A roadside traffic radar installation angle automatic calibration method and system

Abstract

The application provides a roadside traffic radar installation angle automatic calibration method and system, which is applied to automatic calibration of a roadside radar installation angle, and the method comprises the following steps: acquiring original static point cloud data; adopting a RANSAC algorithm based on a dynamic baseline constraint to perform straight line fitting on the static point cloud to obtain a static target straight line; checking and verifying the slope and intercept of the current straight line and the slope and intercept fluctuation of each historical straight line through a sliding window, when each fluctuation is less than a corresponding threshold value, the verification is passed, and a final straight line is output; and obtaining a radar installation angle according to the slope of the static target fitted straight line. The application fits a static target straight line through an improved RANSAC algorithm, verifies the stability of the fitted straight line through a sliding window and historical straight lines, outputs a final straight line, obtains a radar installation angle, and improves the radar deployment efficiency.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent transportation technology, and in particular to a method, system, storage medium, and computer program product for automatic calibration of the installation angle of roadside traffic radar. Background Technology

[0002] In intelligent transportation systems, the installation angle of roadside radar is crucial to the accuracy of vehicle detection and tracking. Traditional manual calibration methods require closed-road operation. According to a 2023 study in the *China Journal of Highway and Transport*, the average calibration time for a single device is approximately 2.5 hours, and the lateral positioning deviation of vehicles exceeds 5 meters at a distance of 200 meters, resulting in low efficiency and significant errors. While vision-based automated calibration schemes can significantly improve calibration speed and accuracy, they still suffer from problems such as failure in low-light conditions and poor adaptability to complex roads.

[0003] Vision-based automated calibration schemes typically rely on camera-radar joint calibration. The camera captures specific calibration targets in the road scene, such as checkerboard patterns, ArUco codes, or natural features like lane lines and traffic signs. Feature extraction is achieved using computer vision algorithms, such as SIFT, ORB, and deep learning keypoint detection. A transformation relationship between the camera and radar coordinate systems is established for coordinate mapping. Angle optimization, such as pitch, yaw, and roll angles, is achieved by using the Perspective-n-Point (PnP) algorithm or iterative nearest ICP registration to optimize the radar mounting angle. However, insufficient light in nighttime or tunnel environments makes it difficult for the camera to extract stable features. Strong light interference (such as headlight glare) leads to feature mismatches; curves, slopes, and non-standard intersections cause deviations in target pose estimation; and dynamic vehicle interference (such as trucks obscuring the target) further exacerbates the problem. Vision-based calibration schemes have a high failure rate in low-light environments, and the failure rate can exceed 30% under complex road geometry conditions, necessitating an automated and high-precision calibration solution. Summary of the Invention

[0004] The purpose of this disclosure is to provide an automatic calibration method, system, storage medium, and computer program product for roadside traffic radar installation angle, thereby solving the aforementioned problems existing in the prior art.

[0005] To achieve the above objectives, the technical solutions adopted in the embodiments of this disclosure are as follows:

[0006] This disclosure provides, in one aspect, a method for automatically calibrating the installation angle of a roadside traffic radar, applicable to the automatic calibration of the installation angle of roadside radars. The method includes:

[0007] Acquire raw stationary point cloud data, which includes: point cloud data of stationary targets on the road;

[0008] The RANSAC algorithm based on dynamic baseline constraints is used to fit a straight line to the stationary point cloud to obtain the stationary target straight line.

[0009] The slope and intercept of the currently fitted static target line are checked and verified by a sliding window, comparing them with the slope and intercept of each historically fitted static target line. When each fluctuation is less than the corresponding threshold, the verification is successful, and the final line is output.

[0010] The radar installation angle is obtained based on the slope of the straight line fitted to the stationary target.

[0011] Optionally, before using the RANSAC algorithm based on dynamic baseline constraints to perform line fitting on the stationary point cloud, the method further includes: determining that the number of stationary point clouds in a single frame is greater than N0; if the number of stationary point clouds in a single frame is greater than N0, then using the RANSAC algorithm based on dynamic baseline constraints to perform line fitting on the stationary point cloud in the single frame.

[0012] If the number of stationary point clouds in a single frame is less than or equal to N0, then the stationary point cloud data from multiple frames is accumulated until the number of point clouds is greater than N0. Then, the RANSAC algorithm based on dynamic baseline constraints is used to fit a straight line to the accumulated stationary point clouds from multiple frames.

[0013] Optionally, the step of using the RANSAC algorithm based on dynamic baseline constraints to perform straight line fitting on the stationary point cloud to obtain the stationary target straight line includes:

[0014] Dynamic baseline sampling is adopted, and any two point pairs with a distance greater than a preset distance M are selected from the static point cloud as the straight line sample point set;

[0015] Calculate the straight line model based on the set of straight line sample points;

[0016] Calculate the distance from each data point in the original point cloud data other than the straight line sample point set to the straight line. If the distance from a point to the straight line is less than the distance error threshold, then the corresponding point is a point inside the straight line.

