Stylized lane changing scene reconstruction method based on natural driving data

Abstract

The invention provides a stylized lane changing scene reconstruction method based on natural driving data, and the method comprises the following steps: collecting the natural driving data, and obtaining the state information of a main vehicle and the state information of other traffic participants; according to a free lane changing scene extraction algorithm, free lane changing scene fragments are mined and extracted from the natural driving data; performing clustering analysis on the extracted free lane changing scene fragments to obtain different types of free lane changing fragment data sets; according to a vehicle trajectory generation method, generating different types of free lane changing fragment trajectory files from the obtained different types of lane changing scene fragment data; and according to the different types of free lane changing fragment trajectory files, generating different types of free lane changing simulation scene files based on a Modeldesk scene generation method. In the creation of the invention, the data set of the free lane changing scene is rich, the scene is diversified, the scene identification extraction and generation efficiency is high, and the simulation precision and the scene reproduction capability are effectively improved.

Description

Technical Field

[0001] The invention relates to the field of simulation test technology in intelligent driving, and in particular to a stylized lane change scene reconstruction method based on natural driving data. Background Art

[0002] Lane changing is the most common operation for vehicles, and nearly 18% of accidents that occur during lane changing are caused by the driver's wrong decision to change lanes. Lane changing behavior carries a greater risk than following a vehicle, because during the lane changing process, you need to pay attention not only to the movement of vehicles in front and behind the current lane, but also to the traffic status of the target lane. In-depth research on this driving behavior will help the development of related intelligent driving assistance technologies. This invention conducts research on this special scenario of free lane changing of vehicles, and proposes a method including data collection, scenario mining, driving behavior clustering and simulation scenario reconstruction, which lays the foundation for the development of intelligent driving assistance system technology in this scenario.

[0003] At present, the research focus of autonomous driving vehicles has gradually shifted from the breakthrough and realization of autonomous driving technology to the testing and evaluation technology of autonomous driving functions. Simulation scenarios are gradually becoming the core means of evaluating the performance and safety of intelligent vehicles due to their significantly improved testing efficiency and greatly reduced cost advantages. At present, there are problems such as limited data range, insufficient scene diversity, low scene recognition extraction and generation efficiency, single scene category, and insufficient consideration of driver characteristics in the natural driving scene library for free lane change scenes. It is not enough to cover the simulation verification of intelligent driving decision-making systems, and cannot meet the needs of existing intelligent driving systems in terms of functional safety, human-likeness, comfort and other related technical development and simulation verification. Summary of the invention

[0004] In view of this, the present invention aims to propose a stylized lane changing scene reconstruction method based on natural driving data, so as to solve the problems of insufficient data set of free lane changing scenes, insufficient scene diversity, low efficiency of scene recognition extraction and generation, single scene category, and insufficient consideration of driver characteristic factors.

[0005] To achieve the above object, the technical solution created by the present invention is implemented as follows:

[0006] A stylized lane-changing scene reconstruction method based on natural driving data comprises the following steps:

[0007] Step S1: Collect natural driving data to obtain the status information of the main vehicle and other traffic participants;

[0008] Step S2: extracting free lane change scene segments from natural driving data according to a free lane change scene extraction algorithm;

[0009] Step S3: performing cluster analysis on the extracted free lane change scene segments to obtain different types of free lane change segment data sets;

[0010] Step S4: generating different types of free lane change segment trajectory files from the acquired different types of lane change scene segment data according to the vehicle trajectory generation method;

[0011] Step S5: generating different types of free lane change simulation scenario files based on the scenario generation method of Modeldesk according to different types of free lane change segment trajectory files.

[0012] Furthermore, in step S1:

[0013] The main vehicle status information includes: time series, main vehicle speed, steering wheel angle, distance between the main vehicle and lane lines on both sides, lateral speed of the main vehicle to lane lines on both sides, confidence of lane lines on both sides of the main vehicle, and width of the main vehicle;

[0014] The other traffic participant status information includes: the distance between the host vehicle and other traffic participants, the angle between the host vehicle and other traffic participants, and the type of traffic participants.

[0015] Furthermore, in step S2, the free lane change scene extraction algorithm is:

[0016] Compare the left lane line confidence and the right lane line confidence. If the left lane line confidence is higher than the right lane line confidence, the left lane line distance is prioritized for identification. If the right lane line confidence is higher than the left lane line confidence, the right lane line distance is prioritized for identification.

