application;

[0022] Figure 3 is a flowchart of a method for determining driving learning samples provided by another embodiment of this application;

[0023] Figure 4 is a flowchart of a method for determining driving learning samples provided by another embodiment of this application;

[0024] Figure 5 is a flowchart of a method for determining driving learning samples provided by another embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] Here, exemplary embodiments will be described in detail, and the examples are shown in the drawings.

[0026] Combined with Figure 1 shown, a method for determining driving learning samples provided by an embodiment of this application includes steps S10 to S30.

[0027] Step S10, obtain a driving data set; the driving data set includes driving data of each vehicle during driving.

[0028] Each piece of driving data is associated with a vehicle. Specifically, the driving data of each vehicle is associated with a vehicle identification number.

[0029] Step S20, based on the driving data, determine target vehicles whose driving conditions meet the predetermined conditions of safe driving habits.

[0030] Here, predetermined conditions that conform to safe driving habits are preset and stored in advance. The driving situation conforms to the predetermined conditions of safe driving habits, indicating that the driving habits of the vehicle are excellent. Thus, determining the driving habit samples is conducive to ensuring the quality of the samples.

[0031] Step S30: Determine a driving learning sample according to the target vehicle.

[0032] Using the method for determining a driving learning sample provided by the embodiments of the present application, a driving data set including the driving data of each vehicle during driving is obtained. The driving data can reflect the driving situation of the vehicle. Screen according to the driving data to determine the target vehicle whose driving situation conforms to the predetermined conditions of safe driving habits, and determine the driving learning sample according to the target vehicle. In this way, it can be ensured that the driving situations of the vehicles in the final driving learning sample all conform to safe habits, which is conducive to improving the quality of the driving learning samples. At the same time, screening is conducive to reducing the number of driving learning samples, thereby improving the data processing efficiency during the driving learning process.

[0033] In some embodiments, each vehicle includes an on-vehicle information collection device. The on-vehicle information collection device is communicatively connected to the cloud server. The on-vehicle collection device is used to upload the original driving data of each vehicle during driving to the cloud server, so that the cloud server processes the original driving data and forwards the processed driving data to a set columnar database management system. Here, obtaining the driving data set includes: regularly querying the processed driving data in the columnar database management system to obtain the driving data of each vehicle during driving. Here, a query message is regularly sent to regularly query the processed driving data in the columnar database management system. Compared with the traditional row-based database management system, the columnar database management system has better query performance, is especially suitable for data aggregation and screening, and has efficient data compression, which is conducive to saving compression space.

[0034] The columnar database management system here is, for example, ClickHouse. More specifically, in some embodiments, the vehicle information collection device realizes communication connection through the MQTT (Message Queuing Telemetry Transport) communication protocol. After collecting the original driving data during the vehicle driving process, the vehicle information collection device uploads it to the cloud server through the MQTT communication protocol. The cloud server here specifically includes a cloud MQTT server. The cloud server processes the original driving data and forwards the processed driving data to the set columnar database management system, specifically: the cloud distributed stream processing platform Kafka pulls the original driving data on the cloud MQTT server in real time. The application side creates a stream processing framework Flink to serve the streaming processing of Kafka data and writes it into the ClickHouse cluster. The analysis side periodically queries the driving data from the ClickHouse cluster. It can be understood that this process is only for example, and specific settings can be made according to the actual situation during the implementation process as long as the functions are achieved.

[0035] In some embodiments, determining the target vehicle whose driving situation meets the set conditions according to the driving data includes: determining the vehicle identification number of the target vehicle whose driving situation meets the set conditions according to the driving data. Determining the driving learning sample according to the target vehicle includes: issuing a driving video data collection instruction to each target vehicle according to the vehicle identification number of the target vehicle; determining the driving learning sample according to the collected driving video data of the target vehicle. The vehicle identification numbers of each vehicle are all unique values, and a unique vehicle can be determined according to the vehicle identification number, ensuring the accuracy of issuing the driving video data collection instruction. After receiving the driving video data collection instruction, if the target vehicle agrees, it will be authorized, and after authorization, the driving video data of the target vehicle can be collected, and these driving video data will be used as driving learning samples for training the intelligent driving model.