[0017] Record the number or proportion of interior points in the current straight line model;

[0018] After multiple iterations, several candidate linear models were generated;

[0019] The line model with the most interior points or the largest proportion of interior points is selected as the optimal line for this fitting.

[0020] Based on the set of interior points of the optimal linear model, the linear parameters are determined by refitting using the least squares method.

[0021] Optionally, the step of checking and verifying the slope and intercept of the currently fitted stationary target line through a sliding window compared with the slope and intercept of each historically fitted stationary target line, and verifying that the verification is successful when all fluctuations are less than the corresponding thresholds, and outputting the final line, includes:

[0022] Maintain a sliding window of length N to store the N most recent line fitting results, each of which includes the slope and intercept.

[0023] Calculate the slope error ratio and intercept error ratio of the current line to each historical line in the sliding window, or the slope error and intercept error.

[0024] If the slope error ratio of the current line and each line in the sliding window is less than the slope ratio threshold and the intercept error ratio is less than the intercept ratio threshold, or if the slope error is less than the slope threshold and the intercept error is less than the intercept threshold, then the current line fitting result is determined to be stable, and the final line equation is output.

[0025] Optionally, obtaining the radar installation angle based on the slope of the straight line fitted from the stationary target includes:

[0026] Calculate the original angle θ based on the slope kˊ of the fitted straight line to the target. line =arctan(kˊ)*180 / π, to calculate the actual installation angle; where kˊ is the slope of the line, and arctan(kˊ) is the returned value in radians;

[0027] If the slope of the fitted straight line is less than 0, then the radar is facing the direction the vehicle is approaching, the road is to the right of the radar, and the radar is installed at an angle θ. install The calculation method is θ install =θ line +90°;

[0028] If the slope of the fitted straight line is greater than 0, then the radar is facing the direction the vehicle is going, the road is to the left of the radar, and the radar is installed at an angle θ. install The calculation method is θ install =θ line -90°.

[0029] Optionally, after obtaining the radar installation angle, the method further includes: calculating the radar straight-line slope kˊ and the preset slope k preset The slope difference is used to determine the installation deflection compensation coefficient λ, λ = f(Δθ, W, L), based on the slope difference, road width, and length, to correct the radar installation angle; when the radar installation angle θ... install =θ line When the radar installation angle is ±90°, the correction formula is θ. install =θ line ±90°+f(Δθ,W,L)μ, where, θ lineTo fit the angle corresponding to the slope of the straight line, μ is the proportionality constant, λ is the installation deflection compensation coefficient, Δθ is the installation angle error, W is the road width, and L is the calibration distance.

[0030] Optionally, determining the installation deflection compensation coefficient λ based on the slope difference, road width, and length includes:

[0031] Based on the radar straight slope kˊ and the preset slope k preset The slope difference Δk is used to calculate the installation angle error Δθ = arctan(kˊ) - arctan(k). preset Then, by using λ=f(Δθ,W,L)=α·Δθ+β·W / L+γ, the installation deflection compensation coefficient λ is determined, where α, β, and γ represent parameters.

[0032] Another aspect of this disclosure provides an automatic calibration system for the installation angle of a roadside traffic radar, the system comprising:

[0033] The acquisition module is used to acquire raw stationary point cloud data; the raw stationary point cloud data includes: point cloud data of stationary targets on the road;

[0034] The line fitting module is used to perform line fitting on the stationary point cloud using the RANSAC algorithm based on dynamic baseline constraints to obtain the stationary target line;

[0035] The verification module is used to check the slope and intercept of the currently fitted static target line with a sliding window to verify the fluctuation of the slope and intercept of each historically fitted static target line. When each fluctuation is less than the corresponding threshold, the verification is passed and the final line is output.

[0036] The determination module is used to obtain the radar installation angle based on the slope of the straight line fitted to the stationary target.

[0037] Another aspect of this disclosure provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described above.

[0038] Another aspect of this disclosure provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described above.

[0039] The beneficial effects of the embodiments disclosed herein are:

[0040] The method of this disclosure is based on the RANSAC algorithm with dynamic baseline constraints. It performs straight line fitting on the acquired stationary target point cloud, checks the slope and intercept of the fitted straight line with the fluctuation of the slope and intercept of the fitted straight line with the historical multiple fitted straight lines through a sliding window, and outputs the final straight line parameters after verification. It automatically calculates the radar installation angle, improves calibration efficiency and accuracy, avoids the low efficiency of manual calibration, and allows calibration at any time without weather restrictions, thereby improving radar deployment efficiency. Attached Figure Description

[0041] Figure 1 This is a schematic flowchart of an automatic calibration method for the installation angle of roadside traffic radar proposed in an embodiment of this disclosure;

[0042] Figure 2 This is a schematic diagram of the structure of an automatic calibration system for the installation angle of roadside traffic radar proposed in an embodiment of this disclosure;

[0043] Figure 3 This is a flowchart illustrating the process of fitting a straight line using the RANSAC algorithm with dynamic baseline constraints in an automatic calibration method for the installation angle of roadside traffic radar proposed in this embodiment of the present disclosure.