[0017] Furthermore, in step S2, the free lane change scene extraction algorithm is:

[0018] If the left lane line confidence is higher than the right lane line confidence:

[0019] When the distance between the main vehicle and the left lane line suddenly changes from a minimum value to a maximum value, it is recognized as a left lane change;

[0020] When the distance between the main vehicle and the left lane line suddenly changes from a maximum value to a minimum value, it is recognized as a lane change to the right;

[0021] If the confidence level of the right lane line is higher than that of the left lane line:

[0022] When the lane line to the right of the main vehicle suddenly changes from a minimum to a maximum, it is recognized as a lane change to the right;

[0023] When the lane line to the right of the main vehicle suddenly changes from a maximum value to a minimum value, it is recognized as a left lane change;

[0024] Furthermore, in step S2, the free lane change scene extraction algorithm further includes: defining the start and end time of each free lane change scene, and based on the lane change start time and the lane change end time, intercepting and saving a data scene segment containing a complete free lane change behavior.

[0025] Furthermore, in step S3, the cluster analysis method includes: free lane change feature parameter extraction, feature parameter dimension reduction, and driving behavior clustering;

[0026] Furthermore, in step S3, the cluster analysis method includes:

[0027] Feature parameter extraction: The vehicle's average lateral speed, maximum speed, minimum speed, average lateral acceleration, maximum acceleration, and minimum acceleration; the vehicle's average longitudinal speed, maximum speed, minimum speed, average longitudinal acceleration, maximum acceleration, and minimum acceleration, as well as the longitudinal distance, lateral distance, and duration of the lane-changing process are taken as feature parameters;

[0028] Feature parameter dimension reduction: principal component analysis method is used to implement feature dimension reduction;

[0029] Driving behavior clustering: Based on the results of dimensionality reduction, the K-means unsupervised clustering method is used to perform cluster analysis on the free lane change segments.

[0030] Furthermore, in step S4, the vehicle trajectory generation method includes the Frenet coordinate system definition, the main vehicle trajectory generation method and the trajectory generation method of other traffic participants;

[0031] Furthermore, in step S4, the vehicle trajectory generation method includes:

[0032] Step S4.1: Frenet coordinate system definition:

[0033] The Frenet coordinate system is based on the centerline of the road and decomposes the vehicle motion into lateral offset and longitudinal offset relative to the centerline. It is easier to describe the trajectory of the vehicle in a simulation environment. The Frenet coordinate system is constructed with the centerline of the road as the longitudinal axis s and the direction perpendicular to the centerline of the road as the transverse axis d. The position of the vehicle or obstacle on the road can be expressed as (s, d), where s is described as the longitudinal offset on the road and d is described as the lateral displacement on the road.

[0034] Step S4.2: Main vehicle trajectory generation:

[0035] The main vehicle trajectory includes the main vehicle longitudinal displacement and the main vehicle lateral displacement;

[0036] First, according to the speed Stlc from the vehicle to the lane line and the speed Ego_v of the vehicle, the longitudinal speed Ego_v_x of the vehicle along the lane line is obtained;

[0037] Secondly, according to the longitudinal velocity Ego_v_x of the main vehicle along the lane line, the longitudinal displacement Ego_d_x of the main vehicle along the lane line is calculated as follows:

[0038]

[0039] The lateral displacement of the main vehicle is obtained according to the characteristic curve of the distance Dtlc_1 from the vehicle to the lane line acquired by the functional camera;

[0040] The calculation method of the lateral displacement Ego_d_y of the main vehicle is as follows:

[0041]

[0042] Where: N is the data length, Dn is the threshold for judging the lateral step, Q is the lateral displacement step difference data set, ID is the id set of the lateral displacement step moment, and M is the number of elements in the Q data set;

[0043] Step S4.3: Target vehicle trajectory generation:

[0044] The target vehicle trajectory includes the target vehicle longitudinal displacement Obj_d_x and the target vehicle lateral displacement Obj_d_y;

[0045] First, according to the relative distance Range of the main vehicle and the relative angle angle relative to the main vehicle, the relative longitudinal distance Obj_dr_x and the relative lateral distance Obj_dr_y of the target vehicle are obtained;

[0046] Secondly, the target vehicle trajectory is obtained according to the main vehicle trajectory, as follows:

