Driving style identification method and device, equipment and storage medium

Abstract

The embodiment of the invention discloses a driving style recognition method and device, equipment and a storage medium, and the method comprises the steps: obtaining the driving data of at least one user, and obtaining a driving data set; extracting a first feature and a second feature from each piece of driving data in the driving data set; wherein the first feature is related to a driving scene, and the second feature is related to a driving style; based on the first feature of each piece of driving data, determining a driving scene corresponding to each piece of driving data; based on the second feature of each piece of driving data, determining a corresponding driving style of each piece of driving data in the corresponding driving scene; and determining the driving style of each user in at least one driving scene based on the corresponding driving style of each driving data in the corresponding driving scene. Therefore, it can be ensured that the driving style recognition of the user is performed in a specific driving scene, so that the output driving style recognition result is more reliable and more persuasive.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile big data analysis, and in particular to a driving style recognition method, device, equipment and storage medium. Background Art

[0002] Today, with the continuous development and maturity of Internet of Vehicles technology, intelligent connected cars are becoming more and more popular, generating a large amount of user driving data. These big data records the various operations of users in daily use of cars and are stored on the cloud platforms of major car companies, waiting to be further analyzed and mined to create value.

[0003] Generally speaking, different drivers have different driving styles, which reflect the driver's judgment criteria for the external dynamic environment and the way of manipulating the car in real time while driving. For example, drivers with more aggressive styles often change lanes frequently and slam on the brakes when driving, while drivers with more conservative styles often maintain a safer following distance and step on the pedal more gently and slowly. By developing a driving style classification method based on the user's real driving data to identify the driving styles of different users, it is helpful to push personalized services such as power matching maps, intelligent driving models and energy management strategies to users in a targeted manner.

[0004] Therefore, how to achieve accurate driving style recognition is a strong demand of major automobile companies. Summary of the invention

[0005] The object of the present invention is to provide a driving style recognition method, device, equipment and storage medium to solve the problem of inaccurate driving style recognition results in the prior art.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] In a first aspect, an embodiment of the present invention provides a driving style recognition method, the method comprising: acquiring driving data of at least one user to obtain a driving data set; extracting a first feature and a second feature for each driving data in the driving data set; wherein the first feature is related to the driving scene, and the second feature is related to the driving style; based on the first feature of each driving data, determining the driving scene corresponding to each driving data; based on the second feature of each driving data, determining the driving style corresponding to each driving data in the corresponding driving scene; based on the driving style corresponding to each driving data in the corresponding driving scene, determining the driving style of each user in at least one driving scene.

[0008] In a second aspect, an embodiment of the present invention provides a driving style recognition device, which includes: an acquisition module, used to acquire driving data of at least one user to obtain a driving data set; a feature extraction module, used to extract a first feature and a second feature for each driving data in the driving data set; wherein the first feature is related to the driving scene, and the second feature is related to the driving style; a first determination module, used to determine the driving scene corresponding to each driving data based on the first feature of each driving data; a second determination module, used to determine the driving style corresponding to each driving data in the corresponding driving scene based on the second feature of each driving data; and a third determination module, used to determine the driving style of each user in at least one driving scene based on the driving style corresponding to each driving data in the corresponding driving scene.

[0009] In a third aspect, an embodiment of the present invention provides a driving style recognition device, which includes a memory and a processor; wherein the memory is used to store computer-executable instructions; the processor is connected to the memory and is used to implement the method described in the first aspect by executing the computer-executable instructions.

[0010] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, the method described in the first aspect is implemented.

[0011] The beneficial effects of the present invention are as follows: the method first associates the driving data of each user with the driving scene, then determines the driving style corresponding to the driving data in each driving scene, and then determines the driving style of each user in at least one specific driving scene based on the driving style corresponding to the driving data in each scene. In this way, it can be ensured that the user's driving style recognition is performed in a specific driving scene, so that the output driving style recognition result is more accurate, reliable, and more convincing.

[0012] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings herein are incorporated into the specification and constitute a part of this specification. These drawings illustrate embodiments consistent with the present invention and are used to explain the technical solutions of the present invention together with the specification.

[0014] Figure 1 A schematic diagram of a driving style identification method provided by an embodiment of the present invention;

[0015] Figure 2A possible implementation flow diagram of the driving style recognition method provided by an embodiment of the present invention;

[0016] Figure 3 A schematic diagram of the structure of a driving style recognition device provided by an embodiment of the present invention;

[0017] Figure 4 A schematic diagram of a hardware entity of a driving style recognition device in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to enable a more detailed understanding of the features and technical contents of the embodiments of the present invention, the implementation of the embodiments of the present invention is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not intended to limit the embodiments of the present invention.

[0019] Unless otherwise defined, all technical and scientific terms used in the embodiments of the present invention have the same meanings as those commonly understood by those skilled in the art of the present invention. The terms used in the embodiments of the present invention are only for the purpose of describing the embodiments of the present invention and are not intended to limit the present invention.

[0020] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. It is also necessary to point out that the terms "first\second\third" involved in the embodiments of the present invention are only used to distinguish similar objects and do not represent a specific order for the objects. It is understood that "first\second\third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of the present invention described herein can be implemented in an order other than that illustrated or described herein.

[0021] It should be understood that the term "and / or" in the embodiments of the present invention is only a description of the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0022] Today, with the continuous development and maturity of Internet of Vehicles technology, intelligent connected cars are becoming more and more popular and entering thousands of households, generating a large amount of user driving data. These big data records the various operations of users in daily use of cars and are stored on the cloud platforms of major car companies, waiting to be further analyzed and mined to create value.

[0023] Generally speaking, different drivers have different driving styles, which reflect the driver's judgment criteria for the external dynamic environment and the way of manipulating the car in real time when driving. For example, drivers with more aggressive styles often change lanes frequently and slam on the brakes when driving, while drivers with more conservative styles often maintain a safer following distance when driving and are gentler and slower when pedaling. Developing a driving style classification method based on the user's real driving data to identify the driving styles of different users is conducive to the subsequent targeted push of personalized services such as power matching maps, intelligent driving models, and energy management strategies to users. Therefore, how to achieve accurate driving style recognition, so as to achieve "thousands of faces" in car driving, is a strong demand of major car companies today.

[0024] In existing driving style recognition technologies, researchers usually analyze all driving data together without considering driving scenario factors. In fact, the data comparability in different scenarios is poor, resulting in insufficient persuasiveness in style recognition results. For example, the same vehicle speed of 80km / h should tend to be dangerous on city streets and conservative on highways. At the same time, the driving style identified by existing methods is fixed, and the impact of driving scenario changes on the user's driving style is not considered. For example, when driving in an unfamiliar environment or when there are other people riding, the driver's driving style tends to be conservative. In addition, some researchers only considered the user's driving style in a single special scenario, and a large amount of driving data and driving scenarios were discarded. The user style results obtained are not comprehensive enough and can only be applied to a few real driving scenarios, which cannot meet the needs of service users.

[0025] In view of this, an embodiment of the present invention provides a driving style recognition method, device, equipment and storage medium, which can be executed by a processor of a computer device. Among them, the computer device may refer to a server, a laptop, a tablet computer, a desktop computer, and other devices with data processing capabilities. In this method, first, the driving data of at least one user can be obtained to obtain a driving data set, and a first feature and a second feature are extracted for each driving data in the driving data set. Among them, the first feature is related to the driving scene, and the second feature is related to the driving style. Subsequently, based on the first feature of each driving data, the driving scene corresponding to each driving data can be determined, and based on the second feature of each driving data, the driving style corresponding to each driving data in the corresponding driving scene can be determined. Furthermore, based on the driving style corresponding to each driving data in the corresponding driving scene, the driving style of each user in at least one driving scene can be determined.