[0044] Figure 4 This is a schematic diagram of the results of detecting stationary targets in an automatic calibration method for the installation angle of roadside traffic radar proposed in an embodiment of this disclosure;

[0045] Figure 5 This is a schematic diagram of the structure of the radar after installation in an automatic calibration method for the installation angle of a roadside traffic radar proposed in an embodiment of this disclosure;

[0046] Figure 6 This is a schematic diagram of the radar data processing structure of an automatic calibration system for the installation angle of roadside traffic radar proposed in an embodiment of this disclosure. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the embodiments of this disclosure will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of this disclosure and are not intended to limit the embodiments of this disclosure. The embodiments of this disclosure relate to a method and system for automatically calibrating the installation angle of a roadside millimeter-wave radar based on radar fence detection data, applicable to the automated calibration and rapid deployment of intelligent transportation infrastructure.

[0048] Example 1, as Figure 1 As shown, this disclosure proposes an automatic calibration method for the installation angle of roadside traffic radar, applicable to the automatic calibration of roadside radar installation angles. The method includes:

[0049] Step S100: Obtain raw stationary point cloud data, wherein the raw stationary point cloud data includes: point cloud data of stationary targets on the road.

[0050] Before the radar acquires point cloud data, the radar to be calibrated is fixedly installed at the corresponding position on the roadside, such as... Figure 5 As shown, the radar is installed at a certain horizontal angle, facing the road side, with a height set between 1.5m and 9m. The horizontal installation angle is best between 10° and 45°. Too small an angle will result in blind spots at close range, while too large an angle will result in blind spots at distant range, affecting subsequent target detection. After installing the radar to be calibrated, power it on. The radar continuously collects point cloud data of the surrounding environment, including point cloud data of stationary targets on the road. Stationary targets can be fences in the center of the road or roadside railings, etc. Stationary targets are straight or nearly straight. The point cloud data of stationary targets is taken from stationary objects on the road within approximately 50 meters of the radar's normal direction. At short distances, the curvature of high-speed or expressways is generally not significant, meaning the radius of curvature is usually large. This is suitable for roads with a radius of curvature ≥ R, where the road is nearly straight. For curved roads, the number of radars is generally increased, and the deployment is more dense to ensure coverage of the lane. R is set based on actual measurement data and can be 1000 meters.

[0051] The automatic radar installation angle calibration method of this disclosure is applicable to millimeter-wave radar. While general lidar does not have a doppler, the point clouds acquired by special lidar and other radars contain information such as Doppler velocity (doppler), sufficient to distinguish between moving and stationary point clouds. Therefore, the method of this disclosure can be used to calibrate the radar installation angle. The acquired raw radar point cloud data undergoes preprocessing, including filtering, ghost image removal, and data accumulation. For millimeter-wave radar simultaneously acquiring moving and stationary point clouds, it is necessary to separate the moving and stationary point clouds from the acquired raw point cloud data. The extracted stationary point clouds are then used to track moving targets. An improved RANSAC (Random Sample Consensus) algorithm can be applied to the preprocessed stationary points for line fitting, line stability verification, and angle calculation. The stationary point cloud includes stationary target point cloud data, which can be point cloud data of fences on roads. Each radar point cloud includes attributes such as range, azimuth, Doppler velocity (doppler), signal-to-noise ratio (SNR), and radar cross section (RCS). Distinguishing between moving and stationary point clouds: Moving point clouds are those with a dopper value of 0, while stationary point clouds are those with a dopper value of 0. In other words, stationary point clouds with a dopper value of 0 are extracted.

[0052] After acquiring still point cloud data, it is determined whether the number of point clouds is greater than a preset threshold N0. If the number of still point clouds in the current frame is greater than N0, line fitting is performed; otherwise, line fitting is skipped and the operation ends. Alternatively, multiple frames of point cloud data are accumulated until the number of still point clouds is greater than N0. If the number of point clouds is too small, line fitting is not performed.

[0053] In step S100 of this embodiment, point cloud data can be acquired frame by frame. When the number of point clouds is greater than a preset threshold N0, the following steps are performed on the corresponding single-frame static point cloud.

[0054] Step S200: Using the RANSAC algorithm with dynamic baseline constraints, a straight line is fitted to the stationary point cloud to obtain the stationary target straight line.

[0055] Step S200 of this embodiment adopts an improved RANSAC algorithm. Specifically, through dynamic baseline sampling, optimal line selection and least squares optimization, robust line fitting of fence points is achieved, and abnormal lines such as those with excessively large or small angles are eliminated.

