Target speed estimation method, device and storage medium

Abstract

The present invention discloses a target speed estimation method, device, control equipment, computer-readable storage medium, computer program product containing instructions, and vehicle. The steps of the target speed estimation method are: calculating a first candidate speed and a second candidate speed according to speed calculation method 1 and speed calculation method 2 respectively; when it is determined that the target is located in an area outside a first boundary with the vehicle as the symmetry center, the first candidate speed obtained according to speed calculation method 1 is determined as the target speed; otherwise, the second candidate speed obtained according to speed calculation method 2 is determined as the target speed. The present invention uses two speed estimation methods to estimate the speed of an obstacle target, and selects a speed estimation result as the final estimated speed of the target according to the different positions of the obstacle target relative to the vehicle, thereby improving the accuracy of target speed estimation.

Description

[0001] This application claims priority to Chinese patent application number 2021103895278, filed on April 12, 2021, and titled “Target Speed ​​Estimation Method, Device and Storage Medium”, the disclosure of which is incorporated herein by reference. Technical Field

[0002] The present invention relates to the field of unmanned driving technology, and in particular to a target speed estimation method, device, control equipment, computer-readable storage medium, a computer program product containing instructions, and a vehicle. Background Art

[0003] In unmanned driving technology, there are currently two main ways to obtain the speed of obstacle targets. The first is to obtain it through direct measurement by millimeter-wave radar. Millimeter-wave radar can obtain the relative speed of the target more accurately, but the currently commonly used millimeter-wave radar configuration has a certain observation blind spot (usually within a close distance to the left and right of the vehicle), so the autonomous driving vehicle cannot obtain the speed information of the target in the blind spot. The second is to infer the speed information of the target through the position change of the obstacle target detected by the laser radar. This method has a small observation blind spot, but the inferred target speed is greatly affected by the position error measured by the laser radar. Summary of the invention

[0004] The object of the present invention is to provide a target speed estimation method, device, control equipment, computer-readable storage medium, computer program product containing instructions and vehicle in view of the technical defects existing in the prior art. The target speed estimation method adopts two different speed estimation methods to estimate the speed of the obstacle target, and selects a speed estimation result as the final estimated speed of the target according to the different positions of the obstacle target relative to the vehicle, thereby improving the accuracy of the target speed estimation.

[0005] According to a first aspect of the present invention, a method for estimating target speed is provided, comprising the following steps:

[0006] Calculate a first candidate speed and a second candidate speed according to the speed calculation method 1 and the speed calculation method 2 respectively;

[0007] When it is determined that the target is located outside the first boundary with the vehicle as the symmetric center, the first candidate speed obtained according to the speed calculation method 1 is determined as the speed of the target;

[0008] When it is determined that the target is located within the first boundary with the vehicle as the symmetric center, the second candidate speed obtained according to the speed calculation method 2 is determined as the speed of the target;

[0009] The first candidate speed is calculated according to the speed calculation method 1, including:

[0010] When it is determined that the target is located in an area outside the second boundary with the vehicle as the symmetric center, the nearest point of the target detected by the laser radar is used as the target tracking point; or, when it is determined that the target is located in an area within the second boundary with the vehicle as the symmetric center, the center point of the target detected by the laser radar is used as the target tracking point;

[0011] Obtaining a first candidate speed based on the target tracking point estimation;

[0012] The second selected speed is calculated according to the second speed calculation method, including:

[0013] Using the position data of the target frame corner points and target tracking points obtained by the laser radar, the moving speed of each target frame corner point and target tracking point is calculated;

[0014] A second candidate speed is estimated according to the moving speed of each target frame corner point and the target tracking point.

[0015] A second aspect of the present invention provides a target speed estimation device, comprising:

[0016] A first speed calculation unit, used for calculating a first speed to be selected according to a speed calculation method 1;

[0017] The second speed calculation unit is used to calculate the second selected speed according to the second speed calculation method.

[0018] A speed determination unit, used to determine that when the target is located outside a first boundary with the vehicle as the symmetric center, the first candidate speed is determined as the target speed, and vice versa, the second candidate speed is determined as the target speed;

[0019] Wherein, the first speed calculation unit includes:

[0020] The tracking point determination module is used to determine that when the target is located outside the second boundary with the vehicle as the symmetric center, the nearest point of the target detected by the laser radar is used as the target tracking point; otherwise, the center point of the target detected by the laser radar is used as the target tracking point;

[0021] A first speed estimation module, used for estimating a first candidate speed based on the target tracking point;

[0022] Wherein, the second speed calculation unit includes:

[0023] The target point speed calculation module is used to calculate the moving speed of each target frame corner point and target tracking point using the position data of the target frame corner point and the target tracking point obtained by the laser radar;

[0024] The second speed estimation module is used to estimate a second candidate speed according to the moving speed of each target frame corner point and the target tracking point.

