WO2022147785A1

WO2022147785A1 - Autonomous driving scenario identifying method and apparatus

Info

Publication number: WO2022147785A1
Application number: PCT/CN2021/070939
Authority: WO
Inventors: 王嘉伟; 曾振华
Original assignee: 华为技术有限公司
Priority date: 2021-01-08
Filing date: 2021-01-08
Publication date: 2022-07-14
Also published as: CN112805724B; CN112805724A

Abstract

The present application provides an autonomous driving scenario identifying method and apparatus. The solution comprises: obtaining a feature sequence of a second vehicle according to driving data of a first vehicle in a current time period; then, according to the obtained feature sequence, identifying a driving behavior of the second vehicle around the first vehicle by using a dynamic time warping algorithm; and finally, matching a driving scenario of the second vehicle by using a scenario library. According to the present solution, the efficiency and accuracy of identifying a driving behavior of a vehicle are improved, and the cost of identifying the driving behavior of the vehicle is reduced. Moreover, the driving scenario of a second vehicle around a first vehicle is actively identified, a reference basis is provided for a driving operation at the next stage of the first vehicle, and the driving safety of the first vehicle is improved.

Description

Vehicle driving scene recognition method and device

technical field

The present application relates to the technical field of intelligent driving, and in particular, to a method and device for recognizing a driving scene of a vehicle.

Background technique

With the development of autonomous driving and assisted driving technologies, more and more companies are developing or mass-producing a driving system that can control the vehicle's own driving in the actual lane, so that the vehicle equipped with the driving system has the ability to operate without human intervention. Autopilot function.

The current intelligent driving solutions for identifying vehicle driving behaviors usually collect vehicle driving data from the fleet to pre-train a deep learning model, and then use the trained deep learning model to identify the driving behavior of the vehicle. For example, when determining whether the front vehicle cuts in, the prior art usually first captures the driving data of the own vehicle itself and vehicles in all the forward lanes, including vehicle position and vehicle speed, and uses a pre-trained recognition module to perform action recognition. If it is recognized that a cut-in action has occurred, the complete data of the cut-in action, including: driving data of the vehicle and the cut-in vehicle and road conditions, is sent to the cloud server, and the cloud identifies whether the cut-in action of the preceding vehicle has occurred according to the uploaded data.

As machine learning models become more complex, such as deeper neural networks, the need for training datasets increases accordingly. Therefore, if you want to more accurately imitate the driver's prediction and analysis level when dealing with the judgment/recognition of the "cut-in" action of the vehicle in the adjacent lane, it is necessary to provide a large-scale actual road data training model, and the accumulation of training data sets is relatively high. difficult and expensive. At the same time, when training the model, there is no distinction between regions, the local correlation is poor, and the accuracy of the identified results is not high.

SUMMARY OF THE INVENTION

The present application provides a method and device for recognizing a driving scene of a vehicle, which uses a dynamic time warping algorithm to calculate the similarity and recognizes the driving area of the vehicle to obtain the driving behavior of the vehicle, so as to solve the problem in the prior art when the driving behavior is recognized by the model. It is difficult and costly to accumulate the training data set of the vehicle, and the driving scene of the vehicle is matched in the scene library to realize the purpose of vehicle driving scene recognition.

In a first aspect, the present application provides a method for identifying a vehicle driving scene, the method comprising:

obtain the driving data of the first vehicle in the current period;

A feature sequence of a second vehicle is obtained according to the driving data, the second vehicle is used to indicate vehicles around the first vehicle, and the feature sequence includes: a speed sequence, a distance between the second vehicle and a reference lane line a sequence and a sequence of angles between the second vehicle and the reference lane line;

Using a dynamic time warping algorithm to obtain the similarity between the feature sequence of the second vehicle and a preset driving behavior reference sequence;

determining the driving behavior of the second vehicle in the current period according to the similarity;

According to the driving behavior of the second vehicle in the current period, a pre-built scene library is used to match the driving scene of the second vehicle in the current period.

From the above, the present application adopts the method of calculating the similarity of the dynamic time warping algorithm to identify the driving behavior of the second vehicle, which improves the accuracy of identifying the driving behavior compared to the recognition through the trained model; the dynamic time warping algorithm can efficiently process the driving behavior. The characteristics of a large number of data streams and short time improve the efficiency of identifying driving behavior; in addition, the present application actively matches the driving scene of the second vehicle through the driving behavior of the vehicle, which can provide a reference for the driving operation of the first vehicle, to a certain extent can improve the safety of the first vehicle.

In a possible embodiment, before obtaining the similarity between the feature sequence of the second vehicle and the driving behavior reference sequence by using a dynamic time warping algorithm, the method further includes:

identifying the area in which the first vehicle travels;

The preset driving behavior reference sequence corresponding to the area where the first vehicle travels is extracted from a database, where the database includes driving behavior reference sequences corresponding to different areas.

From the above, the present application can further improve the speed of identifying the driving behavior of the second vehicle around it by identifying the area where the first vehicle is traveling, and further save the time cost of identifying the driving behavior.

In a possible embodiment, before obtaining the feature sequence of the second vehicle according to the driving data, the method further includes:

Noise reduction processing and smoothing processing are performed on the driving data.

From the above, by processing the driving data in the present application, the influence of the data collection error on the recognition of driving behavior can be reduced, and the recognition accuracy can be improved.

In a possible embodiment, the reference lane line is a lane line of the lane in which the first vehicle travels.

In a possible embodiment, the lane in which the first vehicle travels is a curved lane, and the method further includes:

The distance sequence is curvature compensated according to the curvature of the curved lane.

From the above, when the first vehicle is driving in a curved lane, the present application performs curvature compensation on the distance sequence, which can effectively reduce the influence of road surface features on the collected driving data and improve the accuracy of identifying driving behavior.

In a possible embodiment, the determining, according to the similarity, the driving behavior of the second vehicle in the current period includes:

Judging whether the similarity satisfies a preset condition;

The driving behavior represented by the driving behavior reference sequence corresponding to the degree of similarity satisfying the preset condition is used as the driving behavior of the second vehicle in the current period.

In a possible embodiment, it also includes:

obtaining driving behaviors corresponding to the different scenarios according to the historical driving images corresponding to the different scenarios;

The scene library is constructed according to the different scenes and the driving behaviors corresponding to the different scenes.