[0056] like Figure 3 As shown, for some radars, if the number of stationary point clouds in the current frame is greater than N0, then a straight line fit is performed on the stationary point cloud data; if the number of stationary point clouds in the current frame is less than or equal to N0, then the straight line fit is skipped and the operation ends.

[0057] like Figure 3 As shown, the RANSAC algorithm with dynamic baseline constraints performs line fitting on a stationary point cloud to obtain a stationary target line, including:

[0058] Step S210: Using dynamic baseline sampling, two point pairs with a distance greater than a preset distance M are randomly selected from the static point cloud multiple times as a set of straight line sample points to generate candidate straight line models.

[0059] It should be noted that the dynamic baseline sampling prioritizes selecting any pair of adjacent points with a distance greater than a preset distance M from the static point cloud of the current frame as the data for fitting a static target line. This improves the reliability of fitting in complex scenes and suppresses local noise interference. M is greater than 3m, and can be set to 5m, 10m, etc., depending on the actual situation. The line fitting uses static point cloud data. Dynamic baseline sampling optimizes the point selection strategy for line fitting, excluding point pairs that are too close together and avoiding incorrect lines from small objects. This reduces unnecessary computation and increases the accuracy of the fitting results. Dynamic baseline constraint limits the sampling point spacing by a preset distance threshold M (M≥5 meters), eliminating incorrect line fitting caused by nearby noise points (such as local deformation of the fence or isolated obstacles). Compared to the random sampling of classic RANSAC, this improves the robustness of fitting long strip fence structures.

[0060] Step S220: Based on the set of line sample points, generate one candidate line model each time.

[0061] Step S220 of this embodiment is a model assumption to generate straight line parameters. The general equation of the straight line is calculated as ax + by + c = 0, where a, b, and c are the three coefficients of the straight line equation. In this embodiment, it is converted to the slope-intercept form y = kx + b″, where k is the slope, b″ is the intercept, and x refers to the abscissa of the point cloud, and y is the ordinate. With the radar center as the origin, the y-axis is the radar detection direction, and x is the abscissa. One iteration yields one candidate straight line model; multiple iterations yield multiple candidate straight line models.

[0062] Step S230: Calculate the distance d from each data point j (excluding the line sample point set) in the original point cloud data to the line. j If d j If the distance error threshold is less than ∈, then point j is determined to be a point inside the line.

[0063] In step S230 of this embodiment, it is determined whether each of the remaining data points j falls near the current line based on the distance error threshold ∈. If the distance from point j to the line is d j If <∈, then point j is determined to be a point inside the current line. The remaining data points are point cloud data other than the line sample point set determined in step S210 from the original static point cloud data.

[0064] Calculate interior points: For all other points, calculate the distance d to the line. The general equation ax + by + c = 0 can be used to calculate the distance d from point j to the fitted stationary target line. j , If d j If <∈, then point j is a point inside the line.

[0065] Here, a, b, and c represent the parameters of the straight line equation, and interior points refer to points whose distance from the straight line is less than a set threshold (e.g., M meters). Because a fence usually corresponds to multiple relatively continuous point clouds, it is necessary to ensure that there are enough points falling near the straight line to determine that the fitted straight line is likely to be the line containing the fence. A straight line with a small number of interior points may be fitted with many discrete points, resulting in many incorrect straight lines.

[0066] Step S240: Record the number or proportion of interior points of the straight line model in the current frame.

[0067] Wherein, the ratio of interior points = number of interior points / number of original static point clouds.

[0068] Step S250: After multiple iterations, multiple candidate straight line models are generated.

[0069] Repeat steps S210 to S240N1 times, such that N1 = 100, generating different candidate models L1, L2, ..., L1 each time. N The system records the number or proportion of interior points in the corresponding linear model. In each iteration, the number of interior points is checked. If the number of interior points of any candidate line is less than K, the linear model for that iteration is invalid; if the number of interior points of any candidate line is greater than or equal to K, the linear model for that iteration is valid.

[0070] Step S260: Optimal line selection. Select the line model with the most interior points or the largest proportion of interior points from the effective line models as the optimal line for this fitting.

[0071] It should be noted that, from multiple valid candidate lines, the model L with the most interior points or the largest proportion of interior points is selected. best The optimal line is selected by evaluating metrics such as the number of interior points. Optimal Model Selection identifies the best-fitting line.

[0072] Step S270: For the set of interior points corresponding to the optimal straight line model, refit the straight line parameters using the least squares method.

[0073] The least squares method is used to optimize the interior points, and the line parameters are updated by refitting the set of interior points of the optimal line. Least squares refinement refines the fitting of the interior point set of the candidate line, improving accuracy.