[0025] According to a third aspect of the present invention, a control device is provided, comprising at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the target speed estimation method described in the first aspect of the present invention.

[0026] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, comprising a program or an instruction. When the program or the instruction is executed on a computer, the target speed estimation method according to the first aspect of the present invention is implemented.

[0027] According to a fifth aspect of the present invention, there is provided a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the target speed estimation method according to the first aspect of the present invention.

[0028] A sixth aspect of the present invention provides a vehicle, comprising the control device described in the third aspect of the present invention.

[0029] The present invention adopts two different speed estimation methods to estimate the speed of the obstacle target respectively, and then selects one of the speed estimation results as the final estimated speed of the target according to the different positions of the obstacle target relative to the vehicle. This can overcome the problem that the autonomous driving vehicle cannot obtain the speed information of the target in the blind spot of the millimeter wave radar, and overcome the problem that the laser radar measurement results are greatly affected by the position, thereby improving the accuracy of the target speed estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flow chart of a target speed estimation method according to an embodiment of the present invention;

[0031] Figure 2 is a schematic diagram of a closed first boundary of an embodiment of the present invention;

[0032] Figure 3 is a calculation flow chart of the first selected speed according to an embodiment of the present invention;

[0033] Figure 4 The calculation flow chart of the second selected speed of the embodiment of the present invention is as follows

[0034] Figure 5 is a schematic diagram of the arrangement of a first boundary according to an embodiment of the present invention;

[0035] Figure 6 It is a schematic diagram of determining and switching the tracking point based on the second boundary when calculating the first candidate speed according to an embodiment of the present invention;

[0036] Figure 7 A schematic diagram of velocity calculation of a uniform linear motion model according to an embodiment of the present invention;

[0037] Figure 8 A schematic diagram of a laser radar target corner point and a target center point when calculating a second candidate speed according to an embodiment of the present invention;

[0038] Fig. 9 A schematic diagram of calculating the displacement speed of a laser radar target corner point and a target center point and determining the target speed according to an embodiment of the present invention;

[0039] Fig.10 4 is a schematic diagram of a target speed estimation device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0041] It should be noted that:

[0042] The "self-vehicle" referred to in the present invention refers to the vehicle used to execute the target speed estimation method proposed in the embodiment of the present invention, that is, the current vehicle equipped with sensor equipment such as laser radar and millimeter wave radar.

[0043] The "target speed" referred to in the present invention refers to the speed of the target (obstacle) relative to the vehicle in the vehicle coordinate system, and can also be called the relative speed of the target.

[0044] The lasers commonly used in the market do not have penetrating properties. For larger targets such as vehicles, they can only obtain point clouds facing the vehicle, but not the point clouds facing away from the vehicle. In addition, they may be affected by the self-occlusion of the target or the mutual occlusion between objects. The point clouds obtained by the laser radar often cannot completely cover the entire target. Therefore, the target detection module of the laser radar usually needs to use deep learning to complete the target point cloud based on the original point cloud obtained, and then perform target frame regression based on the completed point cloud data to output information such as the length, width, height, and center point position of the target. However, the farther the target is from the vehicle, the less original point cloud data the laser radar obtains, and the greater the error between the output result and the actual situation. Conversely, the closer the target is to the vehicle, the more original point cloud data the laser radar obtains, and the more accurate the output result will be.

[0045] The millimeter-wave radar configuration commonly used in autonomous driving vehicles currently has certain observation blind spots, usually areas close to the left and right of the vehicle, and it is impossible to obtain speed information of targets in the blind spots.

[0046] In order to solve the above problem, an embodiment of the present invention proposes a target speed estimation method. Based on the characteristics of different perception modules of the ego vehicle, two different speed estimation methods are used to calculate the speed of the target, and then one of the calculated speeds is selected as the target speed according to the different positions of the target relative to the ego vehicle.

[0047] In a first aspect of an embodiment of the present invention, a method for estimating target speed is provided. Figure 1 As shown, the following steps may be adopted:

[0048] S101: Calculate a first candidate speed and a second candidate speed according to speed calculation method 1 and speed calculation method 2 respectively;

[0049] S102: Determine the position of the target relative to the vehicle;

[0050] S103: Determine the first candidate speed or the second candidate speed as the speed of the target according to the position of the target relative to the vehicle:

[0051] When it is determined that the target is located outside the first boundary with the vehicle as the symmetric center, the first candidate speed obtained according to the speed calculation method 1 is determined as the target speed; and when it is determined that the target is located within the first boundary with the vehicle as the symmetric center, the second candidate speed obtained according to the speed calculation method 2 is determined as the target speed;

[0052] In specific implementation, "outside the first boundary" can be set to include the first boundary itself, and "inside the first boundary" can be set to not include the first boundary itself; or, "inside the first boundary" can be set to include the first boundary itself, and "outside the first boundary" can be set to not include the first boundary itself.

