Traffic Flow Machine Learning Modeling System and Method Applied to Vehicles

Abstract

The present invention relates to a traffic flow machine learning modeling system and method applied to vehicles. The system includes: a target fusion module that performs target fusion on radar measurement data and camera measurement data and outputs target attribute information; a lane line model module that outputs an original lane line model according to the camera measurement data; a target selection module that determines the positions of the respective targets in the lanes based on the target attribute information output by the target fusion module and the original lane line model output by the lane line model module and outputs the targets; and a traffic flow calculation module that models the vehicle positions using a clustering algorithm based on the output of the target fusion module, the output of the target selection module, and the output of the lane line model module, and outputs a lane line model based on traffic flow. According to the present invention, accurate lane line parameters can be obtained and relevant lane flow feature information can be provided.

Description

Technical Field

[0001] The present invention relates to autonomous driving technology, and particularly to a traffic flow machine learning modeling system and a traffic flow machine learning modeling method applied to vehicles. Background Art

[0002] Traffic flow refers to the behavior of vehicles running within a transportation network. Current traffic flow research mainly focuses on the mathematical statistics of traffic states, such as studying the probability distributions of speed, traffic volume, traffic density, queue length, and waiting time, etc. The state of traffic flow affects every vehicle on the road.

[0003] Autonomous vehicles are the future development trend, and more and more vehicles will be equipped with advanced driver assistance systems (ADAS) and autonomous driving functions in the future. With the development of autonomous vehicles, if more accurate traffic flow information can be provided, the accuracy of autonomous driving technology can be further improved. For example, abnormal situations in traffic flow information can be identified earlier, and certain control functions can be performed on the vehicle or warning messages can be sent to the driver. Summary of the Invention

[0004] In view of the above problems, the present invention aims to provide a traffic flow machine learning modeling system and a traffic flow machine learning modeling method applied to vehicles that can provide more accurate traffic flow information.

[0005] A traffic flow machine learning modeling system applied to vehicles according to one aspect of the present invention is characterized by including:

[0006] A target fusion module for performing target fusion on radar measurement data from a vehicle radar and camera measurement data from a vehicle camera, and outputting target attribute information;

[0007] A lane line model module for outputting an original lane line model according to camera measurement data from a vehicle camera;

[0008] A target selection module for determining the positions of each target in the lanes according to the target attribute information output by the target fusion module and the original lane line model output by the lane line model module, and outputting targets; and

[0009] A traffic flow calculation module for modeling the vehicle positions by using a clustering algorithm according to the output of the target fusion module, the output of the target selection module, and the output of the lane line model module, and outputting a lane line model based on traffic flow.

[0010] Optionally, in the traffic flow calculation module, the DBSCAN algorithm is used as the clustering algorithm.

[0011] Optionally, the target fusion module outputs the distance between vehicles as the target attribute information, and the traffic flow calculation module uses the distance between vehicles output by the target fusion module as the basis for clustering division, and outputs the traffic flow shape based on vehicle density as the clustering result, and generates a lane line model according to the traffic flow shape.

[0012] Optionally, the traffic flow calculation module further outputs traffic flow feature information.

[0013] Optionally, the traffic flow feature information includes one or more of the following:

[0014] The average speed of vehicles in the traffic flow, the maximum speed of vehicles in the traffic flow, and the minimum speed of vehicles in the traffic flow.

[0015] Optionally, the DBSCAN algorithm includes the following steps:

[0016] Based on the input distance between vehicles, abstract the vehicles into a point, cluster the vehicles, and output clusters;

[0017] Use the principal component analysis algorithm to find the principal direction of the cluster;

[0018] For each cluster, use the least squares method to find the quartic regression equation, where the coefficients of the quartic regression equation are the lane line parameters.

