Blind area target danger level evaluation method and system based on doppler phase frequency characteristics

Abstract

The application provides a blind area target danger level evaluation method and system based on Doppler phase frequency characteristics, and relates to the technical field of vehicle blind area monitoring.The method comprises the following steps: radiating a linear frequency modulation continuous wave signal to a vehicle blind area, capturing a coherent echo scattered by a target, and performing mixing filtering and analog-digital sampling processing to obtain a complex baseband signal sequence; based on the complex baseband signal sequence, analyzing gradient information of phase changes of the complex baseband signal and a coupling relationship with a Doppler effect, and constructing Doppler phase frequency characteristic data; based on the Doppler phase frequency characteristic data, selecting three physical aperture centers in a radar receiving antenna array as spatial sampling nodes; constructing a three-dimensional dynamic Doppler potential field surface through phase differences between the spatial sampling nodes; and performing an isosurface topological segmentation on the three-dimensional dynamic Doppler potential field surface to extract a discrete scattering point set.The application realizes accurate estimation of a blind area target motion state and scientific determination of a danger level.

Description

Technical Field

[0001] This invention relates to the field of vehicle blind spot monitoring technology, and in particular to a method and system for assessing the hazard level of blind spot targets based on Doppler phase frequency characteristics. Background Technology

[0002] With the widespread adoption of advanced driver assistance systems, blind spot monitoring has become a key function for improving driving safety. Currently, most mainstream solutions use millimeter-wave radar to detect the position and speed of targets by transmitting frequency-modulated continuous waves and receiving target echoes.

[0003] There is a specific technical defect in the existing implementation of blind zone target detection based on millimeter-wave radar: the accuracy of target motion state estimation is insufficient in blind zone large-angle detection scenarios.

[0004] Traditional processing methods mainly rely on the amplitude spectrum and frequency information of echo signals for target calculation, ignoring the gradient information of phase changes in complex baseband signals and their coupling relationship with the Doppler effect. This makes it difficult to effectively extract weak target features in blind zone environments with low signal-to-noise ratios or multipath interference. Blind zone targets are usually located at the edge of the radar detection field of view, and radar measurements are easily affected by spatial distortion, resulting in nonlinear errors. Existing technologies mostly use linear models for parameter calculation, lacking spatial distortion correction mechanisms for range drift and angle deviation. The combined effect of insufficient phase information utilization and lack of spatial distortion correction leads to deviations in the calculated target relative velocity, relative range, and azimuth angle, making the hazard level assessment results based on this motion state unreliable and unable to meet the needs of high-precision blind zone early warning. Summary of the Invention

[0005] This invention provides a method and system for assessing the hazard level of blind zone targets based on Doppler phase frequency characteristics, enabling accurate estimation of the motion state of blind zone targets and scientific determination of their hazard level.

[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: Firstly, a method for assessing the hazard level of blind zone targets based on Doppler phase frequency characteristics, the method comprising: A linear frequency modulated continuous wave signal is radiated into the vehicle's blind spot to capture the coherent echo scattered by the target, and then mixed, filtered, and processed by analog-to-digital sampling to obtain a complex baseband signal sequence. Based on complex baseband signal sequences, the gradient information of the phase change of complex baseband signals and their coupling relationship with the Doppler effect are analyzed to construct Doppler phase frequency characteristic data. Based on Doppler phase frequency characteristic data, three physical aperture centers in the radar receiving antenna array are selected as spatial sampling nodes; a three-dimensional dynamic Doppler potential field surface is constructed by the phase difference between each spatial sampling node; and isosurface topological segmentation is performed on the three-dimensional dynamic Doppler potential field surface to extract discrete scattering point sets. Based on the discrete scattering point set, the boundary of the smallest circular region covering the discrete scattering points is determined iteratively. The radius expansion rate and center drift of the boundary of the smallest circular region are extracted as spatial distortion indices. Based on the spatial distortion indices, nonlinear correction processing is performed on the radial relative velocity, relative distance and azimuth parameters to obtain the corrected blind zone target motion state estimation results. Based on the corrected blind spot target motion state estimation results, combined with the vehicle's speed, steering angle and acceleration information, the estimated collision time, predicted minimum approach distance and relative motion convergence trend index are calculated in a unified spatiotemporal coordinate system. Based on the estimated collision time, predicted minimum approach distance, and relative motion convergence trend indicators, the results are compared with preset multi-level hazard thresholds to obtain an assessment result that includes hazard level identification and warning suggestions.

[0007] Furthermore, a linear frequency modulated continuous wave signal is radiated into the vehicle's blind spot to capture the coherent echo scattered by the target, and then mixed, filtered, and processed by analog-to-digital sampling to obtain a complex baseband signal sequence, including: The radar transmitting unit radiates a linear frequency modulated continuous wave signal to the blind spot area to the side and rear of the vehicle. The linear frequency modulated continuous wave signal is blocked by the target in the blind spot on the propagation path and forms a scattered echo. The scattered echo is mixed with the local transmitted reference signal to obtain a beat signal containing range and velocity information. The beat signal is subjected to low-pass filtering to remove high-frequency mixing components, resulting in a baseband analog signal; the baseband analog signal is then sampled by analog-to-digital conversion to obtain a discrete-time complex baseband signal sequence.

[0008] Furthermore, based on the complex baseband signal sequence, the gradient information of the phase change of the complex baseband signal and its coupling relationship with the Doppler effect are analyzed to construct Doppler phase frequency characteristic data, including: Analyze the complex baseband signal sequence and extract the phase value of each sampling point to construct a phase time series; Perform a difference operation on the phase time series to calculate the phase change between adjacent sampling points, and obtain the gradient information of the phase change based on the phase change. The rate of phase change over time is calculated by using the gradient information of phase change. Based on the physical correspondence between the rate of phase change and Doppler frequency shift, the Doppler frequency information of the target is calculated to establish the coupling mapping relationship between phase gradient and Doppler effect. Based on the coupling mapping relationship and Doppler frequency information, the gradient information of phase change is fused with the amplitude information in the complex baseband signal sequence to construct Doppler phase frequency feature data for the motion state characteristics of targets in the blind zone.

[0009] Furthermore, based on Doppler phase-frequency characteristic data, three physical aperture centers in the radar receiving antenna array are selected as spatial sampling nodes; a three-dimensional dynamic Doppler potential field surface is constructed using the phase difference between each spatial sampling node; and isosurface topological segmentation is performed on the three-dimensional dynamic Doppler potential field surface to extract a discrete scattering point set, including: Three physical aperture centers were selected in the radar receiving antenna array as spatial sampling nodes, and the phase information corresponding to each spatial sampling node was extracted from the Doppler phase frequency characteristic data. Calculate the phase difference between the phase information corresponding to each spatial sampling node, and calculate the Doppler potential field intensity value of each sampling node in the spatial dimension based on the phase difference; Step 3.3: Based on the Doppler potential field intensity value, according to the spatial coordinate position of each spatial sampling node, calculate the potential field intensity value of the unknown position between each spatial sampling node and make a smooth connection to construct a continuously changing three-dimensional dynamic Doppler potential field surface. The isosurface topology segmentation process is performed on the three-dimensional dynamic Doppler potential surface to identify energy accumulation regions on the surface, and discrete scattering point sets are extracted from the energy accumulation regions.

[0010] Furthermore, based on the discrete scattering point set, the boundary of the smallest circular region covering the discrete scattering points is iteratively determined, and the radius expansion rate and center drift of the smallest circular region boundary are extracted as spatial distortion indices. Based on the spatial distortion indices, nonlinear correction processing is performed on the radial relative velocity, relative distance, and azimuth parameters to obtain the corrected blind zone target motion state estimation results, including: Using each scattering point in the discrete scattering point set as a geometric constraint, the center coordinates and radius length of the circle are continuously adjusted to verify whether all discrete scattering points are located inside the circular region until the boundary of the circular region with the smallest radius that covers all discrete scattering points is found. The center coordinates and radius length of the smallest circular region are then determined as boundary parameters. Based on boundary parameters, the dynamic change trajectory is tracked within a continuously acquired time frame sequence. The rate of change of the radius length per unit time is calculated to obtain the radius expansion rate. The displacement vector of the center coordinate per unit time is calculated to obtain the center drift. The radius expansion rate and the center drift are defined as spatial distortion indices of the target shape change. A nonlinear correction function is constructed using spatial distortion indices, and a measurement error compensation factor is calculated. The radial relative velocity, relative distance, and azimuth parameters obtained from the original measurements are input into the nonlinear correction function. The measurement deviation is compensated and corrected using the measurement error compensation factor to obtain the corrected motion parameters. These corrected motion parameters are then used as the corrected estimation result of the target motion state in the blind zone. Furthermore, based on the corrected blind spot target motion state estimation results, combined with the vehicle's speed, steering angle, and acceleration information, the estimated collision time, predicted minimum approach distance, and relative motion convergence trend indicators are calculated in a unified spatiotemporal coordinate system, including: Obtain the current driving speed, steering angle and acceleration values ​​of the vehicle, perform time synchronization processing on the values, and construct the real-time motion state vector of the vehicle. By combining the instantaneous motion state vector of the vehicle with the corrected motion state estimation results of the blind spot target, spatiotemporal alignment and fusion are performed in a unified spatiotemporal coordinate system to obtain the relative motion trajectory sequence of the vehicle and the blind spot target. Based on the relative motion trajectory sequence, the positional changes of the vehicle and the target in the blind spot are extrapolated forward along the time axis using a kinematic model. The predicted collision time when the two reach the minimum value is calculated, and the predicted minimum approach distance at the minimum value is recorded. Based on the expected collision time and predicted minimum approach distance, and combined with the rate of change of relative distance over time in the relative motion trajectory sequence, the urgency of the two motions approaching each other is analyzed, and an index of the relative motion convergence trend is obtained.