In a second aspect, the present application further provides a vehicle driving scene recognition device, the device comprising:

a data acquisition unit for acquiring the driving data of the first vehicle in the current period;

a feature acquisition unit, configured to obtain a feature sequence of a second vehicle according to the driving data, the second vehicle is used to indicate vehicles around the first vehicle, and the feature sequence includes: a speed sequence, the second vehicle a sequence of distances from the reference lane line and a sequence of included angles between the second vehicle and the reference lane line;

a similarity calculation unit, configured to obtain the similarity between the feature sequence of the second vehicle and a plurality of preset driving behavior reference sequences by using a dynamic time warping algorithm;

a behavior recognition unit, configured to determine the driving behavior of the second vehicle in the current period according to the similarity;

A scene matching unit, configured to match the driving scene of the second vehicle in the current period by using a pre-built scene library according to the driving behavior of the second vehicle in the current period.

In a possible embodiment, the similarity calculation unit is further used for:

identifying the area in which the first vehicle travels;

In a possible embodiment, the feature acquisition unit is further configured to:

In a possible embodiment, the lane in which the first vehicle travels is a curved lane, and the apparatus further includes:

A curvature compensation unit, configured to perform curvature compensation on the distance sequence according to the curvature of the curved lane.

In a possible embodiment, the behavior identification unit is specifically used for:

Judging whether the similarity satisfies a preset condition;

In a possible embodiment, it also includes a scene library construction unit for:

In a third aspect, the present application also provides a computer storage medium, where instructions are stored in the computer storage medium, and when the instructions are executed on the computer, the computer is made to execute the vehicle driving according to the first aspect of the present application. Scene recognition method.

In a fourth aspect, the present application further provides a computer program product including instructions, which, when the instructions are executed on a computer, cause the computer to execute the method for recognizing a vehicle driving scene as described in the first aspect of the present application.

Description of drawings

Fig. 1 is the application scene system structure diagram provided by this application;

2 is a flowchart of a method for identifying a vehicle driving scene provided by an embodiment of the present application;

3 is a schematic diagram of the effect after noise reduction and smoothing processing provided by an embodiment of the present application;

4a is a schematic diagram of the positional relationship between the first vehicle and the second vehicle in a straight lane provided by an embodiment of the present application;

FIG. 4b is a schematic diagram of the positional relationship between the first vehicle and the second vehicle in a curved lane provided by an embodiment of the present application;

5 is a schematic diagram of calculating similarity by combining a window function and a dynamic time warping algorithm provided by an embodiment of the present application;

Fig. 6 is the similarity comparison diagram of two sequences provided in the embodiment of the present application;

FIG. 7a is a schematic diagram of the distance change between the second vehicle and the reference lane line when the second vehicle changes lanes to the left according to an embodiment of the present application;

7b is a schematic diagram of the distance change between the second vehicle and the reference lane line when the second vehicle changes lanes to the right according to an embodiment of the present application;

FIG. 7c is a schematic diagram of the distance change between the first vehicle and the reference lane line when the first vehicle changes lanes to the left according to an embodiment of the present application;

FIG. 7d is a schematic diagram of the distance change between the first vehicle and the reference lane line when the first vehicle changes lanes to the right according to an embodiment of the present application;

FIG. 8 is a corresponding relationship diagram of driving scenarios and driving behaviors provided by an embodiment of the present application;

9 is a flowchart of a method for recognizing a vehicle driving scene provided by an embodiment of the present application;

10 is a probability distribution diagram of the second vehicle inserting into the driving lane of the first vehicle in the two areas provided by the present application;

FIG. 11 is a schematic diagram of a vehicle driving scene recognition device provided by an embodiment of the present application.

Detailed ways

In order to make the objectives, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions in the embodiments of the present application will be described below with reference to the accompanying drawings.

In the description of the embodiments of the present application, words such as "exemplary", "for example" or "for example" are used to represent examples, illustrations or illustrations. Any embodiments or designs described in the embodiments of the present application as "exemplary," "such as," or "by way of example" should not be construed as preferred or advantageous over other embodiments or designs. Rather, use of words such as "illustrative," "such as," or "for example," is intended to present the related concepts in a specific manner.

In the description of the embodiments of the present application, the term "and/or" is only an association relationship for describing associated objects, indicating that there may be three relationships, for example, A and/or B, which may indicate: A alone exists, A alone exists There is B, and there are three cases of A and B at the same time. Also, unless stated otherwise, the term "plurality" means two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implying the indicated technical features. Thus, a feature defined as "first" or "second" may expressly or implicitly include one or more of that feature. The terms "including", "including", "having" and their variants mean "including but not limited to" unless specifically emphasized otherwise.

FIG. 1 is a structural diagram of an application system for vehicle driving scene recognition provided by the present application. As shown in Figure 1, the application system includes: vehicles running on highways or main roads in urban areas and cloud servers for computing. The vehicle is equipped with sensors, a positioning module (GPS or HD Map), and a wireless gateway T-Box responsible for data transmission between the sensors and the cloud server.

It can be understood that the structure of the application system for vehicle driving scene recognition illustrated in the embodiments of the present application does not constitute a specific limitation to the present application. In other embodiments of the present application, the application system structure of vehicle driving scene recognition may include more or less modules than shown, or some modules are combined, or some modules are split, or different modules are arranged. The illustrated modules may be implemented in hardware, or a combination of software and hardware.

The vehicle in the application system can be an intelligent vehicle that can simulate the operation of a driver, or an ordinary vehicle that requires manual driving. According to the energy consumption, the vehicle can be a new energy vehicle or an ordinary fuel vehicle. The embodiments of the present application do not specifically limit the types of vehicles.

In actual driving, the driving behavior of the vehicle will be affected by the surrounding vehicles. Quickly identify the driving behavior and driving scenes of the surrounding vehicles, which can assist the driver or the driving operating system to make the driving operation of the vehicle. In the embodiment of the present application, for the convenience of description, a certain vehicle in the application system is marked as the first vehicle, and the vehicles around the first vehicle are marked as the second vehicle. Those skilled in the art should know that the settings of the first vehicle and the second vehicle are relative, and the first vehicle and the second vehicle may be each other's surrounding vehicles.