[0036] Specifically, the aforementioned driving data includes the driving speed and the driving data corresponding to the driving speed. During the vehicle driving process, corresponding to the same moment, there is a series of driving data. The driving speed and the other driving data in the series of driving data at the same moment are the driving speed and the driving data corresponding to the driving speed mentioned here.

[0037] Combined with Figure 2 As shown, step S20 of the foregoing, determining the target vehicle whose driving situation meets the predetermined conditions of safe driving habits according to the driving data, includes step S21 and step S22.

[0038] Step S21, determining the driving situation of each vehicle in each set speed range according to the corresponding driving data.

[0039] Filter the driving data according to the driving speed to determine the driving conditions within each set speed interval.

[0040] Step S22: Determine target vehicles whose driving conditions within at least one set speed interval meet the corresponding predetermined conditions according to the driving conditions of each vehicle within each set speed interval.

[0041] Among them, the set speed intervals correspond one-to-one with the set conditions.

[0042] During the driving process of a vehicle, there is often a large fluctuation range in the driving speed, and the operations of users at different driving speeds are also different. Here, the driving speed is divided into intervals, different predetermined conditions are set corresponding to different equipment speed intervals, and target vehicles that meet the corresponding predetermined conditions are determined according to the driving conditions of each vehicle within each speed interval, which is beneficial to improving the accuracy of target vehicle screening and thus beneficial to improving the accuracy of driving learning samples.

[0043] The set speed intervals are preset. For example, in some embodiments, the set speed intervals include a first speed interval and a second speed interval. Among them, the driving speed in the first speed interval is greater than or equal to a first speed threshold; the driving speed in the second speed interval is less than the first speed threshold. For example, the speed interval with a driving speed greater than or equal to 30 km / h is used as the first speed interval, and the speed interval with a driving speed less than 30 km / h is used as the second speed interval. It can be understood that in some embodiments, the first speed interval and the second speed interval can be further divided into more sub-intervals to achieve more detailed control. For example, in some embodiments, the first speed interval includes a high-speed interval and a medium-speed interval, and the second speed interval includes a low-speed interval. The driving speed in the high-speed interval is greater than or equal to 60 km / h, the driving speed in the medium-speed interval is less than 60 km / h and greater than or equal to 30 km / h, and the driving speed in the low-speed interval is less than 30 km / h and greater than or equal to 5 km / h.

[0044] The driving conditions of the target vehicle may only meet the corresponding set conditions within a partial set speed range. When determining the driving learning samples, this partial driving condition of the target vehicle that meets the set conditions should be selected instead of all the driving conditions of the target vehicle. Specifically, according to the vehicle identification number of the target vehicle, a driving video data collection instruction is issued to each target vehicle, including: according to the identification number of the target vehicle, a driving video data collection instruction within the corresponding set speed range is issued to each target vehicle. According to the collected driving video data of the target vehicle, the driving learning samples are determined, including: according to the collected driving video data of the target vehicle within the corresponding set speed range, the driving learning samples are determined. For example, if the driving conditions of a target vehicle within the first speed range meet the set conditions, while the driving conditions within the second speed range do not meet the set conditions, then a driving video data collection instruction within the first speed range is issued to this target vehicle to collect the driving video data of this target vehicle within the first speed range for determining the driving learning samples.

[0045] In some embodiments, the driving data corresponding to the driving speed includes the driving mileage used to represent the distance change during the vehicle driving process. According to the corresponding driving data, the driving conditions of each vehicle within each set speed range are determined, including: according to the driving mileage, the driving conditions of each vehicle within each set speed range are determined. Corresponding to a vehicle, the driving mileage at each driving speed is obtained, and the driving speeds belonging to the same set speed range are combined and calculated to determine the driving conditions of this vehicle within each set speed range.

[0046] Corresponding to different set speed ranges, there are also differences in the specific steps of the method for determining the driving learning samples. Here, the relevant steps when the driving speed is within the first speed range are described. Combining Figure 3 As shown, the embodiment of the present application provides another method for determining the driving learning samples. The method for determining the driving learning samples includes steps S10 to step S30.

[0047] Step S10, obtain the driving data set; the driving data set includes the driving data during the driving process of each vehicle.