[0026] According to the method of the embodiment of the present invention, the driving data of each user is first associated with the driving scene, and then the driving style corresponding to the driving data in the scene is determined for each driving scene. Then, based on the driving style corresponding to the driving data in each scene, the driving style of each user in at least one specific driving scene is determined. In this way, it can be ensured that the user's driving style recognition is performed in a specific driving scene, so that the output driving style recognition result is more reliable and more convincing.

[0027] It should be understood that the “driving style identification” in the embodiment of the present invention can also be understood as “driving style classification”.

[0028] To facilitate understanding of the technical solution of the embodiment of the present invention, the technical solution of the present invention is described in detail below through specific embodiments. The above related technologies can be arbitrarily combined with the technical solution of the embodiment of the present invention as optional solutions, which all belong to the protection scope of the embodiment of the present invention. The embodiment of the present invention includes at least part of the following contents.

[0029] Figure 1 A driving style recognition method provided by an embodiment of the present invention is shown. The method may include:

[0030] S101, acquiring driving data of at least one user to obtain a driving data set.

[0031] The driving data set may also be understood as a driving data set consisting of driving data (or referred to as driving event data) of each user of the at least one user.

[0032] In some embodiments, each item of driving data in the driving data set may correspond to a driving event, or in other words, each item of driving data is data collected during the occurrence of a corresponding driving event, wherein a driving event refers to a short driving process with a change in the state of the vehicle as the starting point and the end point.

[0033] As an example, the driving event type in the embodiment of the present invention may be a start-stop event or a pedal event. A start-stop event refers to a driving process from when the car starts to when it shuts down, and a pedal event refers to a driving process when the driver steps on the pedal. When the driving event type is a start-stop event, the corresponding driving data refers to the data collected from when the car starts to when it shuts down; when the driving event type is a pedal event, the corresponding driving data refers to the data collected during the time period when the car pedal is stepped on.

[0034] It should be understood that the driving event in the embodiment of the present invention may also be other types of driving events, which are not limited in the embodiment of the present invention.

[0035] It should be noted that different driving data in the driving data set may correspond to different driving events of the same type. For example, the driving event type corresponding to the driving data in the driving data set is a start-stop event, and the start time of the start-stop event corresponding to driving data #1 is t1 and the end time is t2, then driving data #1 is the driving data collected between t1 and t2; for another example, the start time of the start-stop event corresponding to driving data #2 is t3 and the end time is t4, then driving data #2 is the driving data collected between t3 and t4.

[0036] Exemplarily, in order to obtain a driving data set, for example, the real driving data of at least one user (hereinafter referred to as driving data) can be downloaded from the automobile big data accumulated by the automobile enterprise cloud platform. Among them, the driving data can be various data uploaded to the cloud by the vehicle end when the user drives the intelligent networked car. For example, the national standard stipulates the uploaded data, fault data, human-computer interaction data, and driving control behavior data. Further, the driving data corresponding to multiple driving events of the same type can be extracted from the driving data of at least one user downloaded to form a driving data set. For example, when extracting the driving data corresponding to the start-stop event, the field signal "vehicle speed" can be used as the basis for determining the start and end time of the start-stop event. For example, the moment when the vehicle speed changes from zero to non-zero can be used as the start time of a start-stop event, and the moment when the vehicle speed changes from non-zero to zero can be used as the end time of the start-stop event, so that the driving data corresponding to the start-stop event of each user can be extracted.

[0037] In some embodiments, before extracting driving data from the user's driving data, the method may further include: performing a preprocessing operation on the user's driving data. As an example, the preprocessing operation on the driving data may include the following steps 1011) to 1013):

[0038] 1011) Data screening: Screening out signal fields required for scene recognition and driving style recognition from driving data. For example, data that directly reflects the user's driving control behavior of the car and / or signals provided by various vehicle sensors may be screened out.

[0039] Among them, the data that can directly reflect the user's driving control behavior of the car may include, but are not limited to, at least one of the following: accelerator pedal travel value, brake pedal travel value, actual gear position, steering wheel angle and speed, horn, front and rear lights, wipers, etc. The signals provided by various vehicle-mounted sensors may include, but are not limited to, at least one of the following: vehicle speed, lateral acceleration, longitudinal acceleration, slope, longitude and latitude, passenger seat seat belt buckle switch, etc.

[0040] As an example, the filtered signal fields may include at least one of the following: vehicle-side time, total mileage, vehicle speed, lateral acceleration, longitudinal acceleration, driving mode, accelerator pedal travel value, brake pedal travel value, steering wheel angle, slope, driving mode, actual gear position, horn switch, low beam switch, high beam switch, wiper switch, headway, head distance, and road curvature.

[0041] 1012) Data cleaning: including duplicate value processing, outlier value removal, missing value filling, etc.

[0042] In some embodiments, duplicate value processing can be achieved, for example, by deleting all duplicate items except the first item. For example, driving data can be composed of a "time" field and other signal fields. Then, if the "time" field and other signal fields contained in driving data A, driving data B, and driving data C are the same, then driving data B and driving data C can be deleted and driving data A can be retained. In some embodiments, abnormal value deletion can be achieved, for example, by removing disordered data. For example, if the "time" field corresponding to a certain driving data is abnormal (such as abnormal format), then the driving data can be removed. In some embodiments, missing values ​​can be filled with zero values, for example.

[0043] 1013) Timestamp sorting: Rearrange driving data in ascending order by the "time" field.

[0044] In some embodiments, the driving data in the driving data set may be in a matrix form of T×N, for example, where T is the duration of the driving event and N is the number of signal fields screened in step 1011).

[0045] S102, extracting a first feature and a second feature from each driving data in the driving data set; wherein the first feature is related to the driving scene, and the second feature is related to the driving style.

[0046] In this step, the first feature and the second feature may be extracted from each driving data in the driving data set, and sample data corresponding to the first feature and sample data corresponding to the second feature may be obtained respectively.

[0047] For example, for a driving data item in a format of T×N, the first feature of the driving data may be extracted to form a sample data item in a format of 1×M1, and the second feature of the driving data may be extracted to form a sample data item in a format of 1×M2, where M1 is the number of the extracted first features and M2 is the number of the extracted second features.

[0048] In some scenarios, the first feature and the second feature may be collectively referred to as high-dimensional features. Exemplarily, the high-dimensional features in the embodiment of the present invention may be selected from at least one of the following: average accelerator pedal stroke value, average accelerator pedal pressing angular velocity, accelerator pedal pressing frequency, average brake pedal stroke value, average brake pedal pressing angular velocity, brake pedal pressing frequency, average steering wheel rotation angular velocity, steering wheel rotation frequency, gear shifting frequency, low gear time proportion, high gear time proportion, driving mode mean, vehicle speed mean, vehicle speed standard deviation, low speed time proportion, high speed time proportion, start-stop event duration, horn frequency, wiper on time proportion, light on time proportion, average following distance, lane change frequency.

[0049] In some embodiments, the first feature is related to the driving scene, and the second feature is related to the driving style. As an example, the first feature may include at least one of the following: average vehicle speed, duration of start-stop events, acceleration and deceleration frequency, pedal pressing frequency, percentage of wiper on time, percentage of fog light on time. The second feature may include at least one of the following: average following distance, lane change frequency, average pedal pressing angular velocity, average steering wheel rotation angular velocity, etc.

