A surrounding target optimization detection system for vehicle radar

By preprocessing data and assessing the environment of vehicle radar and image acquisition tools, adjusting radar operating parameters, and combining the YOLOv4 model for target detection and fusion, the detection performance and safety issues of vehicle radar under environmental interference are resolved, thereby improving target detection accuracy and driver safety warning capabilities.

CN118707512BActive Publication Date: 2024-11-15GUOJING HECHUANG (QINGDAO) TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411186957.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-11-15
Estimated Expiration
2044-08-28

AI Technical Summary

Technical Problem

Existing vehicle radar is susceptible to environmental factors in target detection, leading to detection interference, insufficient detection performance and target detection accuracy, which affects vehicle driving safety.

Method used

By preprocessing the data collected in real time by the preset image acquisition tool and assessing the vehicle's environmental conditions, the working parameters of the vehicle radar are analyzed and adjusted. Target detection and fusion are then performed using the YOLOv4 image detection model to generate driving suggestions and safety warnings.

Benefits of technology

It improves the radar's anti-jamming capabilities and vehicle driving safety, enhances the accuracy of target detection, and improves the driver's safety warning and response capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118707512B_ABST
    Figure CN118707512B_ABST
Patent Text Reader

Abstract

The present invention provides a surrounding target optimization detection system for a vehicle radar, which relates to the field of target detection technology, and includes a data acquisition module, which preprocesses the data collected in real time by the vehicle radar and a preset image acquisition tool to obtain first target data; a parameter analysis module, which evaluates the environmental conditions of the vehicle based on the first target data, and analyzes and adjusts the radar working parameters; a target detection module, which detects the targets around the vehicle to obtain target detection fusion results; and an early warning module, which analyzes the target detection fusion results to generate driving suggestions and safety warnings. The working parameters of the vehicle radar are analyzed and adjusted by preprocessing the data collected in real time by the preset image acquisition tool and evaluating the vehicle environmental conditions; the target detection fusion results obtained by preprocessing the data collected by the vehicle radar and the preset image acquisition tool and then performing target detection and fusion are analyzed to generate driving suggestions and safety warnings, thereby effectively improving the radar's anti-interference ability and vehicle driving safety.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of target detection, and in particular to a surrounding target optimization detection system of a vehicle radar. Background Art

[0002] As the number of vehicles continues to increase, traffic problems are becoming increasingly serious, including traffic congestion and frequent accidents. In order to improve traffic efficiency and safety, accurate perception of the vehicle's surroundings has become particularly important.

[0003] At present, vehicle radar technology is widely used in target detection around vehicles due to its advantages such as high detection accuracy and wide detection range. However, since vehicle radar is easily affected by environmental factors and detection interference frequently occurs, how to improve radar detection performance, target detection accuracy and vehicle driving safety has become one of the current research focuses.

[0004] Therefore, the present invention provides a surrounding target optimization detection system for a vehicle radar. Summary of the invention

[0005] The present invention provides a surrounding target optimization detection system for a vehicle radar, which is used to analyze and adjust the working parameters of the vehicle radar by preprocessing data collected by a preset image acquisition tool in real time and evaluating the vehicle environmental conditions; by analyzing the target detection fusion results obtained by preprocessing the data collected by the vehicle radar and the preset image acquisition tool and then performing target detection and fusion, driving suggestions and safety warnings are generated, and the radar anti-interference capability and vehicle driving safety are effectively improved.

[0006] The present invention provides a surrounding target optimization detection system for a vehicle radar, comprising:

[0007] Data acquisition module: used to acquire the data collected by the radar of the target vehicle and the preset image acquisition tool in real time, and obtain the first target data by preprocessing the data;

[0008] Parameter analysis module: used to evaluate the driving environment of the current target vehicle based on the first target data, and to analyze and adjust the radar working parameters of the target vehicle;

[0009] Target detection module: used to detect the surrounding targets of the target vehicle and obtain the target detection fusion result;

[0010] Warning module: used to generate corresponding driving suggestions and safety warnings to remind the driver in time to avoid potential dangerous situations by analyzing the target detection fusion results;

[0011] Wherein, the parameter analysis module includes:

[0012] Image initial detection unit: used to train the neural network using a preset amount of images collected from the KITTI dataset, and introduce the attention mechanism to establish the YOLOv4 image detection model;

[0013] Using the YOLOv4 image detection model to detect the first target data, to obtain image target detection data;

[0014] An environmental assessment unit: configured to calculate a first environmental condition assessment coefficient based on the image target detection data;

[0015] Parameter analysis unit: used to extract standard radar operating parameters using parameter acquisition block; compare and analyze the radar operating parameters of the current target vehicle with the standard radar operating parameters using analysis block;

[0016] Parameter acquisition block: used to extract the corresponding matching standard radar working parameters of the first environmental condition assessment range to which the first environmental condition assessment coefficient belongs from the set environmental assessment range-working parameter mapping table;

[0017] Analysis block: used to analyze and adjust the radar operating parameters by comparing the radar operating parameters of the current target vehicle with the standard radar operating parameters.