When performing scene recognition on the second vehicle around the first vehicle, first use various sensors installed on the first vehicle to collect the driving data of itself and surrounding vehicles, such as cameras and radars; then use the T-Box installed on the first vehicle The driving data collected by the sensor is uploaded to the cloud server; the cloud server identifies the driving behavior of the second vehicle according to the uploaded data, and uses it as a reference for the next driving operation of the first vehicle.

In actual driving, when the second vehicle is too far away from the first vehicle, the driving operation of the first vehicle will not be affected. Therefore, a distance threshold may be set, and only the driving scene of the second vehicle within the distance threshold is identified. Setting the distance threshold can reduce the amount of data processing on the one hand, and save the cost of arranging sensors for the first vehicle on the other hand. This embodiment of the present application does not specifically limit whether a sensor is installed on the second vehicle.

The camera is used to record image data corresponding to the driving scene of the second vehicle. The working principle of the camera is to collect images through the lens, and then process the collected images by the internal photosensitive components and control components, and then convert them into digital signals that can be recognized by other systems; other systems obtain digital signals through the transmission port of the camera, and then proceed. Image restoration can get images that are consistent with the actual scene. In practical applications, the field of view of the camera to collect image data and the installation number and installation position of the camera can further design a feasible solution according to actual needs. The embodiments of the present application do not specifically limit the field of view, installation quantity, and installation position of the cameras.

The type of camera can be selected according to the different needs of users, as long as it can realize basic functions such as video camera, communication and still image capture. For example, the camera may be one or more types of commonly used in-vehicle cameras, such as a binocular camera and a monocular camera.

If selected according to the signal type, the camera can also be one or two types of digital camera and analog camera, the difference between the two is that the image processing process of the lens is different. The digital camera converts the collected analog signal into a digital signal for storage, while the analog camera uses a specific video capture card to convert the analog signal into a digital mode, compress it and store it. If classified according to the type of image sensor in the camera, the camera can also be one or both of the complementary metal oxide semiconductor (CMOS) type camera and the charge-coupled device (CCD) type camera. kind.

If divided by interface type, the camera can also be one or more types of serial port, parallel port, Universal Serial Bus (Universal Serial Bus, USB) and FireWire interface (IEEE1394). The embodiment of the present application also does not specifically limit the type of the camera.

Radar (radio detection and ranging, Radar) can be used to measure the distance between the first vehicle and different targets in road conditions, and can also be used to measure the speed of the second vehicle. Radar is an electronic device that detects the target by emitting electromagnetic waves. The working principle of the radar is to irradiate the target by emitting electromagnetic waves and receive its echoes, thereby obtaining the distance from the detected target to the electromagnetic wave emission point, and the rate of change of the distance (radial velocity). , bearing and altitude.

The type of radar can be selected according to the needs of the actual application scenario, which can be one or more of commonly used vehicle-mounted radars such as speed radar, lidar, millimeter-wave radar, and ultrasonic radar. Among them, LiDAR has the advantages of high precision, high resolution and the ability to build a 3D model of the surrounding because it sends a laser beam for detection, and is often used in vehicle automatic driving systems or assisted driving systems; for example, adaptive cruise control (Adaptive cruise control, ACC) ), Forward Collision Warning (FCW), Lane Keeping Assist (LKA) and Automatic Parking (AP).

Besides, the radar installed on the first vehicle may also be other types of radars that can satisfy the functions involved in the embodiments of the present application. When using radar to detect the position of the left lane line, a lane line detection method based on the density of radar scanning points is used. This method obtains the coordinates of the radar scanning points and converts them into a grid map, and maps the grid map with the original data. It is a direct coordinate grid map or a polar coordinate grid map. According to the needs of post-processing, the polar coordinate grid map is directly used for lane line recognition, that is, a grid with multiple point mappings is regarded as a lane line point.

The range of radar detection targets is related to its installation quantity, installation position and set detection distance. In specific applications, the installation quantity, installation position and detection distance of radars can be deployed according to actual needs. Likewise, the embodiments of the present application do not specifically limit the factors affecting the radar detection range.

In some embodiments, the first vehicle may also be equipped with an acceleration sensor. The acceleration sensor is used to measure acceleration and centripetal acceleration when the first vehicle is traveling. Acceleration sensors are usually composed of mass blocks, dampers, elastic elements, sensitive elements and adaptive circuits. During the acceleration process, the acceleration value is obtained by using Newton's second law by measuring the inertial force on the mass block. Acceleration sensors commonly used in automobiles are piezoresistive acceleration sensors, and can also be other types of sensors, such as capacitive, inductive, strain, and piezoelectric.

In addition, the first vehicle may also be installed with sensors for collecting environmental data, and the environmental data may be used to judge the next driving scene of the first vehicle. Environmental data includes but is not limited to temperature, humidity, air pressure, weather conditions, number of lanes, distance from traffic lights (traffic lights), ramp locations, prohibited areas, pedestrian locations, traffic light status information, etc. Corresponding to the environmental data, the sensor may also include, but is not limited to, a positioning sensor, an inertial measurement unit (IMU), a temperature sensor, a humidity sensor, a gas detection sensor, an environmental sensor, or other sensors for collecting environmental data etc., which are not limited in the embodiments of the present application.

T-Box has the functions of long-distance wireless communication and CAN communication, and provides long-distance communication interface for the whole vehicle. Specifically, the various data collected by the sensor are uploaded to the cloud server, and the data sent by the cloud server is fed back to the vehicle terminal installed in the vehicle or to the control system of the vehicle, so as to realize the assisted driving of the driver or the vehicle control system to control the vehicle. Autopilot. The in-vehicle terminal may be a mobile phone, a tablet computer or other mobile intelligent terminal used by the user, or an in-vehicle computer installed on the vehicle.

T-Box can also obtain vehicle status data through CAN bus interface, such as: vehicle information, vehicle controller information, motor controller information, battery management system BMS, on-board charger and mileage, average vehicle speed, fuel usage, Average fuel consumption, etc. T-box can also provide computing or storage functions.

The positioning module is used to realize the collection of vehicle position information. The positioning module can be a positioning module based on the global satellite positioning system (GPS), which can locate the vehicle by receiving GPS signals; it can also be a positioning module based on other satellite positioning systems, such as a positioning module based on the Beidou satellite positioning system, a positioning module based on the GLONER The positioning module of the GLONASS GPS and the positioning module based on the Galileo GPS.