[0050] Exemplarily, the first feature is related to the driving scene, and it can also be understood that the corresponding driving scene can be inferred based on the value size / interval of the first feature, or the first feature can be used for driving scene recognition. For example, when the driving scene includes different driving areas, the first feature can be selected as the duration of the start-stop event. Since the duration of the start-stop event in the "highway section" is usually > the duration of the start-stop event in the "rural section" > the duration of the start-stop event in the "urban section", therefore, based on the first feature of the duration of the start-stop event, it can be inferred that the corresponding driving area is a "highway section", "rural section" or "urban section". For another example, when the driving scene includes different congestion levels, the first feature can be selected as the acceleration and deceleration frequency. Since the acceleration and deceleration frequency in "congestion" is usually > the acceleration and deceleration frequency in "normal" > the acceleration and deceleration frequency in "unblocked", therefore, based on the first feature of the acceleration and deceleration frequency, the corresponding congestion level can be inferred as "congestion", "normal" or "unblocked".

[0051] Exemplarily, the second feature is related to the driving style, and it can also be understood that the corresponding driving style can be inferred based on the value size / range of the second feature, or in other words, the second feature can be used for driving style identification. For example, the second feature can be selected as the lane change frequency. Since usually, the lane change frequency of a driver with an "aggressive" driving style> the lane change frequency of a driver with a "normal" driving style> the lane change frequency of a driver with a "conservative" driving style, therefore, based on the second feature of the lane change frequency, it can be inferred that the driver's corresponding driving style is "aggressive", "normal" or "conservative".

[0052] S103: Determine a driving scene corresponding to each driving data based on the first feature of each driving data.

[0053] Since the first feature of the driving data is related to the driving scene, in this step, the driving scene corresponding to each driving data may be determined based on the first feature of each driving data.

[0054] In a possible manner, based on the first feature of each driving data, determining the driving scene corresponding to each driving data may include the following steps 1031) to 1033):

[0055] 1031) Based on the first feature of each driving data, a clustering algorithm is used to divide the driving data in the driving data set into N categories, where N is a positive integer. N can be determined according to the number of driving scenes to be divided, for example.

[0056] For example, suppose the driving data set includes L (L is a positive integer) driving data items, which are denoted as {p1, p2, …, p L}, then, after extracting the first feature of each driving data, the sample data corresponding to the first feature of each driving data can be obtained. The sample data corresponding to the first feature of L items of driving data are, for example, respectively recorded as {x1, x2, …, x L As an implementation method, the clustering algorithm may be, for example, a k-means clustering algorithm. In step 1031), the number of clusters may be set to N, so that {x1, x2, ..., x L} is divided into N categories. Assume that a sample data x l (l=1,2,…,L) is classified into the nth category among N categories, then it means that the driving data p l For example, if the sample data p2 is classified into the first category, it means that the driving data p2 is also classified into the first category.

[0057] 1032) Based on the first prior information related to the driving scene, determine the driving scene corresponding to each category in the N categories.

[0058] The first priori information may be obtained, for example, based on human experience and / or statistical analysis of driving data in different driving scenarios.

[0059] In one example, assuming that the driving scene includes N different driving areas, after the driving data is divided into N categories, the driving area corresponding to each category can be determined based on the first prior information. In another example, assuming that the driving scene includes N different congestion levels, after the driving data is divided into N categories, the congestion level corresponding to each category can be determined based on the first prior information.

[0060] 1033) The driving scene corresponding to the nth category in the N categories is used as the driving scene corresponding to each driving data in the nth category, where n is a positive integer less than or equal to N.

[0061] For example, assuming that the driving data classified into the nth category in step 1031) includes p1, p3, and p6, then, if in step 1032), the driving area corresponding to the nth category is determined to be "urban road section", then the "urban road section" can be used as the driving area corresponding to the driving data p1, p3, and p6. Similarly, if in step 1032), the congestion level corresponding to the nth category is determined to be "congested", then "congested" can be used as the congestion level corresponding to the driving data p1, p3, and p6.

[0062] For ease of understanding, the following takes a driving scenario including different driving areas / congestion levels / weather conditions as an example to introduce a possible implementation method for determining a driving scenario corresponding to each driving data.

[0063] In the first possible case, the driving scene may include a first driving area and a second driving area. In this case, the first prior information may include, for example: the value of the first feature in the first driving area is greater than the value of the first feature in the second driving area. For example, assuming that the first driving area is a "highway section" and the second driving area is a "city section", then the first feature may be, for example, the average vehicle speed or the duration of the start-stop event, and the first prior information may be, for example: the average vehicle speed / duration of the start-stop event in the "highway section"> the average vehicle speed / duration of the start-stop event in the "city section".

[0064] It should be noted that the above driving scenario including the first driving area and the second driving area is only an example for the convenience of explanation. In practice, the driving scenario may also include more (such as 3 or more) driving areas. For example, it may include "highway section", "urban section" and "rural section". In this case, the first prior information may include, for example: the average vehicle speed of the "highway section" > the average vehicle speed of the "urban section" > the average vehicle speed of the "rural section", or the duration of the start-stop event of the "highway section" > the duration of the start-stop event of the "rural section" > the duration of the start-stop event of the "urban section".

[0065] Taking the first feature as the duration of the start-stop event as an example, after obtaining the first prior information, the corresponding driving area can be determined for each driving data based on the first prior information. For example, taking the driving scene including three driving areas of "highway section", "urban section" and "rural section" as an example, the driving data can be divided into three categories through step 1031). Assuming that the duration of the start-stop event corresponding to the driving data in category #1>the duration of the start-stop event corresponding to the driving data in category #2>the duration of the start-stop event corresponding to the driving data in category #3, then, since it is known based on the first prior information that the duration of the start-stop event of the "highway section">the duration of the start-stop event of the "rural section">the duration of the start-stop event of the "urban section", the driving scenes corresponding to category #1, category #2, and category #3 can be determined as "highway section", "rural section" and "urban section" respectively through step 1032). Furthermore, through step 1033), the driving scene corresponding to each driving data in category #1 can be determined as a "highway section", the driving scene corresponding to each driving data in category #2 can be determined as a "rural section", and the driving scene corresponding to each driving data in category #3 can be determined as a "city section".

[0066] It should be understood that the “value of the first feature corresponding to the driving data in category A (such as the duration of the start-stop event) > the value of the first feature corresponding to the driving data in category B” mentioned in the embodiments of the present invention may, for example, mean that the value of the first feature corresponding to the driving data in category A is generally greater than the value of the first feature corresponding to the driving data in category B (such as the average value of the first feature corresponding to the driving data in category A > the average value of the first feature corresponding to the driving data in category B), or may mean that the values ​​of the first feature corresponding to all driving data in category A are greater than the values ​​of the first feature corresponding to the driving data in category B.

[0067] In the second possible scenario, the driving scenario may include a first congestion level and a second congestion level. In this case, the first prior information may include, for example: the value of the first feature when the congestion level is the first congestion level is greater than the value of the first feature when the congestion level is the second congestion level. For example, assuming that the first congestion level is "congested" and the second congestion level is "unimpeded", then the first feature may be, for example, the acceleration / deceleration frequency or the pedal pressing frequency, and the first prior information may be, for example: the acceleration / deceleration frequency / pedal pressing frequency when the congestion level is "congested" > the acceleration / deceleration frequency / pedal pressing frequency when the congestion level is "unimpeded".

[0068] It should be noted that the above driving scenarios including the first congestion level and the second congestion level are only examples for the convenience of explanation. In practice, the driving scenarios may also include more (such as 3 or more) congestion levels. For example, they may include "congested", "normal" and "unimpeded". In this case, the first prior information may include, for example: the acceleration / deceleration frequency / pedal pressing frequency when the congestion level is "congested" > the acceleration / deceleration frequency / pedal pressing frequency when the congestion level is "normal" > the acceleration / deceleration frequency / pedal pressing frequency when the congestion level is "unimpeded".