[0074] The least squares optimization formula for interior point sets is as follows:

[0075] Let the interior point set be The equation of the straight line is y = k′x + b′, and the goal is to minimize the sum of squared errors. m represents the number of interior points.

[0076] Solution

[0077]

[0078] The improved RANSAC algorithm in this embodiment is a customized optimization scheme designed for the specific needs of fence point cloud fitting, such as elongated structures and angular constraints.

[0079] like Figure 4 As shown, the improved RANSAC algorithm is used to fit the straight line l formed by the fence points.

[0080] Step S300: Use a sliding window to check and verify the slope and intercept of the current line and the fluctuation of the slope and intercept of each historically fitted static target line. When each fluctuation is less than the corresponding threshold, the verification is successful and the final line is output.

[0081] In step S300 of this embodiment, a multi-frame sliding window mechanism is used to verify the stability of the fitting result. The fitted straight line is verified by checking the fluctuations of the slope and intercept of the straight line through the sliding window. When the verification is successful, the final straight line is output and stored in the window.

[0082] Maintain a queue of fitting results of length N; set the slope change rate threshold to P. k The threshold for the rate of change of the intercept is set to P. b %; N consecutive successful verifications trigger calibration completion.

[0083] Linear stability verification: Maintain a sliding window of length N to store the fitting results, check the consistency of the slope and intercept between the current result and historical results, and ensure that the slope fluctuates less than P for N consecutive cycles. k % and intercept fluctuate continuously for N times < P b When the percentage reaches %, it indicates that the result is stable, confirming the final straight line.

[0084] Maintain a sliding window of length N to store historical fitting results. Check the consistency of the slope and intercept between the current result and the historical results. When the current result is the straight line fitted to the current frame, assume the equation of the straight line in the current frame is y = kˊ*x + bˊ, where kˊ is the slope of the straight line in the current frame and bˊ is the intercept of the straight line in the current frame. The equations of the straight lines of the N historical fitting results are y i =k i *x+b i , where k i Let b be the slope of the straight line in the i-th frame of history. i The line intercept of the i-th frame in history, i = 1, 2, ... N, and the sliding window stores the k of the corresponding line. i and b i The value is used to calculate the slope error ratio e between the current frame and all historical frames. ki =|k i -kˊ| / kˊ and the ratio of the intercept error between the current frame and historical frames e bi =|b i -bˊ| / bˊ, when the e of N historical frames is satisfied ki All are less than P k % and e bi All are less than P b When the threshold is reached, a stable fence line is considered to have been found, and the current frame is determined to be a valid frame and added to the window; otherwise, the frame is discarded. N, P k % and P b % is preset, N, P k%, and P b % needs to be adjusted according to the actual data, generally it can be set to 5%, or the slope error and intercept error can be directly limited. For example, e bi = |b i - b'|, e ki = |k i - k'|, |e bi | < e b _thr, |e ki | < ek_thr, where b i is the intercept of the i-th historical frame, b' is the intercept of the current frame, k i is the slope of the i-th historical frame, k' is the slope of the current frame, e bi is the intercept error, e ki is the slope error, eb_thr is the maximum intercept error, and ek_thr is the maximum slope error. The slope of the fence line usually changes slowly, so P k % can be set smaller, such as 1%. Also, when the window is full, output the mean value μ k of the slope and μ b of the intercept in the window as the final slope and intercept, and calculate the installation angle θ = arctan(μ k ). It can also be that when the window is full, or when the predetermined calibration times or time is reached, the straight line slope and intercept of the latest frame, or any historical straight line slope and intercept in the window, or the median of the average value of the historical straight line slope and intercept, etc. are used as the final slope and intercept. Through window verification, the straight line converges and tends to be stable. The improved RANSAC algorithm is used to improve the robustness of straight line fitting. This method is especially suitable for the point cloud fitting of long strip structures such as fences and walls. Its advantages are: reducing the influence of short-distance noise through baseline constraints; enhancing semantic rationality by combining angle prior knowledge; ensuring the geometric accuracy of the final model through least squares optimization. Dynamic baseline sampling mechanism: Priority is given to selecting point pairs with a spacing exceeding M meters for straight line fitting to improve the fitting reliability in complex scenarios. Multi-frame sliding window verification: It is confirmed to be valid when the slope fluctuates continuously N times < Pk% and the intercept fluctuates continuously N times < Pb% to ensure the stability of the calibration result.

[0085] Step S400: Obtain the actual installation angle of the radar according to the slope of the stationary target straight line finally fitted.

[0086] That is to say, the radar installation angle is calculated according to the stable straight line parameters. Specifically, according to the straight line fitted to the target, the angle corresponding to the straight line slope k is obtained, and thus the actual installation angle of the radar is obtained. As Figure 5 shown, the installation angle is the angle between the perpendicular line y-axis of the radar plane and the straight line where the road is located, that is, the azimuth angle, and the y-axis is the radar detection direction. Calculate the angle θ corresponding to the straight line slope k' line=arctan(kˊ)*180 / π; Determine the installation angle (θ) based on the angle corresponding to the slope kˊ of the straight line. line ±90°), where θ line Right now Figure 4 θ in the middle.

