Method, system and device for detecting and eliminating false points of vehicle-mounted millimeter wave radar caused by water splashing of tires in rainy days

Abstract

The invention relates to the technical field of intelligent driving environment perception, and discloses a method, a system and a device for detecting and eliminating false points of a vehicle-mounted millimeter wave radar caused by water throwing of tires in rainy days, and the key points of the technical scheme are that the method comprises the following steps: calibrating basic parameters of wheels, and constructing a physical model of water throwing of rear wheels; determining a multi-dimensional theoretical feature of the false water throwing point and a corresponding judgment threshold value; vehicle state data and original point cloud data of the vehicle-mounted rear angle millimeter wave radar are synchronously acquired, timestamps are aligned, and subsequent steps are executed when water throwing detection is triggered; presetting a hydro-extracting detection target area based on the spatial distribution characteristics and a judgment threshold thereof, and screening out original point cloud data in the area as a candidate false point set; performing other multi-dimensional theoretical feature judgment on each point in the candidate false point set in sequence, and recording the points meeting all theoretical feature ranges as tire water throwing false points; and removing the false water throwing points of the tire from the original point cloud data.

Description

Technical Field

[0001] This invention relates to the field of intelligent driving environment perception technology, and more specifically, to a method, system, and device for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar. Background Technology

[0002] Vehicle-mounted millimeter-wave radar is a core sensor for intelligent driving assistance systems. In particular, rear corner millimeter-wave radar undertakes core functions such as blind spot detection (BSD), lane change assist (LCA), and rear collision warning (RCW). Its detection reliability directly determines driving safety.

[0003] When driving on flooded roads during rain, the rear wheels of a vehicle will splash water at high speed, creating a water curtain that sprays towards the side and rear of the vehicle. 77GHz automotive millimeter-wave radar is highly sensitive to reflections from water, detecting a large number of water-splashing point clouds with radial velocity. These point clouds have continuous dynamic velocity characteristics, which can be misidentified by subsequent target tracking algorithms as real dynamic targets, leading to frequent false alarms and erroneous function triggers, seriously affecting the driving experience and safety.

[0004] In existing technologies, radar false point suppression schemes are mostly designed for scenarios such as static clutter, fixed reflections at intersections, virtual images over railings, and general rain noise, which have obvious technical shortcomings: There is no physical model specifically for the rear wheel water-spraying scenario, and it does not take into account the dynamic characteristics of tire water-spraying. Therefore, it is not targeted enough for false water-spraying points and has a low detection accuracy. Relying solely on the characteristics of radar point clouds without integrating vehicle dynamics data (vehicle speed, wheel speed) makes it impossible to distinguish between false water splashing points and real dynamic targets at close range, which can easily lead to the mistaken deletion of real targets or the failure to detect false points. Some solutions have high computational complexity and cannot meet the real-time requirements of automotive embedded platforms.

[0005] Therefore, it is necessary to develop a method for detecting and eliminating false points on vehicle-mounted millimeter-wave radar that is specifically designed for the scenario of tires splashing water in rainy weather, combining physical models and multi-source data fusion, in order to solve the problems existing in the current technology. Summary of the Invention

[0006] The purpose of this invention is to provide a method, system, and device for detecting and eliminating false targets of vehicle-mounted millimeter-wave radar caused by water splashing from tires in rainy weather. This invention can specifically solve the problem of false radar targets caused by water splashing from rear wheels on wet roads in rainy weather. It has high detection accuracy, good real-time performance, and strong robustness, and can effectively improve the perception reliability of vehicle-mounted rear corner radar in rainy weather conditions.

[0007] The above-mentioned technical objective of the present invention is achieved through the following technical solution: a method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar, comprising the following steps: S1. Calibrate the basic parameters of the wheel, construct a physical model of the rear wheel throwing water, and determine the multi-dimensional theoretical characteristics and corresponding judgment thresholds of false water throwing points; S2. Synchronously acquire vehicle status data and raw point cloud data of the vehicle rear corner millimeter-wave radar, align the timestamps, and determine whether water splash detection is triggered. If triggered, execute S3. S3. Based on the spatial distribution characteristics and judgment threshold of the physical model, a target area for water splashing detection is preset, and the original point cloud data in the area is selected as a candidate false point set. S4. For each point in the candidate false point set, the range is determined sequentially by other multi-dimensional theoretical features and judgment thresholds. Points that meet all theoretical feature ranges are recorded as false points of tire water ejection. S5. Remove false points of water splashing from the tires from the original point cloud data, and use the remaining point cloud data as the real points.