[0053] In specific implementation, when it is determined based on the basic characteristics of the laser radar, calibration data, test data, etc. that when the target is located in an area within a certain boundary, the original point cloud obtained by the laser radar is relatively complete compared with the theoretical value (for example, the ratio of the original point cloud to the theoretical value exceeds a preset ratio threshold), and the error of the final output result is smaller than the actual situation of the target. Conversely, when the target is located in an area outside a certain boundary, the original point cloud obtained by the laser radar is incomplete compared with the theoretical value (for example, the ratio of the original point cloud to the theoretical value is lower than the preset ratio threshold), and the error of the final output result is larger than the actual situation of the target. In this case, the boundary can be set to the first boundary in the above embodiment.

[0054] For a specific vehicle and lidar, the first boundary changes dynamically as the vehicle moves, but the relative position of the first boundary and the vehicle is relatively stable. Therefore, the first boundary can be determined in advance in the vehicle coordinate system based on the basic characteristics of the lidar, calibration data, test data, etc.

[0055] When the target is located in the area within the first boundary, the target length, width, height, center point and other information output by the laser radar are relatively accurate. In this case, the calculation result of speed calculation method 2 is used as the target speed; when the target is located in the area outside the first boundary, the target length, width, height, center point and other information output by the laser radar have large errors. In this case, the calculation result of speed calculation method 1 is used as the target speed.

[0056] As an optional embodiment, the first boundary may be a closed boundary, for example, the first boundary is a closed boundary with the origin of the vehicle coordinate system as the symmetry center, such as a circular closed boundary, an elliptical closed boundary or a rectangular closed boundary or a closed boundary of other shapes, such as Figure 2 The figure shows an elliptical boundary 300 with the origin of the vehicle coordinate system as the symmetry center. Relative to the position of the vehicle 100, when the target 200 is within the elliptical boundary 300, the calculation result of the speed calculation method 2 is used as the target speed, and when the target is located in the area outside the elliptical boundary 300, the calculation result of the speed calculation method 1 is used as the target speed.

[0057] As an optional embodiment, the first boundary may also be a non-closed boundary. For example, the first boundary may be two linear boundaries with the origin of the vehicle coordinate system as the symmetry center. The area between the two linear boundaries is within the first boundary, and the area outside the two linear boundaries is outside the first boundary.

[0058] For example, the two linear boundaries may be a first linear boundary 1 and a second linear boundary 2 located in front of and behind the vehicle 100, respectively, and perpendicular to the driving direction of the vehicle. Figure 5 As shown, a vehicle coordinate system is established with the rear axle center of the vehicle 100 as the origin, and the x-axis and y-axis represent the longitudinal axis and the transverse axis of the vehicle coordinate system, respectively. For example, the first straight line boundary 1 is a straight line boundary located 10m in front of the vehicle (i.e. +10m on the x-axis and parallel to the y-axis), and the second straight line boundary 2 is a straight line boundary located 10m behind the vehicle (i.e. -10m on the x-axis and parallel to the y-axis).

[0059] In specific implementation, the first straight line boundary 1 and the second straight line boundary 2 are determined based on the completeness of the point cloud detected by the laser radar for the target. For example, according to the basic characteristics of the laser radar, calibration data and a large amount of test data, when the target is within 10m in front of or behind the vehicle, the laser radar outputs the target length, width, height, center point and other information more accurately, while when the target is more than 10m in front of or behind the vehicle, the laser radar outputs the target length, width, height, center point and other information with large errors. In this case, the first straight line boundary 1 and the second straight line boundary 2 can be set to be 10m in front of the vehicle and 10m behind the vehicle, respectively.

[0060] For details, see Figure 5 As shown, relative to the vehicle 100, the area outside the first straight line boundary 1 and the second straight line boundary 2 is the first area A, and the area within the first straight line boundary 1 and the second straight line boundary 2 is the second area B. When the target 200 (labeled as id#1) is located in the first area A, the speed calculated by the speed calculation method 1 is used as the final target estimated speed; when the target 200 (labeled as id#1) is located in the second area B, the speed calculated by the speed calculation method 2 is used as the final target estimated speed.

[0061] The common laser radars on the market currently use the center point of the detected target as the tracking point for target tracking. Affected by factors such as the observable position and the target clustering effect, when the target is close to the vehicle, the laser radar can continuously, stably and accurately detect the center point of the target. However, when the target is far away from the vehicle, the center point detected by the laser radar cannot be continuous, is not stable enough (for example, the tracking point is lost) or is inaccurate, and the target tracking result is also inaccurate.

[0062] In view of this problem, a second aspect of an embodiment of the present invention provides a target tracking method for a laser radar, comprising:

[0063] Step 1, determine the position of the target relative to the vehicle;

[0064] Step 2, determining the target tracking point according to the position of the target relative to the vehicle, specifically including: when the target is located in an area outside the second boundary with the vehicle as the symmetric center, taking the nearest point of the target detected by the laser radar as the target tracking point; or, when the target is located in an area within the second boundary with the vehicle as the symmetric center, taking the center point of the target detected by the laser radar as the target tracking point;

[0065] Step 3: Track the target based on the target tracking point.