[0087] Angle calculation: Calculate the original angle θ based on the slope kˊ of the fitted line of the target. line =arctan(kˊ)*180 / π, which calculates the actual installation angle. kˊ is the slope of the line, and arctan(kˊ) returns the value in radians.

[0088] If the slope kˊ of the fitted line is less than 0, i.e. θ line If the angle is less than 0, the radar is facing the direction the vehicle is approaching, the road is to the right of the radar, and the radar installation angle is calculated as θ. install =θ line +90°.

[0089] If the slope kˊ of the fitted line is greater than 0, that is, θ line If the angle is greater than 0, the radar is pointing in the direction the vehicle is traveling, the road is to the left of the radar, and the radar installation angle is calculated as θ. install =θ line -90°.

[0090] The angle constraint is the absolute value of the radar installation angle calculated inversely, |θ install |<θ max (e.g., 60°), θ install For the radar installation angle, θ max The maximum angle value is set. Otherwise, recalibrate and repeat step S100.

[0091] Step S500: Automatically set the detection area boundary, emergency lane position, and calculate coordinate transformation parameters according to the installation angle.

[0092] like Figure 2 As shown, the area configuration automatically configures the detection area and calculates coordinate transformation parameters. Automatically setting the detection area improves the accuracy of subsequent vehicle detection. The method in this embodiment reduces installation and debugging costs and improves system deployment efficiency. Figure 5 As shown, this is the result of radar detection after calibrating the installation angle.

[0093] Actual testing has verified that the embodiments disclosed herein have significant advantages over traditional methods:

[0094] index Traditional vision-based methods The method of this disclosure embodiment Calibration time 150 minutes <1 minute Environmental dependence Sunny / Daytime 24 / 7

[0095] 1. Achieve fully automated calibration of roadside radar installation angles without manual intervention; 2. Employ an improved RANSAC algorithm to enhance the robustness of line fitting; 3. Ensure the stability of calibration results through multi-frame verification; 4. Automatically set detection areas to improve the accuracy of subsequent vehicle detection; 5. Reduce installation and debugging costs and improve system deployment efficiency.

[0096] Table 2 shows the radar detection results after calibrating the radar installation angle.

[0097] Serial Number Actual installation angle (°) Radar calibration angle (°) Calibration error (°) 1 5 5.3 0.3 2 10 10.2 0.2 3 15 15.3 0.3 4 20 19.7 -0.3 5 25 25.2 0.2 6 30 29.6 -0.4

[0098] The embodiments disclosed herein are based on an automatic calibration method for roadside radar installation angles using linear fitting of radar-detected fence points. By fitting the fence point data detected by the radar, the radar installation angle is automatically calculated, thereby improving calibration efficiency and accuracy.

[0099] like Figure 6 As shown, another aspect of this disclosure proposes an automatic calibration system for the installation angle of roadside traffic radar, applied to the automatic calibration of the installation angle of roadside radar. The system includes:

[0100] The acquisition module 100 is used to acquire raw stationary point cloud data; wherein, the raw stationary point cloud data includes: point cloud data of stationary targets on the road.

[0101] The line fitting module 200 is used to perform line fitting on the stationary point cloud using the RANSAC algorithm based on dynamic baseline constraints to obtain the stationary target line.

[0102] The verification module 300 is used to check the slope and intercept of the currently fitted static target line with the fluctuation of the slope and intercept of each historically fitted static target line through a sliding window. When each fluctuation is less than the corresponding threshold, the verification is passed and the final line is output.

[0103] The determination module 400 is used to obtain the radar installation angle based on the slope of the straight line fitted to the stationary target.

[0104] like Figure 2As shown, the system in this embodiment further includes: a region configuration module, used to automatically set the detection area boundary, emergency lane position, and calculate coordinate transformation parameters according to the installation angle; a millimeter-wave radio frequency front-end module, used to transmit and receive specific millimeter-wave signal waveforms and perform analog-to-digital conversion; a radar signal processing module, used to detect the raw signal and output raw radar point cloud data, including moving and stationary point cloud data; a radar data processing module, used to preprocess the millimeter-wave raw radar point cloud (filtering, ghost removal, data accumulation, etc.), track moving targets, and perform line fitting, line stability verification, and angle calculation on stationary points using an improved RANSAC (Random Sample Consensus) algorithm. The radar data processing module includes: an acquisition module, a preprocessing module, a line fitting module, a verification module, and a determination module. The preprocessing module preprocesses the millimeter-wave raw radar point cloud (filtering, ghost removal, data accumulation, etc.); and a data output module, used to set the detection area according to the angle calibration results and perform coordinate transformation on the data, outputting target information and traffic event information.