[0008] As a preferred embodiment of the present invention, in S1, the physical model construction method for rear wheel water ejection is as follows: Establish the vehicle coordinate system; Based on the vehicle coordinate system and the pure rolling dynamics of the wheel, the velocity vector relative to the ground when the water droplet leaves the tire is defined. Based on the ground velocity vector of the water droplet when it leaves the tire, and combined with the influence of air resistance and gravity on the flight trajectory of the water droplet, the multidimensional theoretical characteristics of the false water-throwing point are obtained by fitting. Through real-vehicle rain calibration tests, the judgment thresholds corresponding to the theoretical characteristics of each dimension were determined.

[0009] As a preferred embodiment of the present invention, the velocity vector relative to the ground when the water droplet leaves the tire is V. drop =V wheel center +ω×r, where V wheel center The velocity of the rear wheel center relative to the ground is given by r, where r is the position vector of the water droplet relative to the wheel center; ω is the angular velocity of the wheel, ω = v. wheel / R,v wheel R is the rear wheel speed, and R is the calibrated tire rolling radius.

[0010] As a preferred technical solution of the present invention, the multidimensional theoretical features include: spatial distribution features, radial velocity features, energy intensity features, and point cloud distribution features; Spatial distribution characteristics are used to limit the distribution space of false water-throwing points, which are only distributed in the close-range area behind the rear wheel, corresponding to the close-range detection range of the rear corner radar; Radial velocity characteristic: Used to define the correlation between the radar radial velocity and the rear wheel speed at the false water-throwing point; the radial velocity range is 0. <v r <k·v wheel , where v r The radial velocity of the point cloud detected by radar is given by k, which is a pre-calibrated velocity correction factor. Energy intensity characteristics: used to define the range of radar echo energy intensity for decoy water-throwing points; Point cloud distribution characteristics: used to define the spurious points of water splashing as having a discrete band-like or clustered distribution shape, with a large number of point clouds in the neighborhood, small radial velocity variance, and no contour aggregation characteristics of real targets.

[0011] As a preferred technical solution of the present invention, in S2, the vehicle status data includes the vehicle's longitudinal speed, left rear wheel speed, right rear wheel speed, tire rolling radius calibration value, wiper status signal, and road surface adhesion coefficient signal. The raw radar point cloud data includes at least the following: radial velocity, detection range, azimuth angle, echo energy intensity, and timestamp for each detection point.

[0012] As a preferred technical solution of the present invention, in S4, for each point in the candidate false point set, radial velocity feature determination, energy intensity feature determination, and point cloud distribution feature determination are performed sequentially. Radial velocity characteristic determination: Calculate the radial velocity v of the candidate point. r Determine whether it satisfies k1·v wheel <v r <k2·v wheel k1 is the minimum speed correction coefficient, and k2 is the maximum speed correction coefficient. If these conditions are met, proceed to the next judgment. Energy intensity characteristic determination: Determine whether the echo energy intensity P of the candidate point satisfies P min <P<P max , where P min For the minimum detection signal-to-noise ratio of the radar, P max If the pre-calibrated maximum energy threshold for water ejection is met, proceed to the next judgment. Point cloud distribution characteristic determination: Statistically count the number of points in the preset neighborhood of the candidate point and the radial velocity variance. If the number of points in the neighborhood is greater than or equal to the preset number threshold and the radial velocity variance is less than or equal to the preset variance threshold, then it is determined to be a false point of tire water ejection.

[0013] As a preferred technical solution of the present invention, S6 is executed after S5. S6 is: under the standard working condition of no water accumulation and no water splashing, the water splashing characteristic judgment threshold is adaptively calibrated and updated.

[0014] As a preferred technical solution of the present invention, in S3 and S4, when performing multi-dimensional theoretical feature matching and judgment, a weighted voting mechanism is used to replace the serial judgment, and preset weights are assigned to features of different dimensions. Points with a total score exceeding the judgment threshold are judged as false points of water splashing.