[0047]

[0048] Further, in step S5, the scenario generation method based on Modeldesk includes: constructing a simulation scenario of lane-changing driving behavior based on Modeldesk software according to three types of lane-changing scenario data obtained by clustering driving behaviors;

[0049] Furthermore, in step S5, the Modeldesk-based scene generation method includes:

[0050] Road construction: set road geometry, road surface, lane attributes, road markings and other elements according to actual conditions;

[0051] Construction of road facilities and restrictions: Setting road facilities and restriction attributes, including setting road boundaries and traffic sign information;

[0052] Construction of temporary physical restrictions: Setting the properties of temporary objects and facilities on the road surface and above the road section according to actual conditions;

[0053] Meteorological conditions settings: set the lighting conditions, weather conditions and other meteorological conditions attributes;

[0054] Construction of traffic participants: according to the vehicle attributes of the main vehicle, set the main vehicle type, size and other parameters;

[0055] The driving trajectory of the main vehicle is set according to the above-mentioned main vehicle trajectory construction method. Similarly, the parameters of other traffic participants are set according to their types, quantities, characteristics and other attributes. The driving trajectories of other traffic participants are set according to the above-mentioned trajectory construction method of other traffic participants. Finally, the generation of different types of free lane change scene files is completed.

[0056] An electronic device includes a processor and a memory connected to the processor for storing instructions executable by the processor, wherein the processor is used to execute the above-mentioned stylized lane change scene reconstruction method based on natural driving data.

[0057] A server includes at least one processor and a memory in communication with the processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the processor so that the at least one processor executes the above-mentioned stylized lane change scene reconstruction method based on natural driving data.

[0058] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above-mentioned stylized lane-changing scene reconstruction method based on natural driving data.

[0059] Compared with the prior art, the stylized lane change scene reconstruction method based on natural driving data created by the present invention has the following advantages:

[0060] 1. The stylized lane-changing scene reconstruction method based on natural driving data created by the present invention can realize the data collection and storage of the host vehicle and other traffic participants, and can ensure the richness and continuity of the data.

[0061] 2. The stylized lane-changing scene reconstruction method based on natural driving data created by the present invention can realize the automatic extraction of free lane-changing scenes, can quickly and accurately locate the starting time of the scene segment of the host vehicle's free lane change from a large amount of multi-source data, and can automatically extract the free lane-changing scene segment data that meets the conditions efficiently and accurately, greatly improving the recognition accuracy and extraction efficiency of the host vehicle's free lane-changing scenes.

[0062] 3. The stylized lane-changing scene reconstruction method based on natural driving data created by the present invention can realize the automatic clustering of free lane-changing scenes. Compared with the manual classification method, it greatly improves the classification efficiency and accuracy of the free lane-changing scene clustering.

[0063] 4. The stylized lane-changing scene reconstruction method based on natural driving data created by the present invention can integrate real vehicle data with Modeldesk simulation software from aspects such as road environment, weather conditions, and traffic participants, effectively improving simulation accuracy and scene reproduction capabilities, and improving the consistency between simulation tests and real vehicle tests. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The drawings constituting part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments and descriptions of the present invention are used to explain the present invention and do not constitute an improper limitation on the present invention. In the drawings:

[0065] Figure 1 A flow chart of a method for reconstructing a stylized lane change scene based on natural driving data according to an embodiment of the present invention is shown;

[0066] Figure 2 A schematic diagram showing the arrangement of information collection units for a stylized lane change scene reconstruction method based on natural driving data according to an embodiment of the present invention is shown;

[0067] Figure 3 A schematic diagram of a free lane-changing scene extraction algorithm of a stylized lane-changing scene reconstruction method based on natural driving data according to an embodiment of the present invention is shown;

[0068] Figure 4 A schematic diagram of free lane changing according to a stylized lane changing scene reconstruction method based on natural driving data according to an embodiment of the present invention is shown;

[0069] Figure 5 The figure shows a schematic diagram of DTLC changes when the left lane line is used as a reference line in the stylized lane change scene reconstruction method based on natural driving data according to an embodiment of the present invention (including: a case where a lane change to the left is identified (a); a case where a lane change to the right is identified (b));