FIG. 2 is a flowchart of a method for recognizing a driving scene of a vehicle provided by an embodiment of the present application. As shown in FIG. 2 , the method is applied to the first vehicle in the system shown in FIG. 1 , and the specific process of identifying the driving scene of the second vehicle around the first vehicle includes the following steps S1 to S4 .

Step S1. Obtain a feature sequence of the second vehicle in the current period according to the driving data of the first vehicle in the current period.

In the embodiment of the present application, the above-mentioned driving data is collected by a sensor installed on the first vehicle. Wherein, the sensor collects the driving data of the first vehicle in the current period according to the preset sampling frequency.

The driving data includes: the distance between the first vehicle and the reference lane line, the relative distance between the first vehicle and the second vehicle, the speed of the first vehicle, the centripetal acceleration of the first vehicle, the speed of the second vehicle, and the deviation of the second vehicle from the first vehicle. The included angle of the vertical direction of a vehicle.

The distance between the first vehicle and the reference lane line, the relative distance between the first vehicle and the second vehicle, and the speed of the first vehicle can be measured by radar, and the centripetal acceleration of the first vehicle can be measured by the acceleration sensor. The included angle in the vertical direction of a vehicle can be obtained by calculating the image collected by the camera.

After the sensor collects the driving data, it is transmitted to the T-Box of the first vehicle. The T-Box of the first vehicle uploads the driving data to the cloud server. After the cloud server receives the uploaded driving data, it first performs noise reduction and smoothing processing on the driving data. In some embodiments, a pre-designed mean filter, median filter, Gaussian filter or bilateral filter can be used to perform noise reduction and smoothing processing on the driving data. Figure 3 shows a graph of the two types of data without noise reduction and smoothing. As can be seen from Figure 3, due to the interference of the environment to the sensor, there are many glitches in the data sequence before processing. Noise reduction and smoothing are performed to filter out the glitches in the sequence and improve the accuracy of the data. Finally, the cloud server obtains the feature sequence of the second vehicle in the processed driving data.

In the embodiment of the present application, the feature sequence of the second vehicle includes: a distance sequence between the second vehicle and the reference lane line, an angle sequence between the second vehicle and the reference lane line, and a speed sequence of the second vehicle.

Further, the angle between the second vehicle and the reference lane line can be obtained according to the image collected by the camera installed on the first vehicle; the speed of the second vehicle can be obtained by the speed measuring radar of the first vehicle.

In the embodiments of the present application, the left lane line of the lane where the first vehicle travels is used as the reference lane line. In other embodiments, the right side lane line of the lane where the first vehicle travels may be used as the reference lane line

Next, with reference to the accompanying drawings, how to obtain the distance sequence between the second vehicle and the reference lane line in the embodiment of the present application is schematically introduced.

First, according to the centripetal acceleration of the first vehicle, the category of the lane in which the first vehicle travels is obtained. Among them, the categories of lanes are generally divided into straight lanes and curved lanes.

In the embodiment of the present application, when the centripetal acceleration of the first vehicle is zero, it is considered that the type of the lane in which the first vehicle travels is a straight lane, and when it is not zero, the type of the lane in which the first vehicle travels is a curved lane.

Then, a sequence of distances between the second vehicle and the reference lane line is determined according to the obtained driving data. According to different driving lanes of the first vehicle, specific solutions for the two situations are introduced below.

First, when the first vehicle is in the straight lane, a schematic diagram of the positional relationship between the first vehicle and the second vehicle in the straight lane is shown in FIG. 4a.

Substitute the relative distance d _S between the first vehicle and the second vehicle and the included angle α of the second vehicle deviates from the vertical direction of the first vehicle into formula (1) to obtain the difference between the first vehicle and the second vehicle in the first direction. and the distance between the first vehicle and the second vehicle in a second direction, wherein the first direction is the vertical direction of the body of the first vehicle, and the second direction is the horizontal direction of the body of the first vehicle.

In formula (1), d _Sx is the distance between the first vehicle and the second vehicle in the first direction, and d _Sy is the distance between the first vehicle and the second vehicle in the second direction.

Then, according to formula (2), the distance Ego_distoleft between the first vehicle and the reference lane line, and the distance d _Sy between the first vehicle and the second vehicle in the second direction, can be obtained to obtain the distance sequence between the second vehicle and the reference lane line Obj_distoleft.

Obj_distoleft=Ego_distoleft-d _Sy (2)

Second, when the first vehicle is in the curved lane, a schematic diagram of the positional relationship between the first vehicle and the second vehicle in the curved lane is shown in FIG. 4b.

Different from the straight lane, the curvature of the curved lane itself will increase the relative distance between the two vehicles. Therefore, it is necessary to perform curvature compensation on the distance sequence obtained above to eliminate the interference caused by the road surface characteristics to the distance sequence.

Specifically, firstly substitute the speed v of the first vehicle, the centripetal acceleration a _y of the first vehicle, and the distance d _Sx between the first vehicle and the second vehicle in the first direction into formula (3), to obtain the difference between the second vehicle and the second vehicle. The distance compensation sequence for the reference lane lines.

In formula (3), y _off is the distance compensation sequence between the second vehicle and the reference lane line, r is the radius of the lane where the first vehicle travels,

The significance of the sign function sign(a _y ) is to determine whether the second vehicle is on the left or right side of the first vehicle according to the centripetal acceleration of the first vehicle, so as to obtain the sign of the distance y _off , when the second vehicle is on the first vehicle When the vehicle is to the right, the sign of y _off is negative, otherwise, the sign of y _off is positive. In Figure 4b, d is the intermediate quantity,

Then, according to formula (4), the curvature compensation is performed on the reference sequence by using the distance compensation sequence, and the time sequence of the distance between the second vehicle and the reference lane line after compensation is obtained.

Obj_distoleft=Ego_distoleft-d′ _Sy (4)

In formula (4), d' _Sy is the difference between the reference sequence and the distance compensation sequence, d' _Sy =d _Sy -y _off .

Step S2. Calculate the similarity between the feature sequence of the second vehicle in the current period and the preset driving behavior reference sequence.

In the embodiment of the present application, this step firstly uses a preset window function to preprocess the feature sequence of the second vehicle, and then uses the dynamic time warping algorithm to calculate the similarity between the feature sequence and the preset driving behavior reference sequence Spend.