[0048] The driving data includes the driving speed and the driving data corresponding to the driving speed. The driving data corresponding to the driving speed includes the driving mileage used to represent the distance change during the vehicle driving process and the control data used to represent the driving operation input to the vehicle driving control system during the vehicle driving process.

[0049] Step S211, according to the driving mileage and control data of each vehicle within the first speed range, determine the total driving mileage and the average value of the control data of each vehicle within the first speed range.

[0050] The average value of the manipulated data here can be the average value relative to the mileage, that is, the average operation data per set mileage. For example, the average operation data per kilometer, the average operation data per ten kilometers, etc.

[0051] Step S221: Determine target vehicles whose total driving mileage and average value of the manipulated data in the first speed range both meet the first predetermined condition according to the total driving mileage and the average value of the manipulated data of each vehicle in the first speed range.

[0052] Step S30: Determine a driving learning sample according to the target vehicles.

[0053] In this way, the target vehicles are determined by jointly screening according to the driving mileage and the manipulated data of each vehicle in the first speed range, improving the accuracy of target vehicle screening, and thus optimizing the data accuracy of the finally determined driving learning sample.

[0054] In some embodiments, determining target vehicles whose total driving mileage and average value of the manipulated data in the first speed range both meet the first predetermined condition according to the total driving mileage and the average value of the manipulated data of each vehicle in the first speed range includes: screening out vehicles whose total driving mileage in the first speed range is greater than the first mileage threshold as candidate vehicles according to the total driving mileage of each vehicle in the first speed range. Determine target vehicles whose average value of the manipulated data meets the predetermined safe operation screening condition according to the average value of the manipulated data of each candidate vehicle in the first speed range.

[0055] In the process of determining the target vehicles, a primary screening is performed according to the total driving mileage to determine candidate vehicles whose total driving mileage is greater than the first mileage threshold, so as to ensure the validity of the data of the finally determined vehicles and avoid the operation data in the data of vehicles with too small total driving mileage from interfering with the process of subsequently determining the target vehicles. After determining the candidate vehicles, a secondary screening is performed through the operation data of the vehicles to determine vehicles with higher driving safety as the target vehicles, so as to ensure the safety of the finally determined driving learning sample.

[0056] In the case where the first speed range includes multiple sub-ranges, the values of the first mileage thresholds corresponding to the multiple sub-ranges may be different. For example, in the case where the first speed range includes a high-speed range and a medium-speed range, for the high-speed range, screen out vehicles whose total driving mileage in the high-speed range is greater than 3000 km as candidate vehicles; for the medium-speed range, screen out vehicles whose total driving mileage in the medium-speed range is greater than 2000 km as candidate vehicles. The higher the driving speed in the sub-range, the higher the corresponding first mileage threshold.

[0057] Further, in some embodiments, vehicles with a total driving mileage greater than a first mileage threshold within a first speed range are selected as candidate vehicles according to the total driving mileage of each vehicle within the first speed range, including: vehicles with a total driving mileage greater than the first mileage threshold within the first speed range within a first preset time range are selected as candidate vehicles according to the total driving mileage of each vehicle within the first speed range. For example, vehicles with a total driving mileage greater than 2000 km within the first speed range in the past four months are selected. Adding such a time limit condition for calculating the total driving mileage helps to further ensure the validity of the data. Further, in the case where the first speed range includes multiple sub-ranges, the values of the first preset duration range and the first mileage threshold corresponding to the multiple sub-ranges may also be different. For example, in the case where the first speed range includes a high-speed range and a medium-speed range, for the high-speed range, vehicles with a total driving mileage greater than 3000 km within the high-speed range in the past four months are selected as candidate vehicles; for the medium-speed range, vehicles with a total driving mileage greater than 2000 km within the medium-speed range in the past four months are selected as candidate vehicles.

[0058] In some embodiments, the aforementioned average value of the control data includes the average deceleration used to represent the deceleration condition during vehicle driving. Here, the process of obtaining the average deceleration is specifically described. Corresponding to this average deceleration, the control data includes deceleration values. In the driving dataset, driving data with a deceleration greater than or equal to a set deceleration threshold when the brake pedal signal is set is selected as the deceleration value. Then, the average deceleration per set mileage is calculated based on the obtained deceleration values, such as the average deceleration per kilometer. The set deceleration threshold is, for example, 1 m / s 2 .

