Unsupported calibration method for vehicle blind spot monitoring radar

Abstract

The present invention relates to the technical field of vehicle-mounted radar systems, and in particular to a method for calibration of a vehicle blind spot monitoring radar without support. The steps are as follows: after the vehicle is started, the radars located on the left and right sides of the rear of the vehicle are also started synchronously, and the radars on both sides complete self-test. The radars on both sides check the flag information stored in a specific area of ​​their own flash to determine whether the radar on this side has completed the installation angle calibration. If the calibration is not completed, the calibration function is started, and the early warning function is turned off. After the calibration is completed, the calibration result is written into a specific area of ​​the flash, and the flag is changed at the same time; if the calibration has been completed, the calibration function is started, and the early warning function is started. After the calibration is completed, the existing calibration result is corrected according to the new calibration result. The method proposed by the present invention does not require a specific site and equipment, does not require a specific reference target, and does not require information such as vehicle speed. The calibration is fast and accurate, and has anti-interference ability and fault tolerance mechanism.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle-mounted radar systems, and in particular to a method for independently calibrating a vehicle blind spot monitoring radar. Background Art

[0002] With the continuous progress of the semiconductor industry and the electronics industry, many electronic devices have become increasingly miniaturized and their performance has become increasingly powerful. As a result, vehicles are equipped with more sensors to enhance their perception of the vehicle's surrounding environment, making them more intelligent. Blind spot monitoring radar can monitor targets in the driver's blind spot and alert the driver to reduce the risk of collision. Especially at night or in rainy and foggy weather, when the driver's vision is not clear enough, blind spot monitoring radar can especially improve driving safety, and therefore has been increasingly used.

[0003] During the installation process of the blind spot monitoring radar, the angle may deviate from the preset installation angle due to various factors such as the processing error of the mounting bracket and the operation error of the installer. During use, it may also change from the initial value due to the impact of vehicle bumps and vibrations. Therefore, it is necessary to design a simple and efficient calibration method.

[0004] Among the related technologies that have been disclosed, for example, the patent with application publication number CN 110703212 A requires specific sites or equipment for matching calibration, and therefore has certain limitations. The patent with application publication number CN 111190150 B requires obtaining vehicle body information, such as vehicle speed, etc., but there are certain specific models and specific customers that cannot obtain vehicle speed, so there are also certain limitations. Although the patent with application publication number CN 113985370 A does not require matching site equipment and vehicle body information, it requires the car to travel in a straight line. At the same time, it does not fully utilize the association between radar data frames and frames, and does not utilize the information sharing between the left radar and the right radar. There is still room for improvement in fault tolerance mechanism and calibration efficiency. Summary of the invention

[0005] In view of the problems existing in the background technology, a vehicle blind spot monitoring radar independent calibration method is proposed. The method proposed by the present invention does not require a specific site and equipment, does not require a specific reference target, and does not require information such as vehicle speed. The calibration is fast and accurate, and has anti-interference ability and fault tolerance mechanism.

[0006] The present invention proposes a vehicle blind spot monitoring radar independent calibration method, the method steps are as follows:

[0007] S1. After the vehicle is started, the radars on the left and right sides of the rear of the vehicle are started synchronously, and the radars on both sides complete self-test;

[0008] S2. The radars on both sides check the flag information stored in the specific area of ​​their own flash to determine whether the radar on this side has completed the installation angle calibration; if the calibration is not completed, the calibration function is started and the warning function is turned off. After the calibration is completed, the calibration result is written into the specific area of ​​the flash and the flag is changed at the same time; if the calibration is completed, the calibration function and the warning function are started. After the calibration is completed, the existing calibration result is corrected according to the new calibration result;

[0009] S3. Under the calibration function, the radars on the left and right sides of the vehicle respectively record the detection point information measured by each of them, including three dimensional information of distance, angle, and Doppler velocity; the radars on the left and right sides periodically send radar signals to detect targets, and the radar signal sending period is T. The detection point information recorded by the radars on the left and right sides is updated every T time;

[0010] S4. After receiving the detection point information each time, the radars on the left and right sides respectively calculate the two-dimensional estimation matrix of the installation angle vs. the vehicle speed. The value of each element in the matrix represents the estimation probability, the row coordinate of each element represents the estimated value of the installation angle, and the column coordinate of each element represents the estimated value of the vehicle speed.

