Course angle determination method and device, vehicle controller and vehicle

By obtaining point cloud data of the current frame and historical frame, determining the boundary point cloud data and using the weighted least squares method to correct the heading angle, the problem of poor heading angle accuracy in autonomous driving vehicles is solved, and the heading under low computing resources is achieved Angle correction improves the decision-making accuracy and safety of autonomous vehicles.

CN119078852APending Publication Date: 2024-12-06杭州吉利汽车有限公司 +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411233952.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-04
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The accuracy of the target heading angle determined based on lidar in existing autonomous vehicles is poor, resulting in jitter problems. The algorithms of existing improved methods are complex and difficult to meet the real-time requirements of data.

Method used

By obtaining the point cloud data of the current frame and historical frame, determining the boundary point cloud data, using the weighted least squares method for linear fit, correcting the heading angle, reducing the calculation amount, and improving accuracy and smoothness.

Benefits of technology

In the case of consuming less computing resources, the heading angle is accurately corrected, jitter is reduced, and the decision-making accuracy and driving safety of autonomous vehicles are improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119078852A_ABST
    Figure CN119078852A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of automatic driving, and discloses a course angle determination method and device, a vehicle controller and a vehicle, and the method comprises the steps: obtaining a plurality of pieces of first point cloud data and a first course angle of a target under a current frame; determining first boundary point cloud data based on the plurality of first point cloud data and the first course angle, the first boundary point cloud data being at least one piece of first point cloud data facing the direction of the first course angle and located at the boundary of the target close to the laser radar in the plurality of first point cloud data; acquiring second boundary point cloud data of the target under at least one historical frame; correcting the first course angle based on the first boundary point cloud data and the second boundary point cloud data; and determining the corrected first course angle as the course angle of the target under the current frame. According to the method, the smoothness and authenticity of the course angle of the output target under the current frame can be ensured while relatively low computing power is consumed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a method and device for determining a heading angle, a vehicle controller, and a vehicle. Background Art

[0002] Autonomous driving refers to a vehicle's ability to autonomously perceive its surroundings, make decisions, and execute driving tasks without the direct control of a human driver, using sensors, controllers, actuators, computing platforms, and communication technologies, thereby achieving safe and efficient vehicle operation. For highly automated and fully automated driving vehicles, the accuracy of the heading angle of targets (such as other vehicles) determined by lidar affects the vehicle's decision-making and perception, and thus its safety.

[0003] Currently, object detection models used in autonomous vehicles are divided into traditional algorithms and deep learning algorithms. Both traditional and deep learning algorithms determine the target's heading angle based on point cloud data from a LiDAR sensor. However, due to issues such as limited point cloud resolution, target occlusion, and the laser scanning angle, the heading angles output by traditional and deep learning algorithms can fluctuate to varying degrees, resulting in poor accuracy.

[0004] Some improved methods perform some smoothing based on historical heading angles to achieve the desired heading correction. However, without the aid of the target's actual point cloud information, this can lead to significant deviations between the theoretical heading angle and the target's actual heading angle. Some improved methods perform secondary heading corrections based on the laser's raw point cloud data. However, these methods are highly complex and require significant computing power, making them difficult to meet the real-time data requirements of autonomous driving scenarios. Therefore, there is an urgent need to accurately correct heading angles and reduce heading jitter while consuming less computing power. Summary of the Invention

[0005] In view of this, the present invention provides a method and device for determining a heading angle, a vehicle controller, and a vehicle, so as to accurately correct the heading angle while consuming less computing resources and reduce heading angle jitter.

[0006] In a first aspect, the present invention provides a method for determining a heading angle, the method comprising: obtaining multiple first point cloud data and a first heading angle of a target in a current frame; determining first boundary point cloud data based on the multiple first point cloud data and the first heading angle, wherein the first boundary point cloud data is at least one first point cloud data in the multiple first point cloud data that is in a direction toward the first heading angle and is located at a boundary of the target close to a laser radar; obtaining second boundary point cloud data of the target in at least one historical frame, wherein the second boundary point cloud data is at least one second point cloud data in the multiple second point cloud data that is in a direction toward the second heading angle and is located at a boundary of the target close to the laser radar; correcting the first heading angle based on the first boundary point cloud data and the second boundary point cloud data; and determining the corrected first heading angle as the heading angle of the target in the current frame.

[0007] The method for determining the heading angle provided in this embodiment, after obtaining the first point cloud data and the first heading angle of the target in the current frame, determines the first boundary point cloud data based on the first point cloud data and the first heading angle, then obtains the second boundary point cloud data of the target in at least one historical frame, and corrects the first heading angle based on the first boundary point cloud data and at least one second boundary point cloud data. In this embodiment, determining the first boundary point cloud data through the current point cloud data and the current heading angle can reduce the number of point cloud data involved in the correction, thereby reducing the amount of calculation, and then correcting the heading angle output in the current frame through the current boundary point cloud data (i.e., the first boundary point cloud data) and the historical boundary point cloud data (i.e., the second boundary point cloud data), which can reduce the jitter of the heading angle and ensure the smoothness and authenticity of the output heading angle.