[0105] The system of this disclosure collects radar point cloud data of road fences, uses an improved random sampling consensus algorithm for straight line fitting, verifies the stability of the line through a multi-frame sliding window, calculates the installation angle, and finally automatically configures the detection area. Compared with traditional methods, it has high calibration efficiency and is not affected by ambient lighting conditions. This disclosure is suitable for the rapid deployment of roadside sensing devices in intelligent transportation systems.

[0106] Another aspect of this disclosure provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described above.

[0107] Another aspect of this disclosure provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described above.

[0108] In Example 2, for some radars, when the number of single-frame point cloud data is too small, multiple frames of point cloud data can be superimposed and accumulated according to the actual situation. After acquiring stationary point cloud data in step S100 of Example 1, the number of point clouds in a single frame is determined. If the number of stationary point clouds in a single frame is less than or equal to N0, multiple frames of stationary point cloud data can be accumulated until the number of point clouds is greater than N0. Then, the RANSAC algorithm based on dynamic baseline constraints, as described in step S200 of Example 1, is used to perform linear fitting on the accumulated stationary point clouds from multiple frames. The subsequent steps of Example 1 are then executed. In step S300 of Example 1, the corresponding sliding window stores the linear results of historical multi-frame stationary point cloud fitting, which facilitates verification of the slope and intercept of the current multi-frame point cloud data fitting line versus the slope and intercept of each historical multi-frame stationary point cloud fitting line within the sliding window. This will not be elaborated here.

[0109] In Example 3, after obtaining the radar installation angle through step S400 of either step one or step two, the method further includes: calculating the slope k′ of the stationary target line scanned by the radar and a preset slope k. preset The difference is used to set a parameter λ based on the difference, road width, and length, which is used to correct the radar installation angle.

[0110] θ install =θ line ±90°+λμ

[0111] Where, θ line To fit the angle corresponding to the slope of the straight line, μ is a proportionality constant (unit conversion or scaling factor), which is calibrated experimentally; λ is the installation deflection compensation coefficient, which is obtained by looking up a table (which needs to be pre-calibrated) or a corresponding fitting formula based on the difference between the measured error and the preset slope. The radar installation angle refers to the measurement of the angle between the radar normal and true north.

[0112] Table 1 shows the calibration results recorded under various combinations of known installation angle error Δθ, road width W, and calibration distance L, based on the experimental design.

[0113] Δθ(°) W(m) L(m) Measured λ Fitting λ 1.5 3.75 50 0.6 0.62 －2.0 4.50 30 －0.8 －0.79

[0114] Fitting results in λ = f (Δθ, W, L) = α·Δθ+β·W / L+γ

[0115] Where α, β, and γ are calibration constants obtained through fitting in previous experiments. W / L reflects the influence of the road width-to-length ratio on compensation, applicable to scenarios with small errors |Δθ| < 5° and regular road geometry. Preset slope k preset It can be obtained through high-precision maps or manual measurement.

[0116] Preset slope k preset The prior slope for radar installation should be the estimated angle for scanning stationary targets. If k preset =0, Δk=k′-k preset Δθ≈arctan(Δk)≈Δk, Δθ≈arctan(k′)≈k′ (small angle approximation)

[0117] At this point, Δk directly reflects the angular error and can be used to calculate λ.

[0118] If k preset ≠0, Δθ=arctan(k′)-arctan(k preset )

[0119] First, we need to convert the slope difference to an angle difference Δθ using arctangent transformation, and then substitute it into λ=f(Δθ,W,L) to obtain:

[0120] θinstall =θ line ±90°+f(Δθ,W,L)·μ.

[0121] Here, Δθ is the angular deviation between the measured straight line of the stationary target and the preset straight line, directly reflecting the radar installation angle error. Verification of θ install Is it within a reasonable range? (Absolute value of radar installation angle |θ) install |<θ max If yes, output; otherwise, recalibrate.

[0122] The above description is only a preferred embodiment of the present disclosure. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present disclosure, and these improvements and modifications should also be considered within the protection scope of the present disclosure.