[0019] Optionally, the clustering of vehicles includes:

[0020] (1) Arbitrarily select a data object point p from the dataset;

[0021] (2) If for the parameters Eps and MinPts, the selected data object point p is a core point, then find all data object points that are density-reachable from p and form a cluster;

[0022] (3) If the selected data object point p is a border point, select another data object point; and

[0023] (4) Repeat steps (2) and (3) until all points are processed,

[0024] where the parameter Eps represents the neighborhood radius and MinPts represents the threshold of the number of data objects in the neighborhood.

[0025] A traffic flow machine learning modeling method applied to vehicles according to one aspect of the present invention, characterized by including:

[0026] A target fusion step, which performs target fusion on radar measurement data from a vehicle radar and camera measurement data from a vehicle camera, and outputs target attribute information;

[0027] Lane line model step, which outputs an original lane line model according to the camera measurement data from a vehicle camera;

[0028] Target selection step, which is used to determine the positions of the lanes where each target is located and output the targets according to the target attribute information output by the target fusion step and the original lane line model output by the lane line model step;

[0029] Traffic flow calculation step, which models the vehicle positions using a clustering algorithm according to the output of the target fusion step, the output of the target selection step, and the output of the lane line model step, and outputs a lane line model based on traffic flow.

[0030] Optionally, in the traffic flow calculation step, as the clustering algorithm, the DBSCAN algorithm is adopted.

[0031] Optionally, in the target fusion step, the distance between vehicles is output as the target attribute information,

[0032] In the traffic flow calculation step, the distance between vehicles output by the target fusion step is used as the basis for clustering division, and a vehicle density-based traffic flow shape as the clustering result is output, and a lane line model is generated according to the traffic flow shape.

[0033] Optionally, the traffic flow calculation step further outputs traffic flow feature information.

[0034] Optionally, the traffic flow feature information includes one or more of the following:

[0035] The average speed of vehicles in the traffic flow, the maximum speed of vehicles in the traffic flow, the minimum speed of vehicles in the traffic flow.

[0036] Optionally, the DBSCAN algorithm includes the following steps:

[0037] Based on the input distance between vehicles, the vehicles are abstracted into a point, the vehicles are clustered, and clusters are output;

[0038] The principal component analysis algorithm is used to find the principal direction of the cluster;

[0039] For each cluster, the least squares method is used to find a quartic regression equation, where the coefficients of the quartic regression equation are the lane line parameters.

[0040] Optionally, the clustering of the vehicles includes:

[0041] (1) Arbitrarily select a data object point p from the dataset;

[0042] (2) If, for parameters Eps and MinPts, the selected data object point p is a core point, then find all data object points that are density-reachable from p to form a cluster;

[0043] (3) If the selected data object point p is a border point, select another data object point; and

[0044] (4) Repeat steps (2) and (3) until all points are processed,

[0045] where the parameter Eps represents the neighborhood radius and MinPts represents the threshold of the number of data objects in the neighborhood.

[0046] A vehicle according to one aspect of the present invention is characterized in that it includes the traffic flow machine learning modeling system applied to the vehicle as described above.

[0047] A computer-readable medium according to one aspect of the present invention, on which a computer program is stored, is characterized in that when the computer program is executed by a processor, it implements the traffic flow machine learning modeling method applied to the vehicle as described above.

[0048] A computer device according to one aspect of the present invention includes a storage module, a processor, and a computer program stored on the storage module and executable on the processor, and is characterized in that when the processor executes the computer program, it implements the traffic flow machine learning modeling method applied to the vehicle as described above.

[0049] As described above, according to the traffic flow machine learning modeling system and the traffic flow machine learning modeling method of the present invention, by using the DBSCAN clustering algorithm, accurate lane line parameters can be obtained, especially in scenarios where the lane line model module cannot give an accurate lane line model.

[0050] Further, according to the traffic flow machine learning modeling system and the traffic flow machine learning modeling method of the present invention, traffic flow characteristics can also be provided, for example, the average speed of different lanes, and the slow and sudden deceleration behaviors of vehicles in the traffic flow can be monitored. The results can be used as a basis for path planning and warning information can be sent to remind the driver, or control operations can be performed on the vehicle itself, such as decelerating and changing lanes. Thus, more auxiliary assistance can be provided for autonomous driving technology to achieve precise control of the vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a schematic structural diagram of a traffic flow machine learning modeling system applied to a vehicle according to an embodiment of the present invention.