[0008] In an optional embodiment, first boundary point cloud data is determined based on multiple first point cloud data and a first heading angle, including: rasterizing a target set in the direction of the first heading angle to obtain multiple grids; determining first target point cloud data of each grid based on the multiple first point cloud data, wherein the first target point cloud data is the highest first point cloud data in a direction perpendicular to the ground in at least one first point cloud data corresponding to the grid; and determining first boundary point cloud data based on the first target point cloud data of each grid.

[0009] In an optional embodiment, the first boundary point cloud data is determined based on the first target point cloud data of each grid, including: starting from each grid corresponding to the boundary of the target close to the laser radar, searching the grids located in the vertical direction of the first heading angle in sequence, and determining the first target point cloud data closest to the laser radar in each row of grids located in the vertical direction of the first heading angle; and determining the first target point cloud data closest to the laser radar in each row of grids located in the vertical direction of the first heading angle as the first boundary point cloud data.

[0010] In this embodiment, after obtaining the first point cloud data and the first heading angle, the target set in the direction of the first heading angle is rasterized, and then the first target point cloud data of each grid is determined based on multiple first point cloud data. Then, starting from each grid corresponding to the boundary of the target close to the laser radar, the grids located in the vertical direction of the first heading angle are searched in sequence, and the first target point cloud data closest to the laser radar in each row of grids located in the vertical direction of the first heading angle is determined as the first boundary point cloud data, which can improve the efficiency and accuracy of determining the first boundary point cloud data.

[0011] In an optional implementation, the maximum number of grids searched in a direction perpendicular to the first heading angle is less than or equal to 3.

[0012] In this embodiment, the maximum number of grids searched in the vertical direction of the first heading angle is less than or equal to 3, which can improve the quality of the first boundary point cloud data and improve the extraction efficiency of the first boundary point cloud data.

[0013] In an optional embodiment, the first heading angle is corrected based on the first boundary point cloud data and the second boundary point cloud data, including: performing straight line fitting on the first boundary point cloud data and the second boundary point cloud data based on the weighted least squares method to obtain a target straight line; and correcting the first heading angle based on the angle between the target straight line and the preset direction.

[0014] In this embodiment, after determining the first boundary point cloud data and the second boundary point cloud data, a straight line fitting is performed on the first boundary point cloud data and the second boundary point cloud data based on the weighted least squares method to obtain a target straight line, and the first heading angle is corrected based on the angle between the target straight line and the preset direction, which can improve the accuracy of the corrected first heading angle.

[0015] In an optional embodiment, before performing straight line fitting on the first boundary point cloud data and the second boundary point cloud data based on the weighted least squares method to obtain the target straight line, the method also includes: determining the weight coefficient of the first boundary point cloud data to be 1; and determining the weight coefficient of the second boundary point cloud data corresponding to each historical frame based on the time difference between each historical frame and the current frame in at least one historical frame.

[0016] In an optional implementation, the number of historical frames is less than or equal to ten.

[0017] In an optional embodiment, before determining the corrected first heading angle as the heading angle of the target in the current frame, the method also includes: determining the difference between the first heading angle and the corrected first heading angle; determining the corrected first heading angle as the heading angle of the target in the current frame, including: when the difference is less than or equal to a preset angle, determining the corrected first heading angle as the heading angle of the target in the current frame.

[0018] In an optional implementation, the method further includes: when the difference is greater than a preset angle, determining the first heading angle as the heading angle of the target in the current frame.

[0019] In this embodiment, after determining the corrected first heading angle, the difference between the first heading angle and the corrected first heading angle is determined. When the difference is less than or equal to the preset angle, the corrected first heading angle is determined as the heading angle of the target in the current frame. When the difference is greater than the preset angle, the first heading angle is determined as the heading angle of the target in the current frame. This can reduce the interference caused by the calculation error of the corrected first heading angle, improve the accuracy of the heading angle of the target in the current frame that is finally output, thereby improving the decision-making accuracy of the autonomous driving vehicle and further improving driving safety.

[0020] In a second aspect, the present invention provides a device for determining a heading angle, comprising: a first acquisition module for acquiring multiple first point cloud data and a first heading angle of a target in a current frame; a first determination module for determining first boundary point cloud data based on the multiple first point cloud data and the first heading angle, wherein the first boundary point cloud data is at least one first point cloud data in the multiple first point cloud data that is in a direction toward the first heading angle and is located at a boundary of the target close to a laser radar; a second acquisition module for acquiring second boundary point cloud data of the target in at least one historical frame, wherein the second boundary point cloud data is at least one second point cloud data in the multiple second point cloud data that is in a direction toward the second heading angle and is located at a boundary of the target close to the laser radar; a correction module for correcting the first heading angle based on the first boundary point cloud data and the second boundary point cloud data; and a second determination module for determining the corrected first heading angle as the heading angle of the target in the current frame.