[0059] During the implementation process, considering that the values of the deceleration in the actual driving process are relatively random, calculating the average deceleration based on the directly detected deceleration values may result in excessive workload. In the process of calculating the average deceleration per set mileage based on the deceleration values, the data can be simplified. Specifically, multiple deceleration ranges are delimited, and a set deceleration value within the same deceleration range is taken as an approximate value to represent all the deceleration values within that deceleration range. For example, for all the deceleration values within the range of 1 m / s 2 -2.5 m / s 2 range, the mid-value of this range, 1.75 m / s 2 is taken as the representative value, thus avoiding the calculation process for data such as 1.23572 m / s 2 while ensuring accuracy. It should be noted that the deceleration value corresponds one-to-one with the representative value used to represent the deceleration value, and the number of deceleration values remains unchanged.

[0060] In addition, during the implementation process, considering that different deceleration values represent different braking intensities, different weights can be set for the deceleration values within different deceleration intervals to affect the finally determined average deceleration, so that the average deceleration can more accurately reflect the braking intensity during vehicle driving and more accurately reflect driving safety. The greater the deceleration value within the interval, the higher the weight of the interval. For example, for deceleration values within the range of 1 m / s 2 -2.5 m / s 2 no processing is performed on all deceleration values within the interval; for all deceleration values within the range of 2.5 m / s 2 -4 m / s 2 all deceleration values within the interval are multiplied by a set multiple greater than 1. The set multiple is set to 40, for example.

[0061] In some embodiments, when the average value of the control data includes the average deceleration used to represent the deceleration condition during vehicle driving, determining the target vehicle whose average value of the control data meets the predetermined safe operation screening condition according to the average value of the control data of each candidate vehicle within the first speed interval includes: determining the N candidate vehicles with the smallest average deceleration according to the average deceleration of each candidate vehicle within the first speed interval as the target vehicles. The smaller the average deceleration of the vehicle, the less braking behavior and the greater the braking degree the vehicle generates during driving, and the higher the safety and stability during driving. Therefore, using the average deceleration as the screening condition in the process of determining the target vehicle is conducive to screening out vehicles with higher safety as the target vehicles for determining the driving learning samples, thereby improving the quality of the driving learning samples.

[0062] As a feasible implementation manner, during the implementation process, after calculating the average deceleration of each candidate vehicle within the first speed interval, sort the average deceleration of each vehicle to determine the smallest N average decelerations according to the sorting, and then determine the corresponding N candidate vehicles. During the implementation process, ascending ranking can be performed to determine the first N average decelerations in the ranking, and then determine the N candidate vehicles with the smallest average deceleration. N is a set positive integer value, for example, 300.

[0063] In some embodiments, the average value of the control data includes an average steering angle for representing the steering condition of the steering wheel. Here, the process of obtaining the average steering angle is specifically described. Corresponding to the average steering angle, the control data includes the steering angle value of the steering wheel. In the driving dataset, when the steering wheel angle changes, the driving data with a steering angle greater than or equal to the set steering angle threshold is screened out as the steering angle value here. Then, the average steering angle per set mileage, such as the average steering angle per kilometer, is calculated based on the obtained steering angle value. The set steering angle threshold here is, for example, 10°. Here, when the first speed range includes multiple sub-ranges, corresponding to different sub-ranges, the set angle threshold can be set to different values. For example, when the first speed range includes a high-speed range and a medium-speed range, corresponding to the high-speed range, the set steering angle threshold can be taken as 10°; corresponding to the medium-speed range, the set steering angle threshold can be taken as 45°.

[0064] In some embodiments, when the average value of the control data includes an average steering angle for representing the steering condition of the steering wheel, determining a target vehicle whose average value of the control data meets the set safety operation screening conditions based on the average value of the control data of each candidate vehicle in the first speed range includes: determining M candidate vehicles with the smallest average steering angle as the target vehicle according to the average steering angle of each candidate vehicle in the first speed range.

[0065] Specifically, in some embodiments, after calculating the average steering angle of each candidate vehicle in the first speed range, the average steering angular velocity of each vehicle is sorted to determine the smallest M average steering angles according to the sorting, and then the corresponding M candidate vehicles are determined. During the implementation process, ascending ranking can be performed to determine the first M average steering angles in the ranking, and then the M candidate vehicles with the smallest average steering angle are determined. M is a set positive integer value, for example, 300.