[0070] Figure 6The figure shows a schematic diagram of DTLC changes when the stylized lane change scene reconstruction method based on natural driving data described in the embodiment of the invention takes the right lane line as the reference line (including: the case of identifying a lane change to the right (a); the case of identifying a lane change to the left (b));

[0071] Figure 7 A schematic diagram showing clustering results of the stylized lane-changing scene reconstruction method based on natural driving data according to an embodiment of the present invention (including: clustering results of the maximum speed of the vehicle (a); clustering results of the maximum acceleration of the vehicle (b); clustering results of the lane-changing duration of the vehicle (c); and clustering results of the average lateral speed of the vehicle (d));

[0072] Figure 8 A schematic diagram of the Frenet coordinate system structure of the stylized lane change scene reconstruction method based on natural driving data described in an embodiment of the present invention is shown;

[0073] Fig. 9 A schematic diagram of a lane changing simulation scene of a stylized lane changing scene reconstruction method based on natural driving data described in an embodiment of the present invention is shown (including: type 0 lane changing simulation scene (a); type 1 lane changing simulation scene (b); type 2 lane changing simulation scene (c)).

[0074] Description of reference numerals:

[0075] 1-Data storage device; 2-Data recording system; 3-GNSS; 4-Display; 5-Function camera; 6-HD camera; 7-LiDAR; 8-Millimeter wave radar. DETAILED DESCRIPTION

[0076] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0077] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0078] Figure 1 FIG. 1 is a flowchart of a method for reconstructing a stylized lane change scene based on natural driving data according to an embodiment of the present invention. Figure 1 As shown in FIG. 1 , the stylized lane change scene reconstruction method based on natural driving data includes the following steps:

[0079] Step S1: Collect natural driving data to obtain the status information of the main vehicle and other traffic participants;

[0080] Step S2: extracting free lane change scene segments from natural driving data according to a free lane change scene extraction algorithm;

[0081] Step S3: performing cluster analysis on the extracted free lane change scene segments to obtain different types of free lane change segment data sets;

[0082] Step S4: generating different types of free lane change segment trajectory files from the acquired different types of lane change scene segment data according to the vehicle trajectory generation method;

[0083] Step S5: generating different types of free lane change simulation scenario files based on the scenario generation method of Modeldesk according to different types of free lane change segment trajectory files.

[0084] Further, in step S1,

[0085] The main vehicle status information includes: time series t, main vehicle speed, steering wheel angle, distance between the main vehicle and lane lines on both sides, lateral speed of the main vehicle to lane lines on both sides, confidence of lane lines on both sides of the main vehicle, width of the main vehicle and other information; the other traffic participant status information includes: distance between the main vehicle and other traffic participants, angle between the main vehicle and other traffic participants, type of traffic participants and other information. It should be noted that: This paper constructs a comprehensive data acquisition system. The system includes a global navigation satellite system 3 (GNSS), which accurately captures and records key status information such as the vehicle's own position, speed and direction. At the same time, in order to perceive the driving environment in all directions, the vehicle is equipped with high-definition cameras 6, functional cameras 5, millimeter-wave radars 8 and lidars 7, such as Figure 2 As shown. These devices work together to accurately identify target objects on the road (such as other vehicles and pedestrians), and can also clearly capture the dynamic changes of lane lines, traffic signs and the surrounding environment. In addition, the test vehicle is also equipped with a multi-interface high-performance industrial computer, such as data storage device 1 and data recording system 2, which records various types of data in real time and opens the road data acquisition system.

[0086] In step S2,

[0087] The free lane change scene extraction algorithm is as follows Figure 3 As shown,

[0088] Compare the confidence of the left and right lane lines. If the confidence of the left lane line is the highest, the distance to the left lane line is prioritized for identification. The constraints are as follows (the same applies to the right side):

[0089]

[0090] Among them, Dtlc is the distance from the main vehicle to the lane line, Lc min is the shortest distance threshold to the lane line, Lc max is the maximum distance threshold from the lane line, SWA min is the steering wheel angle, SWA minis the steering wheel angle threshold, V ego The speed of the main vehicle.

[0091] like Figure 4 As shown, t0 is the moment when the vehicle crosses the lane line, t1 is the starting moment of free lane change, and t2 is the ending moment of free lane change.