[0018] Preferably, the data acquisition module includes:

[0019] Data acquisition unit: used for real-time acquisition of data from the radar of the target vehicle and the preset image acquisition tool, and obtaining first radar point cloud data and first image data respectively;

[0020] Data processing unit: used for preprocessing the first image data by using a histogram equalization algorithm to obtain target image data, and outputting the data as first target data.

[0021] Preferably, the preset image acquisition tool refers to a camera used to simulate the driver's field of view and a camera installed at the rear of the vehicle.

[0022] Preferably, the environmental assessment unit comprises:

[0023] The image target detection data is combined with the driving state data of the current target vehicle, and the driving environment condition of the current target vehicle is analyzed to obtain a first environmental condition evaluation coefficient;

[0024] The calculation formula of the first environmental condition assessment coefficient is as follows:

[0025] ; Wherein, B represents the first environmental condition assessment coefficient; It is expressed as the influence coefficient of the current weather conditions of the target vehicle on the environmental condition assessment; It is expressed as the influence weight of the image target detection situation with static state attribute on the evaluation of the driving environment condition of the target vehicle; It is represented as the total number of image detection targets with static state attribute at the i-th obstacle level, where i = 1, 2, 3; The state attribute of the i-th obstacle level is represented as the average distance between the static image detection target and the target vehicle; It is expressed as the influence weight of the image target detection situation with dynamic state attribute on the evaluation of the driving environment condition of the target vehicle, where: 1; It is represented as the total number of image detection targets whose state attribute is dynamic at the i-th obstacle level, where i = 1, 2, 3; The state attribute represented as the i-th obstacle level is the average distance between the dynamic image detection target and the target vehicle.

[0026] Preferably, the analysis block includes:

[0027] Obtaining a first operating parameter absolute difference between a radar operating parameter of a current target vehicle and an operating parameter of a standard radar;

[0028] If the current target vehicle does not have a radar operating parameter whose absolute difference value of the first operating parameter is greater than the preset parameter difference threshold, the radar operating parameter of the current target vehicle is not adjusted;

[0029] If the absolute difference between the radar operating parameter of the current target vehicle and the first operating parameter of the standard radar operating parameter is greater than the preset parameter difference threshold, the radar operating parameter of the current target vehicle is adjusted to the corresponding standard radar operating parameter;

[0030] If the current target vehicle has a single or multiple radar operating parameters whose absolute difference of the first operating parameters is greater than a preset parameter difference threshold, the corresponding radar operating parameters are marked as the first parameters;

[0031] treating the remaining radar operating parameters except the first parameter as the reference radar operating parameters;

[0032] The parameter value of the reference radar working parameter of the current target vehicle is combined with the set parameter extension value to obtain the reference working parameter range;

[0033] Extracting all corresponding historical radar operating parameter groups of vehicles within the first environmental condition assessment range from the vehicle radar database, and collecting them to obtain a first historical data set;

[0034] Filtering out corresponding historical radar operating parameter groups from the first historical data set, where the historical parameter values ​​of the corresponding reference radar operating parameters belong to the reference operating parameter range, and marking them as first radar operating parameter groups;

[0035] Obtaining a vehicle radar detection evaluation value of each first radar operating parameter group, and marking a first radar operating parameter group whose vehicle radar detection evaluation value is greater than a set evaluation threshold as a reference radar operating parameter group;

[0036] Constructing a parameter-detection evaluation analysis curve based on the vehicle radar detection evaluation value of the reference radar working parameter group and the historical parameter value corresponding to the first parameter;

[0037] Extracting key features of the parameter-detection evaluation analysis curve, and obtaining parameter evaluation analysis coefficients through weighted calculation and analysis;

[0038] The parameter evaluation and analysis coefficient and the historical environmental condition assessment coefficient are used to calculate a parameter adjustment coefficient, and the first parameter is adjusted to obtain an adjusted radar operating parameter.

[0039] Preferably, the calculation formula of the parameter adjustment coefficient is as follows:

[0040] ; In the formula, P represents the parameter adjustment coefficient; It is represented as the adjustment balance factor of the current first parameter; k is represented as the corresponding parameter evaluation analysis coefficient of the current first parameter; It is expressed as the correlation coefficient between the current first parameter and the remaining x-th radar operating parameter, where: ; Expressed as parameter-related compensation factor; It is expressed as the absolute difference between the historical environmental condition evaluation coefficient of the jth reference radar working parameter group and the first environmental condition evaluation coefficient, where j R; Expressed as the impact weight of environmental conditions; It is expressed as the mean difference of historical environmental status assessment coefficients.