[0011] S5. If there is only one element with the largest value in the two-dimensional estimation matrix, select the estimated probability of the element and its row and column coordinates; if there are multiple elements with the largest value and their positions are continuous, select the element in the middle position; if there are multiple elements with the largest value but their positions are not continuous, select the estimated probabilities and row and column coordinates of all the elements with the largest value; convert the row and column coordinates of the selected element into the estimated value of the installation angle and the estimated value of the vehicle speed, and the conversion method is as follows:

[0012]

[0013] V est =y*N-0.5*N;

[0014] in, Indicates the estimated value of the installation angle, V est represents the estimated value of the vehicle speed, x represents the row coordinate of the two-dimensional estimation array, y represents the column coordinate of the two-dimensional estimation array, K represents the angle step, and N represents the speed step;

[0015] S6, sending the selected element set to a multi-channel tracker, which performs tracking filtering, new creation, deletion and other processing on the estimated result, so that the estimated result is closer and closer to the true value;

[0016] S7, after the above multi-channel tracker tracking process is completed, the left radar and the right radar respectively select the tracker with the highest confidence in their own multi-channel trackers, and then share the relevant information of the selected trackers through the serial port interaction; if the vehicle speed tracking values ​​of the left radar and the right radar are close, the confidence of the tracker is increased by 1;

[0017] S8. Select the tracker with the highest confidence among the multiple trackers. If the confidence is greater than the set threshold, the installation angle calibration of the radar on this side is completed.

[0018] Preferably, the specific steps of S4 are as follows:

[0019] S41, filtering the detection points with negative Doppler velocities and retaining the detection points with positive Doppler velocities;

[0020] S42, the Doppler velocity received by the radar, the vehicle speed, and the angle between the target and the axial direction of the vehicle body are converted using trigonometric formulas;

[0021] v d =v x *cos(θ)+v y *sin(θ);

[0022] where v d is the Doppler velocity, v x is the vehicle velocity component in the axial direction of the vehicle body, v y is the vehicle velocity component in the transverse direction of the vehicle body, θ is the angle between the target and the axial direction of the vehicle;

[0023] S43, using an angle step of K degrees and a speed step of N kilometers per hour to construct a two-dimensional estimation matrix of installation angle vs. vehicle speed, wherein the row coordinates of the two-dimensional estimation matrix represent the estimated value of the installation angle of the radar, the column coordinates represent the estimated value of the driving speed of the vehicle, and the value of each element represents the estimated probability;

[0024] P=N b / N s ;

[0025] Among them, P represents the estimated probability, N b Indicates the total number of detection points that meet the current matrix angle constraint K degrees and speed constraint N kilometers / hour, N s Indicates the total number of detection points remaining after filtering.

[0026] Preferably, the detection point retained in S41 is a stationary target or other moving target far away from the vehicle; the speed of the stationary target relative to the axial direction of the vehicle body is approximately equal to the driving speed of the vehicle, and the speeds of other moving targets relative to the axial direction of the vehicle are usually different.

[0027] Preferably, in S5, each selected element is represented by three information: estimated probability, estimated value of installation angle and estimated value of vehicle speed; the element selection method is used in combination with a window function, or in combination with a two-dimensional CFAR method.

[0028] Preferably, the specific operation steps of S6 are as follows:

[0029] S61, searching whether the next tracker is in an activated state, if the tracker is in an activated state, executing the next step; if the tracker is in an idle state, initializing the tracker, and jumping to S65;

[0030] S62, matching the installation angle tracking value and the vehicle speed tracking value of the road tracker with the elements in the element set selected in S5 one by one, and selecting elements that simultaneously meet the following three conditions;

[0031] Condition 1: |V track -V est |<V threshold ;

[0032] Condition 2:

[0033] Condition 3: P < P track ;

[0034] Among them, V track Represents the vehicle speed tracking value of the tracker, V est Represents the estimated value of the vehicle speed of this element, Vt hreshold Indicates the allowable threshold of vehicle speed estimation error; Indicates the installation angle tracking value of the tracker. Indicates the estimated value of the installation angle of the element; represents the angle estimation error allowable threshold; P represents the estimated probability of the element, P track Indicates the probability threshold for allowing tracking;