[0092] If the left lane line confidence is higher than the right lane line confidence:

[0093] like Figure 5 As shown in a, when the distance between the main vehicle and the left lane line changes from a minimum value to a maximum value, it is identified as a left lane change, and the constraint conditions are as follows: DTLC(T0-1)-DTLC(T1+1)<X<0, where X is the set distance threshold; Figure 5 As shown in b, when the distance from the main vehicle to the left lane line suddenly changes from a maximum value to a minimum value, it is identified as a lane change to the right. The constraint conditions are as follows: DTLC(T0-1)-DTLC(T1+1)>X>0

[0094] If the confidence level of the right lane line is higher than that of the left lane line:

[0095] like Figure 6 As shown in a, when the lane line to the right of the main vehicle suddenly changes from the minimum value to the maximum value, it is recognized as changing lanes to the right. The constraints are as follows: DTLC(T0-1)-DTLC(T1+1)<X<0; Figure 6 As shown in b, when the lane line on the right side of the vehicle suddenly changes from a maximum value to a minimum value, it is identified as a left lane change, and the constraint conditions are as follows: DTLC(T0-1)-DTLC(T1+1)>X>0

[0096] Through in-depth data analysis, the start and end time of each free lane change scenario are accurately defined, and the constraints are as follows:

[0097]

[0098] Among them: EV min is the minimum lateral speed threshold at the end of free lane change, EV max V is the maximum lateral speed threshold at the end of free lane change. y (t1) is the lateral speed of the vehicle at the start of lane change, V y (t2) is the lateral speed of the vehicle at the end of the lane change. SV min is the minimum lateral speed threshold at the start of free lane change, SV max It is the maximum lateral speed threshold at the start of free lane change.

[0099] According to the lane changing start time t1 and the lane changing end time t2, the data scene segment containing the complete free lane changing behavior is intercepted and saved.

[0100] Furthermore, in step S3, the cluster analysis method includes free lane change feature parameter extraction, feature parameter dimension reduction, and driving behavior clustering, which are specifically as follows:

[0101] Feature parameter extraction: the vehicle's average lateral speed, maximum speed, minimum speed, average lateral acceleration, maximum acceleration, and minimum acceleration; the vehicle's average longitudinal speed, maximum speed, minimum speed, average longitudinal acceleration, maximum acceleration, and minimum acceleration as well as the longitudinal distance, lateral distance, and duration of the lane changing process are taken as feature parameters.

[0102] Feature parameter dimensionality reduction: The principal component analysis (PCA) method is used to implement feature dimensionality reduction. The 15-dimensional parameters that characterize the free lane change feature are reduced to h dimensions, and h orthogonal vectors of unit length are selected to ensure that there is no correlation between the new features obtained, that is, their covariance matrix is ​​zero. It is known that the covariance matrix B is expressed as:

[0103]

[0104] Where: xij* is the normalized value of the free lane-changing behavior feature parameter; m and n are the dimensions and number of free lane-changing behavior related data, respectively. The B matrix should ensure that there is no correlation between the new features obtained, that is, their covariance matrix is ​​zero. Diagonalize the B matrix and arrange its diagonal eigenvalues ​​in descending order. Each eigenvalue is called the principal component of the free lane-changing behavior data.

[0105] The corresponding characteristic parameters of the extracted free lane-changing scenario are input into the model, and the expected value of the contribution rate is defined as k%. By calculating the contribution rate of each component, it can be obtained that the first h principal components accumulate to more than k%, so the first h principal components are finally extracted to reduce the free lane-changing characteristic parameter set from 15 dimensions to h dimensions.

[0106] Driving behavior clustering: Based on the results of dimensionality reduction, the K-means unsupervised clustering method is used to perform cluster analysis on the free lane change segments, and its evaluation criterion function is:

[0107]

[0108] In the formula: K is the number of clusters; x is the data point; Ci is the category; ci is the cluster center of category i; dist(,) is the Euclidean distance. K-means clustering is performed on the free lane change dataset after dimensionality reduction to obtain the clustering results of the free lane change scene. Table 1 shows the clustering results of 1136 scene clips.