[0041] Preferably, the target detection module includes:

[0042] A parameter analysis feedback unit: used for treating the first radar point cloud data as first point cloud data if the radar operating parameters of the current target vehicle are not adjusted;

[0043] If the radar operating parameters of the current target vehicle have been adjusted, the radar point cloud data collected by the vehicle radar after the adjusted operating parameters is regarded as the first point cloud data;

[0044] Transmitting the first point cloud data to a point cloud data processing unit for data preprocessing;

[0045] Point cloud processing unit: used for filtering the first point cloud data, traversing each point in the point cloud for each frame of the first point cloud data, and identifying ground points and unexisting points that cannot be projected onto the first image acquired by the preset image acquisition tool;

[0046] Deleting ground points and unstored points from the first point cloud data to obtain target point cloud data;

[0047] Target detection unit: used to perform target detection on target point cloud data using radar detection block; and perform data fusion on the image target detection data and radar target detection data using data fusion block;

[0048] Radar detection block: used to introduce the Pointpillars algorithm to perform target detection on the target point cloud data and obtain radar target detection data;

[0049] Data fusion block: used to perform spatiotemporal synchronization matching and frame fusion on the image target detection data and the radar target detection data to obtain a first detection fusion result;

[0050] The corresponding target confidences of all detected targets in the first detection fusion result are corrected to generate a target detection fusion result.

[0051] Preferably, the early warning module includes:

[0052] Analysis unit: used to evaluate the danger level of each fusion detection target based on the target detection fusion result and adopt a set danger evaluation mechanism to obtain a danger evaluation value;

[0053] Use the risk assessment value to match the corresponding risk level to each fusion detection target, and generate a risk report collectively;

[0054] Warning unit: used to generate corresponding driving suggestions according to the danger report, and take corresponding safety warning methods to remind the driver in time to avoid potential dangerous situations.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] By preprocessing the data collected by the preset image acquisition tool in real time and evaluating the vehicle environment conditions, the working parameters of the vehicle radar are analyzed and adjusted. By preprocessing the data collected by the vehicle radar and the preset image acquisition tool, target detection and fusion are performed, and the target detection fusion results obtained are analyzed to generate driving suggestions and safety warnings, effectively improving the radar's anti-interference ability and vehicle driving safety.

[0057] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0058] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0060] Figure 1 The present invention is a structural diagram of a surrounding target optimization detection system for a vehicle radar in an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0062] The embodiment of the present invention provides a vehicle radar surrounding target optimization detection system, such as Figure 1 As shown, including:

[0063] Data acquisition module: used to acquire the data collected by the radar of the target vehicle and the preset image acquisition tool in real time, and obtain the first target data by preprocessing the data;

[0064] Parameter analysis module: used to evaluate the driving environment of the current target vehicle based on the first target data, and to analyze and adjust the radar working parameters of the target vehicle;

[0065] Target detection module: used to detect the surrounding targets of the target vehicle and obtain the target detection fusion result;

[0066] Warning module: used to generate corresponding driving suggestions and safety warnings to remind the driver in time to avoid potential dangerous situations by analyzing the target detection fusion results;

[0067] Wherein, the parameter analysis module includes:

[0068] Image initial detection unit: used to train the neural network using a preset amount of images collected from the KITTI dataset, and introduce the attention mechanism to establish the YOLOv4 image detection model;

[0069] Using the YOLOv4 image detection model to detect the first target data, to obtain image target detection data;

[0070] An environmental assessment unit: configured to calculate a first environmental condition assessment coefficient based on the image target detection data;

[0071] Parameter analysis unit: used to extract standard radar operating parameters using parameter acquisition block; compare and analyze the radar operating parameters of the current target vehicle with the standard radar operating parameters using analysis block;

[0072] Parameter acquisition block: used to extract the corresponding matching standard radar working parameters of the first environmental condition assessment range to which the first environmental condition assessment coefficient belongs from the set environmental assessment range-working parameter mapping table;

[0073] Analysis block: used to analyze and adjust the radar operating parameters by comparing the radar operating parameters of the current target vehicle with the standard radar operating parameters.

[0074] In this embodiment, the vehicle radar is used to measure distance and detect objects in the surrounding environment based on ultrasonic or electromagnetic waves, and convert the signals into electrical signals to provide relevant information to the driver or the vehicle control system, including lidar, millimeter wave radar, ultrasonic radar, etc.; the preset image acquisition tool refers to a camera used to simulate the driver's field of view and a camera installed at the rear of the vehicle; the first target data is data obtained by pre-processing the image data collected by the preset image acquisition tool using a histogram equalization algorithm.

[0075] In this embodiment, radar operating parameters refer to important indicators of the performance and technical specifications of different types of vehicle radars. For example, the radar operating parameters of lidar include wavelength, detection distance, angular resolution, etc., and the operating parameters of millimeter wave radar include frequency band selection, transmission power, receiving sensitivity, etc.

[0076] In this embodiment, the KITTI dataset is a dataset composed of computer vision algorithm evaluation data under various vehicle driving scenarios; the preset quantity is set in advance; the YOLOv4 image detection model refers to a detection model constructed by YOLOv4 based on deep convolutional neural network training data and introducing an attention mechanism, which is used for high-precision and real-time detection of targets in images, wherein the role of the attention mechanism is to enable the model to focus on important parts of the input data, thereby improving overall performance and efficiency.