[0035] S63, if the road tracker matches an element that meets the conditions, then select the element with the smallest installation angle matching error to update the tracker parameters; the installation angle tracking value and vehicle speed tracking value of the tracker are updated using a Kalman filter, taking into account the dynamic change of the vehicle speed and the smooth approximation of the installation angle; the tracker's confidence +1, decay degree = 0, age +1, state flag = 1; at the same time, delete the element selected by the tracker from the element set;

[0036] S64, if the road tracker does not match an element that meets the conditions, the installation angle tracking value and the vehicle speed tracking value of the tracker remain unchanged, the intrinsic parameters of the Kalman filter are updated automatically, the confidence of the tracker is -1, the attenuation is +1, the age is +1, and the state flag is 1; check whether the attenuation of the tracker is greater than the deletion threshold, if it is greater than the deletion threshold, initialize the road tracker and put the road tracker in an idle state;

[0037] S65, check whether all trackers have been updated, if not all have been updated, execute from S61, if all have been updated, proceed to the next step;

[0038] S67. Check whether there are any remaining elements in the element set. If not, proceed to the next step. If there are any remaining elements, determine whether the following two conditions are met at the same time:

[0039] Condition 1: P>P setup ;

[0040] Condition 2: There are idle trackers in the multi-channel trackers;

[0041] If these two conditions are met at the same time, an idle tracker is activated, and the installation angle estimation value and vehicle speed estimation value of the element are used as the installation angle tracking value and vehicle speed tracking value of the tracker. The tracker's confidence = 1, attenuation = 0, age = 1, and status flag = 1.

[0042] Preferably, the allowable threshold of vehicle speed estimation error in S62 should not only consider the estimation error, but also the acceleration and deceleration capabilities of the vehicle. In each radar test cycle, the maximum change in vehicle speed is: ΔV=a*T, a is the maximum acceleration or maximum deceleration of the vehicle, and T is the radar signal sending period.

[0043] Compared with the prior art, the present invention has the following beneficial technical effects:

[0044] The method steps in the present invention are as follows: after the vehicle is started, the radars located on the left and right sides of the rear of the vehicle are also started synchronously, and the radars on both sides complete self-test. The radars on both sides check the flag information stored in the specific area of ​​their own flash to determine whether the radar on this side has completed the installation angle calibration. If the calibration is not completed, the calibration function is started, and the early warning function is turned off. After the calibration is completed, the calibration result is written into the specific area of ​​the flash, and the flag is changed at the same time; if the calibration has been completed, the calibration function is started and the early warning function is started. After the calibration is completed, the existing calibration result is corrected according to the new calibration result. The method proposed by the present invention does not require a specific site and equipment, does not require a specific reference target, and does not require information such as vehicle speed. The calibration is fast and accurate, and has anti-interference ability and fault tolerance mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a calibration flow chart of the radar in the present invention;

[0046] Figure 2 is a flow chart of the calibration method in the present invention;

[0047] Figure 3 It is a schematic diagram of the position relationship between the radar and the vehicle in the present invention;

[0048] Figure 4 is a two-dimensional estimation matrix diagram of installation angle vs. vehicle speed in the present invention;

[0049] Figure 5 It is the installation angle tracking flow chart of the present invention;

[0050] Figure 6 It is a flow chart of mutual verification between left and right radars in the present invention. DETAILED DESCRIPTION

[0051] The vehicle blind spot monitoring radar independent calibration method proposed by the present invention has the following steps:

[0052] 1. After the vehicle is started, the radars on the left and right sides of the rear of the vehicle are also started synchronously, and the radars on both sides complete self-test. If the self-test is normal, the device indicator light flashes twice and the next step is executed; if the self-test is abnormal, the device indicator light is always on, indicating that the device is faulty and the subsequent steps will not be continued. The device needs to be checked to ensure that it can work normally.

[0053] 2. If Figure 1 In the process shown, the radars on both sides check the flag information stored in the specific area of ​​their own flash to determine whether the radar on this side has completed the installation angle calibration. If the calibration is not completed, the calibration function is started, the early warning function is turned off, and the initial calibration of the radar is performed. After the calibration is completed, the calibration result is written to the specific area of ​​the flash, and the flag information is changed to calibrated. If the calibration has been completed, both the calibration function and the early warning function are started. After the calibration is completed, the existing calibration result is corrected according to the new calibration result.