[0066] Specifically, for distant targets, although the lidar cannot continuously or accurately detect the center point of the target, the visible edge position of the target obtained by the lidar is relatively stable, and the original point cloud data at the edge of the target can be obtained, and then the point where the target is closest to the vehicle can be obtained. Therefore, the nearest point of the target detected by the lidar can be used as the target tracking point. Since the original point cloud data measured directly is used, the target tracking result will be more accurate and stable.

[0067] The target tracking method for laser radar can be applied to calculate the first candidate speed in the speed calculation method 1, such as Figure 3 As shown, calculating the first candidate speed according to the speed calculation method 1 includes:

[0068] S201: Determine the position of the target relative to the vehicle;

[0069] S202: Determine the target tracking point according to the position of the target relative to the vehicle:

[0070] When it is determined that the target is located in an area outside the second boundary with the vehicle as the symmetric center, the nearest point of the target detected by the laser radar is used as the target tracking point; or, when it is determined that the target is located in an area within the second boundary with the vehicle as the symmetric center, the center point of the target detected by the laser radar is used as the target tracking point;

[0071] S203: Obtain a first candidate speed based on the target tracking point estimation.

[0072] In specific implementation, "outside the second boundary" can be set to include the second boundary itself, and "inside the second boundary" can be set to not include the second boundary itself; or, "inside the second boundary" can be set to include the second boundary itself, and "outside the second boundary" can be set to not include the second boundary itself.

[0073] In specific implementation, when it is determined based on the basic characteristics, calibration data, test data, etc. of the laser radar that when the target is located in an area within a certain boundary, the laser radar can continuously and accurately detect the center point of the target. Conversely, when the target is located in an area outside a certain boundary, the laser radar cannot continuously or accurately detect the center point of the target. In this case, the boundary can be set as the second boundary in the above embodiment.

[0074] For a specific vehicle and lidar, the second boundary changes dynamically as the vehicle moves, but the relative position of the second boundary and the vehicle is relatively stable. Therefore, the second boundary can be determined in advance in the vehicle coordinate system based on the basic characteristics of the lidar, calibration data, test data, etc.

[0075] When the target is located in the area within the second boundary, the laser radar can continuously and accurately detect the center point of the target. In this case, the center point of the target detected by the laser radar is used as the target tracking point; when the target is located in the area outside the second boundary, the laser radar cannot continuously or accurately detect the center point of the target. In this case, the nearest point of the target detected by the laser radar is used as the target tracking point.

[0076] As an optional embodiment, the second boundary may be a closed boundary, for example, the second boundary may be a closed boundary with the origin of the vehicle coordinate system as the symmetry center, such as a circular closed boundary, an elliptical closed boundary, a rectangular closed boundary or a closed boundary of other shapes.

[0077] As an optional embodiment, the second boundary may also be a non-closed boundary. For example, the second boundary may also be two linear boundaries with the origin of the vehicle coordinate system as the center of symmetry. The area between the two linear boundaries is the area within the second boundary, and the area outside the two linear boundaries is the area outside the second boundary.

[0078] For example, the two linear boundaries corresponding to the second boundary may be a third linear boundary 3 and a fourth linear boundary 4 respectively located in front of and behind the vehicle 100 and perpendicular to the driving direction of the vehicle, see Figure 6 As shown, a vehicle coordinate system is established with the rear axle center of the vehicle 100 as the origin, and the x-axis and y-axis represent the longitudinal axis and the transverse axis of the vehicle coordinate system, respectively. For example, the third straight line boundary 3 is a straight line boundary located 2m in front of the vehicle (i.e. +2m on the x-axis and parallel to the y-axis), and the fourth straight line boundary 4 is a straight line boundary located 2m behind the vehicle (i.e. -2m on the x-axis and parallel to the y-axis).

[0079] In specific implementation, the third straight line boundary 3 and the fourth straight line boundary 4 are determined based on the stability of the laser radar setting the tracking point for the target. For example, according to the basic characteristics of the laser radar, calibration data and a large amount of test data, when the target is located in the area within 2m in front of or behind the vehicle, the laser radar can continuously and accurately detect the center point of the target, and then the target tracking result using the center point as the tracking point is also continuous and accurate, while when the target is located in the area beyond 2m in front of or behind the vehicle, the laser radar cannot continuously or accurately detect the center point of the target, and then the target tracking result using the center point as the tracking point is not continuous and accurate enough. In this case, the third straight line boundary 3 and the fourth straight line boundary 4 can be set to be located 2m in front of the vehicle and 2m behind the vehicle, respectively.