[0077] In this embodiment, the image target detection data is data obtained by detecting the first target data using the YOLOv4 image detection model, including a target frame and a target confidence; the driving state data includes the vehicle driving speed, position and heading angle data.

[0078] In this embodiment, the first environmental condition assessment coefficient refers to a comprehensive indicator for assessing the driving environment conditions of the current target vehicle; the set environmental assessment range-working parameter mapping table refers to a pre-set table consisting of the first environmental condition assessment coefficient range and the corresponding radar working parameter settings, wherein the first environmental condition assessment range refers to the preset value range of the first environmental condition assessment coefficient; the standard radar working parameters refer to the radar working parameters that match the first environmental condition assessment range in the set environmental assessment range-working parameter mapping table.

[0079] In this embodiment, the target detection fusion result is composed of a target fusion frame, category information, and distance information; driving suggestions include parking, speed increase or decrease, lane change, etc.; safety warning methods include sound alarms, light displays, etc.

[0080] The beneficial effects of the above technical solution are: by preprocessing the data collected by the preset image acquisition tool in real time and evaluating the vehicle environmental conditions, the working parameters of the vehicle radar are analyzed and adjusted; by preprocessing the data collected by the vehicle radar and the preset image acquisition tool and then performing target detection and fusion to obtain the target detection fusion results, driving suggestions and safety warnings are generated, and the radar anti-interference capability and vehicle driving safety are effectively improved.

[0081] An embodiment of the present invention provides a surrounding target optimization detection system for a vehicle radar, wherein the data acquisition module comprises:

[0082] Data acquisition unit: used for real-time acquisition of data from the radar of the target vehicle and the preset image acquisition tool, and obtaining first radar point cloud data and first image data respectively;

[0083] Data processing unit: used for preprocessing the first image data by using a histogram equalization algorithm to obtain target image data, and outputting the data as first target data.

[0084] In this embodiment, the vehicle radar is used to measure distance and detect objects in the surrounding environment based on ultrasonic or electromagnetic waves, and convert the signals into electrical signals to provide relevant information to the driver or the vehicle control system, including lidar, millimeter wave radar, ultrasonic radar, etc.; the preset image acquisition tool refers to a camera used to simulate the driver's field of view and a camera installed at the rear of the vehicle.

[0085] In this embodiment, the first radar point cloud data is data obtained by using the radar of the target vehicle to measure distance and detect objects in the surrounding environment, and is installed on the front and rear parts of the vehicle respectively; the first image data refers to the image collected using the preset image acquisition tool.

[0086] In this embodiment, the first target data is data obtained by preprocessing the first image data using a histogram equalization algorithm, wherein the histogram equalization algorithm refers to transforming the grayscale histogram distribution of the image into an approximately uniform distribution through a cumulative distribution function, thereby expanding the brightness range of the image and achieving an effect of enhancing contrast.

[0087] The beneficial effect of the above technical solution is: by utilizing the target vehicle's radar and preset image acquisition tools to collect data in real time, and preprocessing the data collected by the preset image acquisition tools, the data quality can be improved, and data support can be provided for subsequent vehicle driving environment condition assessment.

[0088] An embodiment of the present invention provides a surrounding target optimization detection system for a vehicle radar, wherein the environment assessment unit comprises:

[0089] The image target detection data is combined with the driving state data of the current target vehicle, and the driving environment condition of the current target vehicle is analyzed to obtain a first environmental condition evaluation coefficient;

[0090] The calculation formula of the first environmental condition assessment coefficient is as follows:

[0091] ; Wherein, B represents the first environmental condition assessment coefficient; It is expressed as the influence coefficient of the current weather conditions of the target vehicle on the environmental condition assessment; It is expressed as the influence weight of the image target detection situation with static state attribute on the evaluation of the driving environment condition of the target vehicle; It is represented as the total number of image detection targets with static state attribute at the i-th obstacle level, where i = 1, 2, 3; The state attribute of the i-th obstacle level is represented as the average distance between the static image detection target and the target vehicle; It is expressed as the influence weight of the image target detection situation with dynamic state attribute on the evaluation of the driving environment condition of the target vehicle, where: 1; It is represented as the total number of image detection targets whose state attribute is dynamic at the i-th obstacle level, where i = 1, 2, 3; The state attribute represented as the i-th obstacle level is the average distance between the dynamic image detection target and the target vehicle.

[0092] In this embodiment, the image target detection data is data obtained by detecting the first target data using the YOLOv4 image detection model, including a target frame and a target confidence; the driving status data includes the vehicle's driving speed, position, and heading angle data; the first environmental condition assessment coefficient refers to a comprehensive indicator for assessing the driving environment conditions of the current target vehicle.

[0093] The beneficial effect of the above technical solution is: by combining the acquired image target detection data with the driving status data of the current target vehicle, the first environmental condition assessment coefficient is calculated to assess the vehicle environmental condition, laying the foundation for the subsequent analysis and automatic adjustment of the radar working parameters.