[0011] Furthermore, based on the estimated collision time, predicted minimum approach distance, and relative motion convergence trend indicators, these are compared with preset multi-level hazard thresholds to obtain an assessment result that includes hazard level indicators and warning suggestions, including: By calling the preset multi-level hazard level threshold configuration table, the configuration table defines the critical threshold ranges of the expected collision time, predicted minimum approach distance, and relative motion convergence trend indicators under different hazard levels. By using a multi-level hazard level threshold configuration table, the predicted collision time, predicted minimum approach distance, and relative motion convergence trend indicators are compared with the critical threshold range to determine the current hazard level range of each indicator. Based on the current danger level range of each indicator, a comprehensive logical judgment is made according to the preset weight priority to obtain the danger level label. Based on the hazard level identification, corresponding early warning strategies are matched to obtain early warning suggestions that include visual warning level, auditory warning frequency, and driving assistance intervention type. The final evaluation result includes hazard level identification and early warning suggestions.

[0012] Secondly, a blind zone target hazard level assessment system based on Doppler phase frequency characteristics includes: The acquisition module is used to radiate linear frequency modulated continuous wave signals into the vehicle's blind spot, capture the coherent echoes scattered by the target, and perform mixing filtering and analog-to-digital sampling processing to obtain a complex baseband signal sequence. The analysis module is used to analyze the gradient information of the phase change of the complex baseband signal and its coupling relationship with the Doppler effect based on the complex baseband signal sequence, and to construct Doppler phase frequency characteristic data. The module is used to select the centers of three physical apertures in the radar receiving antenna array as spatial sampling nodes based on Doppler phase frequency characteristic data; construct a three-dimensional dynamic Doppler potential field surface through the phase difference between each spatial sampling node; and perform isosurface topological segmentation on the three-dimensional dynamic Doppler potential field surface to extract discrete scattering point sets. The execution module is used to iteratively determine the boundary of the smallest circular region covering the discrete scattering points based on the discrete scattering point set, and extract the radius expansion rate and center drift of the boundary of the smallest circular region as spatial distortion indices; according to the spatial distortion indices, nonlinear correction processing is performed on the radial relative velocity, relative distance and azimuth parameters to obtain the corrected blind zone target motion state estimation results; The calculation module is used to calculate the estimated collision time, the predicted minimum approach distance, and the relative motion convergence trend index in a unified spatiotemporal coordinate system based on the corrected blind spot target motion state estimation results and the vehicle's driving speed, steering angle, and acceleration information. The judgment module is used to compare the predicted collision time, predicted minimum approach distance, and relative motion convergence trend indicators with preset multi-level hazard thresholds to obtain an assessment result that includes hazard level identification and warning suggestions.

[0013] Thirdly, a computing device, comprising: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0014] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0015] The above-described solution of the present invention has at least the following beneficial effects: This invention overcomes the technical problems of insufficient phase information utilization, difficulty in extracting weak target features, and lack of spatial distortion correction mechanism in traditional blind spot detection, which lead to nonlinear errors in target motion state estimation and unreliable hazard level assessment results. It effectively improves the ability to extract blind spot target features, accurately compensates for distance drift and angle deviation, and improves the accuracy of blind spot target motion state estimation. This results in more accurate and reliable blind spot target hazard level assessment results, meeting the technical requirements of high-precision blind spot safety warning for vehicles. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the blind zone target hazard level assessment method based on Doppler phase frequency characteristics provided in an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of a blind zone target hazard level assessment system based on Doppler phase frequency characteristics provided in an embodiment of the present invention. Detailed Implementation

[0018] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0019] like Figure 1 As shown, embodiments of the present invention propose a method for assessing the hazard level of blind zone targets based on Doppler phase frequency characteristics. The method includes the following steps: Step 1: Radiate a linear frequency modulated continuous wave signal into the vehicle's blind spot, capture the coherent echo scattered by the target, and perform mixing filtering and analog-to-digital sampling processing to obtain a complex baseband signal sequence; Step 2: Based on the complex baseband signal sequence, analyze the gradient information of the phase change of the complex baseband signal and its coupling relationship with the Doppler effect to construct Doppler phase frequency characteristic data; Step 3: Based on the Doppler phase frequency characteristic data, select the three physical aperture centers in the radar receiving antenna array as spatial sampling nodes; construct a three-dimensional dynamic Doppler potential field surface through the phase difference between each spatial sampling node; perform isosurface topological segmentation on the three-dimensional dynamic Doppler potential field surface to extract discrete scattering point sets; Step 4: Based on the discrete scattering point set, iteratively determine the boundary of the smallest circular region covering the discrete scattering points, and extract the radius expansion rate and center drift of the boundary of the smallest circular region as spatial distortion indices; perform nonlinear correction processing on the radial relative velocity, relative distance and azimuth parameters according to the spatial distortion indices to obtain the corrected blind zone target motion state estimation results; Step 5: Based on the corrected blind spot target motion state estimation results, combined with the vehicle's speed, steering angle, and acceleration information, calculate the estimated collision time, predicted minimum approach distance, and relative motion convergence trend index in a unified spatiotemporal coordinate system. Step 6: Based on the estimated collision time, predicted minimum approach distance, and relative motion convergence trend indicators, compare and determine with the preset multi-level hazard thresholds to obtain an assessment result that includes hazard level identification and warning suggestions.

[0020] In this embodiment of the invention, by radiating a linear frequency modulated continuous wave signal into the vehicle's blind spot and processing it to obtain a complex baseband signal sequence, analyzing the coupling relationship between the phase change gradient and the Doppler effect to construct Doppler phase-frequency characteristic data, selecting the three physical apertures of the radar receiving array as sampling nodes to construct a three-dimensional dynamic Doppler potential field surface and extracting discrete scattering point sets, using the radius expansion rate and center drift of the smallest circular region covering the scattering points as spatial distortion indicators, nonlinear correction is performed on radial relative velocity, relative distance, and azimuth angle, and the collision risk index is calculated by combining the vehicle's motion information and the danger level is determined by using preset multi-level thresholds, the technical means overcome the technical problems of insufficient phase information utilization, difficulty in extracting weak target features, and lack of spatial distortion correction mechanism leading to nonlinear errors in motion parameter measurement and unreliable danger level assessment results in traditional blind spot target detection. This achieves improved blind spot target feature extraction capability and motion state estimation accuracy, effectively compensates for distance drift and angle deviation, makes blind spot target danger level assessment more accurate, and meets the technical effect of high-precision blind spot safety warning for vehicles.

[0021] In a preferred embodiment of the present invention, step 1 above may include: Step 1.1: The radar transmitting unit radiates a linear frequency modulated continuous wave (LFM) signal to the blind zone area behind the vehicle. The LFM signal, upon encountering a target in the blind zone, forms a scattered echo. Specifically, the transmitting unit of the vehicle-mounted millimeter-wave radar continuously radiates a LFM signal to the blind zone area behind the vehicle with fixed modulation parameters. The frequency of this signal changes linearly with time, maintaining coherence during propagation. The signal propagation range completely covers the preset blind zone monitoring area behind the vehicle, ensuring that all targets within the blind zone are illuminated. When the LFM signal encounters vehicles, non-motorized vehicles, pedestrians, or other targets within the blind zone, specular and diffuse reflection occurs on the target surface, forming a scattered echo signal carrying information about the target's distance, speed, and spatial location. This scattered echo signal retains the coherence characteristics of the original transmitted signal and is transmitted to the radar receiving array along the reverse path of the original propagation. The radar receiving unit fully receives and preliminarily conditions this coherent scattered echo signal, effectively capturing the coherent echo of the target in the blind zone.

[0022] Step 1.2 involves mixing the scattered echo with the local transmitted reference signal to obtain a beat signal containing range and velocity information. Specifically, this includes: synchronously inputting the coherent scattered echo signal captured and conditioned by the radar receiving unit and the local transmitted reference signal generated by the radar's internal oscillator into the mixer for mixing. The core operation of the mixing process is multiplying the coherent scattered echo signal with the local transmitted reference signal.

[0023] Let the local transmission reference signal be the original linear frequency modulated continuous wave signal at the transmitting end, denoted as . The captured coherent scattered echo signal, due to propagation delay and Doppler frequency shift, is denoted as... The core calculations performed by the mixer are: Based on the operational characteristics of trigonometric functions and frequency-modulated signals, this multiplication operation can decompose the two signals into two components: frequency addition and frequency subtraction. After expansion, the following conditions are met: , The high-frequency and frequency components are formed by the superposition of the frequencies of the two signals. The low-frequency difference component is formed by subtracting the frequencies of the two signals. The high-frequency sum component is a useless interference component generated by mixing and cannot characterize the target features. The low-frequency difference component is the beat signal required for this processing. Since the coherent scattered echo signal has a propagation delay introduced by the target distance and a Doppler shift introduced by the radial motion of the target compared to the local transmitted reference signal, the frequency characteristics of the beat signal are jointly determined by the propagation delay and the Doppler shift. It directly maps the relative distance information and radial relative motion velocity information of the target in the blind zone. Through mixing and multiplication, the high-frequency radio frequency echo signal can be converted into a low-frequency beat signal, so that the position and motion characteristics of the target can be stably processed by the subsequent filtering, sampling and feature analysis circuits.