Specifically, first, according to the schematic diagram of calculating the similarity by combining the window function and the dynamic time warping algorithm as shown in FIG. 5 , the feature sequence is first discretized by using a preset window function. Common window functions include rectangular window, triangular window, Hanning window, Hamming window and Gaussian window. The window function adopted in the embodiments of the present application is a rectangular window. In Figure 5, winlen represents the width of the scan window, and shift represents the scan offset. The scanning window width is positively related to the length of the reference signal and positively related to the calculation time; the scanning offset is positively related to the sampling frequency and negatively related to the calculation time. The specific values of the two parameters will affect the accuracy of identifying driving behavior and the cost of computing time. In practical applications, the values of the two parameters can be selected and set from the combined values listed in Table 1 according to the identification accuracy of the vehicle driving behavior and the actual requirements of the computing time cost.

Table 1. The combination value table of the scan window width and scan offset of the window function

Then, the feature sequence and the preset driving behavior reference sequence are input into the dynamic time warping algorithm to obtain the similarity between the feature sequence output by the dynamic time warping algorithm and the preset driving behavior reference sequence. Figure 6 shows the similarity comparison of the two sequences. The dark curve in Figure 6 represents the reference sequence, and the light curve represents the sequence to be compared. The similarity between two sequences can be identified from the change trend of the sequences. The sequence on the left side of FIG. 6 is separated and enlarged to obtain the comparison diagram on the right side of FIG. 6 . As shown on the right side of Figure 6, from the parts indicated by the four arrows, it can be seen that there is a certain similarity between the two sequences. The dynamic time warping algorithm can be used to quantify this similarity and obtain the two sequences. similarity between.

In the embodiment of the present application, the preset driving behavior reference sequence is a standard feature sequence of each driving behavior extracted by analyzing the collected historical data.

The similarity is obtained by the dynamic time warping algorithm, which improves the speed of identifying the driving behavior of the vehicle; the use of the window function solves the problem that the dynamic time warping algorithm can only measure the similarity between two discrete time series and cannot handle continuous time series. .

In the embodiment of the present application, the preset driving behaviors include: left lane change, right lane change, left turn, right turn, left turn U-turn, right turn U-turn, acceleration, deceleration, and constant speed.

Further, whether the second vehicle is performing the driving behavior of turning left, turning right, turning left, or turning right can be identified by the sequence of included angles between the second vehicle and the reference lane line.

Among them, taking left turn and left turn as an example, in the reference sequence of left turn, there will be an included angle greater than 90° until equal to 180°, then the change trend of left turn is from an initial value greater than 0 and less than 90. 90. Further, it is possible to distinguish whether the driving behavior of the second vehicle is a left turn or a left turn. In addition, when distinguishing left and right directions, it is necessary to specify a positive direction in advance. For example, when the left side of the reference lane line is positive, when the angle sequence between the second vehicle and the reference lane line does not have an included angle greater than 90° and does not appear an included angle less than 0°, it can be determined that the second vehicle has an angle of less than 0°. The driving behavior is left steering; when the included angle sequence between the second vehicle and the reference lane line does not appear to be greater than 90°, but an included angle less than 0° appears in some time periods, the driving behavior of the second vehicle is determined to be right steering. Whether the second vehicle is performing acceleration, deceleration or constant speed driving behavior can be identified by the speed sequence of the second vehicle.

Whether the second vehicle is performing the driving behavior of changing lanes to the left or changing lanes can be identified through a sequence of distances between the second vehicle and the reference lane line.

Next, taking the reference lane line as the left lane line of the first vehicle as an example, the difference in the distance change from the reference lane line when the second vehicle changes lanes left and right is described. In other embodiments other than the present application, the right lane line of the lane where the first vehicle travels may also be used as the reference lane line.

Taking the second vehicle located in the right lane of the first vehicle as an example, and with reference to the accompanying drawings, the variation trend of the distance from the reference lane line when the second vehicle changes lanes left and right is illustrated schematically.

Fig. 7a shows a schematic diagram of the distance change from the reference lane line when the second vehicle changes lanes left. The horizontal axis of FIG. 7a represents time, and the vertical axis represents the distance between the second vehicle and the reference lane line. As shown in Figure 7a, when the second vehicle changes lanes to the left, when the reference lane line does not change, the second vehicle will gradually approach the reference lane line, and then the distance between the second vehicle and the reference lane line will gradually change from a certain initial value. become smaller. The fact that the reference lane line does not change indicates that the first vehicle is changing lanes.

Fig. 7b shows a schematic diagram of the distance change from the reference lane line when the second vehicle changes lanes right. The horizontal axis of FIG. 7b represents time, and the vertical axis represents the distance between the second vehicle and the reference lane line. As shown in Figure 7b, when the second vehicle changes lanes to the right, when the reference lane line does not change, the change trend is opposite to the change trend of the left lane change. The second vehicle gradually moves away from the reference lane line, and the second vehicle and the reference lane The distance of the line first increases gradually from a certain initial value.

In the following, a variation trend of the distance between the first vehicle and the reference lane line when the first vehicle changes lanes to the right when changing lanes to the right will be schematically described with reference to the accompanying drawings.

Fig. 7c shows a schematic diagram of the change of the distance from the reference lane line when the first vehicle changes lanes to the left. The horizontal axis of FIG. 7c represents time, and the vertical axis represents the distance between the first vehicle and the reference lane line. As shown in Figure 7c, when the first vehicle changes lanes to the left, the distance between the first vehicle and the reference lane line first gradually decreases from a certain initial value to 0, and then the corresponding reference lane line also changes due to the change of the driving lane. , so the distance between the first vehicle and the reference lane line suddenly changes from 0 to the maximum value, and then shows a gradually decreasing trend. The distance maximum in Figure 7a is related to the width of the lane in which the vehicle travels.

FIG. 7d shows a schematic diagram of the distance change from the reference lane line when the first vehicle changes lanes right. Similarly, the horizontal axis of FIG. 7d represents time, and the vertical axis represents the distance between the first vehicle and the reference lane line. As shown in Figure 7d, when the first vehicle changes lanes to the right, the distance between the first vehicle and the reference lane line first gradually increases from a certain initial value to a maximum value, and then due to the change of the driving lane, the corresponding reference lane line also occurs. changes, the distance between the first vehicle and the reference lane line suddenly changes from the maximum value to 0, and then shows a gradually increasing trend. The distance maximum of Figure 7d is also related to the width of the lane in which the vehicle travels.