[0094] An embodiment of the present invention provides a surrounding target optimization detection system for a vehicle radar, wherein the analysis block comprises:

[0095] Obtaining a first operating parameter absolute difference between a radar operating parameter of a current target vehicle and an operating parameter of a standard radar;

[0096] If the current target vehicle does not have a radar operating parameter whose absolute difference value of the first operating parameter is greater than the preset parameter difference threshold, the radar operating parameter of the current target vehicle is not adjusted;

[0097] If the absolute difference between the radar operating parameter of the current target vehicle and the first operating parameter of the standard radar operating parameter is greater than the preset parameter difference threshold, the radar operating parameter of the current target vehicle is adjusted to the corresponding standard radar operating parameter;

[0098] If the current target vehicle has a single or multiple radar operating parameters whose absolute difference of the first operating parameters is greater than a preset parameter difference threshold, the corresponding radar operating parameters are marked as the first parameters;

[0099] treating the remaining radar operating parameters except the first parameter as the reference radar operating parameters;

[0100] The parameter value of the reference radar working parameter of the current target vehicle is combined with the set parameter extension value to obtain the reference working parameter range;

[0101] Extracting all corresponding historical radar operating parameter groups of vehicles within the first environmental condition assessment range from the vehicle radar database, and collecting them to obtain a first historical data set;

[0102] Filtering out corresponding historical radar operating parameter groups from the first historical data set, where the historical parameter values ​​of the corresponding reference radar operating parameters belong to the reference operating parameter range, and marking them as first radar operating parameter groups;

[0103] Obtaining a vehicle radar detection evaluation value of each first radar operating parameter group, and marking a first radar operating parameter group whose vehicle radar detection evaluation value is greater than a set evaluation threshold as a reference radar operating parameter group;

[0104] Constructing a parameter-detection evaluation analysis curve based on the vehicle radar detection evaluation value of the reference radar working parameter group and the historical parameter value corresponding to the first parameter;

[0105] Extracting key features of the parameter-detection evaluation analysis curve, and obtaining parameter evaluation analysis coefficients through weighted calculation and analysis;

[0106] The parameter evaluation and analysis coefficient and the historical environmental condition assessment coefficient are used to calculate a parameter adjustment coefficient, and the first parameter is adjusted to obtain an adjusted radar operating parameter.

[0107] In this embodiment, the absolute difference of the first operating parameter refers to the absolute difference between the radar operating parameter of the target vehicle and the corresponding standard radar operating parameter; the preset parameter difference threshold is set in advance; the first parameter refers to the radar operating parameter of the current target vehicle whose single or multiple first operating parameter absolute difference is greater than the preset parameter difference threshold; the comparison radar operating parameter refers to the remaining radar operating parameters other than the first parameter when the current target vehicle has one or more first operating parameter absolute difference greater than the preset parameter difference threshold.

[0108] In this embodiment, for example, there are first working parameter absolute differences t1, t2, t3 between the radar operating parameters s1, s2, s3 of the target vehicle 1 and the corresponding standard radar operating parameters, wherein the first working parameter absolute differences t1, t2, t3 are not greater than the preset parameter difference threshold, and the radar operating parameters of the target vehicle 1 are not adjusted at this time.

[0109] In this embodiment, the set parameter extension value is set in advance; the comparison working parameter range is a parameter value range obtained by combining the parameter upper limit obtained by adding the parameter value of the comparison radar working parameter of the current target vehicle to the set parameter extension value, and the parameter lower limit obtained by subtracting the set parameter extension value from the parameter value of the comparison radar working parameter of the current target vehicle.

[0110] In this embodiment, the historical radar operating parameter group refers to the radar operating parameters used by the vehicle radar when a vehicle of the same type as the target vehicle was driving in the past; the first historical data set refers to a data set constructed using the historical radar operating parameter groups; the first radar operating parameter group refers to all historical radar operating parameter groups corresponding to the historical parameter values ​​of the reference radar operating parameters that belong to the reference working parameter range; the vehicle radar detection evaluation value is used to quantify the detection effect of the vehicle radar when historically using the first radar operating parameter group for target detection, and is an indicator evaluation value obtained by evaluating the detection effect of the vehicle radar when historically using the first radar operating parameter group for target detection based on the detection evaluation index in advance, and is obtained after weighted averaging, wherein the detection evaluation index includes accuracy, false alarm rate and missed detection rate; setting the evaluation threshold is set in advance.