[0021] In an optional embodiment, the first determination module includes: a processing unit for rasterizing the target set in the direction of the first heading angle to obtain multiple grids; a first determination unit for determining the first target point cloud data of each grid based on multiple first point cloud data, wherein the first target point cloud data is the highest first point cloud data in the direction perpendicular to the ground in at least one first point cloud data corresponding to the grid; and a second determination unit for determining the first boundary point cloud data based on the first target point cloud data of each grid.

[0022] In an optional embodiment, the second determination unit includes: a first determination subunit, which is used to search the grids located in the vertical direction of the first heading angle in sequence with each grid corresponding to the boundary of the target close to the laser radar as the starting point, and determine the first target point cloud data closest to the laser radar in each row of grids located in the vertical direction of the first heading angle; a second determination subunit, which is used to determine the first target point cloud data closest to the laser radar in each row of grids located in the vertical direction of the first heading angle as the first boundary point cloud data.

[0023] In an optional implementation, the maximum number of grids searched in a direction perpendicular to the first heading angle is less than or equal to 3.

[0024] In an optional embodiment, the correction module includes: a fitting unit, used to perform straight line fitting on the first boundary point cloud data and the second boundary point cloud data based on the weighted least squares method to obtain a target straight line; a first correction unit, used to correct the first heading angle based on the angle between the target straight line and the preset direction.

[0025] In an optional embodiment, the device also includes: a third determination module, used to determine the weight coefficient of the first boundary point cloud data to be 1; a fourth determination module, used to determine the weight coefficient of the second boundary point cloud data corresponding to each historical frame based on the time difference between each historical frame and the current frame in at least one historical frame.

[0026] In an optional implementation, the number of historical frames is less than or equal to ten.

[0027] In an optional embodiment, the device also includes: a fifth determination module for determining the difference between the first heading angle and the corrected first heading angle; a correction module including: a third determination unit for determining the corrected first heading angle as the heading angle of the target in the current frame when the difference is less than or equal to a preset angle.

[0028] In an optional implementation, the correction module further includes: a fourth determination unit, configured to determine the first heading angle as the heading angle of the target in the current frame when the difference is greater than a preset angle.

[0029] In a third aspect, the present invention provides a vehicle controller comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0030] In a fourth aspect, the present invention provides a vehicle, which includes the device for determining the heading angle according to the second aspect or any corresponding embodiment thereof, or the vehicle includes the vehicle controller according to the third aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in related technologies, the following briefly introduces the drawings required for use in the specific embodiments or related technical descriptions. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0032] Figure 1 is a schematic flow chart of a method for determining a heading angle according to an embodiment of the present invention;

[0033] Figure 2 is a flow chart of another method for determining a heading angle according to an embodiment of the present invention;

[0034] Figure 3 is a schematic diagram of first boundary point cloud data according to an embodiment of the present invention;

[0035] Figure 4 is a schematic flow chart of another method for determining a heading angle according to an embodiment of the present invention;

[0036] Figure 5 4 is a structural block diagram of a device for determining a heading angle according to an embodiment of the present invention.

[0037] Reference numerals: 310 , target; 320 , first target point; 330 , second target point; 340 , laser radar. DETAILED DESCRIPTION

[0038] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0039] The present invention provides a method for determining a heading angle, which combines the second boundary point cloud data and the first boundary point cloud data to correct the first heading angle. It can ensure the smoothness and authenticity of the heading angle of the target in the output current frame while consuming lower computing power.

[0040] According to an embodiment of the present invention, an embodiment of a method for determining a heading angle is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a vehicle controller, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0041] In this embodiment, a method for determining a heading angle is provided, which can be used in a vehicle controller. For example, the vehicle controller can be an electronic control unit (ECU). Figure 1 FIG. 1 is a flow chart of a method for determining a heading angle according to an embodiment of the present invention. Figure 1 As shown, the method includes the following steps:

[0042] Step S101: Acquire a plurality of first point cloud data and a first heading angle of a target in a current frame.

[0043] Specifically, a laser radar is installed on the vehicle. During driving, the laser radar can measure targets in the surrounding environment by rapidly rotating or scanning the laser beam at preset intervals, thereby obtaining multiple three-dimensional point cloud data of the targets at each moment. For example, the format of the three-dimensional point cloud data can be [x, y, z].

[0044] Among them, the target can be obstacles such as roadblocks, other vehicles or pedestrians, etc. Other vehicles can be bicycles, electric vehicles or cars, etc. The time interval between two adjacent moments is a preset duration, and the preset duration is a preset value, which can be configured by the designer according to needs. For example, the preset duration can be 1s, 5s or 10s.

[0045] For the convenience of distinction, in the present invention, the three-dimensional point cloud data of the target acquired by the vehicle's laser radar at the current moment (current frame) is recorded as the first point cloud data, and the three-dimensional point cloud data of the target acquired by the vehicle's laser radar before the current moment (historical frame) is recorded as the second point cloud data.