Step S3. Obtain the driving behavior of the second vehicle according to the similarity.

In the embodiment of the present application, when the obtained similarity is greater than the preset similarity threshold, it is indicated that the preset driving behavior corresponding to the reference sequence for calculating the similarity is the driving behavior of the second vehicle, according to which the current driving behavior can be obtained. The driving behavior performed by the second vehicle during the time period.

For example, when the obtained similarity between the angle sequence between the second vehicle and the reference lane line and the left-turn reference sequence, the right-turn reference sequence, the left-turn U-turn reference sequence, or the right-turn turn reference sequence is greater than the threshold, it is considered that the second vehicle is in the Perform the appropriate driving behavior.

Similarly, it can be determined whether the second vehicle performs three types of driving behaviors of acceleration, deceleration or constant speed, as well as left lane change or right lane change driving behavior.

Step S4. Match the driving scene of the second vehicle in the preset scene library according to the driving behavior of the second vehicle.

In the embodiment of the present application, different driving scenarios and their corresponding driving behaviors are in the preset scenario library. This step is to use the scene library to classify the driving behavior of the second vehicle obtained in step S3 to obtain the driving scene of the second vehicle.

When building a scene library, firstly, it is necessary to mark multiple driving image data samples of the collected vehicle, and then identify and mark the driving behavior corresponding to each scene from the driving image data corresponding to each scene; Driving behaviors form a scene library.

FIG. 8 shows a corresponding relationship diagram between driving scenarios and driving behaviors. As shown in Figure 8, the driving scenarios include: lane keeping, steering, U-turn, lane change, and overtaking.

In the embodiment of the present application, the corresponding relationship between the identified driving scene and the driving behavior is as follows:

The driving behaviors corresponding to the lane keeping scenario include: vehicle acceleration, vehicle deceleration, and vehicle driving at a constant speed;

The driving behavior corresponding to the steering scene includes: the vehicle turns left and the vehicle turns right;

The driving behaviors corresponding to the U-turn scene include: the vehicle turns left and the vehicle turns right;

The driving behaviors corresponding to the lane change scenarios include: vehicle left lane change and vehicle right lane change;

The driving behaviors corresponding to the overtaking scenario include: the vehicle accelerates, the vehicle changes lanes to the left, the vehicle drives at a constant speed, the vehicle changes lanes to the right, and the vehicle decelerates.

In addition, in other embodiments, the starting and parking scenarios can be obtained by judging the change of the speed of the vehicle. For example, when the speed of the vehicle gradually increases from 0, it can be confirmed that the vehicle executes the start-up scenario; when the vehicle gradually decreases from a certain speed to 0, it can be confirmed that the vehicle executes the parking scenario.

In another embodiment of the present application, calculating the similarity between the feature sequence of the second vehicle and the preset driving behavior reference sequence in step S2 in the above-mentioned vehicle driving scene recognition method may further include: as shown in FIG. 9 Step S21 to Step S23.

Step S21. Identify the driving area of the first vehicle in the current time period.

In a possible implementation manner, in this step, a location module on the first vehicle may be used to determine the driving area of the first vehicle in the current time period. There are certain differences in the characteristics of the driving behavior of vehicles in different areas, which are related to the driving habits and road topology characteristics of drivers in the area. For example, the embodiment of the present application performs statistical analysis on the historical data of vehicle cut-ins in two areas, and obtains a probability distribution diagram of the driving behavior of the second vehicle in the two areas. When the probabilities are the same, the temporal distances between vehicles in each area are different. FIG. 10 shows a probability distribution diagram of the insertion of the second vehicle into the driving lane of the first vehicle in area 1 and area 2. As shown in Figure 10, when the probability of occurrence is the same as 84.135%, the time distance between two vehicles in area 1 is between μ ₁ -σ ₁ and μ ₁ +σ ₁ , and the time between two vehicles in area 2 The distance is between μ ₂ -σ ₂ and μ ₂ +σ ₂ .

Among them, μ ₁ is the mean value of the time distance between the first vehicle and the second vehicle when the second vehicle in area 1 changes lanes to the left, and σ ₁ is the time distance between the first vehicle and the second vehicle when the second vehicle in area 1 changes lanes. The variance of the time distance between the two vehicles, μ ₂ is the mean value of the time distance between the first vehicle and the second vehicle when the second vehicle in area 2 changes lanes, σ ₂ is the second vehicle in area 2 The variance of the temporal distance between the first vehicle and the second vehicle when changing lanes.

Step S22. Extract the driving behavior reference sequence corresponding to the driving area in the database.

In a possible implementation, the database includes driving behavior reference sequences corresponding to different regions. The driving behavior reference sequences corresponding to different regions are obtained by performing statistical analysis on the vehicle driving data corresponding to the corresponding driving behaviors.

Step S23. Combine the window function and the dynamic time warping algorithm to obtain the similarity between the feature sequence and the extracted driving behavior reference sequence.

The embodiment of the present application is based on the precondition that the characteristics of the driving behavior of vehicles in different areas are different. When calculating the similarity between the characteristic sequence and the reference sequence, only the characteristic sequence of the second vehicle and the area corresponding to the driving area of the first vehicle are calculated. The driving behavior benchmark sequence can improve the accuracy of similarity, reduce the amount of calculation, and improve the recognition speed.

Based on the methods of the above embodiments, the present application further provides a vehicle driving scene recognition device. The identification device is deployed in the cloud server, and can be configured to communicate with the vehicle-mounted terminal on the first vehicle, so as to feed back the identification result of the identification device to the user. FIG. 11 shows a vehicle driving scene recognition device provided by an embodiment of the present application. As shown in Figure 11, the identification device specifically includes:

a feature obtaining unit, configured to obtain a feature sequence of the second vehicle according to the driving data;

The scene matching unit is configured to use a pre-built scene library to match the driving scene of the second vehicle in the current period according to the driving behavior of the second vehicle in the current period.

In the embodiment of the present application, the similarity calculation unit is also used for:

Identifying the area where the first vehicle is traveling in the current time period;

A preset driving behavior reference sequence corresponding to the area where the first vehicle travels is extracted from the database.

In the embodiment of the present application, the feature acquisition unit is also used for:

Noise reduction and smoothing are performed on the driving data.

In the embodiment of the present application, the reference lane line is one lane line of the lane in which the first vehicle travels.