[0111] In this embodiment, the reference radar working parameter group refers to the first radar working parameter group whose vehicle radar detection evaluation value is greater than the set evaluation threshold; the parameter-detection evaluation analysis curve is a curve chart constructed in sequence with the vehicle radar detection evaluation value of the reference radar working parameter group as the dependent variable and the historical parameter value corresponding to the first parameter as the independent variable; the parameter evaluation analysis coefficient is used to characterize the influence of the radar working parameters on the radar target detection effect, and is obtained by weighted averaging the key features extracted from the parameter-detection evaluation analysis curve, wherein the key features include the slope, acceleration and curvature of the curve; the historical environmental condition evaluation coefficient refers to the comprehensive evaluation index of the vehicle driving environment conditions when the vehicle history uses the reference radar working parameter group for target detection, and is determined by weighted averaging the index evaluation values ​​obtained by pre-evaluating the vehicle driving environment based on the environmental evaluation index, wherein the environmental evaluation index includes road safety, climate conditions and traffic congestion conditions; the parameter adjustment coefficient is used to adjust the first parameter to obtain the adjusted radar working parameter.

[0112] The beneficial effect of the above technical solution is: by combining and analyzing the radar working parameter group used in the vehicle's historical driving and the corresponding historical radar detection effects, the parameter adjustment coefficient is obtained to flexibly adjust the radar working parameters of the target vehicle based on environmental requirements, which can effectively optimize the radar's detection performance and is beneficial to improving the environmental perception capability of the vehicle radar.

[0113] An embodiment of the present invention provides a vehicle radar surrounding target optimization detection system, and the calculation formula of the parameter adjustment coefficient is as follows:

[0114] ; In the formula, P represents the parameter adjustment coefficient; It is represented as the adjustment balance factor of the current first parameter; k is represented as the corresponding parameter evaluation analysis coefficient of the current first parameter; It is expressed as the correlation coefficient between the current first parameter and the remaining x-th radar operating parameter, where: ; Expressed as parameter-related compensation factor; It is expressed as the absolute difference between the historical environmental condition evaluation coefficient of the jth reference radar working parameter group and the first environmental condition evaluation coefficient, where j R; Expressed as the impact weight of environmental conditions; It is expressed as the mean difference of historical environmental status assessment coefficients.

[0115] The beneficial effect of the above technical solution is: by calculating the parameter adjustment coefficient, the radar operating parameters of the target vehicle are adjusted based on environmental requirements, which can effectively optimize the radar's detection performance and help improve the vehicle radar's environmental perception capability.

[0116] An embodiment of the present invention provides a surrounding target optimization detection system for a vehicle radar, wherein the target detection module comprises:

[0117] A parameter analysis feedback unit: used for treating the first radar point cloud data as first point cloud data if the radar operating parameters of the current target vehicle are not adjusted;

[0118] If the radar operating parameters of the current target vehicle have been adjusted, the radar point cloud data collected by the vehicle radar after the adjusted operating parameters is regarded as the first point cloud data;

[0119] Transmitting the first point cloud data to a point cloud data processing unit for data preprocessing;

[0120] Point cloud processing unit: used for filtering the first point cloud data, traversing each point in the point cloud for each frame of the first point cloud data, and identifying ground points and unexisting points that cannot be projected onto the first image acquired by the preset image acquisition tool;

[0121] Deleting ground points and unstored points from the first point cloud data to obtain target point cloud data;

[0122] Target detection unit: used to perform target detection on target point cloud data using radar detection block; and perform data fusion on the image target detection data and radar target detection data using data fusion block;

[0123] Radar detection block: used to introduce the Pointpillars algorithm to perform target detection on the target point cloud data and obtain radar target detection data;

[0124] Data fusion block: used to perform spatiotemporal synchronization matching and frame fusion on the image target detection data and the radar target detection data to obtain a first detection fusion result;

[0125] The corresponding target confidences of all detected targets in the first detection fusion result are corrected to generate a target detection fusion result.

[0126] In this embodiment, the removal of ground points is achieved by accurately identifying and deleting ground points using a point cloud normal vector method; unexisting points refer to points that cannot be projected onto the first image acquired by a preset image acquisition tool.

[0127] In this embodiment, the target point cloud data is obtained by deleting the ground points and unstored points in the first point cloud data; the Pointpillars algorithm is an algorithm for 3D target detection, which mainly uses point cloud data and deep learning models to realize the detection and positioning of target objects; the radar target detection data includes the target frame and distance information.

[0128] In this embodiment, spatiotemporal synchronization matching includes time synchronization and space synchronization, wherein time synchronization refers to selecting image data collected by the camera closest to the corresponding time based on radar time, with the purpose of ensuring that data from different collection sources can be processed and fused in a unified time frame, and space synchronization is the process of joint calibration of camera and radar; frame fusion is the process of projecting radar detection data and target frames in image target detection data onto the same frame image for edge overlap; the first detection fusion result is a fusion result obtained by spatiotemporal synchronization matching and frame fusion of image target detection data and radar target detection data, including target fusion frame, distance information and category information, wherein category information refers to target category probability and confidence; distance information includes vehicle distance, speed, etc.; the target detection fusion result is a fusion result obtained by correcting the target confidence in the first detection fusion result, and is composed of a target fusion frame, category information and distance information.

[0129] In this embodiment, for example, there is a target confidence A1 of the detection target a1 in the first detection fusion result 1 = 0.65, and the corrected target confidence , c is the adjustment coefficient; e is a mathematical constant, approximately equal to 2.71828.