Claims

1. A roadside traffic radar installation angle automatic calibration method applied to roadside radar installation angle automatic calibration, characterized in that, The method includes: Acquire raw stationary point cloud data; whereby the raw stationary point cloud data includes: point cloud data of stationary targets on the road; The RANSAC algorithm based on dynamic baseline constraints is used to fit a straight line to the stationary point cloud to obtain the stationary target straight line. The slope and intercept of the currently fitted static target line are checked and verified by a sliding window, comparing them with the slope and intercept of each historically fitted static target line. When each fluctuation is less than the corresponding threshold, the verification is successful, and the final line is output. The radar installation angle is obtained based on the slope of the straight line fitted to the stationary target. The process of obtaining the radar installation angle based on the slope of the straight line fitted from the stationary target includes: Based on the slope of the fitted straight line of the target k Calculate the original angle θ line =arctan( k ´) 180 / π The actual installation angle is calculated by reverse calculation; among them, k ´ is the slope of the line, arctan( k ´) returns the value in radians; If the slope of the fitted straight line is less than 0, then the radar is facing the direction the vehicle is approaching, the road is to the right of the radar, and the radar installation angle is... θ install The calculation method is as follows θ install = θ line +90°; If the slope of the fitted straight line is greater than 0, then the radar is facing the direction the vehicle is going, the road is to the left of the radar, and the radar installation angle is... θ install The calculation method is as follows θ install = θ line -90°; After obtaining the radar installation angle, the method further includes: calculating the radar straight-line slope k´ and the preset slope k. preset The slope difference is used to determine the installation deflection compensation coefficient λ based on the slope difference, road width, and length. λ = f (Δ) θ , W , L ), to correct the radar installation angle; When the radar is installed at the angle θ install = θ line When the angle is ±90°, the formula for correcting the radar installation angle is: θ install = θ line ±90°+ f (Δ) θ , W , L ) μ ;in, θ line To fit the angle corresponding to the slope of the straight line, μ Δ is a proportionality constant. θ To account for installation angle error, W For road width, L For calibration distance; The determination of the installation deflection compensation coefficient λ based on the slope difference, road width, and length includes: Based on the radar straight slope k´ and the preset slope k preset Calculate the installation angle error Δk based on the slope difference Δk. θ=arctan ( k ´)- arctan (k) preset ), and then through λ= f (Δ) θ , W , L )= α ·Δ θ + β ·W / L+ γ Determine the installation deflection compensation coefficient λ, where, α, β, γ Indicates a parameter.

2. The method according to claim 1, characterized in that, Before performing line fitting on the static point cloud using the RANSAC algorithm based on dynamic baseline constraints, the method further includes: If the number of stationary point clouds in a single frame is greater than N0, then the RANSAC algorithm based on dynamic baseline constraints is used to fit a straight line to the stationary point cloud in the single frame. If the number of stationary point clouds in a single frame is less than or equal to N0, then the stationary point cloud data from multiple frames is accumulated until the number of point clouds is greater than N0. Then, the RANSAC algorithm based on dynamic baseline constraints is used to fit a straight line to the accumulated stationary point clouds from multiple frames.

3. The method according to claim 1 or 2, characterized in that, The method employs the RANSAC algorithm based on dynamic baseline constraints to perform straight line fitting on the stationary point cloud to obtain the stationary target straight line, including: Dynamic baseline sampling is adopted, and any two point pairs with a distance greater than a preset distance M are selected from the static point cloud as the straight line sample point set; Calculate the straight line model based on the set of straight line sample points; Calculate the distance from each data point in the original point cloud data other than the straight line sample point set to the straight line. If the distance from a point to the straight line is less than the distance error threshold, then the corresponding point is a point inside the straight line. Record the number or proportion of interior points in the current straight line model; After multiple iterations, several candidate linear models were generated; The line model with the most interior points or the largest proportion of interior points is selected as the optimal line for this fitting. Based on the set of interior points of the optimal linear model, the linear parameters are determined by refitting using the least squares method.

4. The method according to claim 1, characterized in that, The process involves checking and verifying the slope and intercept of the currently fitted stationary target line against the slope and intercept of each historically fitted stationary target line using a sliding window. Verification is successful when all fluctuations are less than the corresponding thresholds, and the final line is output, including: Maintain a sliding window of length N to store the N most recent line fitting results, each of which includes the slope and intercept. Calculate the slope error ratio and intercept error ratio of the current line compared to each historical line in the sliding window, or the slope error and intercept error. If the slope error ratio of the current line and each line in the sliding window is less than the slope ratio threshold and the intercept error ratio is less than the intercept ratio threshold, or if the slope error is less than the slope threshold and the intercept error is less than the intercept threshold, then the current line fitting result is determined to be stable, and the final line equation is output.

5. An automatic calibration system for the installation angle of a roadside traffic radar, applied to the automatic calibration of the installation angle of a roadside radar, characterized in that, The system includes: The acquisition module is used to acquire raw stationary point cloud data; the raw stationary point cloud data includes: point cloud data of stationary targets on the road; The line fitting module is used to perform line fitting on the stationary point cloud using the RANSAC algorithm based on dynamic baseline constraints to obtain the stationary target line; The verification module is used to check the slope and intercept of the currently fitted static target line with a sliding window to verify the fluctuation of the slope and intercept of each historically fitted static target line. When each fluctuation is less than the corresponding threshold, the verification is passed and the final line is output. The determination module is used to obtain the radar installation angle based on the slope of the straight line fitted to the stationary target.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 4.

7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method described in any one of claims 1 to 4.

Images