[0066] In some embodiments, the average value of the control data includes an average deceleration for representing the deceleration condition during vehicle driving and an average steering angle for representing the steering condition of the steering wheel; determining a target vehicle whose average value of the control data meets the set safety operation screening conditions based on the average value of the control data of each candidate vehicle in the first speed range includes: determining N candidate vehicles with the smallest average deceleration according to the average deceleration of each candidate vehicle in the first speed range. Determining M candidate vehicles with the smallest average steering angle according to the average steering angle of each candidate vehicle in the first speed range. Determining L candidate vehicles with the smallest comprehensive average deceleration and average steering angle as the target vehicle according to the average deceleration and average steering angle of the N candidate vehicles and the M candidate vehicles in the first speed range. Wherein, N, M, and L are all set positive integer values.

[0067] Specifically, in some embodiments, after calculating the average deceleration and average steering angle of each vehicle within the first speed range, the average deceleration of each vehicle and the average steering angle of each vehicle are sorted respectively, so as to determine the smallest N average decelerations and the smallest M average steering angles according to their respective rankings, and then determine N candidate vehicles corresponding to the average deceleration and M candidate vehicles corresponding to the average steering angle. During the implementation process, the average deceleration of each vehicle and the average steering angle of each vehicle can be ranked in ascending order, and the first N average decelerations in the ascending order of the average deceleration are determined to determine the N vehicles with the smallest average deceleration; the first M average steering angles in the ascending order of the average steering angle are determined to determine the M vehicles with the smallest average steering angle.

[0068] During the process of determining L candidate vehicles with the smallest comprehensive value of the average deceleration and the average steering angle, the comprehensive values of the aforementioned N candidate vehicles and the aforementioned M candidate vehicles are determined respectively, and the comprehensive values are sorted to determine L candidate vehicles with the smallest comprehensive value of the average deceleration and the average steering angle. It can be understood that the comprehensive value is a function with the average deceleration and the average steering angle as independent variables. More specifically, in some embodiments, the comprehensive value is Y = k1·a + k2·θ. Wherein, k1 and k2 are both set weight values, a is the average deceleration of each vehicle, and θ is the average steering angle of each vehicle. In this way, a strong correlation relationship can be established between the comprehensive value and the average deceleration and the average steering angle, so as to accurately determine L candidate vehicles with the smallest comprehensive value of the average deceleration and the average steering angle. Further, k1 is greater than f2. To reflect the priority consideration of the average deceleration. For the possible overlapping part among the N candidate vehicles and the M candidate vehicles, that is, in the case where the same vehicle belongs to both the N candidate vehicles and the M candidate vehicles, the overlapping items are merged and not sorted repeatedly. Here, L is less than N and less than M. For example, N and M are set to 300, and L is set to 100.

[0069] In the case where the set speed range includes a second speed range where the driving speed is less than the first speed threshold, in combination with Figure 4 As shown, the method for determining the driving learning sample includes steps S10 to S30.

[0070] Step S10, obtaining a driving data set; the driving data set includes the driving data of each vehicle during driving.

[0071] The driving data includes the driving speed and the driving data corresponding to the driving speed, and the driving data corresponding to the driving speed includes the driving mileage used to represent the distance change during the driving of the vehicle.

[0072] Step S212: Determine the total driving mileage of the vehicle in the second speed range according to the driving mileage of each vehicle in the second speed range.

[0073] Step S222: Determine the target vehicles whose total driving mileage in the second speed range meets the second predetermined condition according to the total driving mileage of each vehicle in the second speed range.

[0074] Step S30: Determine the driving learning samples according to the target vehicles.

[0075] Vehicles often decelerate to the second speed range with a relatively low driving speed for operations such as braking and steering. At this time, these operations are safety behaviors, and it is relatively inaccurate to use them to evaluate the safety of driving operations. Therefore, in the second speed range with a relatively low driving speed, only the driving mileage is used as the screening condition for target vehicles to save the logical process.