[0069] Further, through steps 1031) to 1033), a corresponding congestion level can be determined for each driving data. It should be understood that the implementation method of determining the corresponding congestion level for each driving data is similar to the implementation method of determining the corresponding driving area for each driving data, which will not be repeated here.

[0070] In a third possible scenario, the driving scenario may include a first weather condition and a second weather condition. In this case, the first prior information may include, for example: the value of the first feature under the first weather condition is greater than the value of the first feature under the second weather condition. For example, assuming that the first weather condition is "bad" and the second weather condition is "good", then the first feature may be, for example, the percentage of time when the wipers are on or the percentage of time when the fog lights are on, and the first prior information may be, for example: the percentage of time when the wipers are on / the percentage of time when the fog lights are on under "bad" weather conditions> the percentage of time when the wipers are on / the percentage of time when the fog lights are on under "good" weather conditions.

[0071] It should be noted that the above-mentioned driving scenario including the first weather condition and the second weather condition is only an example for convenience of explanation. In practice, the driving scenario may also include more (such as 3 or more) weather conditions, which is not limited in the embodiments of the present invention.

[0072] Further, through steps 1031) to 1033), corresponding weather conditions can be determined for each driving data. It should be understood that the implementation method of determining the corresponding weather conditions for each driving data is similar to the implementation method of determining the corresponding driving area for each driving data, which will not be repeated here.

[0073] According to the method of the embodiment of the present invention, the driving scene (such as driving area / congestion level / weather condition) corresponding to each driving data can be determined based on the first feature of each driving data. This process is also the process of scene recognition.

[0074] It should be noted that, in some embodiments, the driving scene may also include scenes of multiple dimensions at the same time. For example, scenes of three dimensions, namely, driving area, congestion level, and weather conditions, may be included at the same time. When the driving scene includes scenes of multiple dimensions at the same time, each driving data may correspond to scenes of multiple different dimensions at the same time. For example, the driving data p l The corresponding driving area is "urban road section", the corresponding congestion level is "congested", and the corresponding weather condition is "good". In this case, the driving scene finally corresponding to the driving data can be a sub-scene composed of the multiple scenes of different dimensions. For example, the driving data p l The final corresponding driving scenario is: urban congestion driving sub-scenario under good weather conditions. In other words, the driving data p l Classified into the "urban congestion driving sub-scenario under good weather conditions".

[0075] S104: Determine a driving style corresponding to each driving data in a corresponding driving scenario based on the second feature of each driving data.

[0076] Since the second feature of the driving data is related to the driving style, in this step, the driving style corresponding to each driving data in the corresponding driving scene (such as a driving sub-scene) can be determined based on the second feature of each driving data. The driving scene corresponding to each driving data is the driving scene determined for each driving data in S103.

[0077] In a possible manner, based on the second feature of each driving data, determining the driving style corresponding to each driving data in the corresponding driving scene may include the following steps 1041) to 1043):

[0078] 1041) Based on the second feature of each driving data corresponding to the same driving scene, a clustering algorithm is used to divide the driving data corresponding to the same driving scene into M categories, where M is a positive integer. M can be determined according to the number of driving styles to be divided, for example.

[0079] The driving data corresponding to the same driving scene can also be understood as all driving data classified into the same driving scene.

[0080] Taking the same driving scenario as driving scenario #1 (such as a city congestion driving sub-scenario under good weather conditions) as an example, assuming that the driving data corresponding to driving scenario #1 includes P (P is a positive integer) items of driving data, then after extracting the second feature for each of the P items of driving data, the sample data corresponding to the second feature of each item of driving data can be obtained. As an implementation method, the clustering algorithm can adopt, for example, the k-means clustering algorithm. In step 1041), the number of clusters can be set to M, so that the P items of driving data corresponding to driving scenario #1 can be divided into M categories by the k-means clustering algorithm. Assuming that the sample data corresponding to the second feature of a certain driving data is divided into the mth category among the M categories, it means that the driving data is also divided into the mth category accordingly.

[0081] 1042) Based on the second prior information related to the driving style, determine the driving style corresponding to each category in the M categories.

[0082] The second prior information may be obtained based on human experience and / or statistical analysis of driving data corresponding to different driving styles. Assuming that the driving style includes M different styles, after the driving data corresponding to the same driving scene (such as driving scene #1) is divided into M categories, the driving style corresponding to each category may be determined based on the second prior information.

[0083] 1043) The driving style corresponding to the m-th category in the M categories is used as the driving style corresponding to each driving data in the m-th category in the same driving scene, where m is a positive integer less than or equal to M.

[0084] For example, assuming that in step 1041), the driving data corresponding to the same driving scene (driving scene #1) is classified into the mth category including p3 and p6; assuming that in step 1042), the driving style corresponding to the mth category is determined to be "aggressive", then "aggressive" can be used as the driving style corresponding to the driving data p3 and p6.

[0085] For ease of understanding, the following takes the driving style including the first driving style and the second driving style as an example to introduce possible implementation methods for determining the driving style corresponding to each driving data in the corresponding driving scenario.

[0086] When the driving style includes the first driving style and the second driving style, the second prior information may include, for example: the value of the second feature corresponding to the first driving style is greater than the value of the second feature corresponding to the second driving style. In one example, assuming that the first driving style is "aggressive" and the second driving style is "conservative", then the second feature may be, for example, the lane change frequency / the average pedal depression angular velocity / the average steering wheel rotation angular velocity, and the second prior information may be, for example: the lane change frequency / the average pedal depression angular velocity / the average steering wheel rotation angular velocity corresponding to "aggressive"> the lane change frequency / the average pedal depression angular velocity / the average steering wheel rotation angular velocity corresponding to "conservative". In another example, assuming that the first driving style is "conservative" and the second driving style is "aggressive", then the second feature may be, for example, the average following distance, and the second prior information may be, for example: the average following distance corresponding to "conservative"> the average following distance corresponding to "aggressive".

[0087] It should be noted that the above-mentioned driving styles include the first driving style and the second driving style for the sake of convenience. In practice, the driving style may include more (such as 3 or more) style types. For example, it may include "conservative", "normal conservative", "normal", "normal aggressive" and "aggressive". In this case, the second prior information may include, for example, the lane change frequency / pedal average pressing angular velocity / steering wheel average turning angular velocity corresponding to "aggressive"> the lane change frequency / pedal average pressing angular velocity / steering wheel average turning angular velocity corresponding to "normal aggressive"> the lane change frequency / pedal average pressing angular velocity / steering wheel average turning angular velocity corresponding to "normal"> the lane change frequency / pedal average pressing angular velocity / steering wheel average turning angular velocity corresponding to "normal conservative"> the lane change frequency / pedal average pressing angular velocity / steering wheel average turning angular velocity corresponding to "conservative". Alternatively, the second prior information may include, for example, the average following distance corresponding to "conservative"> the average following distance corresponding to "normal conservative"> the average following distance corresponding to "normal"> the average following distance corresponding to "normal aggressive"> the average following distance corresponding to "aggressive".

[0088] Taking the second feature as the average following distance as an example, after obtaining the second prior information, the corresponding driving style can be determined for each driving data corresponding to the same driving scene (such as driving scene #1) based on the second prior information. For example, taking the driving style including "aggressive" and "conservative" as an example, through step 1041), the driving data corresponding to driving scene #1 can be divided into 2 categories. Assuming that the average following distance corresponding to the driving data in category #1> the average following distance corresponding to the driving data in category #2, then, since it is known based on the second prior information that the average following distance corresponding to "conservative"> the average following distance corresponding to "aggressive", the driving styles corresponding to category #1 and category #2 can be determined as "conservative" and "aggressive" respectively through step 1042). Furthermore, through step 1043), the driving style "conservative" corresponding to category #1 can be used as the driving style corresponding to each driving data in category #1 under the same driving scene (driving scene #1), and the driving style "aggressive" corresponding to category #2 can be used as the driving style corresponding to each driving data in category #2 under the same driving scene (driving scene #1).