[0015] A system for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar includes: The model building module is used to record the basic parameters of the calibrated wheel, build a physical model of the rear wheel water ejection, and determine the multi-dimensional theoretical characteristics and corresponding judgment thresholds of false water ejection points. The vehicle status information input module is used to acquire vehicle status data; The rear corner millimeter-wave radar point cloud input module is used to acquire the raw point cloud data of the vehicle-mounted rear corner millimeter-wave radar. The time synchronization and coordinate transformation module is used to timestamp-align vehicle status data and raw point cloud data of the vehicle rear angle millimeter-wave radar, and transform the raw point cloud data from the RA polar coordinate system centered on the radar itself to the vehicle coordinate system according to the angle and position of the radar installed on the vehicle. The ROI (Region of Interest) filtering module for water splashing is used to filter out the original point cloud data within the water splashing detection target area based on the spatial distribution characteristics of the physical model and its judgment threshold, as a candidate set of false points. The multi-dimensional false point discrimination module is used to determine the range of each point in the candidate false point set by sequentially using other multi-dimensional theoretical features and judgment thresholds. Points that meet all theoretical feature ranges are recorded as false points of tire water ejection. The false point cloud removal and output module is used to remove false points of water splashed by the tire from the original point cloud data, and output the remaining point cloud data as real points to the subsequent target tracking module.

[0016] A device for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar includes: a processor and a memory, wherein the memory stores a computer program executable by the processor, and the processor implements the above method when executing the computer program.

[0017] In summary, the present invention has the following beneficial effects: Highly targeted and accurate in detection: Based on the dynamic physical model of tire water ejection, this invention establishes a dedicated feature library for rear wheel water ejection scenarios, accurately capturing the core feature that is strongly correlated with rear wheel speed and false water ejection points. Compared with general clutter suppression solutions, the accuracy of identifying false water ejection points is greatly improved, which can effectively solve the problem of false targets caused by water ejection in rainy weather.

[0018] Multi-source fusion, low false detection and false negative rate: This invention integrates the feature dimensions of vehicle dynamics data and radar point cloud, and performs multi-condition fusion judgment. It can effectively eliminate false points and accurately retain real targets at close range, which can significantly reduce the false deletion rate of real targets and the false detection rate of false points.

[0019] Good real-time performance and adaptable to vehicle platforms: This invention uses a water-spraying-specific ROI initial screening to filter out most of the irrelevant point clouds in advance, greatly reducing the amount of subsequent calculations. The overall algorithm has low complexity and can be directly deployed on vehicle radar embedded platforms to meet the real-time requirements of vehicles.

[0020] Robust and adaptable: This invention features adaptive parameter updates, which can adapt to different tire specifications, tire wear levels, rainfall amounts, and road conditions. It also supports independent matching of wheel speed determination for the left and right rear wheels, adapting to complex conditions such as vehicle turning and wheel slippage, and has strong environmental adaptability. Attached Figure Description

[0021] Figure 1 This is a flowchart of the overall method of the present invention; Figure 2 This is a schematic diagram of the physical model of tire water ejection in this invention; Figure 3 This is the system architecture diagram of the present invention. Detailed Implementation

[0022] It is readily understood that, based on the technical solution of this invention, various embodiments of the invention can be conceived by those skilled in the art without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention. Rather, these embodiments are provided to enable those skilled in the art to gain a more thorough understanding of the invention. Preferred embodiments of the invention are described below in conjunction with the accompanying drawings, which form part of this application and, together with the embodiments of the invention, serve to illustrate the innovative concept of the invention.

[0023] like Figure 1 As shown, this invention provides a method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar, comprising the following steps: S1. Pre-calibration and construction of the physical model for tire water ejection: By calibrating basic parameters such as vehicle tire rolling radius and radar installation position, a dynamic physical model of rear wheel water ejection is constructed. The motion trajectory, speed range, spatial distribution, and radar echo characteristics of water droplets after leaving the tire are derived to determine the multi-dimensional theoretical characteristics of false water ejection points. Then, through real vehicle rain calibration tests, the judgment thresholds corresponding to each characteristic are determined. In S1, the physical model for rear wheel water ejection is constructed as follows: Establish a vehicle coordinate system; the vehicle coordinate system takes the center of the rear axle of the vehicle as the origin, with the positive X-axis pointing forward, the positive Y-axis pointing to the right, and the positive Z-axis pointing upward.

[0024] Based on the vehicle coordinate system and the pure rolling dynamics of the wheel, the velocity vector relative to the ground when the water droplet leaves the tire is defined. The velocity vector of the water droplet relative to the ground when it leaves the tire is V. drop =V wheel center +ω×r, where V wheel center The speed of the rear wheel center relative to the ground is v, which is the longitudinal speed of the vehicle. ego In the vehicle coordinate system, the ground velocity vector of the water droplet when it leaves the tire is represented as V. drop =(v ego ,0,0)+ω×r, where r is the position vector of the water droplet relative to the wheel center; ω is the angular velocity of the wheel, ω=v wheel / R, where v wheel R is the rear wheel speed, and R is the calibrated tire rolling radius; The water droplets ejected from the rear wheel mainly detach from the rear half of the tire, with a longitudinal velocity relative to the ground ranging from 0 to v. ego The radial velocity relative to the vehicle-mounted radar (moving with the vehicle) is positive (the target is far from the radar) and has a strong linear correlation with the rear wheel speed. This is the core distinguishing feature between the false water-splashing point and the real target.