[0076] In some embodiments, determining the target vehicles whose total driving mileage in the second speed range meets the second predetermined condition according to the total driving mileage of each vehicle in the second speed range includes: screening out the vehicles whose total driving mileage in the first speed range is greater than the second mileage threshold according to the total driving mileage of each vehicle in the second speed range, and using them as the target vehicles. For example, screening out the vehicles whose total driving mileage in the first speed range is greater than 1000 km as the target vehicles.

[0077] More specifically, screening out the vehicles whose total driving mileage in the first speed range is greater than the second mileage threshold according to the total driving mileage of each vehicle in the second speed range, and using them as the target vehicles, includes: screening out the vehicles whose total driving mileage in the first speed range within the second preset time interval is greater than the second mileage threshold according to the total driving mileage of each vehicle in the second speed range, and using them as the target vehicles. For example, screening out the vehicles whose total driving mileage in the first speed range within the second preset time interval is greater than the second mileage threshold as the target vehicles.

[0078] Combined with Figure 5 As shown, in some embodiments, step S30, determining the driving learning samples according to the target vehicles, includes step S31 and step S32.

[0079] Step S31: Label the driving styles of the target vehicles according to the driving data of each target vehicle.

[0080] Step S32: Determine the driving learning samples according to the labeled target vehicles.

[0081] Among them, the driving data includes the control data of the driving operations during the vehicle driving process.

[0082] In this way, after the target vehicle is labeled, targeted learning can be carried out for different driving styles, which is beneficial to improving the efficiency of the driving learning process. For example, the driving styles can be divided into an efficient type with higher flexibility and a comfortable type with higher stability.

[0083] Specifically, in some embodiments, the driving data includes the number of lane changes and / or deceleration values during the vehicle driving process. Labeling the driving styles of the target vehicles according to the driving data of each target vehicle includes labeling the driving styles of the target vehicles according to the number of lane changes and / or deceleration values of each target vehicle.

[0084] The process of obtaining the number of lane changes is described here. In the driving dataset, the driving data when the turn signal for lane change is set is filtered out to determine the number of lane changes. Further, in at least some embodiments, the driving data when the turn signal for lane change is set and the steering wheel steering angle does not exceed 90° during the continuous setting process is filtered out to determine the number of lane changes. In this way, it is possible to avoid misidentifying normal vehicle turning and vehicle U-turn processes as lane changes, resulting in inaccurate number of lane changes.

[0085] In some embodiments, labeling the driving styles of the target vehicles according to the number of lane changes of each target vehicle includes: determining the average number of lane changes of each target vehicle according to the number of lane changes of each target vehicle, determining that the driving style of the A target vehicles with the least average number of lane changes is the first driving style, for example, the comfortable type; determining that the driving style of the B target vehicles with the most average number of lane changes is the second driving style, for example, the efficient type. The average number of lane changes is the number of lane changes per set mileage, for example, the number of lane changes per kilometer. The first driving style is different from the second driving style, and both A and B are positive integer values less than the total number of target vehicles.

[0086] In some embodiments, labeling the driving styles of target vehicles according to the deceleration values of the target vehicles includes: determining the average deceleration of each target vehicle according to the deceleration values of the target vehicles, determining that the driving styles of C target vehicles with the minimum average deceleration are the third driving style, for example, a comfort style; determining that the driving styles of D target vehicles with the maximum average deceleration are the fourth driving style, for example, an efficiency style. The third driving style is different from the fourth driving style, and both C and D are positive integer values less than the total number of target vehicles. The specific calculation process of the average deceleration has been described in detail in the previous embodiments and will not be elaborated here. More specifically, according to the foregoing analysis, when the set speed range includes a first speed range and a second speed range, a vehicle that meets the set conditions within any set speed range can be used as a target vehicle. To improve the accuracy of style calibration, different style labeling methods can be used for vehicles that meet different set conditions. For example: In some embodiments, labeling the driving styles of target vehicles according to the driving data of the target vehicles includes: if the driving conditions of the target vehicle in the first speed range where the driving speed is greater than or equal to the first speed threshold meet the set conditions, then labeling the driving style of the target vehicle according to the number of lane changes during the driving process of the vehicle. In some embodiments, labeling the driving styles of target vehicles according to the driving data of the target vehicles includes: if the driving conditions of the target vehicle in the second speed range where the driving speed is less than the first speed threshold meet the set conditions, then labeling the driving style of the target vehicle according to the deceleration value during the driving process of the vehicle.