[0089] It should be understood that the “value of the second feature corresponding to the driving data in category A (such as the average following distance) > the value of the second feature corresponding to the driving data in category B” mentioned in the embodiments of the present invention may, for example, mean that the value of the second feature corresponding to the driving data in category A is generally greater than the value of the second feature corresponding to the driving data in category B (such as the average value of the second feature corresponding to the driving data in category A > the average value of the second feature corresponding to the driving data in category B), or may mean that the values ​​of the second feature corresponding to all driving data in category A are greater than the values ​​of the second feature corresponding to the driving data in category B.

[0090] By executing the above steps 1041) to 1043) for each driving scene (driving sub-scene), the driving style corresponding to each driving data in each driving scene (driving sub-scene) can be obtained.

[0091] S105: Determine the driving style of each user in at least one driving scenario based on the driving style corresponding to each driving data in the corresponding driving scenario.

[0092] In a possible manner, the at least one user may include, for example, user #1, and the at least one driving scenario may include, for example, driving scenario #1. In this case, based on the driving style corresponding to each driving data in the corresponding driving scenario, determining the driving style of each user in at least one driving scenario may include, for example, the following steps 1051) and 1052):

[0093] 1051) In the driving data corresponding to driving scene #1, determine the driving data belonging to user #1 (for example, recorded as the first driving data).

[0094] For example, assuming that the driving data corresponding to driving scenario #1 includes P items of driving data, among which Q items of driving data belong to user #1, then the Q items of driving data are the first driving data.

[0095] 1052) Determine the driving style of user #1 in driving scenario #1 based on the driving style corresponding to each first driving data.

[0096] Exemplarily, after determining the first driving data, the number of first driving data corresponding to each driving style can be counted according to the driving style corresponding to each first driving data, and the maximum number of the first driving data corresponding to each driving style can be determined as the first number, so that the driving style corresponding to the first number can be used as the driving style of user #1 in driving scene #1.

[0097] Taking the driving styles including "conservative", "normal" and "aggressive" as an example, the number of first driving data corresponding to "conservative", "normal" and "aggressive" can be counted respectively. For example, the number of first driving data corresponding to "conservative" is 10, the number of first driving data corresponding to "normal" is 15, and the number of first driving data corresponding to "aggressive" is 25. Subsequently, the maximum number (i.e., 25) among 10, 15, and 25 can be determined as the first number, so that the driving style corresponding to the first number 25 (i.e., "aggressive") can be used as the driving style of user #1 in driving scene #1.

[0098] It is understandable that for other driving scenarios, the driving style of the user #1 in other driving scenarios can also be determined by the methods of steps 1051) and 1052). For example, when the at least one driving scenario also includes driving scenario #2 and driving scenario #3, the driving style of the user #1 in driving scenario #2 and the driving style in driving scenario #3 can also be determined. This process is also a process of identifying (classifying) the user's driving style.

[0099] It is also understandable that by performing the above steps for each user, the driving style of each user in the specified driving scene can be determined. In this way, it can be ensured that the user's driving style recognition is performed in a specific driving scene, so that the output driving style recognition result is more reliable and more convincing.

[0100] Combination of the above Figure 1 A driving style recognition method provided by an embodiment of the present invention is introduced. Figure 2A possible implementation process of the driving style recognition method provided by an embodiment of the present invention is introduced.

[0101] Figure 2 A possible implementation flow diagram of the driving style recognition method provided by an embodiment of the present invention is shown in FIG. Figure 2 As shown, the implementation process may include:

[0102] S201, downloading the real driving data of at least one user from the automobile big data accumulated by the automobile enterprise cloud platform.

[0103] For example, in order to obtain the real driving data of at least one user, the automobile big data accumulated by the automobile enterprise cloud platform can be fully utilized, and the real driving data of a large number of users can be downloaded from it. Among them, the real driving data can be various data uploaded from the vehicle to the cloud when the user drives the intelligent connected car. For example, the uploaded data, fault data, human-computer interaction data, and driving control behavior data specified by the national standard, etc.

[0104] In some embodiments, the implementation process may further include: performing a preprocessing operation on the user's real driving data. As an example, the preprocessing operation may include the following steps 2011) to 2012):

[0105] 2011)Data screening.

[0106] Through data screening, the signal fields required for building algorithm models (such as scene recognition models and style classification models) can be screened out from the driving data. For example, data that directly reflects the user's driving control behavior of the car and / or signals provided by various on-board sensors can be screened out.

[0107] 2012)Data cleaning.

[0108] Data cleaning may include duplicate value processing, outlier value removal, missing value filling, etc.

[0109] 2013) Timestamp sorting.

[0110] For example, you can rearrange the driving data in ascending order by the Time field.

[0111] The implementation of steps 2011) to 2012) may refer to steps 1011) to 1013) in the aforementioned embodiment, and will not be described again here to avoid repetition.

[0112] S202: extract driving event data from real driving data of at least one user.

[0113] In this step, driving event data (corresponding to the driving data in the aforementioned embodiment) may be extracted from the downloaded real driving data of at least one user to form a driving event data set (corresponding to the driving data set in the aforementioned embodiment).

[0114] Among them, a driving event refers to a short driving process with a change in the state of the car as the starting point and the end point, including but not limited to start-stop events and pedal events. Driving event data refers to the data collected / recorded during the driving event. For example, a start-stop event refers to a driving process from the start of a car to the shutdown of the car, and the event data of a start-stop event refers to the data collected from the start of a car to the shutdown of the car; for another example, a pedal event refers to a driving process in which the driver steps on the pedal, and the event data of a pedal event refers to the data collected during the time period when the car pedal is stepped on. Among them, pedal events can be further subdivided into accelerator pedal events, brake pedal events, and no pedal pressed events.

[0115] As an example, the driving event in this embodiment may be a start-stop event. For example, the start-stop event may use the field signal "vehicle speed" as a judgment basis, and the change of the vehicle speed from zero to non-zero is used as the event start point, and the change of the vehicle speed from non-zero to zero is used as the event end point, thereby extracting the start-stop events of all users.

[0116] In some embodiments, in order to discard invalid data that is not helpful for driving style identification as much as possible, the vehicle state with a speed less than 3 km / h can be regarded as stationary, and users with a total of less than 30 events in a single month (such as a user's start-stop events in a month) are removed from the driving data set.

[0117] S203: Perform feature engineering on the extracted driving event data to obtain corresponding sample data.

[0118] In this step, feature engineering can be performed on the extracted driving event data to obtain sample data corresponding to the driving data. For example, for a driving event data in the format of T×N, high-dimensional features of the driving event data can be extracted to form a sample data in the format of 1×M. Where T is the time length of the driving event, N is the number of signal fields screened in S201, and M is the number of high-dimensional features extracted.

[0119] As an example, high-dimensional features include but are not limited to at least one of the following: mean (such as speed mean; such as acceleration mean), range, standard deviation, first-order time derivative (such as the first-order time derivative of speed is acceleration), second-order time derivative, time integral, wavelet energy, average frequency, number of peaks (such as the number of peaks in the speed waveform), pedal depression frequency, low-speed time proportion, high-speed time proportion, event duration, etc.