[0046] For example, currently four frames of target three-dimensional point cloud data are acquired. If the fourth frame is the current frame, the first, second and third frames are all historical frames. At this time, the three-dimensional point cloud data of the target acquired in the fourth frame is the first point cloud data, and the three-dimensional point cloud data of the target acquired in the first frame, the three-dimensional point cloud data of the target acquired in the second frame and the three-dimensional point cloud data of the target acquired in the third frame are all second point cloud data.

[0047] Exemplarily, the first heading angle may be determined based on corresponding three-dimensional point cloud data and an artificial intelligence (AI) model.

[0048] Specifically, the vehicle controller is pre-configured with a trained AI model, which is used to determine the target's heading angle. The AI ​​model's input is the target's three-dimensional point cloud data at each moment, and its output is the target's heading angle at that moment. After acquiring the first point cloud data, the first point cloud data is input into the AI ​​model, and the output of the AI ​​model is the first heading angle.

[0049] The heading angle represents the target's direction of motion. For example, if the target is another vehicle, the heading angle can be the deviation between the vehicle's front and a preset direction. The preset direction is a set value that can be set by the designer. For example, the preset direction could be due north. In this case, if the vehicle's front is facing due north, the heading angle could be 0°; if the vehicle's front is facing due east, the heading angle could be 90°.

[0050] Step S102: determining first boundary point cloud data based on the plurality of first point cloud data and the first heading angle.

[0051] The first boundary point cloud data is at least one first point cloud data in the plurality of first point cloud data that is oriented toward the first heading angle and is located at the boundary of the target close to the laser radar.

[0052] Specifically, the point cloud data at the boundary of the target close to the laser radar is richer and can accurately reflect the actual heading angle of the target. After determining multiple first point cloud data and the first heading angle, the first point cloud data located at the boundary of the target close to the laser radar in the multiple first point cloud data are screened based on the first heading angle, and then the first heading angle is corrected based on the first boundary point cloud data. This can reduce the amount of calculation while ensuring the accuracy of the correction result.

[0053] For example, the area formed by multiple first point cloud data of the target is divided into a first area and a second area with a straight line parallel to the first heading angle as the center line, and the first point cloud data in the area close to the laser radar in the two areas (the first area and the second area) can be determined as the first boundary point cloud data.

[0054] Step S103: Acquire second boundary point cloud data of the target in at least one historical frame.

[0055] The second boundary point cloud data is at least one second point cloud data in the plurality of second point cloud data that is oriented toward the second heading angle and is located at the boundary of the target close to the laser radar.

[0056] Specifically, the first boundary point cloud data and the second boundary point cloud data are determined in the same way, and the current frame and the historical frame are relative. If the vehicle controller obtains the three-dimensional point cloud data of the target in the 4th frame at the current moment, the current frame is the 4th frame. At this time, the boundary point cloud data determined based on the three-dimensional point cloud data and the heading angle of the target in the 4th frame is the first boundary point cloud data; if the vehicle controller obtains the three-dimensional point cloud data of the target in the 5th frame at the current moment, the current frame is the 5th frame, and the 4th frame is the historical frame. At this time, the boundary point cloud data determined based on the three-dimensional point cloud data and the heading angle of the target in the 5th frame is the first boundary point cloud data, and the boundary point cloud data determined based on the three-dimensional point cloud data and the heading angle of the target in the 4th frame is the second boundary point cloud data.

[0057] Step S104: correcting the first heading angle based on the first boundary point cloud data and the second boundary point cloud data.

[0058] Specifically, after determining the first boundary point cloud data and the at least one second boundary point cloud data, a straight line fitting may be performed on the first boundary point cloud data and the at least one second boundary point cloud data to obtain a target straight line, and the first heading angle may be corrected based on the target straight line. For example, the deviation angle between the target straight line and the preset direction may be determined as the corrected first heading angle.

[0059] For example, a linear fitting may be performed on the first boundary point cloud data and the at least one second boundary point cloud data by using a fitting method such as least squares method, weighted squares method, random sampling consensus algorithm or Hough transform.

[0060] Step S105: determining the corrected first heading angle as the heading angle of the target in the current frame.

[0061] Specifically, after correcting the first heading angle, the corrected first heading angle is determined as the heading angle of the target in the current frame, and the vehicle's decision planning and perception presentation are performed using the corrected first heading angle.

[0062] The method for determining the heading angle provided in this embodiment, after obtaining the first point cloud data and the first heading angle of the target in the current frame, determines the first boundary point cloud data based on the first point cloud data and the first heading angle, then obtains the second boundary point cloud data of the target in at least one historical frame, and corrects the first heading angle based on the first boundary point cloud data and at least one second boundary point cloud data. In this embodiment, determining the first boundary point cloud data through the current point cloud data and the current heading angle can reduce the number of point cloud data involved in the correction, thereby reducing the amount of calculation, and then correcting the heading angle output in the current frame through the current boundary point cloud data (i.e., the first boundary point cloud data) and the historical boundary point cloud data (i.e., the second boundary point cloud data), which can reduce the jitter of the heading angle and ensure the smoothness and authenticity of the output heading angle.