[0024] Step 1.3: Perform a low-pass filter operation on the beat signal to remove high-frequency mixing components, obtaining the baseband analog signal; perform analog-to-digital conversion sampling on the baseband analog signal to obtain a discrete-time complex baseband signal sequence. Specifically, the mixed signal output by the mixer contains both high-frequency sum-frequency components and low-frequency difference-frequency components. These two types of components differ in frequency. The high-frequency sum-frequency components do not carry any effective target information and will severely interfere with subsequent feature analysis. Therefore, the mixed signal needs to be sent to a low-pass filter in the radar signal processing link for filtering. This low-pass filter has been pre-configured with a matching cutoff frequency based on the beat signal frequency range corresponding to blind zone target detection. This cutoff frequency is higher than... The highest frequency of all possible low-frequency difference components is much lower than that of high-frequency sum and saturation components. During signal filtering, the low-pass filter completely suppresses the amplitude and blocks the transmission of high-frequency sum and saturation components whose frequencies exceed the cutoff frequency, making it difficult for them to enter the processing link. For low-frequency difference components whose frequencies are lower than the cutoff frequency, distortion-free transmission is maintained, and their amplitude and phase characteristics are fully preserved. After low-pass filtering, useless high-frequency interference components in the mixed signal are completely filtered out, leaving only the low-frequency difference components that can accurately reflect the relative distance and radial relative motion velocity information of the target in the blind zone. The continuous time-domain signal output by the filter is an interference-free, pure baseband analog signal. This filtering process can be expressed as: In the formula, The mixed signal output by the mixer. This is a low-pass filter operation. This is the baseband analog signal obtained after filtering.

[0025] After obtaining the clean baseband analog signal, the analog-to-digital conversion (ADC) sampling stage begins. The ADC module performs synchronous sampling at equal intervals according to a pre-set fixed sampling frequency. The time interval between two adjacent samples, i.e., the sampling period, is calculated based on the set sampling frequency. The sampling period and sampling frequency are reciprocals of each other, and the calculation formula is as follows: In the formula The sampling frequency for analog-to-digital conversion. For the corresponding sampling period, at each discrete sampling time ,in The analog-to-digital converter (ADC) acquires the instantaneous amplitude of the baseband analog signal with high precision, starting from 0 and incrementing sequentially as non-negative integers. At the same time, it completely preserves the phase information carried by the signal, converting the baseband analog signal, which originally varied in the continuous time domain, into a digital signal distributed in the discrete time domain. The entire sampling process follows the equal interval rule and does not lose the amplitude and phase complex features in the baseband analog signal. After sampling and quantization at each time step, a discrete-time complex baseband signal sequence with continuous time sequence, complete features, and no interference is finally obtained. This signal sequence can be used for phase change gradient analysis and the construction of Doppler phase frequency characteristic data.

[0026] In this embodiment of the invention, by transmitting a coherent linear frequency modulated continuous wave signal and capturing the target's scattered coherent echo, the beat signal containing target range and velocity information is separated by multiplying and mixing the echo signal with a local reference signal. High-frequency interference components are eliminated through low-pass filtering, and analog signals are converted into discrete digital signals through analog-to-digital conversion. This overcomes the technical problems of insufficient coherence utilization, difficulty in suppressing high-frequency interference, inability to separate the coupling of target range and velocity information, and the inability of analog signals to be directly used for phase and Doppler feature analysis in traditional blind zone radar signal preprocessing. This results in a complex baseband signal sequence free from high-frequency clutter interference, with complete feature information, and retaining all phase and amplitude information. This provides a clean and reliable data foundation for subsequent analysis of phase change gradients and establishment of the coupling relationship between phase and Doppler effects. It improves the extractability of weak target features from the signal acquisition and preprocessing stages, laying the data prerequisite for spatial distortion correction and accurate hazard level assessment.

[0027] In a preferred embodiment of the present invention, step 2 above may include: Step 2.1 involves analyzing the complex baseband signal sequence and extracting the phase value of each sampling point to construct a phase time series. Specifically, after obtaining the discrete-time complex baseband signal sequence, point-by-point analysis is immediately performed. Each sampling point in the discrete-time complex baseband signal sequence exists in complex form, containing both amplitude information characterizing signal strength and phase information reflecting target motion characteristics. Phase information is more sensitive to Doppler effect changes caused by the target's radial motion and is the core foundation for subsequently constructing Doppler phase-frequency characteristics and analyzing the target's motion state. It is also a key step in overcoming the insufficient utilization of phase information in traditional techniques. During the analysis, a complex signal analysis algorithm is used to accurately analyze each sampling point. By separating the real and imaginary parts of the complex signal, the phase value corresponding to each sampling point is extracted. The analysis process follows the physical rules of complex phase extraction to ensure the accuracy of phase value extraction at each sampling point, without losing any phase details, while avoiding potential phase ambiguity issues during phase extraction. For the extracted phase values, if there are any values ​​exceeding... In cases where the phase value is within a certain range, timely phase unwrapping should be performed to correct the phase value to a reasonable range and ensure the continuity of phase information.

[0028] After extraction, the phase values ​​of all sampling points are arranged sequentially according to the temporal order of the sampling points to construct a continuous phase time series. This phase time series is completely synchronized with the time series of the discrete-time complex baseband signal. Each time series node has a unique phase value that matches it, and the trend of phase value change is consistent with the target motion state. This provides orderly, complete, and accurate phase data support for subsequent phase change analysis of adjacent sampling points and phase gradient calculation, ensuring the reliability of the calculation results.

[0029] Step 2.2 involves performing a difference operation on the phase time series to calculate the phase change between adjacent sampling points and then calculating the gradient information of the phase change based on the phase change. Specifically, this includes performing a difference operation on the constructed phase time series. The core purpose of the difference operation is to capture the phase change pattern between adjacent sampling points, and then calculate the phase gradient over time, laying the foundation for correlating the Doppler effect. The difference operation is performed point-by-point, traversing the entire phase time series, selecting any two adjacent sampling points in the series, and subtracting the phase value of the previous sampling point from the phase value of the subsequent sampling point to obtain the phase change between these two adjacent sampling points. Each pair of adjacent sampling points corresponds to an independent phase change, calculated using the following formula: In the formula, For the first Phase value at each sampling point For the first Phase value at each sampling point The phase change between two adjacent sampling points is denoted as , where The positive and negative values ​​represent the direction of phase change. A positive value indicates that the phase increases with time, while a negative value indicates that the phase decreases with time. The absolute value represents the magnitude of the phase change, which directly reflects the instantaneous change in the target's motion state.

[0030] After obtaining the phase change at all adjacent sampling points, the gradient information of the phase change is further calculated based on the phase change. Since the sampling process of the discrete-time complex baseband signal sequence is performed at a fixed sampling frequency, the time interval between two adjacent sampling points is a fixed sampling period. The sampling period is exactly the same as the sampling period of the analog-to-digital conversion, that is... The sampling period is consistent with the sampling period of the analog-to-digital converter. Since the sampling frequency is constant, the gradient information of the phase change is essentially the amount of phase change per unit time. Its calculation process involves dividing the phase change between adjacent sampling points by the sampling period. The formula is: In the formula, For gradient information of phase change, This represents the phase change between adjacent sampling points. The sampling period is the time interval between two adjacent samples. Through this calculation, the originally discrete phase change is transformed into continuous gradient information, which accurately represents the trend of phase change over time. The magnitude of the gradient value directly reflects the rate of phase change, and thus indirectly reflects the change in the target's velocity, providing data for calculating the Doppler frequency.

[0031] Step 2.3: Calculate the rate of phase change over time using the gradient information of the phase change. Based on the physical correspondence between the rate of phase change and the Doppler frequency shift, calculate the Doppler frequency information of the target to establish the coupling mapping relationship between the phase gradient and the Doppler effect. Specifically, according to the basic physical principles of radar signal processing, there is a fixed quantitative physical correspondence between the rate of phase change over time and the Doppler frequency shift. The Doppler frequency shift is the signal frequency shift caused by the radial relative motion of the target, and the rate of phase change over time is exactly equal to... The product of the Doppler frequency shift and the phase change rate can be used to calculate the Doppler frequency information of the target. Based on this correspondence, the Doppler frequency information can be obtained by inverse calculation. The specific calculation formula is as follows: In the formula, Doppler frequency information for the target The gradient information of phase change is calculated using this formula, which yields the Doppler frequency of the target corresponding to each adjacent sampling point, forming a continuous Doppler frequency sequence. This sequence can reflect the changes in the radial relative motion velocity of the target in real time.

[0032] Through the above calculations, the quantitative correspondence between the phase change gradient and the Doppler frequency shift was clarified, thus establishing the coupled mapping relationship between the phase gradient and the Doppler effect: an increase or decrease in the phase change gradient directly leads to a synchronous increase or decrease in the Doppler frequency shift, and the change in the Doppler frequency shift directly corresponds to the change in the radial relative motion velocity of the target. That is, the larger the phase gradient, the larger the Doppler frequency shift, and the faster the radial relative motion velocity of the target; the smaller the phase gradient, the smaller the Doppler frequency shift, and the slower the radial relative motion velocity of the target. This coupled mapping relationship closely links the previously ignored phase characteristics with the target motion characteristics, breaking the limitation of traditional techniques that rely solely on amplitude and frequency information, and providing core theoretical support and quantitative calculation basis for the subsequent accurate analysis of the target motion state.