[0052] Figure 2 is a schematic diagram of the clustering algorithm adopted in the present invention.

[0053] Figure 3 It is a schematic flowchart of a traffic flow machine learning modeling method applied to vehicles according to an embodiment of the present invention. Detailed implementation manners

[0054] The following describes some of the multiple embodiments of the present invention, aiming to provide a basic understanding of the present invention. It is not intended to identify the key or decisive elements of the present invention or to limit the scope to be protected.

[0055] For the sake of simplicity and illustrative purposes, the principles of the present invention are mainly described herein with reference to its exemplary embodiments. However, those skilled in the art will readily recognize that the same principles can be equivalently applied to all types of traffic flow machine learning modeling systems and traffic flow machine learning modeling methods applied to vehicles, and these same principles can be implemented therein, and any such variations do not depart from the true spirit and scope of this patent application.

[0056] Moreover, in the following description, reference is made to the accompanying drawings, which illustrate specific exemplary embodiments. Electrical, mechanical, logical, and structural changes can be made to these embodiments without departing from the spirit and scope of the present invention. In addition, although a feature of the present invention is disclosed in conjunction with only one of several embodiments / embodiments, this feature can be combined with one or more other features of other embodiments / embodiments as may be desired and / or advantageous for any given or identifiable function. Therefore, the following description should not be construed in a limiting sense, and the scope of the present invention is defined by the appended claims and their equivalents.

[0057] Terms such as "comprising" and "including" indicate that in addition to having the units (modules) and steps directly and clearly stated in the specification and claims, the technical solutions of the present invention do not exclude the situation of having other units (modules) and steps not directly or clearly stated.

[0058] Figure 1 It is a schematic structural diagram of a traffic flow machine learning modeling system applied to vehicles according to an embodiment of the present invention.

[0059] As Figure 1 shown, a traffic flow machine learning modeling system applied to vehicles according to an embodiment of the present invention includes:

[0060] A target fusion module 100, which receives radar measurement data from a vehicle radar and camera measurement data from a vehicle camera, performs target fusion, and outputs target attribute information, for example, including: information related to a target and a lane line model, etc.;

[0061] The lane line model module 200 outputs an original lane line model based on the camera measurement data from the vehicle camera. Specifically, it converts the camera measurement data collected by the camera into lane line parameters, such as the distances of the vehicle from the left and right lane lines, the curvature of the lane, etc.

[0062] The target selection module 300 determines the positions of the targets in each lane and outputs the targets based on the target attribute information output by the target fusion module 100 and the original lane line model output by the lane line model module 200.

[0063] The traffic flow calculation module 400 models the vehicle positions based on the outputs of the target fusion module 100, the target selection module 200, and the lane line model module 300 using a clustering algorithm, and outputs a lane line model based on traffic flow.

[0064] Here, the target fusion module 100 is used to fuse the target information of multiple sensors to generate more accurate target attributes than a single sensor. The present invention does not involve the target fusion logic. As long as there is accurate target attribute information, it can also be the output of a single sensor or the output of a fusion algorithm.

[0065] The lane line model module 200 determines the positions of the targets in each lane based on the targets and the lane line model output by the target fusion module 100, and outputs the targets in a specified order. In the present invention, the specified order is not limited and can be set as needed as long as the lane where the target is located and its front and rear positions can be determined. As an example, the specified order is: (1) the nearest target in the current lane; (2) the second nearest target in the current lane; (3) the nearest target in the left lane; (4) the second nearest target in the left lane; (5) the nearest target in the right lane; (6) the second nearest target in the right lane.

[0066] The main function of the lane line model module 200 is to convert the lane image seen by the camera into lane line parameters, such as the distances of the vehicle from the left and right lane lines, the curvature of the lane, etc., thereby obtaining the original lane line model.