[0080] for Figure 6 In the scenario shown, it is considered that the target may move back and forth near the second boundary (third straight boundary 3, fourth straight boundary 4) (for example, the target vehicle suddenly decelerates, accelerates, changes lanes, etc.), and according to the aforementioned target tracking method for lidar, whenever the target moves from an area outside or inside the second boundary to an area inside or outside the second boundary, the tracking point needs to be switched. However, each time the tracking point is switched, the Kalman filter needs to be reinitialized, which affects the tracking effect and tracking efficiency to a certain extent.

[0081] In order to avoid the situation where the tracking point switches back and forth at the boundary, an embodiment of the present invention provides a more optimized target tracking method for laser radar, including:

[0082] When it is determined that the target enters the area between the third straight line boundary 3 and the fourth straight line boundary 4 from the area outside the third straight line boundary 3 in front of the vehicle or from the area outside the fourth straight line boundary 4 behind the vehicle, and then enters the area outside the fourth straight line boundary 4 behind the vehicle or enters the area outside the third straight line boundary 3 in front of the vehicle, when the target has not left the area between the third straight line boundary 3 and the fourth straight line boundary 4, the target tracking point before entering the area is still used. When the target leaves the area, the target tracking point is updated to the nearest point of the target currently detected by the lidar.

[0083] For details, see Figure 6 As shown, the area outside the third straight line boundary 3 is the first area, the area between the third straight line boundary 3 and the fourth straight line boundary 4 is the second area, and the area outside the fourth straight line boundary 4 is the third area.

[0084] For the target 200 labeled with id#0, when the target 200 is generated as a target (i.e., as a valid obstacle target for subsequent tracking), since the target is located in the third area, the tracking point is selected as the nearest point of the target detected by the lidar, i.e., the dot marked at the top of the rectangular block corresponding to id#0 in the figure.

[0085] For the target 200 labeled with id#1, when the target 200 is generated as a target, since the target is located in the first area, the tracking point is also selected as the nearest point of the target detected by the laser radar, that is, the dot marked at the top of the rectangular block corresponding to id#1 in the figure.

[0086] For the target 200 labeled as id#2, during the tracking process, the target enters the second area from the third area and then the first area. In the third area, the target tracking point is the nearest point of the target detected by the laser radar, that is, the dot marked at the top of the rectangular block corresponding to id#2 in the figure (for example, the center point of the front of the target vehicle); when crossing the fourth straight line boundary 4 to enter the second area, but not reaching the third straight line boundary 3, the target tracking point is still the nearest point of the target detected by the laser radar (the dot marked at the top of the rectangular block corresponding to id#2 in the figure); when the target crosses the third straight line boundary 3 and enters the first area, the target tracking point is switched to the nearest point of the target detected by the current laser radar, that is, the dot marked at the bottom of the rectangular block corresponding to id#2 in the figure (for example, the center point of the rear of the target vehicle); if the target is to enter the second area again and then enter the third area, the tracking point is switched at the fourth straight line boundary 4. By setting in this way, the situation where the target moves back and forth near the third straight line boundary 3 or the fourth straight line boundary 4 but does not completely enter other areas can be filtered out, thereby avoiding the switching of tracking points back and forth caused by this, and ensuring tracking efficiency and accuracy.

[0087] As an optional embodiment, the first candidate speed is estimated based on the target tracking point in the above embodiment, which can be obtained by using the point cloud data of the target tracking point updated in real time by the laser radar, and / or the target position and speed information updated in real time by the millimeter wave radar, and using the Kalman filter to estimate the first candidate speed. This step can adopt the multi-sensor information fusion technology and target tracking technology commonly used in the field, and the embodiments of the present invention will not be repeated.

[0088] For the target within the first boundary, the second candidate speed is calculated according to the speed calculation method 2, such as Figure 4 As shown, specifically including:

[0089] S301: Calculate the moving speed of each target frame corner point and target tracking point using the position data of the target frame corner point and the target tracking point obtained by the laser radar;

[0090] S302: Estimate a second candidate speed according to the moving speed of each target frame corner point and the target tracking point.

[0091] like Figure 8 As shown, in the ego-vehicle coordinate system of the ego-vehicle 100, the target 200 (gray rectangular box) is located in the area outside the second boundary, and the target tracking point is set to the closest point of the target detected by the laser radar; the target frame 7 detected by the laser radar for the target 200 is a rectangular frame shown by a dotted line, with four target frame corner points 5 at the upper left, lower left, upper right, and lower right, and the dot at the center of the bottom edge of the rectangular frame is the target tracking point 6. If the target 200 is located in the area within the second boundary, the target tracking point 6 is the target center point detected by the laser radar.

[0092] like Fig. 9 As shown, the laser radar is recorded at multiple consecutive moments ( Fig. 9 The position data of each target box corner point 5 (upper left corner point, upper right corner point, lower left corner point, lower right corner point) and the position data of the target tracking point 6 are obtained for 5 consecutive time stamps.