[0033] Step 2.4: Based on the coupling mapping relationship and Doppler frequency information, the gradient information of phase change is fused with the amplitude information in the complex baseband signal sequence to construct Doppler phase-frequency feature data for the motion state characteristics of targets in the blind zone. Specifically, this includes: based on the coupling mapping relationship between phase gradient and Doppler effect, and the calculated continuous Doppler frequency information, further deep fusion processing is performed on the gradient information of phase change, Doppler frequency information, and amplitude information in the discrete-time complex baseband signal sequence. The core purpose of fusion is to integrate target motion characteristics and signal intensity characteristics to construct feature data that can comprehensively and accurately characterize the motion state of targets in the blind zone, solving the problems of incomplete target feature characterization and difficulty in identifying weak targets in traditional technologies. Before fusion processing, the amplitude information corresponding to each sampling point is extracted from the discrete-time complex baseband signal sequence. The calculation process of amplitude information is the square root of the sum of the squares of the real part and the squares of the imaginary part of the complex signal. The calculation formula is: In the formula, This represents the amplitude information of the complex baseband signal sampling points. For the first The real part of each sampling point For the first The imaginary part of each sampling point, the amplitude information can reflect the signal strength, and can help determine the size and distance of the target as well as the signal-to-noise ratio of the signal, providing support for the identification of weak targets.

[0034] The fusion process is executed sequentially according to the sampling points. It correlates the three core pieces of information corresponding to each sampling point: phase gradient, Doppler frequency, and amplitude. Each sampling point corresponds to a set of feature data containing these three pieces of information. The core logic of the fusion process is to combine the phase gradient, Doppler frequency, and amplitude information of the same sampling point to form a three-dimensional feature vector. Each feature vector corresponds to the target feature state at a given sampling time. During the fusion process, the timing of the three pieces of information is synchronized to prevent misalignment or omission. Simultaneously, the fused feature data undergoes preliminary verification to remove anomalies caused by signal interference. To ensure the integrity and reliability of the feature data, this deep fusion process integrates the originally scattered and independent phase, frequency, and amplitude information into an organic whole, ultimately constructing Doppler phase-frequency feature data for analyzing the motion state characteristics of targets in the blind zone. This data includes both phase and frequency changes caused by target motion to accurately reflect the radial relative motion velocity and motion trend of the target, and amplitude information related to signal strength, which can help identify weak targets and judge signal reliability. Thus, it comprehensively and accurately reflects the motion state characteristics of targets in the blind zone, providing high-quality feature support for three-dimensional Doppler potential field modeling, discrete scattering point extraction, and spatial distortion correction.

[0035] In this embodiment of the invention, phase information is extracted by parsing complex baseband signal sequences and a phase time series is constructed. The phase gradient is calculated through differential operations, and the Doppler frequency is calculated by combining physical correspondences and establishing coupling mapping relationships. Finally, the phase gradient, Doppler frequency, and amplitude information are fused to construct Doppler phase-frequency feature data. This technique overcomes the technical problems of traditional blind zone target detection, which ignores the coupling relationship between the phase change gradient of complex baseband signals and the Doppler effect and relies only on amplitude spectrum and frequency information for target calculation. This results in difficulty in extracting weak target features and incomplete representation of target motion features in low signal-to-noise ratio and multipath interference scenarios. As a result, comprehensive capture of target motion features in blind zones is achieved, effectively improving the extractability of weak target features. The constructed Doppler phase-frequency feature data accurately reflects the target's motion state and provides comprehensive and reliable feature support for three-dimensional Doppler potential field modeling, spatial distortion correction, and hazard level assessment. This further improves the accuracy and reliability of blind zone target hazard level assessment.

[0036] In a preferred embodiment of the present invention, step 3 above may include: Step 3.1 involves selecting three physical aperture centers in the radar receiving antenna array as spatial sampling nodes and extracting the phase information corresponding to each spatial sampling node from the Doppler phase frequency characteristic data. Specifically, this includes selecting three physical aperture centers in the radar receiving antenna array as spatial sampling nodes. The selection criteria are that the three nodes can form a stable spatial triangle structure, accurately covering the detection range of the vehicle's blind spot, while balancing sampling efficiency and feature extraction accuracy. This avoids excessive sampling nodes leading to increased computation, or insufficient sampling nodes causing distortion in spatial feature modeling. During the selection process, it is ensured that the distance between the three physical aperture centers is uniform and that all are within the effective detection range of the radar receiving antenna array. The spatial coordinates of each sampling node are clearly defined and pre-calibrated to provide an accurate spatial reference for subsequent spatial potential field modeling. After selecting the spatial sampling nodes, the phase information corresponding to each spatial sampling node is extracted from the Doppler phase frequency characteristic data.

[0037] Since Doppler phase frequency characteristic data contains phase, frequency, and amplitude information for each sampling time and spatial location, during the extraction process, the Doppler phase frequency characteristic data is traversed based on the spatial coordinates of the three spatial sampling nodes. The timing phase information that matches the spatial coordinates of each node is selected to ensure that the phase information of each sampling node is completely synchronized with the timing of the Doppler phase frequency characteristic data. During extraction, abnormal phase values ​​caused by multipath interference and signal attenuation are removed, and missing phase data are supplemented by linear interpolation to ensure that the phase information corresponding to each spatial sampling node is continuous and accurate, laying a reliable foundation for phase difference calculation.

[0038] Step 3.2: Calculate the phase difference between the phase information corresponding to each spatial sampling node, and calculate the Doppler potential field intensity value of each sampling node in the spatial dimension based on the phase difference. Specifically, this includes: based on the extracted phase information of the three spatial sampling nodes, combining the three nodes pairwise, and calculating the phase difference between each pair of nodes to ensure comprehensive capture of the phase distribution differences at different spatial locations, providing data support for the Doppler potential field intensity calculation. The specific calculation process is as follows: Let the three spatial sampling nodes be node 1, node 2, and node 3, and their corresponding phase values ​​be respectively... , , The phase difference is calculated by combining each pair of pairs, and the calculation formulas are as follows: , , In the formula, The phase difference between node 1 and node 2. The phase difference between node 2 and node 3. Let be the phase difference between node 1 and node 3. The sign of the phase difference reflects the lead or lag relationship between the two nodes' phases, and its absolute value reflects the magnitude of the phase change caused by the spatial position difference. After calculating the phase differences of all pairs of nodes, the Doppler potential field intensity value of each sampling node in the spatial dimension is calculated based on the phase difference. According to the physical principles of radar signal processing, the Doppler potential field intensity characterizes the signal energy strength at a certain point in space, is proportional to the phase difference, and is related to the wavelength of the radar transmitted signal. The formula for calculating the wavelength of the radar transmitted signal is: In the formula, The wavelength of the linear frequency modulated continuous wave signal transmitted by the radar. The speed of light (value 3 × 10⁻⁶) 8 m / s), This is the center frequency of the radar's transmitted signal.

[0039] Combining phase difference and wavelength, the formula for calculating the Doppler potential field intensity is as follows: In the formula, The value of the Doppler potential field is... Let be the phase difference between any two sampling nodes. For the radar signal wavelength, respectively , , Substituting into the formula, three sets of potential field strength values ​​are calculated, corresponding to the spatial potential field distribution between the three sampling nodes. At the same time, the calculated potential field strength values ​​are verified, and outliers exceeding the reasonable range are eliminated to ensure that the potential field strength values ​​can truly reflect the spatial signal energy distribution characteristics.

[0040] Step 3.3: Based on the Doppler potential field intensity values, and according to the spatial coordinates of each spatial sampling node, calculate the potential field intensity values ​​at unknown locations between the spatial sampling nodes and smoothly connect them to construct a continuously changing three-dimensional dynamic Doppler potential field surface. Specifically, this includes: based on the Doppler potential field intensity values ​​and combined with the calibrated spatial coordinates of the three spatial sampling nodes, calculating and smoothly connecting the potential field intensity values ​​at unknown locations to construct a continuously changing three-dimensional dynamic Doppler potential field surface, and defining the three-dimensional spatial coordinates of the three spatial sampling nodes. Let the coordinates of node 1 be... The coordinates of node 2 are The coordinates of node 3 are The coordinates are set based on the vehicle coordinate system to ensure the accuracy of spatial positioning.

[0041] The potential field strength at unknown locations is calculated using a bilinear interpolation algorithm. This algorithm can accurately calculate the potential field strength at any unknown location within the spatial plane formed by the three nodes, based on the known coordinates and potential field strength of the three nodes, balancing computational accuracy and efficiency. The specific calculation process is as follows: for any unknown point within the spatial plane formed by the three nodes... First, calculate the spatial distance from the unknown point to the three sampling nodes. The closer the distance, the greater the influence of the potential field strength of the node on the unknown point. Based on this, determine the weighting coefficient for each sampling node. , , The weighting coefficients satisfy Furthermore, the closer the distance, the larger the weighting coefficient. The formula for calculating the potential field strength at an unknown point is: In the formula, The value of the Doppler potential field at the unknown point. , , These represent the potential field strength values ​​at the three sampling nodes. , , These are the weighting coefficients for the corresponding sampling nodes.