In the embodiment of the present application, the lane in which the first vehicle travels is a curved lane, and the device further includes:

The curvature compensation unit is used to perform curvature compensation on the distance sequence according to the curvature of the curved lane.

In the embodiment of this application, the behavior recognition unit is specifically used for:

Determine whether the similarity satisfies the preset conditions;

The driving behavior represented by the driving behavior reference sequence corresponding to the similarity satisfying the preset condition is taken as the driving behavior of the second vehicle in the current period.

In the embodiment of the present application, the apparatus further includes a scene library construction unit for:

According to the historical driving images corresponding to different scenarios, the driving behaviors corresponding to different scenarios are obtained;

The scene library is constructed according to different scenes and driving behaviors corresponding to the different scenes.

It should be understood that the above-mentioned apparatus is used to execute the method in the above-mentioned embodiment, the implementation principle and technical effect of the corresponding program module in the apparatus are similar to those described in the above-mentioned method embodiment, and the working process of the apparatus may refer to the above-mentioned method. The corresponding process will not be repeated here.

The vehicle-mounted terminal may be one of a smart phone, a tablet computer, or a vehicle-mounted computer. The vehicle terminal includes a processor, a display module and a data interface. The processor may be a general purpose processor or a special purpose processor. For example, a processor may include a central processing unit (CPU) and/or a baseband processor. The baseband processor may be used to process communication data, and the CPU may be used to implement corresponding control and processing functions, execute software programs, and process data of software programs.

The data interface is used to receive data, and the display module is used to display driving behavior and scene recognition results. The processor receives data through a data interface, and sends a display instruction to the display module after calculation, so as to display the result sent by the cloud.

In addition, the vehicle terminal may also include: a charging management module, a power management module, a battery, an antenna, a mobile communication module, a wireless communication module, an audio module, a speaker, an earphone interface, an audio Bluetooth module, a display screen, a modem, and a baseband processor.

The charging management module can receive the charging input of the wired charger through the USB interface. In some wireless charging embodiments, the charging management module may receive wireless charging input through the wireless charging coil of the terminal device. While the charging management module charges the battery, it can also supply power to other devices through the power management module.

The power management module is used to connect the battery, the charge management module and the processor. The power management module receives input from the battery and/or charging management module, and supplies power to the processor, the display screen, and the wireless communication module. The power management module can also be used to monitor parameters such as battery capacity, battery cycle times, battery health status (leakage, impedance). In some other embodiments, the power management module may also be provided in the processor. In other embodiments, the power management module and the charging management module may also be provided in the same device.

The wireless communication function of the vehicle terminal can realize the communication with the server through the antenna, mobile communication module, wireless communication module, modem and baseband processor.

The mobile communication module can provide wireless communication solutions including 2G/3G/4G/5G applied on the vehicle terminal. The mobile communication module may include at least one filter, switch, power amplifier, low noise amplifier (LNA) and the like. The mobile communication module can receive electromagnetic waves by at least two antennas, filter and amplify the received electromagnetic waves, and transmit them to the modem for demodulation. The mobile communication module can also amplify the signal modulated by the modem and radiate it into electromagnetic waves through the antenna. In some embodiments, at least part of the functional modules of the mobile communication module may be provided in the same device as at least part of the modules of the processor.

A modem may include a modulator and a demodulator. Wherein, the modulator is used to modulate the low frequency baseband signal to be sent into a medium and high frequency signal. The demodulator is used to demodulate the received electromagnetic wave signal into a low frequency baseband signal. Then the demodulator transmits the demodulated low-frequency baseband signal to the baseband processor for processing. The low frequency baseband signal is processed by the baseband processor and passed to the application processor. In some embodiments, the modem may be a stand-alone device. In other embodiments, the modem may be independent of the processor, and may be provided in the same device as the mobile communication module or other functional modules. In other embodiments, the mobile communication module may be a module in a modem.

The wireless communication module can provide applications on terminal devices including wireless local area networks (WLAN) (such as wireless fidelity (Wi-Fi) networks), global navigation satellite systems (GNSS) , frequency modulation (frequency modulation, FM), near field communication technology (near field communication, NFC), infrared technology (infrared, IR) and other wireless communication solutions. The wireless communication module may be one or more devices integrating at least one communication processing module. The wireless communication module receives electromagnetic waves through the antenna, frequency modulates and filters the electromagnetic wave signals, and sends the processed signals to the processor. The wireless communication module can also receive the signal to be sent from the processor, frequency-modulate it, amplify it, and radiate it into electromagnetic waves through the antenna.

The vehicle-mounted terminal can communicate with the server through the mobile communication module and the wireless communication module, receive the identification result issued by the server, or transmit data to the server.

The display screen is used to display the recognized driving behavior and driving scenarios in the form of images or videos. The display includes a display panel. The display panel can be a liquid crystal display (LCD), an organic light-emitting diode (OLED), an active-matrix organic light-emitting diode or an active-matrix organic light-emitting diode (active-matrix organic light). emitting diode, AMOLED), flexible light-emitting diode (flex light-emitting diode, FLED), Miniled, MicroLed, Micro-oLed, quantum dot light-emitting diode (quantum dot light emitting diodes, QLED) and so on. In some embodiments, the vehicle-mounted terminal may include one or more display screens. In one example, the display screen may also be used to display the interface of the application, displaying visual controls in the interface of the application.

The audio module is used to convert digital audio information into analog audio signal output, and also used to convert analog audio input to digital audio signal. The audio module can also be used to encode and decode audio signals. In some embodiments, the audio module may be provided in the processor, or some functional modules of the audio module may be provided in the processor. In some embodiments, the audio module is used to feed back the recognition result to the user in the form of speech. The loudspeaker is used for feedback of the external sound to the user.

The headphone jack is used to connect wired headphones. The headphone interface can be a USB interface, or a 3.5mm open mobile terminal platform (OMTP) standard interface or a cellular telecommunications industry association of the USA (CTIA) standard interface. The audio Bluetooth module is used to connect the user's Bluetooth headset. This embodiment of the present application does not limit the Bluetooth technology version applied to the audio Bluetooth module, and the Bluetooth chip may be a chip applying any version of the Bluetooth technology.

The user can receive the voice corresponding to the recognition result in wired or wireless form through the headphone interface or the audio Bluetooth module.