[0063] For example, if the first point cloud data is the 3D point cloud data of the target in the first frame, and the first heading angle is the heading angle of the target in the first frame, and there is no second boundary point cloud data, then the second boundary point cloud data is 0. The first heading angle can be corrected directly based on the first boundary point cloud data. That is, a straight line is directly fitted to the first boundary point cloud data to obtain the target line, and the first heading angle is corrected based on the target line.

[0064] In this embodiment, a method for determining a heading angle is provided, which can be used in a vehicle controller. Figure 2 FIG. 1 is a flow chart of another method for determining a heading angle according to an embodiment of the present invention. Figure 2 As shown, the method includes the following steps:

[0065] Step S201: Acquire a plurality of first point cloud data and a first heading angle of the target in the current frame.

[0066] For details, please see Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0067] Step S202: Determine first boundary point cloud data based on the plurality of first point cloud data and the first heading angle.

[0068] Specifically, the above step S202 includes:

[0069] Step S2021: performing rasterization processing on the target set toward the first heading angle direction to obtain a plurality of grids.

[0070] For example, Figure 3 As shown, the shape of the target 310 is simplified into a rectangle, and the direction of the long side is made the same as the direction of the first heading angle. Then, the target set toward the first heading angle is rasterized to obtain a plurality of grids.

[0071] Among them, the first heading angle direction can be Figure 3 In the direction of arrow S1, the size and shape of the grid can be set by the designer according to the requirements. For example, the shape of the grid can be square and the side length of the grid can be 20 cm. The gridding processing method in the present invention is a common gridding processing method in the art and will not be described in detail here.

[0072] Step S2022: Determine first target point cloud data of each grid based on the plurality of first point cloud data.

[0073] The first target point cloud data is the highest first point cloud data in a direction perpendicular to the ground among the at least one first point cloud data corresponding to the grid.

[0074] Specifically, a plurality of first point cloud data come from the surface of the target, such as Figure 3 There is a corresponding relationship between the target 310 shown in FIG. , after matching the first point cloud data to the corresponding grid, the z coordinate of at least one first point cloud data in each grid is compared, and the first point cloud data with the largest z value in each grid is the first target point cloud data of the corresponding grid. Figure 3 The first target point 320 and the second target point 330 shown are both first target point cloud data of corresponding grids.

[0075] In this embodiment, the highest first point cloud data in the direction perpendicular to the ground is retained in each grid, which can exclude non-target point cloud data as much as possible, reduce interference, and improve the accuracy of the determined first boundary point cloud data.

[0076] Step S2023: Determine first boundary point cloud data based on the first target point cloud data of each grid.

[0077] Exemplarily, the above step S2023 includes step a1 and step a2:

[0078] Step a1, starting from each grid corresponding to the boundary of the target close to the laser radar, searches the grids located in the vertical direction of the first heading angle in sequence, and determines the first target point cloud data closest to the laser radar in each row of grids located in the vertical direction of the first heading angle.

[0079] Step a2: determine the first target point cloud data closest to the laser radar in each row of grids in the vertical direction of the first heading angle as the first boundary point cloud data.

[0080] Specifically, if Figure 3 As shown, the vertical direction of the first heading angle can be the direction of arrow S2, that is, the short side of the area formed by the target 310. Starting from each grid corresponding to the long side of the target 310 close to the laser radar 340, the grids in the direction of S2 are searched in sequence until the searched grid contains the first target point cloud data, and the searched first target point cloud data is determined as the first boundary point cloud data. Figure 3 As shown, the first boundary point cloud data is the first target point 320 .

[0081] Exemplarily, in order to improve the quality of the first boundary point cloud data and improve the extraction efficiency of the first boundary point cloud data, the maximum number of grids searched in the vertical direction of the first heading angle is less than or equal to 3.

[0082] Step S203: Acquire second boundary point cloud data of the target in at least one historical frame.

[0083] For details, please see Figure 1Step S103 of the illustrated embodiment will not be described in detail here.

[0084] Step S204: Correct the first heading angle based on the first boundary point cloud data and the second boundary point cloud data.

[0085] For details, please see Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.

[0086] Step S205: Determine the corrected first heading angle as the heading angle of the target in the current frame.

[0087] For details, please see Figure 1 Step S105 of the illustrated embodiment will not be described in detail here.

[0088] The method for determining the heading angle provided in this embodiment, after obtaining the first point cloud data and the first heading angle, performs rasterization processing on the target set in the direction of the first heading angle, and then determines the first target point cloud data of each grid based on multiple first point cloud data. Then, starting from each grid corresponding to the boundary of the target close to the laser radar, the grids located in the vertical direction of the first heading angle are searched in sequence, and the first target point cloud data closest to the laser radar in each row of grids located in the vertical direction of the first heading angle is determined as the first boundary point cloud data, which can improve the efficiency and accuracy of determining the first boundary point cloud data.