[0093] As an optional embodiment, step S301 may use the following uniform linear motion model to calculate the moving speed of the target frame corner point and the target tracking point at each time stamp according to the position data of the target frame corner point and the target tracking point at multiple consecutive time stamps:

[0094]

[0095]

[0096] Wherein, N is an integer greater than or equal to 2, indicating the Nth timestamp;

[0097] Vx is the X-axis velocity component of the target frame corner point\target tracking point; Vy is the Y-axis velocity component of the target frame corner point\target tracking point; N The X-axis position data of the target frame corner point\target tracking point at the Nth timestamp; Y N The Y-axis position data of the target frame corner point\target tracking point at the Nth timestamp; T N The time of the Nth timestamp;

[0098] X0 is the X-axis position data of the target box corner point\target tracking point collected at the first timestamp among N consecutive timestamps; Y0 is the Y-axis position data of the target box corner point\target tracking point collected at the first timestamp among N consecutive timestamps; T0 is the time corresponding to the first timestamp of N consecutive timestamps.

[0099] In the above uniform linear motion model, data collected at least at three consecutive time stamps (N is an integer greater than or equal to 2) are used for calculation, so that the adverse effects of the same system errors in data collected at adjacent time stamps can be filtered out, and more accurate results can be obtained.

[0100] In specific implementation, the target frame 7 obtained by the laser radar may also be a non-rectangular frame, and its corner points are certainly not limited to the above four corner points. Optionally, several target frame corner points may be arbitrarily selected from all target frame corner points for calculating the second selected speed.

[0101] As an optional embodiment, step S302 may be to determine at least one of the moving speeds of each target frame corner point and the target tracking point as the second selected speed, such as determining the moving speed of the four corner points and the target tracking point that best reflects the current speed of the target as the second selected speed.

[0102] Optionally, step S302 determines the second candidate speed of the current timestamp that is closest to the moving speed of the target frame corner point and the target tracking point at the current timestamp as the second candidate speed of the current timestamp; wherein the second candidate speed of the initial timestamp is the speed of the target when the target is newly established. This step adopts an iterative calculation method to select the second candidate speed of the current timestamp that is closest to the moving speed of the target frame corner point and the target tracking point at the current timestamp as the second candidate speed of the current timestamp. This method can filter the speed estimation deviation caused by the jump of the target frame corner point detected by the laser radar and obtain a smoother speed estimation value.

[0103] like Fig. 9 As shown in the figure, after storing the position data of the target frame corner points and target tracking points obtained by the laser radar for 5 consecutive moments, the moving speed (including the speed component of the X axis and the speed component of the Y axis) of each target frame corner point and target tracking point at the time T-1 is calculated using the uniform linear motion model:

[0104] Moving speed of target tracking point: vxrel_0,vyrel_0

[0105] The movement speed of the upper left corner: vxrel_1,vyrel_1

[0106] The movement speed of the upper right corner: vxrel_2,vyrel_2

[0107] The movement speed of the lower left corner: vxrel_3,vyrel_3

[0108] Movement speed of the lower right corner: vxrel_4,vyrel_4

[0109] After that, a group of speeds is selected from the above five groups of speed data corresponding to the time T-1 as the second candidate speed at the time T, such as the second candidate speed (vxrel, vyrel) closest to the time T-1 as the second candidate speed (vxrel', vyrel') at the time T. The second candidate speed at the time T-1 is determined based on the second candidate speed at the time T-2 according to the above method, that is, the second candidate speed at each moment is determined by iteration. Among them, the second candidate speed of the initial timestamp is the speed component of the X-axis and Y-axis when the target is established as a new target, which can also be calculated according to the uniform linear motion model, which will not be repeated here.

[0110] In an embodiment of the present invention, a newly established target means that for a certain virtual obstacle target, based on multiple associated sensor measurement values ​​(such as lidar, millimeter-wave radar or visual sensor measurement data) obtained, when it is determined that the virtual obstacle target meets the requirements of the new target generation rule (such as the lateral or longitudinal distance from the vehicle reaches a safety threshold), a new target is established using the obtained measurement information.

[0111] The embodiment of the present invention adopts a target speed calculation method based on the tracking points determined by the laser radar, and a target speed calculation method based on the displacement information of the corner points and center points of the target frame measured by the laser radar, respectively, to estimate the speed of the obstacle target, and then select a speed estimation result as the final estimated speed of the target according to the different positions of the obstacle target relative to the vehicle. By selecting the output target speed, the calculated target speed can be guaranteed to have high accuracy, overcoming the technical problem that the laser radar's target speed estimation is greatly affected by the position of the target, resulting in inaccurate estimation. In addition, by utilizing the characteristics of the laser radar's 360-degree full coverage scanning of the surrounding area of ​​the vehicle, the embodiment of the present invention solves the blind spot coverage problem of the millimeter wave radar from the hardware configuration.