[0087] This application provides a system for determining driving learning samples, including a data acquisition module, a target vehicle determination module, and a sample determination module. Among them, the data acquisition module is used to obtain a driving data set. The driving data set includes the driving data of each vehicle during the driving process. The target vehicle determination module is used to determine the target vehicles whose driving conditions meet the predetermined conditions of safe driving habits according to the driving data. The sample determination module is used to determine the driving learning samples according to the target vehicles.

[0088] This application also provides an electronic device, including one or more processors, for implementing the foregoing method for determining driving learning samples. Specifically, the electronic device can be: a desktop computer, a portable computer, a smart mobile terminal, a server, a PDA (Personal Digital Assistant), a handheld terminal, etc. Among them, the PDA can include an industrial PDA and a consumer PDA. Any electronic device that can implement the embodiments of the present invention belongs to the protection scope of this application and will not be limited here.

[0089] The present application also provides a computer-readable storage medium, on which a program is stored. When the program is executed by a processor, the foregoing method for determining driving learning samples is implemented.

[0090] In the description of the present application, it should be understood that the terms "first", "second", etc. are only used for descriptive purposes, and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", etc. may explicitly or implicitly include at least one such feature. In the description of the present application, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise specifically defined.

Claims

1. A method for determining driving learning samples, characterized in that, including: obtaining a driving data set; the driving data set includes driving data during the driving process of each vehicle; determining, according to the driving data, target vehicles whose driving conditions meet a predetermined condition of safe driving habits; determining driving learning samples according to the target vehicles.

2. The method according to claim 1, wherein the driving data includes a driving speed and driving data corresponding to the driving speed; the determining, according to the driving data, target vehicles whose driving conditions meet a predetermined condition of safe driving habits includes: determining the driving conditions of each vehicle in each set speed interval according to the corresponding driving data; determining, according to the driving conditions of each vehicle in each set speed interval, target vehicles whose driving conditions in at least one of the set speed intervals meet the corresponding predetermined conditions; wherein, the set speed intervals correspond to set conditions one by one.

3. The method according to claim 2, wherein the driving data corresponding to the driving speed includes a driving mileage for indicating a distance change during the driving process of the vehicle; the determining, according to the corresponding driving data, the driving conditions of each vehicle in each set speed interval includes: determining the driving conditions of each vehicle in each set speed interval according to the driving mileage.

4. The method according to claim 3, wherein the set speed interval includes a first speed interval where the driving speed is greater than or equal to a first speed threshold; the driving data corresponding to the driving speed further includes control data for indicating that a vehicle driving control system is input by a driving operation during the driving process of the vehicle; the determining, according to the driving mileage, the driving conditions of each vehicle in each set speed interval includes: determining an average value of the total driving mileage and the control data of each vehicle in the first speed interval according to the driving mileage and the control data; the determining, according to the driving conditions of each vehicle in each set speed interval, target vehicles whose driving conditions in at least one of the set speed intervals meet the corresponding predetermined conditions includes: determining, according to the total driving mileage and the average value of the control data of each vehicle in the first speed interval, target vehicles whose average values of the total driving mileage and the control data in the first speed interval both meet a first predetermined condition.

5. The method according to claim 4, wherein the determining, according to the total driving mileage and the average value of the control data of each vehicle in the first speed interval, target vehicles whose average values of the total driving mileage and the control data in the first speed interval both meet a first predetermined condition includes: screening out vehicles whose total driving mileage in the first speed interval is greater than a first mileage threshold as candidate vehicles according to the total driving mileage of each vehicle in the first speed interval; determining target vehicles whose average value of the control data meets a predetermined safe operation screening condition according to the average value of the control data of each candidate vehicle in the first speed interval.