[0054] In this implementation case, the termination condition of the initial calibration is that the maximum confidence of the multi-channel tracker is greater than 600; the termination condition of the non-initial calibration is that the maximum confidence of the multi-channel tracker is greater than 1200. This is because the early warning function is turned off during the initial calibration, and it is hoped that the calibration can be completed quickly so that the early warning function can be turned on as soon as possible; the non-initial calibration is calibrated while the normal early warning is issued. At this time, the purpose of calibration is to correct the previous calibration results, hoping that the new calibration results will be more accurate, so a longer time is used for more accurate calibration.

[0055] In this implementation case, the calibration result is corrected in the following way when it is not the first calibration: if the difference between the new calibration result and the original calibration result is less than 0.5 degrees, the original result is used; otherwise, the new calibration result is used. The purpose of this is to reduce the number of flash writes, because the number of flash erases and writes is limited.

[0056] 3. Under the calibration function, the radars on the left and right sides of the vehicle respectively record the detection point information measured by each of them, including three dimensional information: distance, angle, and Doppler velocity. The radars on the left and right sides periodically send radar signals to detect targets. The transmission period of the radar signal is T, so the detection point information recorded by the radars on the left and right sides is updated every T time. Blind spot monitoring radars usually require that the transmission period of radar signals is less than 80 milliseconds. In this implementation case, the transmission period of radar signals is 50 milliseconds, that is, T = 50 milliseconds.

[0057] 4. After receiving the detection point information, the left and right radars calculate the two-dimensional estimation matrix of installation angle vs. vehicle speed. The value of each element in the matrix represents the estimated probability, the row coordinate of each element represents the estimated value of the installation angle, and the column coordinate of each element represents the estimated value of the vehicle speed. The three-dimensional plane projection of the two-dimensional estimation matrix is ​​as follows: Figure 4 As shown, the specific calculation steps are as follows:

[0058] 4.1. Filter out the detection points with negative Doppler speed and keep the detection points with positive Doppler speed. The detection points that are kept may be stationary targets or other moving targets far away from the vehicle. The speed of the stationary target relative to the axial direction of the vehicle is approximately equal to the vehicle's driving speed. The speed of other moving targets relative to the axial direction of the vehicle is usually different. The relationship between the axial direction and the transverse direction and the vehicle is as follows: Figure 3 shown.

[0059] 4.2. Due to the characteristics of the wheel structure of the car, the car moves forward and backward or turns, and cannot move laterally. Therefore, the Doppler velocity received by the radar, the speed of the car, and the angle between the target and the axial direction of the car body can be converted by trigonometric formulas.

[0060] v d =v x *cos(θ)+v y *sin(θ);

[0061] where v d is the Doppler velocity, v x is the vehicle velocity component in the axial direction of the vehicle body, v y is the vehicle velocity component in the transverse direction of the vehicle body, and θ is the angle between the target and the axial direction of the vehicle. When the vehicle is traveling in a straight line, the velocity v in the transverse direction of the vehicle body yClose to zero; when the vehicle turns, the body differential controls the inner tire and the outer tire to rotate at different speeds, thus generating a certain transverse speed during the turn, but the transverse speed v y Usually much smaller than the axial velocity v x , which can be solved by the multi-path tracker described in the subsequent step 6.

[0062] 4.3. An angle step of K degrees and a speed step of N kilometers per hour are used to construct a two-dimensional estimation matrix of installation angle vs. vehicle speed. The row coordinates of the two-dimensional estimation matrix represent the estimated value of the radar installation angle, the column coordinates represent the estimated value of the vehicle's driving speed, and the value of each element represents the estimated probability.

[0063] P=N b / N s ;

[0064] Among them, P represents the estimated probability, N b Indicates the total number of detection points that meet the current matrix angle constraint K degrees and speed constraint N kilometers / hour, N s Indicates the total number of detection points remaining after filtering. In this implementation case, the angle step K is 0.5 degrees and the speed step N is 2 kilometers per hour.