[0089] In this embodiment, a method for determining a heading angle is provided, which can be used in a vehicle controller. Figure 4 FIG. 1 is a flow chart of another method for determining a heading angle according to an embodiment of the present invention. Figure 4 As shown, the method includes the following steps:

[0090] Step S401: Acquire a plurality of first point cloud data and a first heading angle of the target in the current frame.

[0091] For details, please see Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0092] Step S402: Determine first boundary point cloud data based on the plurality of first point cloud data and the first heading angle.

[0093] For details, please see Figure 2 Step S202 of the illustrated embodiment will not be described in detail here.

[0094] Step S403: Acquire second boundary point cloud data of the target in at least one historical frame.

[0095] For details, please see Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.

[0096] Step S404: Correct the first heading angle based on the first boundary point cloud data and the second boundary point cloud data.

[0097] Specifically, the above step S404 includes:

[0098] Step S4041 : performing straight line fitting on the first boundary point cloud data and the second boundary point cloud data based on the weighted least squares method to obtain a target straight line.

[0099] Specifically, the weighted least squares method is an improvement on the least squares method. The goal of the least squares method is to minimize the sum of squared errors between the observations (the first boundary point cloud data and the second boundary point cloud data in this application) and the fitting values. Different observations may have different importance. The weighted least squares method assigns a weight to each observation, and the weight reflects the relative importance of the observation.

[0100] For example, the target straight line can be expressed as formula (1):

[0101] y=a0+a1×x (1)

[0102] Among them, x represents the coordinate of the three-dimensional point cloud data in the x-axis direction, y represents the coordinate of the three-dimensional point cloud data in the y-axis direction, a0 and a1 are fixed values ​​determined based on the first boundary point cloud data and the second boundary point cloud data.

[0103] The matrix expression of the target line can be shown as formula (2):

[0104] AX=B (2)

[0105] Wherein, A represents the matrix of coordinates of the first boundary point cloud data and the second boundary point cloud data in the x-axis direction, X represents the matrix of unknown quantities, B represents the matrix of coordinates of the first boundary point cloud data and the second boundary point cloud data in the y-axis direction,

[0106] The unknown quantity of the target line can be determined by the following formula (3):

[0107] X=[A T W T WA] -1 A T W T WB (3)

[0108] Among them, A T represents the transpose of matrix A, W T represents the transpose of the matrix W, W represents the matrix of the weight coefficients of the first boundary point cloud data and the second boundary point cloud data, Exemplarily, the weight coefficients of the first boundary point cloud data and the second boundary point cloud data may be determined by a designer.

[0109] After determining the first boundary point cloud data and at least one second boundary point cloud data, the unknown quantity in the target straight line can be determined by formula (3), thereby determining the target straight line.

[0110] Step S4042: Correct the first heading angle based on the angle between the target line and the preset direction.

[0111] The preset direction may be a horizontal direction or a vertical direction.

[0112] Specifically, after the target straight line is determined, the angle between the target straight line and the preset direction is determined as the corrected first heading angle.

[0113] In some optional implementations, before step S4041, the method for determining the heading angle further includes steps b1 and b2:

[0114] Step b1: determine the weight coefficient of the first boundary point cloud data to be 1.

[0115] Step b2: determining a weight coefficient of the second boundary point cloud data corresponding to each historical frame based on a time difference between each historical frame and the current frame in at least one historical frame.

[0116] Specifically, the weight coefficients of the first boundary point cloud data and the second boundary point cloud data can be determined by the following formula (4):

[0117]

[0118] Among them, t represents the time difference between the historical frame corresponding to the second boundary point cloud data and the current frame corresponding to the first boundary point cloud data. The unit of t is seconds (s). The larger the time difference between the historical frame corresponding to the second boundary point cloud data and the current frame, the smaller the weight.

[0119] For example, the number of history frames may be determined by a designer according to requirements. For example, the number of history frames may be less than or equal to 10.

[0120] Step S405: Determine the difference between the first heading angle and the corrected first heading angle.

[0121] Specifically, after determining the modified first heading angle, the difference between the first heading angle and the corrected first heading angle is determined. If the difference is less than or equal to a preset angle, the heading angle of the target in the current frame is the corrected first heading angle, and step S406 is executed. If the difference is greater than the preset angle, the heading angle of the target in the current frame is the first heading angle, and step S407 is executed. The preset angle can be determined by the designer according to requirements.

[0122] Step S406: When the difference is less than or equal to the preset angle, the corrected first heading angle is determined as the heading angle of the target in the current frame.

[0123] Step S407: When the difference is greater than the preset angle, the first heading angle is determined as the heading angle of the target in the current frame.