[0109]

[0110] The characteristic parameter distribution after clustering is as follows: Figure 7As shown in the figure, according to the maximum speed range of the vehicle, the maximum acceleration range of the vehicle, the lane changing time of the vehicle, and the average lateral speed of the vehicle, the lane changing style can be divided into three categories. Category 0 is the conservative type (maximum speed range of the vehicle: 70-120 km / h, maximum acceleration range of the vehicle: 0-1.7 m / s-2, lane changing time of the vehicle: 4.8-9.1 s, average lateral speed of the vehicle: 0.1-0.78 m / s), and category 1 is the aggressive type (maximum speed range of the vehicle: 70-120 km / h, maximum acceleration range of the vehicle: 0-1.7 m / s-2, lane changing time of the vehicle: 4.8-9.1 s, average lateral speed of the vehicle: 0.1-0.78 m / s). Category 2 is conservative (maximum speed range of the vehicle: 50-90 km / h, maximum acceleration range of the vehicle: -0.5-1.5 m / s-2, lane change time of the vehicle: 2.5-7 s, average lateral speed of the vehicle: 0.05-0.5 m / s).

[0111] In step S4, the vehicle trajectory generation method includes the Frenet coordinate system definition, the main vehicle trajectory generation method and the trajectory generation method of other traffic participants, which are specifically as follows:

[0112] Step S4.1: The Frenet coordinate system definition method is as follows:

[0113] The Frenet coordinate system is based on the centerline of the road and decomposes the vehicle motion into lateral and longitudinal offsets relative to the centerline. It is easier to describe the vehicle trajectory in the simulation environment, as shown below: Figure 8 As shown in the figure, the Frenet coordinate system is constructed with the road centerline as the longitudinal axis s and the direction perpendicular to the road centerline as the horizontal axis d. The position of the vehicle or obstacle on the road can be expressed as (s, d), where s is described as the longitudinal offset on the road and d is described as the lateral displacement on the road.

[0114] Step S4.2: The method for generating the main vehicle trajectory is as follows:

[0115] The main vehicle trajectory includes the main vehicle longitudinal displacement and the main vehicle lateral displacement. First, according to the speed Stlc from the vehicle to the lane line and the vehicle speed Ego_v, the longitudinal speed Ego_v_x of the main vehicle along the lane line is obtained. Secondly, according to the longitudinal speed Ego_v_x of the main vehicle along the lane line, the longitudinal displacement Ego_d_x of the main vehicle along the lane line is calculated. The calculation method is as follows:

[0116]

[0117] The lateral displacement of the main vehicle is obtained according to the characteristic curve of the distance Dtlc_1 from the vehicle to the lane line obtained by the functional camera.

[0118] The calculation method of the lateral displacement Ego_d_y of the main vehicle is as follows:

[0119]

[0120] Where: N is the data length, Dn is the threshold for judging the lateral step, Q is the lateral displacement step difference data set, ID is the ID set of the lateral displacement step moment, and M is the number of elements in the Q data set.

[0121] Step S4.3: The target vehicle trajectory generation method is as follows:

[0122] The target vehicle trajectory includes the target vehicle longitudinal displacement Obj_d_x and the target vehicle lateral displacement Obj_d_y. First, according to the relative distance Range of the main vehicle and the relative angle angle relative to the main vehicle, the target vehicle relative longitudinal distance Obj_dr_x and the target vehicle relative lateral distance Obj_dr_y are obtained. Then, the target vehicle trajectory is obtained according to the main vehicle trajectory, as follows:

[0123]

[0124] Further, in step S5,

[0125] The Modeldesk-based scene generation method is as follows:

[0126] Based on the three types of lane-changing scenario data obtained by clustering driving behaviors, a simulation scenario of lane-changing driving behaviors is built based on the Modeldesk software, including:

[0127] Road construction: set road geometry, road surface, lane attributes, road markings and other elements according to actual conditions;

[0128] Construction of road facilities and restrictions: Setting road facilities and restriction attributes, including setting road boundaries and traffic sign information;

[0129] Construction of temporary physical restrictions: Setting the properties of temporary objects and facilities on the road surface and above the road section according to actual conditions;

[0130] Meteorological conditions settings: set the lighting conditions, weather conditions and other meteorological conditions attributes;

[0131] Construction of traffic participants: according to the vehicle attributes of the main vehicle, set the main vehicle type, size and other parameters;

[0132] The driving trajectory of the main vehicle is set according to the above-mentioned main vehicle trajectory construction method. Similarly, the parameters of other traffic participants are set according to their types, quantities, characteristics and other attributes. The driving trajectories of other traffic participants are set according to the above-mentioned trajectory construction method of other traffic participants. Finally, the generation of different types of free lane change scene files is completed.