[0120] It should be noted that the selection of high-dimensional features in the embodiment of the present invention needs to comprehensively consider the needs of subsequent driving scene recognition and driving style classification. As an example, based on human experience and / or prior knowledge (prior information), the high-dimensional features in the embodiment of the present invention can be selected from at least one of the following: average accelerator pedal stroke value, average accelerator pedal pressing angular velocity, accelerator pedal pressing frequency, average brake pedal stroke value, average brake pedal pressing angular velocity, brake pedal pressing frequency, steering wheel average rotation angular velocity, steering wheel rotation frequency, gear shift frequency, low gear time proportion, high gear time proportion, driving mode mean, vehicle speed mean, vehicle speed standard deviation, low speed time proportion, high speed time proportion, start-stop event duration, horn frequency, wiper on time proportion, light on time proportion, average following distance, lane change frequency.

[0121] As an implementation method, the low gear in the high-dimensional feature can refer to gears 1, 2, and 3, for example, and the high gear can refer to gears 7, 8, and 9, for example; the low speed can refer to a speed less than 40 km / h, for example, and the high speed can refer to a speed greater than 80 km / h.

[0122] In some embodiments, after extracting the high-dimensional features of the driving event data, the same type of high-dimensional features may be normalized using the maximum and minimum normalization method based on the following formula (1):

[0123]

[0124] Among them, x represents the eigenvalue before standardization, y represents the eigenvalue after standardization, and x min Represents the minimum value of the high-dimensional feature of this type before normalization, x max Indicates the maximum value of the high-dimensional feature of this type before normalization.

[0125] S204: Divide the driving event data into various driving sub-scenarios based on the driving scenario recognition model.

[0126] The division of the driving event data into various driving sub-scenarios may also be understood as the division of the driving events associated with the driving event data into various driving sub-scenarios.

[0127] In this step, a driving scene recognition model can be trained based on a machine learning algorithm, so that the sample data can be divided into various driving sub-scenes using the driving scene recognition model.

[0128] As an example, first, multiple different driving scenarios can be defined. Among them, the driving scenario refers to the objective driving environment that is unrelated to the driver's subjective style, and can be described in multiple different dimensions. Exemplarily, the description dimensions include but are not limited to: driving area (highway, city, rural), road congestion (congested, unobstructed, normal), weather conditions (good, bad), road slope (uphill, flat road, downhill), whether there are other passengers (yes, no), etc.

[0129] The driving sub-scenario in this embodiment refers to a specific instance of the above driving scenarios, such as a driving sub-scenario on a congested flat road in a city with no passengers in good weather.

[0130] For ease of understanding, the driving scene recognition solution of this embodiment is introduced below by taking the driving event type being a start-stop event and the driving scene including three dimensions as an example.

[0131] As an example, the three dimensions of the driving scene may be:

[0132] (1) Driving area, which is divided into three categories: highway, urban, and rural. Based on human experience and / or prior knowledge, on highway sections, the vehicle is continuously at high speed; on urban sections, due to the large number of intersections, the duration of the start-stop event is shorter; and on rural sections, it is somewhere in between;

[0133] (2) Road congestion, which is divided into three categories: congested, normal, and unobstructed. Based on human experience and / or prior knowledge, when the road is congested, the vehicle accelerates and decelerates frequently, while when the road is unobstructed, the pedal is pressed less frequently.

[0134] (3) Weather conditions are divided into two categories: good and bad. Bad weather, such as rain, snow and fog, can be reflected by turning on the vehicle's windshield wipers and fog lights.

[0135] In summary, in this example, there are 3×3×2=18 driving sub-scenarios.

[0136] As an implementation method, the scene recognition model in this embodiment may be, for example, a k-means clustering model. Exemplarily, the steps of the k-means clustering algorithm may include the following steps a) to e):

[0137] a) Set the maximum number of iterations (e.g. 300) and the number of clusters k;

[0138] b) Initialize cluster center c i , i=1,2,…,k;

[0139] c) For a given sample data {x1,x2,…,x n}, assign it to the nearest cluster center c according to the distancei . Where n is the total number of start-stop events (i.e., the total number of driving event data of start-stop events). As an example, the distance may be, for example, the Euclidean distance;

[0140] d) Calculate the mean of each cluster as the new cluster center c i ;

[0141] e) Repeat steps c) and d) until the maximum number of iterations is reached.

[0142] The purpose of the iteration is to minimize the total distance D between the sample data and its nearest cluster center. The calculation formula of D is shown in formula (2):

[0143]

[0144] Among them, x j is the sample data, c i is the cluster center closest to the sample data.

[0145] For example, in order to determine the driving area corresponding to a certain driving event data (or driving event), the high-dimensional feature can be determined as the mean vehicle speed, so that the corresponding sample data can be obtained by extracting the mean vehicle speed feature from the driving event data. Among them, extracting the mean vehicle speed feature from the driving event data can also be understood as calculating the mean vehicle speed based on the driving event data, thereby obtaining sample data including the mean vehicle speed.

[0146] Assuming that the driving area includes "highway", "city" and "rural", the number of clusters k can be set to 3. Therefore, by adopting the above k-means clustering model, the sample data can be divided into 3 clusters based on the mean vehicle speed. For ease of explanation, it is assumed that the mean vehicle speed corresponding to cluster #1> the mean vehicle speed corresponding to cluster #2> the mean vehicle speed corresponding to cluster #3. Since under normal circumstances, the mean vehicle speed of the "highway" section> the mean vehicle speed of the "city" section> the mean vehicle speed of the "rural" section, the driving event data (or driving events) corresponding to the sample data in cluster #1 can be divided into "highway", the driving event data corresponding to the sample data in cluster #2 can be divided into "city", and the driving event data corresponding to the sample data in cluster #3 can be divided into "rural".

[0147] It should be noted that, in some embodiments, in order to determine the driving area corresponding to the driving event data, the high-dimensional feature may also be determined as the duration of the start-stop event, for example. In this case, the driving event data may be divided into corresponding driving areas based on the prior information that the duration of the start-stop event in the "highway" section is greater than the duration of the start-stop event in the "rural" section and the duration of the start-stop event in the "urban" section.

[0148] Similarly, assuming that the road congestion conditions include "congested", "normal" and "unblocked", the number of clusters k can be set to 3. At this time, in order to determine the road congestion condition corresponding to a certain driving event data, for example, the high-dimensional feature can be determined as the acceleration / deceleration frequency / pedal pressing frequency. In this case, the driving event data can be divided into the corresponding road congestion conditions based on the prior information that the acceleration / deceleration frequency / pedal pressing frequency in "congested" > the acceleration / deceleration frequency / pedal pressing frequency in "normal" > the acceleration / deceleration frequency / pedal pressing frequency in "unblocked".

[0149] Similarly, assuming that the weather conditions include "good" and "bad", the number of clusters k can be set to 2. At this time, in order to determine the weather conditions corresponding to a certain driving event data, the high-dimensional features can be determined as the percentage of wiper on time / fog light on time, for example. In this case, the driving event data can be divided into corresponding weather conditions based on the prior information that the percentage of wiper on time / fog light on time under "bad" weather conditions > the percentage of wiper on time / fog light on time under "good" weather conditions.

[0150] It can be understood that if the driving area corresponding to a driving event data is a city, the corresponding road congestion is congested, and the corresponding weather condition is good, then the driving sub-scenario corresponding to the driving event data is the "city congestion driving sub-scenario in good weather conditions". In other words, the driving event data can be classified into the "city congestion driving sub-scenario in good weather conditions".

[0151] S205 : For each driving sub-scenario, determine the driving style of each user in the driving sub-scenario based on the driving style classification model.

[0152] In this step, the driving event data in each individual driving sub-scenario may be classified based on the driving style classification model, thereby obtaining the user's driving style segmented according to the scenario.

[0153] Among them, the user driving style segmented by scenario refers to the user's driving style performance in different driving sub-scenarios. For example, for a certain driver, the algorithm's final recognition result is: the driver exhibits an aggressive driving style in driving sub-scenario A, an ordinary to aggressive driving style in driving sub-scenario B, and an aggressive driving style in driving sub-scenario C, etc.