[0067] The traffic flow calculation module 400 models the vehicle positions using a clustering algorithm based on the outputs of the target fusion module 100, the target selection module 200, and the lane line model module 300, and outputs a lane line model based on traffic flow, and can also calculate traffic flow characteristics. Here, as the clustering algorithm applied in the traffic flow calculation module 400, the DBSCAN clustering algorithm is preferably used because the DBSCAN algorithm does not require specifying the number of clusters, while other clustering algorithms (such as K-means, K-median) require specifying the number of clusters, and the number of clusters cannot be predicted in advance in the actual scenario, so the DBSCAN algorithm is more suitable.

[0068] Next, the specific calculation process using the DBSCAN algorithm will be described.

[0069] DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a relatively effective density-based clustering algorithm. It defines a cluster as the largest set of density-connected points, can divide regions with sufficient high density into clusters, and can discover clusters of arbitrary shapes in a spatial database with noise.

[0070] DBSCAN requires two parameters: eps and the minimum number of points (minPts) required to form a high-density region. It starts with an arbitrarily unvisited point and then explores the eps neighborhood of this point. If there are enough points in the eps neighborhood, a new cluster is established; otherwise, this point is labeled as noise. If a point is in the dense region of a cluster, the points in its eps neighborhood also belong to this cluster. When these new points are added to the cluster, if they are also in the dense region, the points in their eps neighborhoods will also be added to the cluster. This process will be repeated until no more points can be added, and thus a density-connected cluster is completely found. Then, an unvisited point will be explored to discover a new cluster or noise. The advantages of DBSCAN are that it can discover clusters of arbitrary shapes, does not require manual specification of the number of clusters, and has a mechanism to handle outliers.

[0071] In the present invention, the distance between vehicles (the distance between vehicles is provided by the target fusion module 100) is used as the basis for clustering division. Finally, a density-based traffic flow shape is obtained. Based on the traffic flow shape, the lane shape can be obtained, and at the same time, information such as the average speed of the traffic flow and the maximum / minimum speed of the vehicles in the traffic flow can also be calculated.

[0072] Figure 2 It is a schematic diagram showing the clustering algorithm adopted in the present invention.

[0073] As Figure 2 shown, the horizontal axis represents the longitudinal distance, and the vertical axis represents the lateral distance. Figure 2 The numbered squares in it represent the vehicles ahead detected by the camera or radar. According to the distance between vehicles, three clustering results can be obtained using the DBSCAN algorithm: Vehicle 1 and Vehicle 2 are clustering result 1; Vehicle 3 and Vehicle 4 are clustering result 2; Vehicle 7, Vehicle 8, and Vehicle 9 are clustering result 3. In addition, Vehicle 5 and Vehicle 6 are outliers.

[0074] By continuously collecting data and based on the clustering results, the lane line parameters can be inferred. Meanwhile, it can be compared with the output of the lane line model module 200 to improve the accuracy of the lane line model, and it can also provide some status information that the lane line model module 200 cannot provide. For example, as an example, it provides information on whether the lane is passable. Specifically, if there is a traffic flow cluster on a certain lane, it means that the lane is passable. This is just an example, and other status information can also be provided according to specific situations.

[0075] Next, the specific content of using the DBSCAN algorithm in the present invention will be described.

[0076] The input of the DBSCAN algorithm is the distance between vehicles (the distance between vehicles is provided by the target fusion module 100). Here, we can abstract the vehicle into a point, and the output is a cluster. The calculation process is as follows:

[0077] (1) Arbitrarily select a data object point p from the dataset;

[0078] (2) If for the parameters Eps and MinPts, the selected data object point p is a core point, then find all the data object points that are density-reachable from p to form a cluster;

[0079] (3) If the selected data object point p is a border point, select another data object point;

[0080] (4) Repeat (2) and (3) until all points are processed, where the parameter Eps represents the neighborhood radius and MinPts represents the threshold of the number of data objects in the neighborhood.