[0133] Fig. 9 Three types of lane-changing simulation scenarios are constructed: Type 0, the road type is highway, 3 lanes; the road boundary is a highway guardrail; the weather is sunny; the maximum speed of the vehicle is 101km / h, and the maximum acceleration of the vehicle is 0.6m / s -2 , the lane change time of the vehicle is 5.3s, and the average lateral speed of the vehicle is 0.58m / s. Type 1, the lane type is high-speed, 3 lanes; the road boundary is a high-speed railing; the weather is sunny; the maximum speed of the vehicle is 135km / h, and the maximum acceleration of the vehicle is 1.5m / s -2 , the lane change time of the vehicle is 3.8s, and the average lateral speed of the vehicle is 0.98m / s. Type 2, the lane type is high-speed, 3 lanes; the road boundary is a high-speed railing; the weather is sunny; the maximum speed of the vehicle is 87km / h, and the maximum acceleration of the vehicle is 0.1m / s -2 The lane changing time of the vehicle is 6.7s, and the average lateral speed of the vehicle is 0.48m / s.

[0134] Those of ordinary skill in the art will appreciate that the units and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0135] In the several embodiments provided in the present application, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the division of the units described above is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. The above-mentioned units may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.

[0136] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

[0137] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A stylized lane-changing scene reconstruction method based on natural driving data, characterized in that: The steps include: Step S1: Collect natural driving data to obtain the status information of the main vehicle and other traffic participants; Step S2: extracting free lane change scene segments from natural driving data according to a free lane change scene extraction algorithm; Step S3: performing cluster analysis on the extracted free lane change scene segments to obtain different types of free lane change segment data sets; Step S4: generating different types of free lane change segment trajectory files from the acquired different types of lane change scene segment data according to the vehicle trajectory generation method; Step S5: generating different types of free lane change simulation scenario files based on the scenario generation method of Modeldesk according to different types of free lane change segment trajectory files.

2. The stylized lane-changing scene reconstruction method based on natural driving data according to claim 1, characterized in that: In step S1: The main vehicle status information includes: time series, main vehicle speed, steering wheel angle, distance between the main vehicle and lane lines on both sides, lateral speed of the main vehicle to lane lines on both sides, confidence of lane lines on both sides of the main vehicle, and width of the main vehicle; The other traffic participant status information includes: the distance between the host vehicle and other traffic participants, the angle between the host vehicle and other traffic participants, and the type of traffic participants.

3. The stylized lane-changing scene reconstruction method based on natural driving data according to claim 1, characterized in that: In step S2, the free lane change scene extraction algorithm is: Compare the left lane line confidence and the right lane line confidence. If the left lane line confidence is higher than the right lane line confidence, the left lane line distance is prioritized for identification. If the right lane line confidence is higher than the left lane line confidence, the right lane line distance is prioritized for identification. Furthermore, in step S2, the free lane change scene extraction algorithm is: If the left lane line confidence is higher than the right lane line confidence: When the distance between the main vehicle and the left lane line suddenly changes from a minimum value to a maximum value, it is recognized as a left lane change; When the distance between the main vehicle and the left lane line suddenly changes from a maximum value to a minimum value, it is recognized as a lane change to the right; If the confidence level of the right lane line is higher than that of the left lane line: When the lane line to the right of the main vehicle suddenly changes from a minimum to a maximum, it is recognized as a lane change to the right; When the lane line to the right of the main vehicle suddenly changes from a maximum value to a minimum value, it is recognized as a left lane change; Furthermore, in step S2, the free lane change scene extraction algorithm further includes: defining the start and end time of each free lane change scene, and based on the lane change start time and the lane change end time, intercepting and saving a data scene segment containing a complete free lane change behavior.

4. The method for reconstructing a stylized lane-changing scene based on natural driving data according to claim 1, characterized in that: In step S3, the cluster analysis method includes: free lane change feature parameter extraction, feature parameter dimension reduction, and driving behavior clustering; Furthermore, in step S3, the cluster analysis method includes: Feature parameter extraction: The vehicle's average lateral speed, maximum speed, minimum speed, average lateral acceleration, maximum acceleration, and minimum acceleration; the vehicle's average longitudinal speed, maximum speed, minimum speed, average longitudinal acceleration, maximum acceleration, and minimum acceleration, as well as the longitudinal distance, lateral distance, and duration of the lane-changing process are taken as feature parameters; Feature parameter dimension reduction: principal component analysis method is used to implement feature dimension reduction; Driving behavior clustering: Based on the results of dimensionality reduction, the K-means unsupervised clustering method is used to perform cluster analysis on the free lane change segments.