[0154] For ease of understanding, the driving style classification scheme of this embodiment is introduced below by taking the driving style as “conservative”, “normal conservative”, “normal”, “normal aggressive” and “aggressive” according to the degree of aggressiveness when driving a car as an example.

[0155] As an implementation, the driving style classification model in this embodiment can be, for example, a k-means clustering model. It is understandable that when the driving style is aggressive, it is usually manifested as a small average following distance, a high frequency of lane changes, a large average pedal depression angular velocity, a large average steering wheel rotation angular velocity, etc., while when the driving style is conservative, it is the opposite. Therefore, when the k-means clustering model is used to classify sample data (or driving event data) under a driving sub-scenario (such as a city congestion driving sub-scenario under good weather conditions), the optional high-dimensional features can include, for example, but are not limited to at least one of the following: average following distance, lane change frequency, average pedal depression angular velocity, and average steering wheel rotation angular velocity. Since driving styles can be divided into five types, namely "conservative", "normal conservative", "normal", "normal aggressive" and "aggressive", the number of clusters k can be set to 5, so that the driving event data under the driving sub-scenario can be divided into five clusters based on the selected high-dimensional features.

[0156] Furthermore, taking the high-dimensional feature of average following distance as an example, the driving event data in the driving sub-scenario can be divided into corresponding driving styles based on the prior information that the average following distance corresponding to "conservative" > the average following distance corresponding to "normal and conservative" > the average following distance corresponding to "normal" > the average following distance corresponding to "normal and aggressive" > the average following distance corresponding to "aggressive". Alternatively, taking the high-dimensional feature of lane change frequency / average pedal depression angular velocity / average steering wheel rotation angular velocity (assuming it is recorded as the first high-dimensional feature) as an example, the driving event data in the driving sub-scenario can be divided into corresponding driving styles based on the prior information that the first high-dimensional feature corresponding to "aggressive" > the first high-dimensional feature corresponding to "normal and aggressive" > the first high-dimensional feature corresponding to "normal" > the first high-dimensional feature corresponding to "normal and conservative" > the first high-dimensional feature corresponding to "conservative".

[0157] It is understandable that the driving style classification in S205 is different from the k-means model used in the driving scene recognition in S204 in that:

[0158] (1) The training data of the models are different. The training data in driving scene recognition is the start-stop event data of all users, while the training data in driving style classification is the start-stop event data of all users in a single driving sub-scenario.

[0159] (2) Different high-dimensional features are used. In driving scene recognition, the high-dimensional features selected are related to the driving scene, such as: the duration of the start-stop event, the frequency of pedal pressing, the proportion of wiper on time, etc.; while in driving style classification, the high-dimensional features selected are related to the driving style, such as: average following distance, lane change frequency, average pedal pressing angular velocity, average steering wheel rotation angular velocity, etc.

[0160] According to the method of this embodiment, for each user, the final output driving style classification (recognition) result is in the form of {e i :s j}. Among them, e i represents the i-th driving sub-scenario, for example, in this embodiment, i=1, 2, ... 18; s j Indicates the jth driving style. For example, in this embodiment, j=1, 2, 3, 4, 5, corresponding to five driving styles in sequence. For example, assuming that for user A, the output driving style classification result includes {e1:s2}, then it indicates that the driving style of user A in the first driving sub-scenario is the second driving style.

[0161] It should be noted that during the implementation, the e i There may be missing cases because not all users can cover all driving sub-scenarios. For example, for a city-operated vehicle, the driving data of the vehicle may be limited to the city for a long period of time, without involving highways and rural areas. Therefore, the driving style classification results of the vehicle may not include the classification results corresponding to the driving sub-scenarios related to highways and rural areas.

[0162] According to the above implementation process, the solution of the embodiment of the present invention can bring the following advantages:

[0163] (1) Based on the user's real driving data, driving scene recognition is performed first, and then driving style classification is completed, thereby ensuring that the driving style clustering algorithm is performed under the same driving sub-scene. This makes the output driving style recognition results more reliable and convincing.

[0164] (2) The driving style output of the model can be refined to different driving scenarios, taking into full account the fact that the driver's driving style may change with the change of driving scenarios, thereby making the output style recognition results more accurate, which in turn helps to provide a reference for better serving users in the future.

[0165] Based on the above embodiments, the embodiments of the present invention provide a corresponding driving style recognition device. The device includes the modules included, and the submodules included in each module, which can be implemented by a processor in a computer device with information processing capabilities; of course, it can also be implemented by a specific logic circuit; in the implementation process, the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP) or a field programmable gate array (FPGA), etc.

[0166] Figure 3 FIG. 3 shows the structure of a driving style recognition device 300 provided by an embodiment of the present invention. Figure 3 As shown, the apparatus 300 may include:

[0167] The acquisition module 301 is used to acquire driving data of at least one user to obtain a driving data set; the feature extraction module 302 is used to extract a first feature and a second feature for each driving data in the driving data set; wherein the first feature is related to the driving scene, and the second feature is related to the driving style; the first determination module 303 is used to determine the driving scene corresponding to each driving data based on the first feature of each driving data; the second determination module 304 is used to determine the driving style corresponding to each driving data in the corresponding driving scene based on the second feature of each driving data; the third determination module 305 is used to determine the driving style of each user in at least one driving scene based on the driving style corresponding to each driving data in the corresponding driving scene.

[0168] In some embodiments, the first determination module 303 is specifically used to: based on the first feature of each driving data, use a clustering algorithm to divide the driving data in the driving data set into N categories, where N is a positive integer; based on the first prior information related to the driving scene, determine the driving scene corresponding to each category in the N categories; use the driving scene corresponding to the nth category in the N categories as the driving scene corresponding to each driving data in the nth category, where n is a positive integer less than or equal to N.

[0169] In some embodiments, when the driving scene includes a first driving area and a second driving area, the first prior information includes: a value of the first feature in the first driving area is greater than a value of the first feature in the second driving area; when the driving scene includes a first congestion level and a second congestion level, the first prior information includes: a value of the first feature when the congestion level is the first congestion level is greater than a value of the first feature when the congestion level is the second congestion level; when the driving scene includes a first weather condition and a second weather condition, the first prior information includes: a value of the first feature under the first weather condition is greater than a value of the first feature under the second weather condition.

[0170] In some embodiments, the second determination module 304 is specifically used to: based on the second feature of each driving data corresponding to the same driving scene, use a clustering algorithm to divide the driving data corresponding to the same driving scene into M categories, where M is a positive integer; based on the second prior information related to the driving style, determine the driving style corresponding to each category in the M categories; use the driving style corresponding to the mth category in the M categories as the driving style corresponding to each driving data in the mth category in the same driving scene, where m is a positive integer less than or equal to M.

[0171] In some embodiments, when the driving style includes a first driving style and a second driving style, the second priori information includes: a value of the second feature corresponding to the first driving style is greater than a value of the second feature corresponding to the second driving style.

[0172] In some embodiments, at least one user includes a first user, and at least one driving scene includes a first scene; the third determination module 305 includes: a first determination submodule, used to determine the first driving data belonging to the first user in the driving data corresponding to the first scene; a second determination submodule, used to determine the driving style of the first user in the first scene based on the driving style corresponding to each first driving data.

[0173] In some embodiments, the second determination submodule is specifically used to: count the number of first driving data corresponding to each driving style; determine the maximum number among the numbers of first driving data corresponding to each driving style as the first number; and use the driving style corresponding to the first number as the driving style of the first user in the first scenario.