6. The method according to claim 5, wherein The average value of the control data includes the average deceleration used to represent the deceleration condition during the vehicle driving; determining the target vehicle whose average value of the control data meets the predetermined safe operation screening condition according to the average value of the control data of each candidate vehicle within the first speed range includes: Determining the N candidate vehicles with the minimum average deceleration as the target vehicles according to the average deceleration of each candidate vehicle within the first speed range; or The average value of the control data includes the average steering angle used to represent the steering condition of the steering wheel; determining the target vehicle whose average value of the control data meets the set safe operation screening condition according to the average value of the control data of each candidate vehicle within the first speed range includes: Determining the M candidate vehicles with the minimum average steering angle as the target vehicles according to the average steering angle of each candidate vehicle within the first speed range; or The average value of the control data includes the average deceleration used to represent the deceleration condition during the vehicle driving and the average steering angle used to represent the steering condition of the steering wheel; determining the target vehicle whose average value of the control data meets the set safe operation screening condition according to the average value of the control data of each candidate vehicle within the first speed range includes: Determining the N candidate vehicles with the minimum average deceleration according to the average deceleration of each candidate vehicle within the first speed range; Determining the M candidate vehicles with the minimum average steering angle according to the average steering angle of each candidate vehicle within the first speed range; Determining the L candidate vehicles with the minimum comprehensive average deceleration and average steering angle as the target vehicles according to the average deceleration and average steering angle of the N candidate vehicles and the M candidate vehicles within the first speed range; wherein, N, M, and L are all set positive integer values.

7. The method according to claim 3, wherein The set speed range includes a second speed range where the driving speed is less than the first speed threshold; Determining the driving condition of the vehicle within each set speed range according to the driving mileage includes: Determining the total driving mileage of the vehicle within the second speed range according to the driving mileage; Determining the target vehicle whose driving condition within at least one of the set speed ranges meets the corresponding predetermined condition according to the driving condition of each vehicle within each set speed range includes: Determining the target vehicle whose total driving mileage within the second speed range meets the second predetermined condition according to the total driving mileage of each vehicle within the second speed range.

8. The method according to claim 7, wherein Determining the target vehicle whose total driving mileage within the second speed range meets the second predetermined condition according to the total driving mileage of each vehicle within the second speed range includes: Screening out the vehicles whose total driving mileage within the first speed range is greater than the second mileage threshold as the target vehicles according to the total driving mileage of each vehicle within the second speed range.

9. The method according to claim 1, wherein Determining a driving learning sample according to the target vehicle includes: Labeling the driving styles of the target vehicles according to the driving data of each target vehicle, and determining a driving learning sample according to the labeled target vehicles; Among them, the driving data includes the control data of driving operations during vehicle driving.

10. The method according to claim 9, wherein the driving data includes the number of lane changes and / or deceleration values during vehicle driving; and the labeling of the driving styles of the target vehicles according to the driving data of each target vehicle includes: Labeling the driving styles of the target vehicles according to the number of lane changes and / or deceleration values of each target vehicle.

11. The method according to claim 1, wherein determining the target vehicles whose driving conditions meet the set conditions according to the driving data includes: Determining the vehicle identification numbers of the target vehicles whose driving conditions meet the set conditions according to the driving data; Determining a driving learning sample according to the target vehicle includes: Issuing a driving video data acquisition instruction to each target vehicle according to the vehicle identification number of the target vehicle; Determining a driving learning sample according to the acquired driving video data of the target vehicle.

12. The method according to any one of claims 1 to 11, wherein each vehicle includes an in-vehicle information acquisition device, the in-vehicle information acquisition device is communicatively connected to a cloud server, and the in-vehicle acquisition device is used to upload the original driving data of each vehicle during driving to the cloud server, so that the cloud server processes the original driving data and forwards the processed driving data to a set columnar database management system; Obtaining a driving data set includes: Regularly querying the processed driving data in the columnar database management system to obtain the driving data of each vehicle during driving.

13. A system for determining driving learning samples, characterized in that, Includes: A data acquisition module for obtaining a driving data set; The driving data set includes the driving data of each vehicle during driving; A target vehicle determination module for determining target vehicles whose driving conditions meet a predetermined condition of safe driving habits according to the driving data; A sample determination module for determining a driving learning sample according to the target vehicle.

14. An electronic device, characterized in that, Includes one or more processors for implementing the method for determining a driving learning sample according to any one of claims 1 to 12.

15. A computer-readable storage medium, characterized in that, A program is stored thereon, and when the program is executed by the processor, the method for determining a driving learning sample according to any one of claims 1 to 12 is implemented.

Images