[0025] Based on the ground velocity vector of the water droplet when it leaves the tire, and combined with the influence of air resistance and gravity on the water droplet's flight trajectory, a multi-dimensional theoretical feature library of false water droplet ejection points is obtained through fitting. The multi-dimensional theoretical features include: spatial distribution features, radial velocity features, energy intensity features, and point cloud distribution features. Spatial distribution characteristics: The false points of water splashing are only distributed in the close-range area behind the rear wheel, corresponding to the close-range detection range of the rear corner radar; Radial velocity characteristics: The radar radial velocity of the spurious water-splashing point is strongly correlated with the corresponding rear wheel speed, and the radial velocity range is 0. <v r <k·v wheel , where v r v is the radial velocity of the point cloud detected by radar, k is the pre-calibrated velocity correction factor, and v wheel This corresponds to the wheel speed of the rear wheel; Energy intensity characteristics: The radar echo energy intensity (signal-to-noise ratio SNR) of the water-splashing dummy point is in the preset low range, which is much smaller than the echo energy of real vehicle and pedestrian targets; Point cloud distribution characteristics: The false points of the water splash are distributed in discrete bands and clusters. There are many point clouds in the neighborhood, the radial velocity variance is small, and there are no contour clustering characteristics of real targets.

[0026] Through real-vehicle rain calibration tests, the judgment thresholds corresponding to the theoretical characteristics of each dimension were determined.

[0027] S2. Through the vehicle CAN bus, synchronously acquire vehicle status data and raw point cloud data of the vehicle rear corner millimeter-wave radar, align the timestamps, and determine whether water splash detection is triggered. If triggered, proceed to S3. Water splash detection is triggered when the windshield wipers are on and the vehicle speed is ≥A, where A is the minimum trigger speed. The false water splash detection and removal process is only initiated when both the windshield wipers are on and the vehicle speed is ≥10km / h; otherwise, the original point cloud is output directly without performing the false point detection operation, further reducing invalid calculations and avoiding misjudgments under dry conditions.

[0028] Vehicle status data includes the vehicle's longitudinal speed v ego Left rear wheel speed v rl Right rear wheel speed v rr Tire rolling radius calibration value, wiper status signal, road adhesion coefficient signal; Raw radar point cloud data should include at least: the radial velocity v of each detection point. r Detection distance d, azimuth angle θ, echo energy intensity P (signal-to-noise ratio SNR), timestamp.

[0029] When aligning the timestamps of vehicle status data and the raw point cloud data of the vehicle rear corner millimeter-wave radar, ensure that the error between the two is less than a preset time error threshold, for example, limit the data time synchronization error to no more than 10ms.

[0030] S3. Spatial distribution characteristics based on physical models and their judgment thresholds, such as Figure 2 As shown, a preset target area for water splash detection (ROI) is used to filter out the original point cloud data within the area as a candidate set of false points; In S3, the preset rules for the water-spraying detection target area are as follows: Distance range: d∈[d min ,d max For example, d can be used. min Set to 0.1m, d max The setting is 15m, but d can be adjusted according to the actual situation. min and d max The set value. Used to cover the maximum flight distance of the rear wheel throwing water; Azimuth range: The left rear radar corresponds to θ∈[-b1,b2], and the right rear radar corresponds to θ∈[-b2,b1], where b1 is the detection angle of the radar on the outside of the vehicle body, and b2 is the detection angle of the radar on the rear side of the vehicle body. The radar normal direction is 0°, with a positive angle towards the rear of the vehicle and a negative angle towards the vehicle body, matching the side and rear splash area of ​​water thrown by the rear wheels; for example, based on the actual rear radar characteristics, the obtained azimuth range is: the left rear radar corresponds to θ∈[-30°,60°], and the right rear radar corresponds to θ∈[-60°,30°].

[0031] Point clouds falling within the ROI are filtered as candidate false point sets, while point clouds outside the ROI are directly identified as real points, requiring no further processing and significantly reducing computational load.