5. The stylized lane-changing scene reconstruction method based on natural driving data according to claim 1, characterized in that: In step S4, the vehicle trajectory generation method includes the Frenet coordinate system definition, the main vehicle trajectory generation method and the trajectory generation method of other traffic participants; Furthermore, in step S4, the vehicle trajectory generation method includes: Step S4.1: Frenet coordinate system definition: The Frenet coordinate system is based on the centerline of the road and decomposes the vehicle motion into lateral offset and longitudinal offset relative to the centerline. It is easier to describe the trajectory of the vehicle in a simulation environment. The centerline direction of the road is the longitudinal axis s, and the direction perpendicular to the centerline of the road is the transverse axis d. The Frenet coordinate system is constructed, and the position of the vehicle or obstacle on the road can be expressed as (s, d), where s is described as the longitudinal offset on the road and d is described as the lateral displacement on the road. Step S4.2: Main vehicle trajectory generation: The main vehicle trajectory includes the main vehicle longitudinal displacement and the main vehicle lateral displacement; First, according to the speed Stlc from the vehicle to the lane line and the speed Ego_v of the vehicle, the longitudinal speed Ego_v_x of the vehicle along the lane line is obtained; Secondly, according to the longitudinal velocity Ego_v_x of the main vehicle along the lane line, the longitudinal displacement Ego_d_x of the main vehicle along the lane line is calculated as follows: The lateral displacement of the main vehicle is obtained according to the characteristic curve of the distance Dtlc_1 from the vehicle to the lane line acquired by the functional camera; The calculation method of the lateral displacement Ego_d_y of the main vehicle is as follows: Where: N is the data length, Dn is the threshold for judging the lateral step, Q is the lateral displacement step difference data set, ID is the id set of the lateral displacement step moment, and M is the number of elements in the Q data set; Step S4.3: Target vehicle trajectory generation: The target vehicle trajectory includes the target vehicle longitudinal displacement Obj_d_x and the target vehicle lateral displacement Obj_d_y; First, according to the relative distance Range of the main vehicle and the relative angle angle relative to the main vehicle, the relative longitudinal distance Obj_dr_x and the relative lateral distance Obj_dr_y of the target vehicle are obtained; Secondly, the target vehicle trajectory is obtained according to the main vehicle trajectory, as follows:

6. The stylized lane-changing scene reconstruction method based on natural driving data according to claim 1, characterized in that: In step S5, the scenario generation method based on Modeldesk includes: constructing a simulation scenario of lane-changing driving behavior based on Modeldesk software according to three types of lane-changing scenario data obtained by clustering driving behaviors; Furthermore, in step S5, the Modeldesk-based scene generation method includes: Road construction: set road geometry, road surface, lane attributes, road markings and other elements according to actual conditions; Construction of road facilities and restrictions: Setting road facilities and restriction attributes, including setting road boundaries and traffic sign information; Construction of temporary physical restrictions: Setting the properties of temporary objects and facilities on the road surface and above the road section according to actual conditions; Meteorological conditions settings: set the lighting conditions, weather conditions and other meteorological conditions attributes; Construction of traffic participants: according to the vehicle attributes of the main vehicle, set the main vehicle type, size and other parameters; The driving trajectory of the main vehicle is set according to the above-mentioned main vehicle trajectory construction method. Similarly, the parameters of other traffic participants are set according to their types, quantities, characteristics and other attributes. The driving trajectories of other traffic participants are set according to the above-mentioned trajectory construction method of other traffic participants. Finally, the generation of different types of free lane change scene files is completed.

7. An electronic device, comprising a processor and a memory connected to the processor for storing instructions executable by the processor, characterized in that: The processor is used to execute the stylized lane change scene reconstruction method based on natural driving data as described in any one of claims 1-6 above.

8. A server, characterized in that: The invention comprises at least one processor and a memory in communication with the processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the processor so that the at least one processor executes the stylized lane change scene reconstruction method based on natural driving data as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the stylized lane change scene reconstruction method based on natural driving data described in any one of claims 1-6 is implemented.

Images