[0065] 5. If there is only one element with the largest value in the two-dimensional estimation matrix, select the estimated probability of the element and its row and column coordinates; if there are multiple elements with the largest value and their positions are continuous, select the element in the middle position; if there are multiple elements with the largest value but their positions are not continuous, select the estimated probability of all elements with the largest value and their row and column coordinates. Convert the row and column coordinates of the selected element into the estimated value of the installation angle and the estimated value of the vehicle speed in the following way:

[0066]

[0067] V est =y*N-0.5*N;

[0068] in, Indicates the estimated value of the installation angle, V est represents the estimated value of the vehicle speed, x represents the row coordinate of the two-dimensional estimation array, y represents the column coordinate of the two-dimensional estimation array, K represents the angle step, and N represents the speed step. Each selected element can be represented by three pieces of information: estimated probability, installation angle estimate, and vehicle speed estimate. In this implementation case, the angle step K is 0.5 degrees, and the speed step N is 2 kilometers per hour. It should be noted that the above element selection method can also be used in combination with a window function, or in combination with methods such as two-dimensional cFAR.

[0069] 6. Send the selected element set to the multi-channel tracker, which performs tracking filtering, new creation, deletion and other processing on the estimated results, so that the estimated results are closer and closer to the true value. The flowchart of the multi-channel tracker is as follows Figure 5 The specific process is as follows:

[0070] 6.1. Search whether the next tracker is in an active state. If the tracker is in an active state, execute the next step. If the tracker is in an idle state, initialize the tracker and jump to step 6.5. After the tracker is initialized, the confidence = 0, the decay = 0, the age = 0, and the state flag = 0.

[0071] 6.2. Match the installation angle tracking value and vehicle speed tracking value of the road tracker with the elements in the element set selected in step 5 above one by one, and select the elements that meet the following three conditions at the same time.

[0072] Condition 1: |V track -V est |<V threshold ;

[0073] Condition 2:

[0074] Condition 3: P < P track ;

[0075] Among them, V track Represents the vehicle speed tracking value of the tracker, V est Represents the estimated value of the vehicle speed of this element, V threshold Indicates the allowable threshold of vehicle speed estimation error; Indicates the installation angle tracking value of the tracker. represents the estimated value of the installation angle of the element, represents the angle estimation error allowable threshold; P represents the estimated probability of the element, P track Indicates the probability threshold for allowing tracking.

[0076] It should be noted that the vehicle speed estimation error allowable threshold should not only consider the estimation error, but also the acceleration and deceleration capabilities of the vehicle. In each radar test cycle, the maximum change in vehicle speed is: ΔV = a*T, a is the maximum acceleration or deceleration of the vehicle, and T is the radar signal transmission period. In this implementation case, the maximum acceleration or deceleration of the vehicle a = 5 m / s, the radar signal transmission period T = 50 ms, V threshold =5 km / h, P track =0.3.

[0077] 6.3. If the road tracker matches an element that meets the conditions, the element with the smallest installation angle matching error is selected to update the tracker parameters. The installation angle tracking value and vehicle speed tracking value of the tracker are updated using the Kalman filter, taking into account the dynamic changes of vehicle speed and the smooth approximation of the installation angle. The tracker's confidence +1, decay degree = 0, age +1, and status flag = 1. At the same time, the element selected by the tracker is deleted from the element set.

[0078] Among them, the installation angle tracking value filtering can use relatively simple filters such as α-β filters. However, for the vehicle speed tracking value filtering, it is still recommended to use the Kalman filter to take into account both dynamics and stability.

[0079] 6.4. If the road tracker does not match any qualified element, the installation angle tracking value and vehicle speed tracking value of the tracker remain unchanged, the intrinsic parameters of the Kalman filter are updated automatically, the tracker's confidence is -1, the attenuation is +1, the age is +1, and the status flag is = 1.

[0080] Check whether the attenuation of the tracker is greater than the deletion threshold. If it is greater than the deletion threshold, initialize the tracker and put it in an idle state. In this implementation case, the deletion threshold of the tracker is 10.

[0081] 6.5. Check whether all trackers have been updated. If not, proceed to step 6.1. If all have been updated, proceed to the next step.

[0082] 6.6. Check whether there are any remaining elements in the element set. If not, execute step 7. If there are any remaining elements, determine whether the following two conditions are met at the same time:

[0083] Condition 1: P>P setup ;

[0084] Condition 2: There are idle trackers in the multi-path trackers.