[0032] The left rear corner radar matches the speed of the left rear wheel for subsequent determination, and the right rear corner radar matches the speed of the right rear wheel for subsequent determination.

[0033] S4. For each point in the candidate false point set, the range is determined sequentially by other multi-dimensional theoretical features and judgment thresholds. Points that meet all theoretical feature ranges are recorded as false points of tire water ejection. In S4, for each point in the candidate false point set, radial velocity feature determination, energy intensity feature determination, and point cloud distribution feature determination are performed sequentially. Radial velocity characteristic determination: Calculate the radial velocity v of the candidate point. r Determine whether it satisfies k1·v wheel <v r <k2·v wheel k1 is the minimum speed correction coefficient, and k2 is the maximum speed correction coefficient. If these conditions are met, proceed to the next judgment. Specifically, radial velocity characteristics are the core judgment dimension. Based on the radial velocity characteristics of the water-throwing model, the left rear corner radar matches the left rear wheel speed v. rl The radar in the right rear corner matches the wheel speed of the right rear wheel (v). rr Determine the radial velocity v of the candidate point r Does it satisfy 0.2·v? wheel <v r <1.2·v wheel , where 0.2 and 1.2 are pre-calibrated speed correction coefficients, covering speed attenuation caused by air resistance and radial projection error; if satisfied, proceed to the next judgment, otherwise judged as the true point.

[0034] Energy intensity characteristic determination: Determine whether the echo energy intensity P of the candidate point satisfies P min <P<P max , where P min For the minimum detection signal-to-noise ratio of the radar, P maxThe maximum energy threshold for water splashing is pre-calibrated, for example, it can be set to 12dB. The RCS of the water droplet is much smaller than that of the real target, and the echo energy is in the low range. If this condition is met, proceed to the next judgment; otherwise, it is judged as a real point. Point cloud distribution characteristic determination: Statistically count the number of points and radial velocity variance within the preset neighborhood of the candidate point (preferably within ±0.3m of distance and ±3° of azimuth). If the number of points in the neighborhood is greater than or equal to a preset threshold (preferably 5) and the radial velocity variance is less than or equal to a preset variance threshold (preferably 0.5m / s), it is determined to be a false point of tire water ejection; otherwise, it is determined to be a real point. False points of water ejection are distributed in clusters or strips, and the point cloud velocity within the neighborhood is highly consistent, forming a clear distinction from the contour aggregation characteristics of the real target.

[0035] S5. Remove false points of water splashing from the tires from the original point cloud data, and use the remaining point cloud data as the real points.

[0036] Specifically, all false points of tire water ejection identified in S4 are removed from the original radar point cloud, and the remaining real point cloud is output to the subsequent target tracking and target classification algorithm modules. This fundamentally prevents false points of water ejection from being tracked as false dynamic targets and eliminates the possibility of false function triggering.

[0037] After S5, perform the adaptive update step for model parameters in S6.

[0038] S6 is: Under standard operating conditions with no water accumulation and no water splashing, the threshold for judging water splashing characteristics is adaptively calibrated and updated.

[0039] The adaptive update is triggered when the wipers are off and the vehicle stability system detects a road surface adhesion coefficient ≥ 0.8, which is considered a dry condition without water splashing. The updates include the speed correction factor k and the maximum energy threshold P for water ejection. max ROI distance maximum d max The system optimizes the judgment threshold based on near-range point cloud data under dry conditions, adapts to differences in tire wear, radar attenuation, and different vehicle models, continuously optimizes the detection accuracy of the model, and reduces the false judgment rate.

[0040] In S3 and S4, when performing multi-dimensional theoretical feature matching judgment, a weighted voting mechanism can be used to replace the serial judgment. Preset weights are assigned to features of different dimensions, and points with a total score exceeding the judgment threshold are judged as false points.

[0041] Corresponding to the above methods, such as Figure 3 As shown, the present invention also provides a system for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar, comprising: The model building module is used to record the basic parameters of the calibrated wheel, build a physical model of the rear wheel water ejection, and determine the multi-dimensional theoretical characteristics and corresponding judgment thresholds of false water ejection points. The vehicle status information input module is used to acquire vehicle status data; The rear corner millimeter-wave radar point cloud input module is used to acquire the raw point cloud data of the vehicle-mounted rear corner millimeter-wave radar. The time synchronization and coordinate transformation module is used to timestamp-align the vehicle status data and the raw point cloud data of the vehicle rear angle millimeter-wave radar, and transform the raw point cloud data from the RA polar coordinate system centered on the radar itself to the vehicle coordinate system according to the angle and position of the radar installed on the vehicle, so as to facilitate the calculation of subsequent modules. The ROI (Region of Interest) filtering module for water splashing is used to filter out the original point cloud data within the water splashing detection target area based on the spatial distribution characteristics of the physical model and its judgment threshold, as a candidate set of false points. The multi-dimensional false point discrimination module is used to determine the range of each point in the candidate false point set by sequentially using other multi-dimensional theoretical features and judgment thresholds. Points that meet all theoretical feature ranges are recorded as false points of tire water ejection. The false point cloud removal and output module is used to remove false points of water splashed by the tire from the original point cloud data, and output the remaining point cloud data as real points to the subsequent target tracking module.