[0081] After clustering the vehicles, use the principal component analysis (PCA) algorithm to find the principal direction of the cluster, and use the least squares method (LR) for each cluster to obtain a quartic regression equation. The coefficients of the quartic regression equation are the lane line parameters.

[0082] The traffic flow machine learning modeling system according to the present invention can obtain accurate lane line parameters by using the DBSCAN clustering algorithm. The present invention is particularly suitable for scenarios where the lane line model module cannot give an accurate lane line model under abnormal lane line conditions, such as: there are no lane lines on the road; the lane lines have been redrawn many times due to road construction, leaving multiple old lane lines on the road; there are Chinese or English characters on the road, interfering with lane line recognition; due to weather reasons such as rain, snow, fog, strong wind and sand, and night conditions, the camera cannot recognize the lane lines; or the camera cannot recognize the lane lines for other reasons, such as the camera being blocked, or the image system hardware or software failure resulting in the inability to output lane lines, etc. In several similar situations like these, the traffic flow machine learning modeling system according to the present invention can obtain accurate lane line parameters.

[0083] Furthermore, the traffic flow machine learning modeling system according to the present invention can further provide traffic flow characteristic information, such as the average speed of traffic flow vehicles, the maximum speed of vehicles in the traffic flow, the minimum speed of vehicles in the traffic flow, etc. By using the DBSCAN clustering algorithm, calculate the average speed of the traffic flow clusters of each lane. If the average vehicle speed is lower than a certain threshold, it can be determined that there is a slow behavior in this lane. Similarly, the average speed at different times can also be calculated and compared. If the decrease in the average speed exceeds a certain threshold, it can be determined that there is a sharp deceleration behavior in a certain lane. In this way, it is possible to monitor the slow or accelerating behavior of vehicles based on the traffic flow characteristic information and output warning information.

[0084] The above has described the traffic flow machine learning modeling system applied to vehicles according to an embodiment of the present invention. Next, the traffic flow machine learning modeling method applied to vehicles according to an embodiment of the present invention is described.

[0085] Figure 3 It is a flowchart showing the traffic flow machine learning modeling method applied to vehicles according to an embodiment of the present invention.

[0086] As Figure 3 shown, the traffic flow machine learning modeling method applied to vehicles according to an embodiment of the present invention includes:

[0087] Target fusion step S100: Perform target fusion on the radar measurement data from the vehicle radar and the camera measurement data from the vehicle camera, and output target attribute information;

[0088] Lane line acquisition step S200: Output an original lane line model according to the camera measurement data from the vehicle camera;

[0089] Target selection step S300: Determine the positions of the lanes where each target is located based on the target attribute information output by the target fusion step and the original lane line model output by the lane line acquisition step, and output the targets.

[0090] Traffic flow calculation step S400: Based on the output of the target fusion step S100, the output of the lane line acquisition step S200, and the output of the target selection step S300, use a clustering algorithm to model the vehicle positions and output a lane line model based on the traffic flow.

[0091] Among them, in the traffic flow calculation step S400, as the clustering algorithm, the DBSCAN algorithm is adopted. In the target fusion step S100, the distance between vehicles is output as the target attribute information. In this way, in the traffic flow calculation step S100, the distance between vehicles output by the target fusion step S100 is used as the basis for clustering division, and the vehicle density-based traffic flow shape as the clustering result is output, and the lane line model is generated according to the traffic flow shape.

[0092] Furthermore, in the traffic flow calculation step, traffic flow feature information can be further output. The traffic flow feature information includes one or more of the following: the average speed of vehicles in the traffic flow, the maximum speed of vehicles in the traffic flow, the minimum speed of vehicles in the traffic flow, etc.