[0130] The beneficial effect of the above technical solution is: by performing target detection and then fusing the data collected by the vehicle radar and camera that have been processed separately, the target detection fusion result is obtained, which can effectively improve the accuracy of target detection, lay a favorable foundation for subsequent driving decisions, and improve the safety and efficiency of the target vehicle's driving.

[0131] An embodiment of the present invention provides a surrounding target optimization detection system for a vehicle radar, wherein the early warning module comprises:

[0132] Analysis unit: used to evaluate the danger level of each fusion detection target based on the target detection fusion result and adopt a set danger evaluation mechanism to obtain a danger evaluation value;

[0133] Use the risk assessment value to match the corresponding risk level to each fusion detection target, and generate a risk report collectively;

[0134] Warning unit: used to generate corresponding driving suggestions according to the danger report, and take corresponding safety warning methods to remind the driver in time to avoid potential dangerous situations.

[0135] In this embodiment, the risk assessment mechanism is set in advance and is used to evaluate the risk level of the fused detection target according to the risk assessment index, wherein the risk assessment index includes vehicle distance, speed change, etc.; the risk assessment value is obtained by evaluating the risk level of the fused detection target using the risk assessment index, and then obtaining the index evaluation value after weighted averaging, and is used to quantify the risk level of the fused detection target to the target vehicle, wherein the fused detection target refers to the target obtained after fusing the camera and radar target detection results; the risk level is used to express the threat level of the fused detection target to the target vehicle; the risk report is a report generated by the risk assessment values ​​of all fused detection targets and the corresponding risk levels; driving suggestions include parking, speed increase or decrease, lane change, etc.; safety warning methods include sound alarms, light displays, etc.

[0136] The beneficial effect of the above technical solution is: by generating corresponding driving suggestions and safety warnings based on the analysis of the target detection fusion results, the driver can be reminded in time to help him make safe decisions, avoid potential dangerous situations, and improve the safety and efficiency of the target vehicle's driving.

[0137] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A vehicle radar surrounding target optimization detection system, characterized in that: include: Data acquisition module: used to acquire the data collected by the radar of the target vehicle and the preset image acquisition tool in real time, and obtain the first target data by preprocessing the data; Parameter analysis module: used to evaluate the driving environment of the current target vehicle based on the first target data, and to analyze and adjust the radar working parameters of the target vehicle; Target detection module: used to detect the surrounding targets of the target vehicle and obtain the target detection fusion result; Warning module: used to generate corresponding driving suggestions and safety warnings to remind the driver in time to avoid potential dangerous situations by analyzing the target detection fusion results; Wherein, the parameter analysis module includes: Image initial detection unit: used to train the neural network using a preset amount of images collected from the KITTI dataset, and introduce the attention mechanism to establish the YOLOv4 image detection model; Using the YOLOv4 image detection model to detect the first target data, to obtain image target detection data; An environmental assessment unit: configured to calculate a first environmental condition assessment coefficient based on the image target detection data; Parameter analysis unit: used to extract standard radar operating parameters using parameter acquisition block; compare and analyze the radar operating parameters of the current target vehicle with the standard radar operating parameters using analysis block; Parameter acquisition block: used to extract the corresponding matching standard radar working parameters of the first environmental condition assessment range to which the first environmental condition assessment coefficient belongs from the set environmental assessment range-working parameter mapping table; Analysis block: used to analyze and adjust the radar operating parameters by comparing the radar operating parameters of the current target vehicle with the standard radar operating parameters.

2. The vehicle radar surrounding target optimization detection system according to claim 1, characterized in that: The data acquisition module comprises: Data acquisition unit: used for real-time acquisition of data from the radar of the target vehicle and the preset image acquisition tool, and obtaining first radar point cloud data and first image data respectively; Data processing unit: used for preprocessing the first image data by using a histogram equalization algorithm to obtain target image data, and outputting the data as first target data.

3. The vehicle radar surrounding target optimization detection system according to claim 2, characterized in that: The preset image acquisition tool refers to a camera used to simulate the driver's field of view and a camera installed at the rear of the vehicle.

4. The vehicle radar surrounding target optimization detection system according to claim 1, characterized in that: The environmental assessment unit includes: The image target detection data is combined with the driving state data of the current target vehicle, and the driving environment condition of the current target vehicle is analyzed to obtain a first environmental condition evaluation coefficient; The calculation formula of the first environmental condition assessment coefficient is as follows: ; Wherein, B represents the first environmental condition assessment coefficient; It is expressed as the influence coefficient of the current weather conditions of the target vehicle on the environmental condition assessment; It is expressed as the influence weight of the image target detection situation with static state attribute on the evaluation of the driving environment condition of the target vehicle; It is represented as the total number of image detection targets with static state attribute at the i-th obstacle level, where i = 1, 2, 3; The state attribute of the i-th obstacle level is represented as the average distance between the static image detection target and the target vehicle; It is expressed as the influence weight of the image target detection situation with dynamic state attribute on the evaluation of the driving environment condition of the target vehicle, where: 1; It is represented as the total number of image detection targets whose state attribute is dynamic at the i-th obstacle level, where i=1, 2, 3; The state attribute represented as the i-th obstacle level is the average distance between the dynamic image detection target and the target vehicle.