[0124] The method for determining the heading angle provided in this embodiment, after determining the first boundary point cloud data and the second boundary point cloud data, performs straight line fitting on the first boundary point cloud data and the second boundary point cloud data based on the weighted least squares method to obtain a target straight line, and corrects the first heading angle based on the angle between the target straight line and the preset direction, which can improve the accuracy of the corrected first heading angle. After determining the corrected first heading angle, the difference between the first heading angle and the corrected first heading angle is determined. If the difference is less than or equal to the preset angle, the corrected first heading angle is determined as the heading angle of the target in the current frame. If the difference is greater than the preset angle, the first heading angle is determined as the heading angle of the target in the current frame. This can reduce interference caused by errors in the calculation of the corrected first heading angle, improve the accuracy of the heading angle of the target in the current frame that is finally output, thereby improving the decision-making accuracy of the autonomous driving vehicle and further improving driving safety.

[0125] This embodiment also provides a device for determining a heading angle, which is used to implement the above-mentioned embodiments and preferred embodiments. Details already described will not be repeated here. As used below, the term "module" may refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0126] This embodiment provides a device for determining a heading angle, such as Figure 5 Shown, including:

[0127] A first acquisition module 501 is used to acquire a plurality of first point cloud data and a first heading angle of the target in the current frame;

[0128] A first determining module 502 is configured to determine first boundary point cloud data based on the plurality of first point cloud data and the first heading angle, wherein the first boundary point cloud data is at least one first point cloud data in the plurality of first point cloud data that is oriented in a direction of the first heading angle and is located at a boundary of the target close to the laser radar;

[0129] A second acquisition module 503 is configured to acquire second boundary point cloud data of the target in at least one historical frame, wherein the second boundary point cloud data is at least one second point cloud data in the plurality of second point cloud data that is oriented toward the second heading angle and is located at a boundary of the target close to the laser radar;

[0130] A correction module 504 is configured to correct the first heading angle based on the first boundary point cloud data and the second boundary point cloud data;

[0131] The second determining module 505 is configured to determine the corrected first heading angle as the heading angle of the target in the current frame.

[0132] In some optional implementations, the first determining module 502 includes:

[0133] a processing unit, configured to perform a rasterization process on a target set in the direction of the first heading angle to obtain a plurality of grids;

[0134] A first determining unit is configured to determine first target point cloud data of each grid based on the plurality of first point cloud data, wherein the first target point cloud data is the highest first point cloud data in a direction perpendicular to the ground among the at least one first point cloud data corresponding to the grid;

[0135] The second determining unit is configured to determine the first boundary point cloud data based on the first target point cloud data of each grid.

[0136] In some optional implementations, the second determining unit includes:

[0137] A first determining subunit is configured to sequentially search the grids located in a vertical direction of the first heading angle, starting from each grid corresponding to the boundary of the target close to the laser radar, and determine the first target point cloud data closest to the laser radar in each row of grids located in the vertical direction of the first heading angle;

[0138] The second determining subunit is used to determine the first target point cloud data closest to the laser radar in each row of grids in the vertical direction of the first heading angle as the first boundary point cloud data.

[0139] In some optional implementations, the maximum number of grids searched in a direction perpendicular to the first heading angle is less than or equal to 3.

[0140] In some optional implementations, the correction module 504 includes:

[0141] A fitting unit, configured to perform straight line fitting on the first boundary point cloud data and the second boundary point cloud data based on a weighted least squares method to obtain a target straight line;

[0142] The first correction unit is configured to correct the first heading angle based on an angle between the target straight line and a preset direction.

[0143] In some optional embodiments, the device further comprises:

[0144] A third determining module, configured to determine the weight coefficient of the first boundary point cloud data to be 1;

[0145] The fourth determining module is configured to determine a weight coefficient of the second boundary point cloud data corresponding to each historical frame based on a time difference between each historical frame and the current frame in at least one historical frame.

[0146] In some optional implementations, the number of historical frames is less than or equal to ten.

[0147] In some optional embodiments, the device further comprises:

[0148] a fifth determining module, configured to determine a difference between the first heading angle and the corrected first heading angle;

[0149] The correction module 504 includes:

[0150] The third determining unit is configured to determine the corrected first heading angle as the heading angle of the target in the current frame when the difference is less than or equal to a preset angle.

[0151] In some optional implementations, the correction module 504 further includes:

[0152] The fourth determining unit is configured to determine the first heading angle as the heading angle of the target in the current frame when the difference is greater than a preset angle.

[0153] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0154] In this embodiment, a device for determining a heading angle is presented in the form of a functional unit, where the unit refers to an application-specific integrated circuit (ASIC), a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above-mentioned functions.

[0155] An embodiment of the present invention also provides a vehicle controller, which includes: one or more processors, a memory, and interfaces for connecting various components, including a high-speed interface and a low-speed interface. The various components are connected to each other using different buses and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the vehicle controller, including instructions stored in or on the memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories.

[0156] The processor may be a central processing unit, a network processor, or a combination thereof. The processor may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0157] The memory stores instructions that can be executed by at least one processor, so that the at least one processor executes the method shown in the above embodiment.