[0174] In some embodiments, each driving data is data collected from when the car starts to when the car is turned off, or is data collected during a time period when a car pedal is pressed.

[0175] The description of the above device embodiment is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. In some embodiments, the functions or modules included in the device provided in the embodiment of the present invention can be used to execute the method described in the above method embodiment. For technical details not disclosed in the device embodiment of the present invention, please refer to the description of the method embodiment of the present invention for understanding.

[0176] It should be noted that, in the embodiment of the present invention, if the above method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the relevant technology. The software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk. In this way, the embodiment of the present invention is not limited to any specific hardware, software or firmware, or any combination of hardware, software, and firmware.

[0177] An embodiment of the present invention further provides a driving style recognition device, comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0178] The embodiment of the present invention further provides a chip, which includes: a processor, which is used to call and run a computer program from a memory, so that a device equipped with the chip executes part or all of the steps in the above method.

[0179] The embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, some or all of the steps in the above method are implemented. The computer-readable storage medium may be transient or non-transient.

[0180] An embodiment of the present invention further provides a computer program, comprising computer-readable codes, and when the computer-readable codes are executed in a device (such as a driving style recognition device), a processor in the device executes part or all of the steps in the above method.

[0181] The embodiment of the present invention further provides a computer program product, the computer program product includes a non-transitory computer-readable storage medium storing a computer program, and when the computer program is read and executed by a computer, some or all of the steps in the above method are implemented. The computer program product can be implemented specifically by hardware, software or a combination thereof. In some embodiments, the computer program product is specifically embodied as a computer storage medium, and in other embodiments, the computer program product is specifically embodied as a software product, such as a software development kit (SDK), etc.

[0182] It should be noted here that the description of the various embodiments above tends to emphasize the differences between the various embodiments, and the same or similar aspects can be referenced to each other. The description of the above device, chip, storage medium, computer program and computer program product embodiments is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the embodiments of the device, storage medium, computer program and computer program product of the present invention, please refer to the description of the method embodiment of the present invention for understanding.

[0183] Figure 4 A schematic diagram of a hardware entity of a driving style recognition device in an embodiment of the present invention. Figure 4 The driving style recognition device 400 shown includes a processor 410 , which can call and run a computer program from a memory to implement the method in the embodiment of the present invention.

[0184] In some embodiments, Figure 4 As shown, the driving style recognition device 400 may further include a memory 420. The processor 410 may call and run a computer program from the memory 420 to implement the method in the embodiment of the present invention. The memory 420 may be a separate device independent of the processor 410, or may be integrated in the processor 410.

[0185] In some embodiments, Figure 4 As shown, the driving style recognition device 400 may further include a transceiver 430, and the processor 410 may control the transceiver 430 to communicate with other devices, specifically, to send information or data to other devices, or to receive information or data sent by other devices. The transceiver 430 may include a transmitter and a receiver. The transceiver 430 may further include an antenna, and the number of antennas may be one or more.

[0186] It should be understood that "one embodiment", "an embodiment" or "some embodiments" mentioned throughout the specification means that specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present invention. Therefore, "in one embodiment", "in an embodiment" or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present invention, the size of the serial number of each step / process mentioned above does not mean the order of execution. The execution order of each step / process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention. The serial numbers of the embodiments of the present invention mentioned above are for description only and do not represent the advantages and disadvantages of the embodiments.

[0187] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0188] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units or modules is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0189] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0190] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0191] A person skilled in the art can understand that all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM), magnetic disks or optical disks, etc., various media that can store program codes.

[0192] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the relevant technology. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0193] The above description is only an implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A driving style recognition method, characterized in that: The method comprises: Acquire driving data of at least one user to obtain a driving data set; Extracting a first feature and a second feature from each driving data in the driving data set; wherein the first feature is related to the driving scene, and the second feature is related to the driving style; Determine a driving scene corresponding to each driving data based on the first feature of each driving data; Determining a driving style corresponding to each driving data in a corresponding driving scenario based on a second feature of each driving data; Based on the driving style corresponding to each driving data in the corresponding driving scenario, the driving style of each user in at least one driving scenario is determined.

2. The method according to claim 1, characterized in that The determining, based on the first feature of each driving data, a driving scene corresponding to each driving data includes: Based on the first feature of each driving data, using a clustering algorithm to divide the driving data in the driving data set into N categories, where N is a positive integer; Determining, based on first prior information related to the driving scene, a driving scene corresponding to each of the N categories; The driving scene corresponding to the nth category in the N categories is used as the driving scene corresponding to each driving data in the nth category, where n is a positive integer less than or equal to N.

3. The method according to claim 2, characterized in that In a case where the driving scene includes a first driving area and a second driving area, the first priori information includes: a value of the first feature in the first driving area is greater than a value of the first feature in the second driving area; In the case where the driving scenario includes a first congestion level and a second congestion level, the first priori information includes: a value of the first feature when the congestion level is the first congestion level is greater than a value of the first feature when the congestion level is the second congestion level; In the case where the driving scenario includes a first weather condition and a second weather condition, the first priori information includes: a value of the first feature under the first weather condition is greater than a value of the first feature under the second weather condition.

4. The method according to any one of claims 1 to 3, characterized in that The determining, based on the second feature of each driving data, a driving style corresponding to each driving data in a corresponding driving scenario includes: Based on the second feature of each driving data corresponding to the same driving scene, a clustering algorithm is used to divide the driving data corresponding to the same driving scene into M categories, where M is a positive integer; Determining a driving style corresponding to each of the M categories based on second prior information related to the driving style; The driving style corresponding to the m-th category in the M categories is used as the driving style corresponding to each driving data in the m-th category in the same driving scene, where m is a positive integer less than or equal to M.

5. The method according to claim 4, characterized in that In the case where the driving style includes a first driving style and a second driving style, the second priori information includes: a value of the second feature corresponding to the first driving style is greater than a value of the second feature corresponding to the second driving style.

6. The method according to any one of claims 1 to 3, characterized in that The at least one user includes a first user, and the at least one driving scenario includes a first scenario; The determining the driving style of each user in at least one driving scenario based on the driving style corresponding to each driving data in the corresponding driving scenario includes: Determining first driving data belonging to the first user from the driving data corresponding to the first scenario; Based on the driving style corresponding to each first driving data, a driving style of the first user in the first scenario is determined.

7. The method according to claim 6, characterized in that The determining, based on the driving style corresponding to each first driving data, the driving style of the first user in the first scenario includes: Counting the number of first driving data corresponding to each driving style; determining a maximum number of the numbers of first driving data corresponding to each driving style as a first number; The driving style corresponding to the first number is used as the driving style of the first user in the first scenario.

8. The method according to any one of claims 1 to 3, characterized in that Each of the driving data is data collected from the time when the car starts to the time when the car is turned off, or data collected during the time when the car pedal is stepped on.

9. A driving style recognition device, characterized in that: The device comprises: An acquisition module, used to acquire driving data of at least one user to obtain a driving data set; A feature extraction module, configured to extract a first feature and a second feature from each driving data in the driving data set; wherein the first feature is related to the driving scene, and the second feature is related to the driving style; A first determination module, configured to determine a driving scene corresponding to each driving data based on a first feature of each driving data; A second determination module, configured to determine a driving style corresponding to each driving data in a corresponding driving scenario based on a second feature of each driving data; The third determination module is used to determine the driving style of each user in at least one driving scenario based on the driving style corresponding to each driving data in the corresponding driving scenario.

10. A driving style recognition device, characterized in that: The device comprises: A memory for storing computer executable instructions; A processor, connected to the memory, configured to implement the method according to any one of claims 1 to 8 by executing the computer executable instructions.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, the method according to any one of claims 1 to 8 is implemented.

Images