[0112] Optionally, in an embodiment of the present invention, different strategies are adopted for estimating the speed of the newly established target according to the associated sensor measurement category. For example, when the measurement value associated with the virtual target includes lidar measurement, the displacement and time change of the virtual target are used to estimate the speed of the newly established target; and when the measurement value associated with the virtual target includes millimeter-wave radar measurement, the speed measurement value of the millimeter-wave radar can also be used as the speed estimate of the newly established target.

[0113] See also Fig.10 As shown, in a second aspect of an embodiment of the present invention, a target speed estimation device is provided, comprising:

[0114] A first speed calculation unit, used for calculating a first speed to be selected according to a speed calculation method 1;

[0115] A second speed calculation unit, used for calculating a second selected speed according to a second speed calculation method;

[0116] A speed determination unit, used to determine that when the target is located outside a first boundary with the vehicle as the symmetric center, the first candidate speed is determined as the target speed, and vice versa, the second candidate speed is determined as the target speed;

[0117] Wherein, the first speed calculation unit includes:

[0118] The tracking point determination module is used to determine that when the target is located outside the second boundary with the vehicle as the symmetric center, the nearest point of the target detected by the laser radar is used as the target tracking point; otherwise, the center point of the target detected by the laser radar is used as the target tracking point;

[0119] A first speed estimation module, used for estimating a first candidate speed based on the target tracking point;

[0120] Wherein, the second speed calculation unit includes:

[0121] The target point speed calculation module is used to calculate the moving speed of each target frame corner point and target tracking point using the displacement information of the target frame corner point and target tracking point obtained by the laser radar;

[0122] The second speed estimation module is used to estimate a second candidate speed according to the moving speed of each target frame corner point and the target tracking point.

[0123] The target speed estimation device of the embodiment of the present invention and its method for estimating the target speed can specifically adopt the target speed estimation technology disclosed in the target speed estimation method described in the first aspect of the embodiment of the present invention to estimate the obstacle target speed. This will not be repeated here. Please refer to the content of the target speed estimation method disclosed in the target speed estimation method of the first aspect of the embodiment of the present invention.

[0124] In a third aspect of an embodiment of the present invention, a control device is provided, comprising at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the target speed estimation method described in the first aspect of the embodiment of the present invention to estimate the target speed of an obstacle, and the target speed estimation method will not be described in detail.

[0125] In a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, comprising a program or an instruction. When the program or the instruction is executed on a computer, the target speed estimation method described in the first aspect of the embodiment of the present invention is implemented to estimate the obstacle target speed. The target speed estimation method will not be described in detail.

[0126] In a fifth aspect of an embodiment of the present invention, a computer program product comprising instructions is provided. When the computer program product is run on a computer, the computer executes the target speed estimation method described in the first aspect of the embodiment of the present invention to estimate the obstacle target speed. The target speed estimation method is not described in detail here.

[0127] According to a sixth aspect of the embodiments of the present invention, a vehicle is provided, comprising the control device as described in the third aspect of the embodiments of the present invention.

[0128] The present invention adopts two different speed estimation methods to estimate the speed of the obstacle target respectively, and then selects a speed estimation result as the final estimated speed of the target according to the different positions of the obstacle target relative to the vehicle, thereby overcoming the problem that the autonomous driving vehicle cannot obtain the speed information of the target in the blind spot of the millimeter wave radar, and overcoming the problem that the laser radar measurement result is greatly affected by the position, thereby improving the accuracy of the target speed estimation.

[0129] The above is only a preferred embodiment of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A target speed estimation method, characterized in that: The following steps are involved: Calculate a first candidate speed and a second candidate speed according to the speed calculation method 1 and the speed calculation method 2 respectively; When it is determined that the target is located outside the first boundary with the vehicle as the symmetric center, the first candidate speed obtained according to the speed calculation method 1 is determined as the speed of the target; When it is determined that the target is located within the first boundary with the vehicle as the symmetric center, the second candidate speed obtained according to the speed calculation method 2 is determined as the speed of the target; The first candidate speed is calculated according to the speed calculation method 1, including: When it is determined that the target is located in an area outside the second boundary with the vehicle as the symmetric center, the nearest point of the target detected by the laser radar is used as the target tracking point; or, when it is determined that the target is located in an area within the second boundary with the vehicle as the symmetric center, the center point of the target detected by the laser radar is used as the target tracking point; Obtaining a first candidate speed based on the target tracking point estimation; The second selected speed is calculated according to the second speed calculation method, including: Using the position data of the target frame corner points and target tracking points obtained by the laser radar, the moving speed of each target frame corner point and target tracking point is calculated; A second candidate speed is estimated according to the moving speed of each target frame corner point and the target tracking point.

2. The target speed estimation method according to claim 1, characterized in that: The first boundary is a circular closed boundary, an elliptical closed boundary or a rectangular closed boundary with the origin of the vehicle coordinate system as the symmetry center.

3. The target speed estimation method according to claim 1, characterized in that: The first boundary includes two linear boundaries with the origin of the vehicle coordinate system as the symmetry center.