[0158] The memory may include a program storage area and a data storage area. The program storage area may store an operating system and application programs required for at least one function, while the data storage area may store data generated based on the use of the vehicle controller. Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage device.

[0159] The memory may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid state drive; the memory may also include a combination of the above types of memory.

[0160] An embodiment of the present invention further provides a vehicle, the vehicle including the device for determining a heading angle provided by any of the above embodiments, or the vehicle including the vehicle controller provided by any of the above embodiments.

[0161] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations shall all fall within the scope defined by the present invention.

Claims

1. A method for determining a heading angle, characterized in that: The method comprises: Acquire multiple first point cloud data and first heading angle of the target in the current frame; Determine first boundary point cloud data based on the multiple first point cloud data and the first heading angle, wherein the first boundary point cloud data is at least one first point cloud data in the multiple first point cloud data that is oriented in the direction of the first heading angle and is located at a boundary of the target close to the laser radar; Acquire second boundary point cloud data of the target in at least one historical frame, wherein the second boundary point cloud data is at least one second point cloud data in a direction toward a second heading angle and located at a boundary of the target close to a laser radar among a plurality of second point cloud data; Correcting the first heading angle based on the first boundary point cloud data and the second boundary point cloud data; The corrected first heading angle is determined as the heading angle of the target in the current frame.

2. The method according to claim 1, characterized in that The determining of first boundary point cloud data based on the plurality of first point cloud data and the first heading angle comprises: Performing a rasterization process on the target set in the direction of the first heading angle to obtain a plurality of grids; Determine first target point cloud data of each grid based on the multiple first point cloud data, wherein the first target point cloud data is the highest first point cloud data in a direction perpendicular to the ground among at least one first point cloud data corresponding to the grid; The first boundary point cloud data is determined based on the first target point cloud data of each grid.

3. The method according to claim 2, characterized in that The determining the first boundary point cloud data based on the first target point cloud data of each grid includes: Taking each grid corresponding to the boundary of the target close to the laser radar as a starting point, sequentially searching the grids located in the vertical direction of the first heading angle, and determining the first target point cloud data closest to the laser radar in each row of grids located in the vertical direction of the first heading angle; The first target point cloud data closest to the laser radar in each row of grids in the vertical direction of the first heading angle is determined as the first boundary point cloud data.

4. The method according to claim 3, characterized in that: The maximum number of grids to be searched in the vertical direction of the first heading angle is less than or equal to 3.

5. The method according to any one of claims 1 to 4, characterized in that The correcting the first heading angle based on the first boundary point cloud data and the second boundary point cloud data includes: Performing straight line fitting on the first boundary point cloud data and the second boundary point cloud data based on a weighted least squares method to obtain a target straight line; The first heading angle is corrected based on an angle between the target straight line and a preset direction.

6. The method according to claim 5, characterized in that Before performing straight line fitting on the first boundary point cloud data and the second boundary point cloud data based on the weighted least squares method to obtain a target straight line, the method further includes: Determine the weight coefficient of the first boundary point cloud data as 1; Based on the time difference between each historical frame in at least one historical frame and the current frame, a weight coefficient of the second boundary point cloud data corresponding to each historical frame is determined.

7. The method according to claim 6, characterized in that The number of the historical frames is less than or equal to 10.

8. The method according to any one of claims 1 to 4, characterized in that Before determining the corrected first heading angle as the heading angle of the target in the current frame, the method further includes: determining a difference between the first heading angle and the corrected first heading angle; The step of determining the corrected first heading angle as the heading angle of the target in the current frame includes: When the difference is less than or equal to the preset angle, the corrected first heading angle is determined as the heading angle of the target in the current frame.

9. The method according to claim 8, characterized in that The method further comprises: When the difference is greater than the preset angle, the first heading angle is determined as the heading angle of the target in the current frame.

10. A device for determining a heading angle, characterized in that: The device comprises: A first acquisition module, used to acquire a plurality of first point cloud data and a first heading angle of the target in the current frame; A first determination module is configured to determine first boundary point cloud data based on the plurality of first point cloud data and the first heading angle, wherein the first boundary point cloud data is at least one first point cloud data in the plurality of first point cloud data that is oriented in a direction toward the first heading angle and is located at a boundary of the target close to the laser radar; A second acquisition module is used to acquire second boundary point cloud data of the target in at least one historical frame, wherein the second boundary point cloud data is at least one second point cloud data in a direction toward a second heading angle and located at a boundary of the target close to the laser radar among the plurality of second point cloud data; A correction module, configured to correct the first heading angle based on the first boundary point cloud data and the second boundary point cloud data; The second determination module is used to determine the corrected first heading angle as the heading angle of the target in the current frame.

11. A vehicle controller, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 9 by executing the computer instructions.

12. A vehicle, characterized in that: The vehicle comprises the device for determining a heading angle according to claim 10 or the vehicle controller according to claim 11 .