[0042] Corresponding to the above methods and systems, the present invention also provides a device for detecting and eliminating false points of vehicle-mounted millimeter-wave radar caused by water splashing from tires in rainy weather, comprising: a processor and a memory, wherein the memory stores a computer program executable by the processor, and the processor implements the above methods when executing the computer program.

[0043] As an embodiment of the present invention, a specific implementation method for detecting and eliminating false points of vehicle-mounted millimeter-wave radar caused by water splashing from tires in rainy weather is provided, which is applied to a passenger vehicle equipped with an L2-level intelligent driving assistance system, and the specific configuration is as follows: The vehicle is equipped with two 77GHz automotive millimeter-wave radars, with the left and right rear corner radars mounted on the left and right sides of the rear bumper at a height of 0.35m and a longitudinal distance of 0.8m from the rear axle. The detection range is 0.2m-100m, with an azimuth angle of ±60°, a distance resolution of 0.1m, an angular resolution of 1°, and a point cloud output frequency of 20Hz. The vehicle also features four wheel speed sensors with a wheel speed sampling frequency of 100Hz. The vehicle speed is output by the body controller and sampled at 100Hz. All data is transmitted via the vehicle's CAN bus, with a time synchronization error of ≤5ms. The vehicle's tire specifications are 225 / 55R18, with a calibrated rolling radius R=0.35m.

[0044] The specific implementation steps are as follows: Step 1: Pre-calibration and tire water ejection model construction: Through real-vehicle rain calibration tests, rear wheel water ejection point cloud data were collected at different vehicle speeds (20km / h, 40km / h, 60km / h, 80km / h). A physical model of tire water ejection was established by fitting the data, and the feature library and judgment threshold for false water ejection points were determined. Water-spraying ROI area: distance range d∈[0.1m,12m], left rear radar azimuth θ∈[-20°,50°], right rear radar azimuth θ∈[-50°,20°]; Speed ​​correction factor: k∈[0.3,1.1], covering air resistance attenuation and radial projection error; Energy intensity threshold: P min =3dB (Minimum Detection NR of Radar), P max =12dB; Distribution characteristic thresholds: neighborhood range distance ±0.3m, azimuth angle ±3°, point cloud number threshold ≥5, radial velocity variance threshold ≤0.5m / s.

[0045] Step 2: Activation determination of operating conditions: Real-time vehicle status monitoring: windshield wipers are on (rainy weather condition), vehicle speed v ego =60km / h=16.67m / s≥10km / h, meeting the activation conditions, and the water splash false point detection process is initiated.

[0046] Step 3: Real-time synchronized data acquisition: Simultaneously collect vehicle status data and raw point cloud data from the left rear corner radar at the current moment: Vehicle status data: Vehicle speed v ego =16.67m / s, left rear wheel speed v rl =16.7m / s, right rear wheel speed v rr =16.6m / s, timestamp t1; Raw point cloud data from the rear left radar: a total of 120 detection points, each point containing distance d, azimuth angle θ, and radial velocity v. r SNR value P, timestamp t1, time synchronization error 2ms.

[0047] Step 4: Preliminary ROI screening for water removal: For the 120 original point clouds of the left rear corner radar, 42 points that fall within the ROI region (d∈[0.1,12]m, θ∈[-20°,50°]) are selected as candidate false point sets; the remaining 78 points that are not within the ROI are directly determined as real points.

[0048] Step 5: Multi-dimensional feature matching determination: For the 42 candidate false points, each one was sequentially evaluated: Radial velocity feature determination: The wheel speed v of the left rear wheel rl = 16.7 m / s, and the speed determination range is 0.3×16.7 = 5.01 m / s < v r <1.1×16.7 = 18.37 m / s; among the 42 candidate points, 36 points meet this condition and enter the next determination, and the remaining 6 points are determined as real points.