4. The target speed estimation method according to claim 3, characterized in that: The two linear boundaries include a first linear boundary and a second linear boundary which are respectively located in front of and behind the vehicle and are perpendicular to the driving direction of the vehicle.

5. The target speed estimation method according to claim 1, characterized in that: The second boundary is a circular closed boundary, an elliptical closed boundary or a rectangular closed boundary with the origin of the vehicle coordinate system as the symmetry center.

6. The target speed estimation method according to claim 1, characterized in that: The second boundary includes two linear boundaries with the origin of the vehicle coordinate system as the symmetry center.

7. The target speed estimation method according to claim 6, characterized in that: The two linear boundaries include a third linear boundary and a fourth linear boundary respectively located in front of and behind the vehicle and perpendicular to the driving direction of the vehicle.

8. The target speed estimation method according to claim 7, characterized in that: When it is determined that the target is located in an area outside the second boundary with the vehicle as the symmetric center, the nearest point of the target detected by the laser radar is used as the target tracking point; or when it is determined that the target is located in an area within the second boundary with the vehicle as the symmetric center, the center point of the target detected by the laser radar is used as the target tracking point, including: It is determined that the target enters the area between the third straight line boundary and the fourth straight line boundary from the area outside the third straight line boundary in front of the vehicle or from the area outside the fourth straight line boundary behind the vehicle, and then enters the area outside the fourth straight line boundary behind the vehicle or enters the area outside the third straight line boundary in front of the vehicle; then, When the target has not left the area between the third straight line boundary and the fourth straight line boundary, the target tracking point before entering the area is still used; when the target leaves the area, the target tracking point is updated to the nearest point of the target currently detected by the laser radar.

9. The target speed estimation method according to claim 1, characterized in that: The first candidate speed is estimated based on the target tracking point, including: The first candidate speed is estimated by using the point cloud data of the target tracking points updated in real time by the laser radar and / or the target position and speed information updated in real time by the millimeter wave radar, and applying the Kalman filter.

10. The target speed estimation method according to claim 1, characterized in that: Using the position data of the target frame corner points and target tracking points obtained by the laser radar, the moving speed of each target frame corner point and target tracking point is calculated, including: Using the uniform linear motion model, according to the position data of the target frame corner points and the target tracking points at multiple consecutive time stamps, the moving speed of the target frame corner points and the target tracking points at each time stamp is calculated.

11. The target speed estimation method according to claim 10, characterized in that: The multiple consecutive time stamps are at least 3 consecutive time stamps.

12. The target speed estimation method according to claim 1, characterized in that: Estimating a second candidate speed according to the moving speed of each target frame corner point and the target tracking point includes: At least one of the moving speeds of each target frame corner point and the target tracking point is determined as a second selected speed.

13. The target speed estimation method according to claim 12, characterized in that: Determining at least one of the moving speeds of each target frame corner point and the target tracking point as a second selected speed includes: The second candidate speed of the current timestamp is determined as the second candidate speed of the current timestamp, among the moving speeds of the target frame corner point and the target tracking point at the current timestamp, wherein the second candidate speed of the initial timestamp is the speed of the target when the target is newly established.

14. A target speed estimation device, characterized in that: include: A first speed calculation unit, used for calculating a first speed to be selected according to a speed calculation method 1; A second speed calculation unit, used for calculating a second selected speed according to a second speed calculation method; A speed determination unit, used to determine that when the target is located outside a first boundary with the vehicle as the symmetric center, the first candidate speed is determined as the target speed, and vice versa, the second candidate speed is determined as the target speed; Wherein, the first speed calculation unit includes: The tracking point determination module is used to determine that when the target is located outside the second boundary with the vehicle as the symmetric center, the nearest point of the target detected by the laser radar is used as the target tracking point; otherwise, the center point of the target detected by the laser radar is used as the target tracking point; A first speed estimation module, used for estimating a first candidate speed based on the target tracking point; Wherein, the second speed calculation unit includes: The target point speed calculation module is used to calculate the moving speed of each target frame corner point and target tracking point using the position data of the target frame corner point and the target tracking point obtained by the laser radar; The second speed estimation module is used to estimate a second candidate speed according to the moving speed of each target frame corner point and the target tracking point.

15. A control device, characterized in that: It comprises at least one processor; and a memory which is communicatively connected to the at least one processor; wherein the memory stores instructions which can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the target speed estimation method as described in any one of claims 1 to 13.

16. A computer-readable storage medium, characterized in that: The method comprises a program or an instruction, and when the program or the instruction is executed on a computer, the target speed estimation method according to any one of claims 1 to 13 is implemented.

17. A computer program product comprising instructions, characterized in that When the computer program product is run on a computer, the computer is enabled to execute the target speed estimation method according to any one of claims 1 to 13.

18. A vehicle, characterized in that: Comprising a control device as claimed in claim 15.

Images