[0042] Following the above method, all unknown spatial locations between the three sampling nodes are traversed, and the potential field strength value at each location is calculated one by one, completing the full coverage calculation of the potential field strength at unknown locations. Subsequently, the potential field strength values ​​of all known sampling nodes and the calculated potential field strength values ​​at unknown locations are smoothly connected. A smoothing filtering algorithm is used to eliminate possible abrupt changes in potential field strength during interpolation, ensuring that the potential field strength values ​​throughout the entire spatial range exhibit a continuous and stable trend, thus constructing a three-dimensional dynamic Doppler potential field surface: using spatial coordinates... , , The surface is divided into three dimensions, with the Doppler potential field intensity value as the height of the surface. All known and calculated potential field intensity values ​​are mapped to the corresponding spatial coordinates to form a continuously changing three-dimensional surface. At the same time, as the sampling time sequence is updated, the potential field intensity value of each spatial position is updated in real time, realizing the dynamic refresh of the three-dimensional dynamic Doppler potential field surface, ensuring that the surface can reflect the changes in signal spatial distribution caused by the movement of the target in the blind zone in real time.

[0043] Step 3.4 involves performing isosurface topology segmentation on the three-dimensional dynamic Doppler potential field surface to identify energy accumulation regions on the surface and extract discrete scattering point sets from these regions. Specifically, this includes performing isosurface topology segmentation on the three-dimensional dynamic Doppler potential field surface. The core purpose is to distinguish between the energy accumulation region of the target scattering signal and the background clutter region, accurately extract the discrete scattering point set corresponding to the target in the blind zone, and set an isosurface segmentation threshold. The threshold is set based on the maximum potential field strength of the three-dimensional dynamic Doppler potential field surface, typically set to 70%–80% of the maximum potential field strength. This threshold effectively preserves the energy accumulation region formed by the target scattering signal while eliminating low potential field strength regions formed by background clutter and signal interference, preventing clutter from being misjudged as target features. During the isosurface topology segmentation process, a topology segmentation algorithm is used to extract all points on the three-dimensional dynamic Doppler potential field surface whose potential field strength value is equal to the set threshold. These points are then connected sequentially to form a closed isosurface. The area enclosed by this isosurface is the signal energy accumulation region, which is the region where the target scattering signal is concentrated in the blind zone. After segmentation, the resulting isosurfaces are analyzed to select energy accumulation regions with areas that match the target size range in the blind zone and have continuous and stable potential field strength. Clutter regions with excessively small areas and large fluctuations in potential field strength are removed to ensure that the identified energy accumulation regions correspond to the real blind zone targets.

[0044] Discrete scattering point sets are extracted from the identified energy accumulation regions. During the extraction process, the energy accumulation regions are sampled at uniform spatial intervals, prioritizing the extraction of local maxima of potential field intensity within the regions. These points correspond to strong scattering locations on the target surface and can accurately reflect the target's spatial morphology. During sampling, the sampling interval is controlled to avoid excessively dense scattering points leading to increased computation, or excessively sparse scattering points resulting in incomplete target morphology representation. Abnormal scattering points appearing during the sampling process are removed to ensure the integrity and reliability of the discrete scattering point set, forming a discrete scattering point set composed of multiple discrete scattering points. Each scattering point corresponds to a scattering location on the target surface. This point set can accurately reflect the target's spatial morphology and positional characteristics, providing reliable geometric constraints for subsequent spatial distortion correction and target motion state estimation based on the scattering point set.

[0045] In this embodiment of the invention, by selecting three spatial sampling nodes in the radar receiving antenna array and extracting synchronization phase information, calculating the phase difference between each pair of nodes and combining it with the radar signal wavelength to calculate the Doppler potential field strength, using bilinear interpolation to calculate the potential field strength at unknown locations and smoothly connecting them to construct a three-dimensional dynamic Doppler potential field surface, and using isosurface topological segmentation to identify energy accumulation regions and extract discrete scattering point sets, this invention overcomes the technical problems of traditional blind zone target detection, which lacks spatial dimension feature modeling, relies solely on single signal features, and leads to difficulties in extracting weak target scattering features and effectively eliminating clutter interference in low signal-to-noise ratio and multipath interference scenarios, as well as the inability to obtain target spatial morphology information. This technology enables the spatial and precise extraction of scattering features from targets in blind zones. The three-dimensional dynamic Doppler potential surface can reflect the changes in signal spatial distribution caused by target motion in real time. Isosurface topological segmentation can effectively distinguish targets from clutter. The extracted discrete scattering point set can accurately characterize the spatial morphology and positional features of targets, effectively improving the identification capability of weak targets. It solves the problems of incomplete and inaccurate target feature extraction in traditional technologies. The discrete scattering point set provides reliable geometric constraints for determining the boundary of the smallest circular region, extracting spatial distortion indicators, and correcting motion parameters. This further lays a solid spatial feature foundation for improving the accuracy of target motion state estimation in blind zones and ensuring the accuracy of hazard level assessment.

[0046] In a preferred embodiment of the present invention, step 4 above may include: Step 4.1: Using the discrete scattering points in the set as geometric constraints, continuously adjust the center coordinates and radius to verify whether all discrete scattering points are located inside the circular region, until the boundary of the circular region with the smallest radius that covers all discrete scattering points is found. The center coordinates and radius of the smallest circular region are then determined as boundary parameters. Specifically, this involves using the discrete scattering point set as all geometric constraints and employing an iterative search algorithm to find the boundary of the smallest circular region that completely encloses all scattering points. During the solution process, a set of center coordinates and radius lengths are randomly initialized, and the position of each scattering point in the discrete scattering point set is checked one by one. The check rule is to calculate the Euclidean distance between a single scattering point and the current center, and compare this distance with the current radius. The calculation formula is: In the formula, Let be the distance from the scattering point to the center of the circle. , The spatial coordinates of the scattering point. , Using the current center coordinates, if the calculated distance is less than or equal to the radius, the scattering point is determined to be inside the circular region; if the distance is greater than the radius, the scattering point is determined to be outside the circular region. Whenever a scattering point is detected outside the circular region, the center coordinates are translated and adjusted, and the radius is increased. After the adjustment, the position verification of all scattering points is performed again, and the above adjustment and verification process is repeated. During the iteration process, the optimization goal is always to obtain the minimum radius. The radius is increased only when necessary, until all discrete scattering points meet the constraint condition of being inside the circular region, and the radius cannot be further reduced. This completes the determination of the minimum circular region boundary. Finally, the spatial coordinates of the center and the radius of the corresponding circular region are output as core boundary parameters to provide reference geometric data for spatial distortion analysis.

[0047] Step 4.2: Based on boundary parameters, track the dynamic trajectory within a continuously acquired time frame sequence, calculate the rate of change of the radius length per unit time to obtain the radius expansion rate, and calculate the displacement vector of the center coordinates per unit time to obtain the center drift. Define the radius expansion rate and center drift as spatial distortion indices of target morphology changes. Specifically, this includes: based on the boundary parameters of the minimum circular region, dynamically track the minimum circular region corresponding to each frame within a multi-frame time sequence continuously acquired by the radar, and obtain the center coordinates and radius length sequences of multiple consecutive frames. Based on this, calculate the radius expansion rate and center drift respectively, and define both parameters together as spatial distortion indices of target morphology changes.

[0048] The radius expansion rate is used to characterize the size change of the smallest circular region per unit time. It is calculated as the ratio of the radius change between two adjacent frames to the frame interval, and the formula is as follows: In the formula, The radius expansion rate, For the first The minimum circle radius of the frame, For the first The minimum circle radius of the frame, The time interval between two adjacent frames of data is used. The center drift is used to calculate the spatial displacement of the center of the smallest circular region per unit time. First, the horizontal and vertical offsets of the center coordinates between two adjacent frames are calculated. Then, the total displacement is synthesized and divided by the time interval. The calculation formula is: In the formula, For center drift, , For the first Frame center coordinates, , For the first Frame center coordinates, The time interval between two adjacent frames. Radius expansion rate and center drift together reflect the degree of spatial distortion of targets in the blind zone under large-angle radar detection scenarios. The higher the degree of distortion, the larger the values ​​of the two indicators.

[0049] Step 4.3: Construct a nonlinear correction function using spatial distortion indices, calculate the measurement error compensation factor, and input the original measured radial relative velocity, relative distance, and azimuth parameters into the nonlinear correction function. The measurement error compensation factor is used to compensate for and correct the measurement deviation, resulting in the corrected motion parameters. These corrected motion parameters are then used as the estimated motion state of the target in the blind zone. Specifically, this involves constructing a nonlinear correction function for the target motion parameters in the blind zone using radius expansion rate and center drift as input variables. This function is used to calculate the measurement error compensation factor corresponding to the three original measured parameters: radial relative velocity, relative distance, and azimuth. The error compensation factor is obtained by weighted summation of spatial distortion indices. The weighting coefficients are determined through previous system calibration experiments to ensure that the compensation value matches the degree of spatial distortion. The original measured radial relative velocity, relative distance, and azimuth are substituted into the nonlinear correction function, and the corresponding measurement error compensation factor is used to compensate for and correct the original measurement deviation in real time. The correction calculation formula is as follows: In the modified original expression, The corrected motion parameters, These are the original radar measurement parameters. The weighting coefficients obtained from the calibration. The nonlinear distortion mapping function includes the radius expansion rate and the center drift. After nonlinear compensation corrections for radial relative velocity, relative distance and azimuth angle are performed respectively, the three corrected motion parameters are integrated to form a complete estimation result of the motion state of the target in the blind zone. This result has eliminated the nonlinear measurement error caused by spatial distortion and can truly reflect the actual motion state of the target in the blind zone.