[0049] Energy intensity feature determination: For the 36 candidate points, determine whether the SNR value satisfies 3 dB < P < 12 dB; among them, 32 points meet this condition and enter the next determination, and the remaining 4 points have P > 12 dB and are determined as real points.

[0050] Point cloud distribution feature determination: For the 32 candidate points, count the number of point clouds in the neighborhood and the radial velocity variance one by one; among them, 28 points satisfy that the number of points in the neighborhood ≥ 5 and the radial velocity variance ≤ 0.5 m / s, and are determined as false tire water splashing points; the remaining 4 points do not meet the conditions and are determined as real points.

[0051] Step 6: False point elimination and result output: Eliminate the 28 false water splashing points obtained from the determination from the original 120 point clouds, and output the remaining 92 real point clouds to the subsequent target tracking and classification module, effectively avoiding the tracking of water splashing point clouds as false dynamic targets and eliminating the false triggering of lane change assistance and blind spot monitoring functions.

[0052] Step 7: Adaptive update of model parameters: When the vehicle travels on a dry road surface, the windshield wiper is turned off, and the vehicle body stability system detects that the road surface adhesion coefficient is 0.85, which meets the adaptive update trigger condition; collect the radar near-range point cloud data under this working condition, and perform adaptive optimization updates on the energy threshold P max and the speed correction coefficient k to correct the parameter deviation caused by tire wear and continuously improve the model detection accuracy.

[0053] As described above, only the preferred specific embodiments of the present invention are provided, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

[0054] It should be understood that in order to streamline the present invention and help those skilled in the art understand various aspects of the present invention, in the above description of the exemplary embodiments of the present invention, various features of the present invention are sometimes described in a single embodiment or with reference to a single figure. However, the present invention should not be construed as meaning that the features included in the exemplary embodiments are all essential technical features of the claims of the present invention.

[0055] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0056] It should be understood that the modules, units, components, etc., included in the device of one embodiment of the present invention can be adaptively changed to be placed in a device different from that embodiment. Different modules, units, or components included in the device of the embodiment can be combined into a single module, unit, or component, or they can be divided into multiple sub-modules, sub-units, or sub-components.

[0057] The modules, units, or components in the embodiments of the present invention can be implemented in hardware, in software running on one or more processors, or in a combination thereof. Those skilled in the art should understand that... In practice, microprocessors or digital signal processors (DSPs) can be used to implement embodiments of the invention. The invention can also be implemented on computer program products or computer-readable media for performing some or all of the methods described herein.

Claims

1. A method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar, characterized by: Includes the following steps: S1. Calibrate the basic parameters of the wheel, construct a physical model of the rear wheel throwing water, and determine the multi-dimensional theoretical characteristics and corresponding judgment thresholds of false water throwing points; S2. Synchronously acquire vehicle status data and raw point cloud data of the vehicle rear corner millimeter-wave radar, align the timestamps, and determine whether water splash detection is triggered. If triggered, execute S3. S3. Based on the spatial distribution characteristics and judgment threshold of the physical model, a target area for water splashing detection is preset, and the original point cloud data in the area is selected as a candidate false point set. S4. For each point in the candidate false point set, the range is determined sequentially by other multi-dimensional theoretical features and judgment thresholds. Points that meet all theoretical feature ranges are recorded as false points of tire water ejection. S5. Remove false points of water splashing from the tires from the original point cloud data, and use the remaining point cloud data as the real points.

2. The method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar according to claim 1, characterized in that: In S1, the physical model for rear wheel water ejection is constructed as follows: Establish the vehicle coordinate system; Based on the vehicle coordinate system and the pure rolling dynamics of the wheel, the velocity vector relative to the ground when the water droplet leaves the tire is defined. Based on the ground velocity vector of the water droplet when it leaves the tire, and combined with the influence of air resistance and gravity on the flight trajectory of the water droplet, the multidimensional theoretical characteristics of the false water-throwing point are obtained by fitting. Through real-vehicle rain calibration tests, the judgment thresholds corresponding to the theoretical characteristics of each dimension were determined.

3. The method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar, as described in claim 2, is characterized in that: The velocity vector of the water droplet relative to the ground when it leaves the tire is V. drop =V wheel center +ω×r, where V wheel center The velocity of the rear wheel center relative to the ground is given by r, where r is the position vector of the water droplet relative to the wheel center; ω is the angular velocity of the wheel, ω = v. wheel / R,v wheel R is the rear wheel speed, and R is the calibrated tire rolling radius.