5. The vehicle radar surrounding target optimization detection system according to claim 1, characterized in that: The analysis block includes: Obtaining a first operating parameter absolute difference between a radar operating parameter of a current target vehicle and an operating parameter of a standard radar; If the current target vehicle does not have a radar operating parameter whose absolute difference value of the first operating parameter is greater than the preset parameter difference threshold, the radar operating parameter of the current target vehicle is not adjusted; If the absolute difference between the radar operating parameter of the current target vehicle and the first operating parameter of the standard radar operating parameter is greater than the preset parameter difference threshold, the radar operating parameter of the current target vehicle is adjusted to the corresponding standard radar operating parameter; If the current target vehicle has a single or multiple radar operating parameters whose absolute difference of the first operating parameters is greater than a preset parameter difference threshold, the corresponding radar operating parameters are marked as the first parameters; treating the remaining radar operating parameters except the first parameter as the reference radar operating parameters; The parameter value of the reference radar working parameter of the current target vehicle is combined with the set parameter extension value to obtain the reference working parameter range; Extracting all corresponding historical radar operating parameter groups of vehicles within the first environmental condition assessment range from the vehicle radar database, and collecting them to obtain a first historical data set; Filtering out corresponding historical radar operating parameter groups from the first historical data set, where the historical parameter values ​​of the corresponding reference radar operating parameters belong to the reference operating parameter range, and marking them as first radar operating parameter groups; Obtaining a vehicle radar detection evaluation value of each first radar operating parameter group, and marking a first radar operating parameter group whose vehicle radar detection evaluation value is greater than a set evaluation threshold as a reference radar operating parameter group; Constructing a parameter-detection evaluation analysis curve based on the vehicle radar detection evaluation value of the reference radar working parameter group and the historical parameter value corresponding to the first parameter; Extracting key features of the parameter-detection evaluation analysis curve, and obtaining parameter evaluation analysis coefficients through weighted calculation and analysis; The parameter evaluation and analysis coefficient and the historical environmental condition assessment coefficient are used to calculate a parameter adjustment coefficient, and the first parameter is adjusted to obtain an adjusted radar operating parameter.

6. The vehicle radar surrounding target optimization detection system according to claim 5, characterized in that: The calculation formula of the parameter adjustment coefficient is as follows: ; In the formula, P represents the parameter adjustment coefficient; It is represented as the adjustment balance factor of the current first parameter; k is represented as the corresponding parameter evaluation analysis coefficient of the current first parameter; It is expressed as the correlation coefficient between the current first parameter and the remaining x-th radar operating parameter, where: ; Expressed as parameter-related compensation factor; It is expressed as the absolute difference between the historical environmental condition evaluation coefficient of the jth reference radar working parameter group and the first environmental condition evaluation coefficient, where j R; Expressed as the impact weight of environmental conditions; It is expressed as the mean difference of historical environmental status assessment coefficients.

7. The vehicle radar surrounding target optimization detection system according to claim 2, characterized in that: The target detection module comprises: A parameter analysis feedback unit: used for treating the first radar point cloud data as first point cloud data if the radar operating parameters of the current target vehicle are not adjusted; If the radar operating parameters of the current target vehicle have been adjusted, the radar point cloud data collected by the vehicle radar after the adjusted operating parameters is regarded as the first point cloud data; Transmitting the first point cloud data to a point cloud data processing unit for data preprocessing; Point cloud processing unit: used for filtering the first point cloud data, traversing each point in the point cloud for each frame of the first point cloud data, and identifying ground points and unexisting points that cannot be projected onto the first image acquired by the preset image acquisition tool; Deleting ground points and unstored points from the first point cloud data to obtain target point cloud data; Target detection unit: used to perform target detection on target point cloud data using radar detection block; and perform data fusion on the image target detection data and radar target detection data using data fusion block; Radar detection block: used to introduce the Pointpillars algorithm to perform target detection on the target point cloud data and obtain radar target detection data; Data fusion block: used to perform spatiotemporal synchronization matching and frame fusion on the image target detection data and the radar target detection data to obtain a first detection fusion result; The corresponding target confidences of all detected targets in the first detection fusion result are corrected to generate a target detection fusion result.

8. The vehicle radar surrounding target optimization detection system according to claim 1, characterized in that: The early warning module comprises: Analysis unit: used to evaluate the danger level of each fusion detection target based on the target detection fusion result and adopt a set danger evaluation mechanism to obtain a danger evaluation value; Use the risk assessment value to match the corresponding risk level to each fusion detection target, and generate a risk report collectively; Warning unit: used to generate corresponding driving suggestions according to the danger report, and take corresponding safety warning methods to remind the driver in time to avoid potential dangerous situations.