[0050] In this embodiment of the invention, this step overcomes the technical problems of traditional blind zone target detection, which lack a spatial distortion correction mechanism and is prone to range drift and azimuth deviation when using linear models to solve parameters, ultimately leading to large nonlinear errors in target motion state estimation. This is achieved by iteratively solving the minimum coverage circular region boundary with discrete scattering point set as constraints, calculating the radius expansion rate and center drift in continuous time frames and using them as spatial distortion indicators, constructing a nonlinear correction function to calculate error compensation factors, and compensating and correcting the original motion parameters. This achieves accurate error compensation for the original measurement parameters of blind zone targets, effectively weakens the impact of spatial distortion on radar measurement results, and significantly improves the estimation accuracy of radial relative velocity, relative distance, and azimuth. It provides real and reliable target motion state data for subsequent collision risk index calculation and hazard level assessment, fundamentally improving the credibility of blind zone hazard level assessment results.

[0051] In a preferred embodiment of the present invention, step 5 above may include: Step 5.1: Obtain the vehicle's current speed, steering angle, and acceleration values. Perform time synchronization processing on these values ​​to construct the vehicle's real-time motion state vector. This includes: real-time acquisition of the vehicle's core operating parameters, including speed, steering angle, and acceleration values, via the vehicle's CAN bus. These parameters are collected and output by the vehicle's wheel speed sensor, steering angle sensor, and acceleration sensor, respectively, to accurately reflect the vehicle's instantaneous driving state and ensure high accuracy and real-time performance of each acquired parameter. After acquisition, time synchronization processing is required on the data output from multiple sensors. Using the radar signal sampling timestamp as a unified benchmark, the acquisition times of different sensor output data are aligned to the same timing node, completely eliminating timing deviations caused by transmission delays and differences in acquisition frequencies between different sensors. This ensures that all vehicle motion parameters remain completely consistent in the time dimension, preventing timing misalignment from affecting the accuracy of relative motion analysis.

[0052] After time synchronization is completed, the driving speed, steering angle, and acceleration are decomposed into lateral and longitudinal motion components in a unified coordinate system according to the decomposition rules of vehicle kinematics. The decomposed motion components are then integrated and combined to construct the instantaneous motion state vector of the vehicle. This vector contains both translational and rotational characteristics of the vehicle, which can comprehensively and accurately represent the current instantaneous motion state of the vehicle. The translational characteristics are represented by the lateral and longitudinal components of driving speed and acceleration, and the rotational characteristics are represented by the rotational component corresponding to the steering angle. This provides a unified and reliable benchmark for the subsequent relative motion analysis between the vehicle and targets in the blind spot.

[0053] Step 5.2: Using the vehicle's instantaneous motion state vector and the corrected blind spot target motion state estimation results, perform spatiotemporal alignment and fusion in a unified spatiotemporal coordinate system to obtain the relative motion trajectory sequence between the vehicle and the blind spot target. Specifically, this includes: using the vehicle's geodetic coordinate system as the unified spatiotemporal coordinate system, performing spatiotemporal alignment and data fusion between the vehicle's instantaneous motion state vector and the corrected blind spot target motion state estimation results to eliminate calculation errors caused by differences in spatial coordinate systems and temporal deviations; transforming the radial relative velocity, relative distance, and azimuth correction parameters of the blind spot target to the unified spatiotemporal coordinate system to obtain the target's position coordinates and velocity components within this coordinate system; and then, combining the vehicle's motion state vector, calculating the relative position vector and relative velocity vector between the vehicle and the blind spot target at each time step. The formula for calculating the relative position vector is: The formula for calculating the relative velocity vector is: , It is a relative position vector. The coordinates of the target location in the blind spot. These are the coordinates of the vehicle's position. It is a relative velocity vector. For the velocity components of the target in the blind zone, The relative position vector and relative velocity vector are calculated and recorded frame by frame based on the continuous sampling time sequence to obtain a complete sequence of relative motion trajectories between the vehicle and the target in the blind spot. This sequence fully records the relative motion law of the two changing over time.

[0054] Step 5.3: Based on the relative motion trajectory sequence, the kinematic model is used to extrapolate the position changes of the vehicle and the blind spot target along the time axis, calculate the estimated collision time at the moment when the two reach their minimum values, and record the predicted minimum approach distance at the moment of minimum values. Specifically, this includes: based on the relative motion trajectory sequence, a kinematic model combining uniform velocity and uniform acceleration is used. This model is based on the application scenario of short-term prediction for blind spot warning. It assumes that within a short prediction time range, neither the vehicle nor the blind spot target will experience sudden acceleration, sudden braking, or sharp turns, and that the current speed, acceleration, and direction of motion remain unchanged. Position extrapolation is performed under this premise, ensuring prediction accuracy while meeting the real-time calculation requirements of the onboard system. The future position changes of the vehicle and the blind spot target are extrapolated along the time axis, maintaining the current motion state during the extrapolation process. The estimated collision time and the predicted minimum approach distance are calculated based on this. The estimated collision time is the theoretical moment when the relative distance between the two decreases to zero, and the calculation formula is: In the formula, To predict the collision time, dot operations represent vector dot products. A positive result indicates a possible collision, while a negative result indicates that the two objects are moving further apart. In extrapolating the changes in relative position, all future relative distance values ​​are iterated through, and the minimum relative distance is selected. This minimum value is the predicted minimum approach distance, calculated using the following formula: In the formula, To predict the minimum approach distance, the estimated collision time and the predicted minimum approach distance are recorded simultaneously after the calculation is completed, serving as the core quantitative indicators for collision risk assessment.

[0055] Step 5.4: Based on the estimated collision time and predicted minimum approach distance, and combined with the rate of change of relative distance over time in the relative motion trajectory sequence, analyze the urgency of their approach and obtain an index of the relative motion convergence trend. Specifically, this includes: combining the obtained estimated collision time and predicted minimum approach distance, performing numerical analysis on the relative motion trajectory sequence, and calculating the rate of change of relative distance over time. The formula for calculating the rate of change is: In the formula, The relative distance change rate, along with the estimated collision time and predicted minimum approach distance, is used to comprehensively determine the urgency of the approach between the two vehicles, resulting in a relative motion convergence trend index: when the relative distance continuously and rapidly decreases and the estimated collision time is short, it is considered a strong convergence state; when the relative distance first decreases and then increases, it is considered a weak convergence state; and when the relative distance continuously increases, it is considered a divergence state. The convergence trend index quantifies the urgency of the approach between the target and the vehicle using numerical methods, and together with the estimated collision time and predicted minimum approach distance, constitutes a complete hazard level assessment index system.

[0056] In this embodiment of the invention, this step involves collecting the vehicle's motion parameters and synchronizing them with time to construct an instantaneous motion state vector. Under a unified spatiotemporal coordinate system, the motion states of the vehicle and the target are aligned and fused to obtain a relative motion trajectory sequence. A kinematic model is used to extrapolate position changes, calculate the estimated collision time and the predicted minimum approach distance, and then the relative motion convergence trend index is obtained by analyzing the rate of change of relative distance. This overcomes the technical problems of traditional blind spot hazard assessment, such as the failure to unify and integrate the vehicle's motion state with the target's motion state, the lack of spatiotemporal alignment leading to distorted trajectory extrapolation, and the reliance on a single parameter for risk assessment resulting in a single assessment dimension. This achieves accurate extrapolation of the relative motion patterns between the vehicle and the blind spot target, constructing a multi-dimensional, quantitative collision risk index system. This provides real, comprehensive, and time-consistent quantitative data support for subsequent multi-level hazard assessment, effectively improving the scientific rigor and accuracy of hazard level assessment.

[0057] In a preferred embodiment of the present invention, step 6 above may include: Step 6.1 involves calling a pre-defined multi-level hazard level threshold configuration table. This table defines the critical threshold ranges for the estimated collision time, predicted minimum approach distance, and relative motion convergence trend indicators under different hazard levels. Specifically, the system's onboard computing unit pre-stores this multi-level hazard level threshold configuration table, which has been verified through extensive real-vehicle road testing, traffic scenario calibration, and vehicle safety regulations. This configuration table uses the collision risk level between the target in the blind spot and the vehicle as the core classification criterion, dividing the hazard levels from low to high into four gradients: safe level, general warning level, higher hazard level, and emergency hazard level. For each hazard level, the configuration table clearly defines the quantitative critical threshold ranges for the three core indicators: estimated collision time, predicted minimum approach distance, and relative motion convergence trend. The threshold for estimated collision time is based on the urgency of the collision, the threshold for predicted minimum approach distance is based on the standard for safe driving distance in the blind spot of motor vehicles, and the relative motion convergence trend indicator is based on the approach rate and the strength of motion convergence between the target and the vehicle. The specific threshold division rules are as follows: the smaller the predicted collision time, the higher the collision risk; the smaller the predicted minimum approach distance, the less safe distance; and the stronger the relative motion convergence trend, the faster the target approaches. All thresholds are fixed quantification ranges that can be directly called without real-time adjustment, providing a unified, standardized, and traceable basis for comparing the risk levels of various indicators and determining comprehensive risks.