[0093] Among them, the DBSCAN algorithm adopted in the traffic flow calculation step S400 mainly includes the following steps:

[0094] Based on the input distance between vehicles, abstract the vehicles into points, cluster the vehicles, and output clusters;

[0095] Use the principal component analysis algorithm to find the principal direction of the cluster; and

[0096] For each cluster, use the least squares method to find the quartic regression equation, where the coefficients of the quartic regression equation are the lane line parameters.

[0097] Here, the "clustering of vehicles" includes:

[0098] (1) Arbitrarily select a data object point p from the dataset;

[0099] (2) If for the parameters Eps and MinPts, the selected data object point p is a core point, then find all the data object points that are density-reachable from p and form a cluster;

[0100] (3) If the selected data object point p is a border point, select another data object point; and

[0101] Repeat steps (2) and (3) until all points are processed.

[0102] Among them, the parameter Eps represents the neighborhood radius, and MinPts represents the threshold of the number of data objects in the neighborhood.

[0103] As described above, according to the traffic flow machine learning modeling system and the traffic flow machine learning modeling method of the present invention, by using the DBSCAN clustering algorithm, accurate lane line parameters can be obtained, and accurate lane line parameters can be given in scenarios where the lane line model module cannot give an accurate lane line model. Further, according to the traffic flow machine learning modeling system and the traffic flow machine learning modeling method of the present invention, traffic flow characteristics can also be provided, for example, the average speed of different lanes, and the slow and sudden deceleration behaviors of vehicles in the traffic flow can be monitored. The results can be used as a basis for path planning, and warning information can be sent to remind the driver, or control operations can be performed on the vehicle itself, such as decelerating and changing lanes. Therefore, it can provide more auxiliary help for autonomous driving technology and achieve precise control of the vehicle.

[0104] Furthermore, the present invention also provides a vehicle, which includes the above-mentioned traffic flow machine learning modeling system applied to the vehicle.

[0105] The present invention also provides a computer-readable medium, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned traffic flow machine learning modeling method applied to the vehicle is implemented.

[0106] The present invention also provides a computer device, which includes a storage module, a processor, and a computer program stored on the storage module and executable on the processor. When the processor executes the computer program, the above-mentioned traffic flow machine learning modeling method applied to the vehicle is implemented.

[0107] The above examples mainly illustrate the traffic flow machine learning modeling system and the traffic flow machine learning modeling method of the present invention applied to vehicles. Although only some specific embodiments of the present invention are described, those of ordinary skill in the art should understand that the present invention can be implemented in many other forms without departing from its gist and scope. Therefore, the examples and embodiments shown are regarded as illustrative rather than restrictive, and the present invention may cover various modifications and substitutions without departing from the spirit and scope of the present invention as defined by the appended claims.

Claims

1. A traffic flow machine learning modeling system applied to vehicles, characterized in that, Comprising: A target fusion module, configured to perform target fusion on radar measurement data from a vehicle radar and camera measurement data from a vehicle camera, and output target attribute information; A lane line model module, configured to output an original lane line model according to camera measurement data from a vehicle camera; A target selection module, configured to determine the positions of each target in the lanes according to the target attribute information output by the target fusion module and the original lane line model output by the lane line model module, and output targets; And A traffic flow calculation module, configured to perform modeling on vehicle positions by using a clustering algorithm according to the output of the target fusion module, the output of the target selection module, and the output of the lane line model module, and output a lane line model based on traffic flow. Wherein, in the traffic flow calculation module, the DBSCAN algorithm is used as the clustering algorithm. The DBSCAN algorithm comprises the following steps: Based on the distance between input vehicles, abstract the vehicles into points, cluster the vehicles, and output clusters; Use the principal component analysis algorithm to find the principal direction of the clusters; For each cluster, use the least squares method to obtain a quartic regression equation, where the coefficients of the quartic regression equation are the lane line parameters.

2. The traffic flow machine learning modeling system applied to vehicles according to claim 1, characterized in that, In the target fusion module, the distance between vehicles is output as the target attribute information. In the traffic flow calculation module, the distance between vehicles output by the target fusion module is used as the basis for clustering division, and a traffic flow shape based on vehicle density as the clustering result is output, and a lane line model is generated according to the traffic flow shape.