Based on the above method embodiments, the present application further provides a computer storage medium, where instructions are stored in the computer storage medium, and when the instructions are executed on the computer, the computer is made to execute a vehicle driving scene recognition method as in the embodiments of the present application .

Based on the above method embodiments, the present application further provides a computer program product containing instructions, when the instructions are run on a computer, the computer executes a vehicle driving scene recognition method as in the embodiments of the present application.

The method steps in the embodiments of the present application may be implemented in a hardware manner, or may be implemented in a manner in which a processor executes software instructions. The software instructions may be composed of corresponding software modules, and the software modules may be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), Erasable Programmable Read-Only Memory (erasablePROM, EPROM), Electrically Erasable Programmable Read-Only Memory (electrically EPROM, EEPROM), registers, hard disks, removable hard disks, CD-ROMs, or any other form of storage medium known in the art middle. An illustrative storage medium is coupled to the processor, such that the processor can read information from, and write information to, the storage medium. Of course, the storage medium can also be an integral part of the processor. The processor and storage medium may reside in an ASIC.

In the above-mentioned embodiments, it may be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented in software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer may be a general purpose computer, special purpose computer, computer network, or other programmable device. The computer instructions may be stored in or transmitted over a computer-readable storage medium. The computer instructions can be sent from one website site, computer, server, or data center to another website site by wire (eg, coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (eg, infrared, wireless, microwave, etc.) , computer, server or data center. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, a data center, or the like that includes an integration of one or more available media. The usable media may be magnetic media (eg, floppy disks, hard disks, magnetic tapes), optical media (eg, DVDs), or semiconductor media (eg, solid state disks (SSDs)), and the like.

It can be understood that, the various numbers and numbers involved in the embodiments of the present application are only for the convenience of description, and are not used to limit the scope of the embodiments of the present application.

Claims

A vehicle driving scene recognition method, characterized in that the method comprises:

obtain the driving data of the first vehicle in the current period;

A feature sequence of a second vehicle is obtained according to the driving data, the second vehicle is used to indicate vehicles around the first vehicle, and the feature sequence includes: a speed sequence, a distance between the second vehicle and a reference lane line a sequence and a sequence of angles between the second vehicle and the reference lane line;

Using a dynamic time warping algorithm to obtain the similarity between the feature sequence of the second vehicle and a preset driving behavior reference sequence;

determining the driving behavior of the second vehicle in the current period according to the similarity;

According to the driving behavior of the second vehicle in the current period, a pre-built scene library is used to match the driving scene of the second vehicle in the current period.
The method according to claim 1, characterized in that, before obtaining the similarity between the characteristic sequence of the second vehicle and the driving behavior reference sequence by using a dynamic time warping algorithm, the method further comprises:

identifying the area in which the first vehicle travels;

The preset driving behavior reference sequence corresponding to the area where the first vehicle travels is extracted from a database, where the database includes driving behavior reference sequences corresponding to different areas.
The method according to claim 1 or 2, wherein before obtaining the characteristic sequence of the second vehicle according to the driving data, the method further comprises:

Noise reduction processing and smoothing processing are performed on the driving data.
The method according to any one of claims 1 to 3, wherein the reference lane line is a lane line of a lane in which the first vehicle travels.
The method according to any one of claims 1 to 4, wherein the lane in which the first vehicle travels is a curved lane, and the method further comprises:

The distance sequence is curvature compensated according to the curvature of the curved lane.
The method according to any one of claims 1 to 5, wherein the determining the driving behavior of the second vehicle in the current period according to the similarity comprises:

Judging whether the similarity satisfies a preset condition;

The driving behavior represented by the driving behavior reference sequence corresponding to the degree of similarity satisfying the preset condition is used as the driving behavior of the second vehicle in the current period.
The method according to any one of claims 1 to 6, further comprising:

obtaining driving behaviors corresponding to the different scenarios according to the historical driving images corresponding to the different scenarios;

The scene library is constructed according to the different scenes and the driving behaviors corresponding to the different scenes.
A vehicle driving scene recognition device, characterized in that the device comprises:

a data acquisition unit for acquiring the driving data of the first vehicle in the current period;

a feature acquisition unit, configured to obtain a feature sequence of a second vehicle according to the driving data, the second vehicle is used to indicate vehicles around the first vehicle, and the feature sequence includes: a speed sequence, the second vehicle a sequence of distances from the reference lane line and a sequence of included angles between the second vehicle and the reference lane line;

a similarity calculation unit, configured to obtain the similarity between the feature sequence of the second vehicle and a plurality of preset driving behavior reference sequences by using a dynamic time warping algorithm;

a behavior recognition unit, configured to determine the driving behavior of the second vehicle in the current period according to the similarity;

A scene matching unit, configured to match the driving scene of the second vehicle in the current period by using a pre-built scene library according to the driving behavior of the second vehicle in the current period.
The device according to claim 8, wherein the similarity calculation unit is further configured to:

identifying the area in which the first vehicle travels;

The preset driving behavior reference sequence corresponding to the area where the first vehicle travels is extracted from a database, where the database includes driving behavior reference sequences corresponding to different areas.
The device according to claim 8 or 9, wherein the feature acquisition unit is further configured to:

Noise reduction processing and smoothing processing are performed on the driving data.
The device according to any one of claims 8 to 10, wherein the reference lane line is a lane line of a lane in which the first vehicle travels.
The device according to any one of claims 8 to 11, wherein the lane in which the first vehicle travels is a curved lane, and the device further comprises:

A curvature compensation unit, configured to perform curvature compensation on the distance sequence according to the curvature of the curved lane.
The device according to any one of claims 8 to 12, wherein the behavior recognition unit is specifically configured to:

Judging whether the similarity satisfies a preset condition;

The driving behavior represented by the driving behavior reference sequence corresponding to the degree of similarity satisfying the preset condition is used as the driving behavior of the second vehicle in the current period.
The device according to any one of claims 8 to 13, further comprising a scene library construction unit for:

obtaining driving behaviors corresponding to the different scenarios according to the historical driving images corresponding to the different scenarios;

The scene library is constructed according to the different scenes and the driving behaviors corresponding to the different scenes.
A computer storage medium in which instructions are stored, and when the instructions are executed on a computer, cause the computer to execute the method according to any one of claims 1-7.
A computer program product comprising instructions which, when executed on a computer, cause the computer to perform the method of any of claims 1-7.