[0058] Step 6.2: Using a multi-level hazard level threshold configuration table, the predicted collision time, predicted minimum approach distance, and relative motion convergence trend indicators are compared with the critical threshold ranges to determine the current hazard level range of each indicator. Specifically, this involves comparing the predicted collision time, predicted minimum approach distance, and relative motion convergence trend indicators with the corresponding critical threshold ranges in the multi-level hazard level threshold configuration table in a predetermined order. The predetermined order is a fixed sequence of indicator comparisons, consistent with the weight priority of the subsequent comprehensive hazard level determination. The specific predetermined order is: comparing the predicted collision time, comparing the predicted minimum approach distance, and comparing the relative motion convergence trend indicator. The core purpose of setting this fixed predetermined order is to ensure that the hazard level determination process is consistent and the logic is reproducible in all blind zone detection scenarios, avoiding inconsistencies in assessment results due to random changes in the comparison order, and aligning with the impact of each indicator on collision risk, ensuring the rationality and standardization of the comparison process.

[0059] The multi-level hazard threshold configuration table is a standardized quantitative comparison table pre-stored within the system and verified through real-vehicle calibration and safety specifications. This configuration table focuses on the degree of collision risk, clearly dividing hazard levels into four gradients: safe, general warning, higher risk, and emergency risk. Each hazard level corresponds to three specified critical threshold ranges: predicted collision time corresponds to four quantitative threshold intervals, predicted minimum approach distance corresponds to four quantitative threshold intervals, and relative motion convergence trend corresponds to three states of divergence, weak convergence, and strong convergence. All thresholds and judgment criteria are fixed and can be directly called for comparison. The comparison process is carried out item by item in the above preset order: comparing the predicted collision time and extracting the specific number of predicted collision time. The value is matched one by one with the four critical threshold intervals corresponding to the expected collision time in the multi-level hazard level threshold configuration table to determine which threshold interval of the four hazard levels the value falls into. Based on the matching results, the initial hazard level of the first indicator is assigned. Next, the predicted minimum approach distance is compared. According to the same comparison rules as the expected collision time, the specific value of the predicted minimum approach distance is matched with the critical threshold interval corresponding to the predicted minimum approach distance in the configuration table to determine the hazard level interval corresponding to the indicator and complete the assignment. The relative motion convergence trend indicator is compared. Based on the actual state of the indicator, it is matched with the state judgment standard corresponding to the relative motion convergence trend in the configuration table to determine the hazard level interval corresponding to the indicator and complete the assignment.

[0060] Throughout the comparison process, numerical matching rules are executed, following the principle that higher risk corresponds to higher risk values: the smaller the predicted collision time, the higher the risk level of the corresponding individual indicator; the smaller the predicted minimum approach distance, the higher the risk level of the corresponding individual indicator; and the stronger the relative motion convergence trend, the higher the risk level of the corresponding individual indicator. Through this comparison method, which involves looking up values ​​one by one in a preset order, the independent risk level ranges of the three indicators are located and numerically assigned, ensuring that the value of each indicator accurately corresponds to its actual risk level.

[0061] Step 6.3: Based on the current danger level range of each indicator, a comprehensive logical judgment is performed according to the preset weight priority to obtain the danger level label. Specifically, this includes: the system pre-calibrates the weight coefficients of the three risk indicators through multi-scenario blind spot collision risk tests. Among them, the predicted collision time directly determines the urgency of the collision and has the highest impact on driving safety, with a weight coefficient set to 0.5; the predicted minimum approach distance reflects the safe distance between vehicles, with a weight coefficient set to 0.3; and the relative motion convergence trend characterizes the dynamic changes in the approach of the target, with a weight coefficient set to 0.2. The sum of the three weight coefficients equals 1, calculated using the following formula: In the formula, The weighting coefficient for the expected collision time. To predict the minimum proximity distance weighting coefficient, As the weighting coefficient for the relative motion convergence trend, after assigning numerical values ​​to the risk levels of individual indicators, a weighted summation algorithm is used to calculate the comprehensive risk score. The calculation formula is as follows: In the formula, For comprehensive risk assessment, Assign a value to the expected collision time level. Assign a value to the predicted minimum proximity level. The relative motion convergence trend level is assigned a value. After the calculation is completed, the comprehensive hazard score is compared with the preset level boundary value. A unique corresponding hazard level label is matched according to the score range. By using a multi-indicator weighted comprehensive judgment method, the bias of judgment caused by the fluctuation or anomaly of a single indicator is avoided, and the final hazard level result is objectively consistent with the actual collision risk of the blind spot target.

[0062] Step 6.4: Match the corresponding warning strategy according to the hazard level identifier to obtain warning suggestions including visual warning level, auditory warning frequency, and driving assistance intervention type. The final evaluation result, including hazard level identifier and warning suggestions, is obtained. Specifically, based on the hazard level identifier determined by comprehensive logic, the system automatically matches the corresponding standardized warning execution plan from the preset graded warning strategy library. Different hazard levels correspond to differentiated visual warnings, auditory warnings, and driving assistance intervention strategies. At the safety level, the blind spot monitoring indicator on the vehicle's dashboard remains off, with no auditory prompts and no driving assistance intervention activated. At the general warning level, a yellow low-brightness visual warning is triggered, accompanied by low-frequency intermittent auditory prompts. At the higher hazard level, a red high-brightness visual warning is triggered, accompanied by mid-frequency continuous auditory warnings, and a mild driving assistance reminder is activated. At the emergency hazard level, a red high-frequency flashing visual warning is triggered, accompanied by a rapid, continuous buzzing auditory alarm, and an active driving assistance intervention prompt is activated. The final determined hazard level labels are integrated with the matched visual warning levels, auditory warning frequencies, and driving assistance intervention types to form standardized warning suggestions. The final combination yields a blind spot target hazard level assessment result that includes hazard level labels and complete warning suggestions. This result is output in real time to warning terminals such as the vehicle dashboard, in-vehicle central control, and head-up display to execute corresponding warning and intervention operations.

[0063] In this embodiment of the invention, this step uses a multi-level hazard level threshold configuration table calibrated by actual vehicles to compare and assign values ​​to the three core risk indicators one by one. A weighted summation algorithm is used to calculate the comprehensive hazard score of multiple indicators. Then, a differentiated graded warning strategy is matched according to the hazard level identifier. This overcomes the technical problems in traditional blind spot target hazard assessment, such as relying on a single parameter, which easily leads to misjudgment and missed judgment; lack of quantitative graded thresholds, which leads to ambiguous assessment results; and lack of differentiated warning strategies. This achieves quantitative, accurate, and comprehensive judgment of collision risk of blind spot targets. The graded warning strategy can provide appropriate warnings and intervention suggestions according to the actual degree of danger, effectively improving the rationality and effectiveness of blind spot warning, significantly reducing the false alarm rate and missed alarm rate of blind spot monitoring, providing drivers with reliable blind spot safety prompts, and fully meeting the practical application needs of advanced driver assistance systems for high-precision blind spot safety warnings.

[0064] like Figure 2 As shown, embodiments of the present invention also provide a blind zone target hazard level assessment system based on Doppler phase frequency characteristics, including: The acquisition module is used to radiate linear frequency modulated continuous wave signals into the vehicle's blind spot, capture the coherent echoes scattered by the target, and perform mixing filtering and analog-to-digital sampling processing to obtain a complex baseband signal sequence. The analysis module is used to analyze the gradient information of the phase change of the complex baseband signal and its coupling relationship with the Doppler effect based on the complex baseband signal sequence, and to construct Doppler phase frequency characteristic data. The module is used to select the centers of three physical apertures in the radar receiving antenna array as spatial sampling nodes based on Doppler phase frequency characteristic data; construct a three-dimensional dynamic Doppler potential field surface through the phase difference between each spatial sampling node; and perform isosurface topological segmentation on the three-dimensional dynamic Doppler potential field surface to extract discrete scattering point sets. The execution module is used to iteratively determine the boundary of the smallest circular region covering the discrete scattering points based on the discrete scattering point set, and extract the radius expansion rate and center drift of the boundary of the smallest circular region as spatial distortion indices; according to the spatial distortion indices, nonlinear correction processing is performed on the radial relative velocity, relative distance and azimuth parameters to obtain the corrected blind zone target motion state estimation results; The calculation module is used to calculate the estimated collision time, the predicted minimum approach distance, and the relative motion convergence trend index in a unified spatiotemporal coordinate system based on the corrected blind spot target motion state estimation results and the vehicle's driving speed, steering angle, and acceleration information. The judgment module is used to compare the predicted collision time, predicted minimum approach distance, and relative motion convergence trend indicators with preset multi-level hazard thresholds to obtain an assessment result that includes hazard level identification and warning suggestions.

[0065] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0066] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0067] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0068] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for assessing the hazard level of blind zone targets based on Doppler phase frequency characteristics, characterized in that, The method includes: A linear frequency modulated continuous wave signal is radiated into the vehicle's blind spot to capture the coherent echo scattered by the target, and then mixed, filtered, and processed by analog-to-digital sampling to obtain a complex baseband signal sequence. Based on complex baseband signal sequences, the gradient information of the phase change of complex baseband signals and their coupling relationship with the Doppler effect are analyzed to construct Doppler phase frequency characteristic data. Based on Doppler phase frequency characteristic data, three physical aperture centers in the radar receiving antenna array are selected as spatial sampling nodes; a three-dimensional dynamic Doppler potential field surface is constructed by the phase difference between each spatial sampling node; and isosurface topological segmentation is performed on the three-dimensional dynamic Doppler potential field surface to extract discrete scattering point sets. Based on the discrete scattering point set, the boundary of the smallest circular region covering the discrete scattering points is determined iteratively. The radius expansion rate and center drift of the boundary of the smallest circular region are extracted as spatial distortion indices. Based on the spatial distortion indices, nonlinear correction processing is performed on the radial relative velocity, relative distance and azimuth parameters to obtain the corrected blind zone target motion state estimation results. Based on the corrected blind spot target motion state estimation results, combined with the vehicle's speed, steering angle and acceleration information, the estimated collision time, predicted minimum approach distance and relative motion convergence trend index are calculated in a unified spatiotemporal coordinate system. Based on the estimated collision time, predicted minimum approach distance, and relative motion convergence trend indicators, the results are compared with preset multi-level hazard thresholds to obtain an assessment result that includes hazard level identification and warning suggestions.