4. The method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar according to claim 1, characterized in that: The multidimensional theoretical characteristics include: spatial distribution characteristics, radial velocity characteristics, energy intensity characteristics, and point cloud distribution characteristics; Spatial distribution characteristics are used to limit the distribution space of false water-throwing points, which are only distributed in the close-range area behind the rear wheel, corresponding to the close-range detection range of the rear corner radar; Radial velocity characteristic: Used to define the correlation between the radar radial velocity and the rear wheel speed at the false water-throwing point; the radial velocity range is 0. <v r <k·v wheel , where v r The radial velocity of the point cloud detected by radar is given by k, which is a pre-calibrated velocity correction factor. Energy intensity characteristics: used to define the range of radar echo energy intensity for decoy water-throwing points; Point cloud distribution characteristics: used to define the spurious points of water splashing as having a discrete band-like or clustered distribution shape, with a large number of point clouds in the neighborhood, small radial velocity variance, and no contour aggregation characteristics of real targets.

5. The method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar, as described in claim 1. Its features are: In S2, the vehicle status data includes the vehicle's longitudinal speed, left rear wheel speed, right rear wheel speed, tire rolling radius calibration value, wiper status signal, and road surface adhesion coefficient signal; The raw radar point cloud data includes at least the following: radial velocity, detection range, azimuth angle, echo energy intensity, and timestamp for each detection point.

6. The method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar according to claim 4, characterized in that: In S4, for each point in the candidate false point set, radial velocity feature determination, energy intensity feature determination, and point cloud distribution feature determination are performed sequentially. Radial velocity characteristic determination: Calculate the radial velocity v of the candidate point. r Determine whether it satisfies k1·v wheel <v r <k2·v wheel k1 is the minimum speed correction coefficient, and k2 is the maximum speed correction coefficient. If these conditions are met, proceed to the next judgment. Energy intensity characteristic determination: Determine whether the echo energy intensity P of the candidate point satisfies P min <P<P max , where P min For the minimum detection signal-to-noise ratio of the radar, P max If the pre-calibrated maximum energy threshold for water ejection is met, proceed to the next judgment. Point cloud distribution characteristic determination: Statistically count the number of points in the preset neighborhood of the candidate point and the radial velocity variance. If the number of points in the neighborhood is greater than or equal to the preset number threshold and the radial velocity variance is less than or equal to the preset variance threshold, then it is determined to be a false point of tire water ejection.

7. The method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar according to claim 1, characterized in that: After S5, S6 is executed. S6 is: under the standard working condition of no water accumulation and no water splashing, the threshold for judging water splashing characteristics is adaptively calibrated and updated.

8. The method for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar according to claim 1, characterized in that: In S3 and S4, when performing multi-dimensional theoretical feature matching and judgment, a weighted voting mechanism is used instead of serial judgment. Preset weights are assigned to features of different dimensions, and points with total scores exceeding the judgment threshold are judged as false points.

9. A system for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar, characterized in that: include: The model building module is used to record the basic parameters of the calibrated wheel, build a physical model of the rear wheel water ejection, and determine the multi-dimensional theoretical characteristics and corresponding judgment thresholds of false water ejection points. The vehicle status information input module is used to acquire vehicle status data; The rear corner millimeter-wave radar point cloud input module is used to acquire the raw point cloud data of the vehicle-mounted rear corner millimeter-wave radar. The time synchronization and coordinate transformation module is used to timestamp-align vehicle status data and raw point cloud data of the vehicle rear angle millimeter-wave radar, and transform the raw point cloud data from the RA polar coordinate system centered on the radar itself to the vehicle coordinate system according to the angle and position of the radar installed on the vehicle. The ROI (Region of Interest) filtering module for water splashing is used to filter out the original point cloud data within the water splashing detection target area based on the spatial distribution characteristics of the physical model and its judgment threshold, as a candidate set of false points. The multi-dimensional false point discrimination module is used to determine the range of each point in the candidate false point set by sequentially using other multi-dimensional theoretical features and judgment thresholds. Points that meet all theoretical feature ranges are recorded as false points of tire water ejection. The false point cloud removal and output module is used to remove false points of water splashed by the tire from the original point cloud data, and output the remaining point cloud data as real points to the subsequent target tracking module.

10. A device for detecting and eliminating false points caused by tire water splashing in rainy weather using vehicle-mounted millimeter-wave radar, characterized in that: include: A processor and a memory, the memory storing a computer program executable by the processor, wherein the processor, when executing the computer program, implements the method of any one of claims 1-8.

Images