3. The traffic flow machine learning modeling system applied to vehicles according to claim 1, characterized in that, The traffic flow calculation module further outputs traffic flow feature information.

4. The traffic flow machine learning modeling system applied to vehicles according to claim 3, characterized in that, The traffic flow feature information includes one or more of the following: The average speed of vehicles in the traffic flow, the maximum speed of vehicles in the traffic flow, and the minimum speed of vehicles in the traffic flow.

5. The traffic flow machine learning modeling system applied to vehicles according to claim 1, characterized in that, The clustering of the vehicles includes: (1) Arbitrarily select a data object point p from the dataset; (2) If, for parameters Eps and MinPts, the selected data object point p is a core point, find all data object points that are density-reachable from p, and form a cluster; (3) If the selected data object point p is a border point, select another data object point; and (4) Repeat steps (2) and (3) until all points are processed. Wherein, the parameter Eps represents the neighborhood radius, and the MinPts represents the threshold of the number of data objects in the neighborhood.

6. A traffic flow machine learning modeling method applied to vehicles, characterized in that, Comprising: A target fusion step, performing target fusion on radar measurement data from a vehicle radar and camera measurement data from a vehicle camera, and outputting target attribute information; A lane line model step, outputting an original lane line model according to camera measurement data from a vehicle camera; A target selection step, configured to determine the positions of each target in the lanes according to the target attribute information output by the target fusion step and the original lane line model output by the lane line model step, and output targets; Traffic flow calculation steps: Based on the output of the target fusion step, the output of the target selection step, and the output of the lane line model step, use a clustering algorithm to model the vehicle positions and output a lane line model based on traffic flow. In the traffic flow calculation step, as the clustering algorithm, the DBSCAN algorithm is adopted. The DBSCAN algorithm includes the following steps: Based on the distances between the input vehicles, abstract the vehicles into points, cluster the vehicles, and output clusters. Use the principal component analysis algorithm to find the principal direction of the clusters. For each cluster, use the least squares method to obtain a quartic regression equation, where the coefficients of the quartic regression equation are the lane line parameters.

7. The traffic flow machine learning modeling method applied to vehicles according to claim 6, wherein, In the target fusion step, the distances between vehicles are output as the target attribute information. In the traffic flow calculation step, use the distances between vehicles output by the target fusion step as the basis for clustering division, output the traffic flow shape based on vehicle density as the clustering result, and generate a lane line model according to the traffic flow shape.

8. The traffic flow machine learning modeling method applied to vehicles according to claim 6, wherein, The traffic flow calculation step further outputs traffic flow feature information.

9. The traffic flow machine learning modeling method applied to vehicles according to claim 8, wherein, The traffic flow feature information includes one or more of the following: The average speed of vehicles in the traffic flow, the maximum speed of vehicles in the traffic flow, and the minimum speed of vehicles in the traffic flow.

10. The traffic flow machine learning modeling method applied to vehicles according to claim 6, wherein, The clustering of the vehicles includes: (1) Arbitrarily select a data object point p from the dataset. (2) If for the parameters Eps and MinPts, the selected data object point p is a core point, then find all the data object points that are density-reachable from p and form a cluster. (3) If the selected data object point p is a border point, select another data object point; and (4) Repeat steps (2) and (3) until all points are processed. Among them, the parameter Eps represents the neighborhood radius, and MinPts represents the threshold of the number of data objects in the neighborhood.

11. A vehicle, wherein, Including the traffic flow machine learning modeling system for vehicles according to any one of claims 1 to 5.

12. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the traffic flow machine learning modeling method for vehicles according to any one of claims 6 to 10.

13. A computer device, comprising a storage module, a processor, and a computer program stored on the storage module and executable on the processor, wherein, When the processor executes the computer program, it implements the traffic flow machine learning modeling method for vehicles according to any one of claims 6 to 10.

Images