[0085] 6.7. If both conditions are met, an idle tracker is activated, and the estimated installation angle and vehicle speed of the element are used as the installation angle tracking value and vehicle speed tracking value of the tracker. The tracker's confidence = 1, decay = 0, age = 1, and status flag = 1. In this implementation case, P setup =0.6.

[0086] 7. After the above multi-channel tracking process is completed, the left radar and the right radar respectively select the tracker with the highest confidence in their own multi-channel trackers, and then share the relevant information of the selected tracker through the serial port. If the vehicle speed tracking values ​​of the left radar and the right radar are close, the confidence of the tracker is +1.

[0087] In this implementation case, the left radar and the right radar share information through the serial port, but in other design cases, other interfaces suitable for their own design can be selected to share information. If the difference between the vehicle speed tracking values ​​of the left radar and the right radar is less than 3.6 km / h, the tracker confidence is +1. The process of interactive verification is as follows Figure 6 shown.

[0088] 8. Select the one with the highest confidence in the multi-channel tracker. If the confidence is greater than the set threshold, the installation angle calibration of the radar on this side is completed. The results of multiple engineering measurements are shown in Table 1. The results of multiple calibrations are relatively stable and the error jitter is small. Table 1 is the test statistical data of the present invention in practical application.

[0089]

[0090]

[0091] Table 1

[0092] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A vehicle blind spot monitoring radar independent calibration method, characterized in that: The steps are as follows: S1. After the vehicle is started, the radars on the left and right sides of the rear of the vehicle are started synchronously, and the radars on both sides complete self-test; S2, the radars on both sides check the flag information stored in the specific area of ​​their own flash to determine whether the radar on this side has completed the installation angle calibration; If the calibration is not completed, the calibration function is started and the warning function is turned off. After the calibration is completed, the calibration result is written into the specific area of ​​the flash and the flag is changed at the same time. If the calibration is completed, both the calibration function and the warning function are started. After the calibration is completed, the existing calibration result is corrected according to the new calibration result. S3. Under the calibration function, the radars on the left and right sides of the vehicle respectively record the detection point information measured by each of them, including three dimensional information of distance, angle, and Doppler velocity; the radars on the left and right sides periodically send radar signals to detect targets, and the radar signal sending period is T. The detection point information recorded by the radars on the left and right sides is updated every T time; S4. After receiving the detection point information each time, the radars on the left and right sides respectively calculate a two-dimensional estimation matrix, where the value of each element in the matrix represents the estimation probability, the row coordinate of each element represents the estimated value of the installation angle, and the column coordinate of each element represents the estimated value of the vehicle speed; S5. If there is only one element with the largest value in the two-dimensional estimation matrix, select the estimated probability of the element and its row and column coordinates; if there are multiple elements with the largest value and their positions are continuous, select the element in the middle position; if there are multiple elements with the largest value but their positions are not continuous, select the estimated probabilities and row and column coordinates of all the elements with the largest value; convert the row and column coordinates of the selected element into the estimated value of the installation angle and the estimated value of the vehicle speed, and the conversion method is as follows: V est =y*N-0.5*N; in, Indicates the estimated value of the installation angle, V est represents the estimated value of the vehicle speed, x represents the row coordinate of the two-dimensional estimation array, y represents the column coordinate of the two-dimensional estimation array, K represents the angle step, and N represents the speed step; S6, sending the set of elements selected in S5 to a multi-channel tracker, which performs tracking filtering, new creation, and deletion processing on the estimated results, so that the estimated results are closer and closer to the true value; S7, after the above multi-channel tracker tracking process is completed, the left radar and the right radar respectively select the tracker with the highest confidence in their own multi-channel trackers, and then share the relevant information of the selected trackers through the serial port interaction; if the vehicle speed tracking values ​​of the left radar and the right radar are close, the confidence of the tracker is increased by 1; S8. Select the tracker with the highest confidence among the multiple trackers. If the confidence is greater than the set threshold, the installation angle calibration of the radar on this side is completed.