2. The blind zone target hazard level assessment method based on Doppler phase frequency characteristics according to claim 1, characterized in that, A linear frequency modulated continuous wave signal is radiated into the vehicle's blind spot to capture the coherent echo scattered by the target. After mixing, filtering, and analog-to-digital sampling, a complex baseband signal sequence is obtained, including: The radar transmitting unit radiates a linear frequency modulated continuous wave signal to the blind spot area to the side and rear of the vehicle. The linear frequency modulated continuous wave signal is blocked by the target in the blind spot on the propagation path and forms a scattered echo. The scattered echo is mixed with the local transmitted reference signal to obtain a beat signal containing range and velocity information. The beat signal is subjected to low-pass filtering to remove high-frequency mixing components, resulting in a baseband analog signal; the baseband analog signal is then sampled by analog-to-digital conversion to obtain a discrete-time complex baseband signal sequence.

3. The blind zone target hazard level assessment method based on Doppler phase frequency characteristics according to claim 2, characterized in that, Based on complex baseband signal sequences, the gradient information of the phase change of the complex baseband signal and its coupling relationship with the Doppler effect are analyzed to construct Doppler phase frequency characteristic data, including: Analyze the complex baseband signal sequence and extract the phase value of each sampling point to construct a phase time series; Perform a difference operation on the phase time series to calculate the phase change between adjacent sampling points, and obtain the gradient information of the phase change based on the phase change. The rate of phase change over time is calculated by using the gradient information of phase change. Based on the physical correspondence between the rate of phase change and Doppler frequency shift, the Doppler frequency information of the target is calculated to establish the coupling mapping relationship between phase gradient and Doppler effect. Based on the coupling mapping relationship and Doppler frequency information, the gradient information of phase change is fused with the amplitude information in the complex baseband signal sequence to construct Doppler phase frequency feature data for the motion state characteristics of targets in the blind zone.

4. The blind zone target hazard level assessment method based on Doppler phase frequency characteristics according to claim 3, characterized in that, Based on Doppler phase frequency characteristic data, three physical aperture centers in the radar receiving antenna array are selected as spatial sampling nodes; a three-dimensional dynamic Doppler potential field surface is constructed by the phase difference between each spatial sampling node. The three-dimensional dynamic Doppler potential field surface is subjected to isosurface topological segmentation to extract discrete scattering point sets, including: Three physical aperture centers were selected in the radar receiving antenna array as spatial sampling nodes, and the phase information corresponding to each spatial sampling node was extracted from the Doppler phase frequency characteristic data. Calculate the phase difference between the phase information corresponding to each spatial sampling node, and calculate the Doppler potential field intensity value of each sampling node in the spatial dimension based on the phase difference; Based on the Doppler potential field intensity value, according to the spatial coordinate position of each spatial sampling node, the potential field intensity value of unknown positions between each spatial sampling node is calculated and smoothly connected to construct a continuously changing three-dimensional dynamic Doppler potential field surface. The isosurface topology segmentation process is performed on the three-dimensional dynamic Doppler potential surface to identify energy accumulation regions on the surface, and discrete scattering point sets are extracted from the energy accumulation regions.

5. The blind zone target hazard level assessment method based on Doppler phase frequency characteristics according to claim 4, characterized in that, Based on the discrete scattering point set, the boundary of the smallest circular region covering the discrete scattering points is determined iteratively, and the radius expansion rate and center drift of the boundary of the smallest circular region are extracted as spatial distortion indicators. Based on the spatial distortion index, nonlinear correction processing is performed on the radial relative velocity, relative distance, and azimuth parameters to obtain the corrected blind zone target motion state estimation results, including: Using each scattering point in the discrete scattering point set as a geometric constraint, the center coordinates and radius length of the circle are continuously adjusted to verify whether all discrete scattering points are located inside the circular region until the boundary of the circular region with the smallest radius that covers all discrete scattering points is found. The center coordinates and radius length of the smallest circular region are then determined as boundary parameters. Based on boundary parameters, the dynamic change trajectory is tracked within a continuously acquired time frame sequence. The rate of change of the radius length per unit time is calculated to obtain the radius expansion rate. The displacement vector of the center coordinate per unit time is calculated to obtain the center drift. The radius expansion rate and the center drift are defined as spatial distortion indices of the target shape change. A nonlinear correction function is constructed using spatial distortion indices. A measurement error compensation factor is calculated. The radial relative velocity, relative distance, and azimuth parameters obtained from the original measurements are input into the nonlinear correction function. The measurement deviation is compensated and corrected by the measurement error compensation factor to obtain the corrected motion parameters. The corrected motion parameters are used as the result of the corrected blind zone target motion state estimation.

6. The blind zone target hazard level assessment method based on Doppler phase frequency characteristics according to claim 5, characterized in that, Based on the corrected blind spot target motion state estimation results, combined with the vehicle's speed, steering angle, and acceleration information, the estimated collision time, predicted minimum approach distance, and relative motion convergence trend indicators are calculated in a unified spatiotemporal coordinate system, including: Obtain the current driving speed, steering angle and acceleration values ​​of the vehicle, perform time synchronization processing on the values, and construct the real-time motion state vector of the vehicle. By combining the instantaneous motion state vector of the vehicle with the corrected motion state estimation results of the blind spot target, spatiotemporal alignment and fusion are performed in a unified spatiotemporal coordinate system to obtain the relative motion trajectory sequence of the vehicle and the blind spot target. Based on the relative motion trajectory sequence, the positional changes of the vehicle and the target in the blind spot are extrapolated forward along the time axis using a kinematic model. The predicted collision time when the two reach the minimum value is calculated, and the predicted minimum approach distance at the minimum value is recorded. Based on the expected collision time and predicted minimum approach distance, and combined with the rate of change of relative distance over time in the relative motion trajectory sequence, the urgency of the two motions approaching each other is analyzed, and an index of the relative motion convergence trend is obtained.

7. The blind zone target hazard level assessment method based on Doppler phase frequency characteristics according to claim 6, characterized in that, Based on the estimated collision time, predicted minimum approach distance, and relative motion convergence trend indicators, the results are compared with preset multi-level hazard thresholds to obtain an assessment result that includes hazard level identification and warning recommendations, including: By calling the preset multi-level hazard level threshold configuration table, the configuration table defines the critical threshold ranges of the expected collision time, predicted minimum approach distance, and relative motion convergence trend indicators under different hazard levels. By using a multi-level hazard level threshold configuration table, the predicted collision time, predicted minimum approach distance, and relative motion convergence trend indicators are compared with the critical threshold range to determine the current hazard level range of each indicator. Based on the current danger level range of each indicator, a comprehensive logical judgment is made according to the preset weight priority to obtain the danger level label. Based on the hazard level identification, corresponding early warning strategies are matched to obtain early warning suggestions that include visual warning level, auditory warning frequency, and driving assistance intervention type. The final evaluation result includes hazard level identification and early warning suggestions.

8. A blind zone target hazard level assessment system based on Doppler phase frequency characteristics, wherein the system implements the method as described in any one of claims 1 to 7, characterized in that, include: The acquisition module is used to radiate linear frequency modulated continuous wave signals into the vehicle's blind spot, capture the coherent echoes scattered by the target, and perform mixing filtering and analog-to-digital sampling processing to obtain a complex baseband signal sequence. The analysis module is used to analyze the gradient information of the phase change of the complex baseband signal and its coupling relationship with the Doppler effect based on the complex baseband signal sequence, and to construct Doppler phase frequency characteristic data. The module is used to select the centers of three physical apertures in the radar receiving antenna array as spatial sampling nodes based on Doppler phase frequency characteristic data; construct a three-dimensional dynamic Doppler potential field surface through the phase difference between each spatial sampling node; and perform isosurface topological segmentation on the three-dimensional dynamic Doppler potential field surface to extract discrete scattering point sets. The execution module is used to iteratively determine the boundary of the smallest circular region covering the discrete scattering points based on the discrete scattering point set, and extract the radius expansion rate and center drift of the boundary of the smallest circular region as spatial distortion indices; according to the spatial distortion indices, nonlinear correction processing is performed on the radial relative velocity, relative distance and azimuth parameters to obtain the corrected blind zone target motion state estimation results; The calculation module is used to calculate the estimated collision time, the predicted minimum approach distance, and the relative motion convergence trend index in a unified spatiotemporal coordinate system based on the corrected blind spot target motion state estimation results and the vehicle's driving speed, steering angle, and acceleration information. The judgment module is used to compare the predicted collision time, predicted minimum approach distance, and relative motion convergence trend indicators with preset multi-level hazard thresholds to obtain an assessment result that includes hazard level identification and warning suggestions.

9. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images