2. The vehicle blind spot monitoring radar independent calibration method according to claim 1, characterized in that: The specific steps of S4 are as follows: S41, filtering the detection points with negative Doppler velocities and retaining the detection points with positive Doppler velocities; S42, the Doppler velocity received by the radar, the vehicle speed, and the angle between the target and the axial direction of the vehicle body are converted using trigonometric formulas; v d =v x *cos(θ)+v y *sin(θ); where v d is the Doppler velocity, v x is the vehicle velocity component in the axial direction of the vehicle body, v y is the vehicle velocity component in the transverse direction of the vehicle body, θ is the angle between the target and the axial direction of the vehicle; S43, constructing a two-dimensional estimation matrix with an angle step of K degrees and a speed step of N kilometers per hour, wherein the row coordinates of the two-dimensional estimation matrix represent the estimated value of the installation angle of the radar, the column coordinates represent the estimated value of the driving speed of the vehicle, and the value of each element represents the estimated probability; P=N b / N s ; Among them, P represents the estimated probability, N b Indicates the total number of detection points that meet the current matrix angle constraint K degrees and speed constraint N kilometers / hour, N s Indicates the total number of detection points remaining after filtering.

3. The vehicle blind spot monitoring radar independent calibration method according to claim 2, characterized in that: The detection points retained in S41 are stationary targets or other moving targets far away from the vehicle; the speed of the stationary targets relative to the axial direction of the vehicle body is approximately equal to the vehicle's driving speed, and the speeds of other moving targets relative to the axial direction of the vehicle are different.

4. The vehicle blind spot monitoring radar independent calibration method according to claim 1, characterized in that: In S5, each selected element is represented by three pieces of information: estimated probability, estimated value of installation angle and estimated value of vehicle speed; the element selection method is used in combination with a window function, or in combination with a two-dimensional CFAR method.

5. The vehicle blind spot monitoring radar independent calibration method according to claim 1, characterized in that: The specific operation steps of S6 are as follows: S61, searching whether the next tracker is in an activated state, if the tracker is in an activated state, executing the next step; if the tracker is in an idle state, initializing the tracker, and jumping to S65; S62, matching the installation angle tracking value and the vehicle speed tracking value of the road tracker with the elements in the element set selected in S5 one by one, and selecting elements that simultaneously meet the following three conditions; Condition 1: |V track -V est | <V threshold ; Condition 2: Condition 3: P <P track ; Among them, V track Represents the vehicle speed tracking value of the tracker, V est Represents the estimated value of the vehicle speed of this element, V threshold Indicates the allowable threshold of vehicle speed estimation error; Indicates the installation angle tracking value of the tracker. Indicates the estimated value of the installation angle of the element; represents the angle estimation error allowable threshold; P represents the estimated probability of the element, P track Indicates the probability threshold for allowing tracking; S63, if the road tracker matches an element that meets the conditions, then select the element with the smallest installation angle matching error to update the tracker parameters; the installation angle tracking value and vehicle speed tracking value of the tracker are updated using a Kalman filter, taking into account the dynamic change of the vehicle speed and the smooth approximation of the installation angle; the tracker's confidence +1, decay degree = 0, age +1, state flag = 1; at the same time, delete the element selected by the tracker from the element set; S64, if the road tracker does not match an element that meets the conditions, the installation angle tracking value and the vehicle speed tracking value of the tracker remain unchanged, the intrinsic parameters of the Kalman filter are updated automatically, the confidence of the tracker is -1, the attenuation is +1, the age is +1, and the state flag is 1; check whether the attenuation of the tracker is greater than the deletion threshold, if it is greater than the deletion threshold, initialize the road tracker and put the road tracker in an idle state; S65, check whether all trackers have been updated, if not all have been updated, execute from S61, if all have been updated, proceed to the next step; S67. Check whether there are any remaining elements in the element set. If not, proceed to the next step. If there are any remaining elements, determine whether the following two conditions are met at the same time: Condition 1: P>P setup ; Condition 2: There are idle trackers in the multi-channel trackers; If these two conditions are met at the same time, an idle tracker is activated, and the installation angle estimation value and vehicle speed estimation value of the element are used as the installation angle tracking value and vehicle speed tracking value of the tracker. The tracker's confidence = 1, attenuation = 0, age = 1, and status flag = 1.

6. The vehicle blind spot monitoring radar independent calibration method according to claim 1, characterized in that: In addition to the estimation error, the allowable threshold of the vehicle speed estimation error in S62 should also take into account the acceleration and deceleration capabilities of the vehicle. In each radar test cycle, the maximum change in vehicle speed is: ΔV = a*T, a is the maximum acceleration or maximum deceleration of the vehicle, and T is the radar signal sending cycle.

Images