A multi-view fusion method for simultaneous positioning and mapping of automotive radar

Abstract

The present invention relates to a multi-perspective fused automobile radar synchronous positioning and mapping method, which belongs to the field of synchronous positioning and mapping technology, and solves the problem of RCS flickering in the prior art. By acquiring the original radar data and corresponding two-dimensional radar images acquired by each automobile radar located at different positions of the automobile; and based on this, obtaining the time and space synchronization between multiple automobile radars, that is, obtaining the radar image in the same spatial coordinate system at the same time; based on the time and space synchronization results, the imaging results of each radar are fused; according to the fused radar image, the radar point cloud at the current moment is extracted; by matching the radar point cloud at the current moment with the radar point cloud at the previous moment, the posture transformation relationship between the vehicle at the current moment and the previous moment is obtained; the posture transformation relationship is solved to perform positioning and mapping. It is achieved that while solving the RCS flicker, accurate positioning results are obtained and a map corresponding to the real environment is reconstructed.

Description

Technical Field

[0001] The present invention relates to the technical field of synchronous positioning and mapping, and in particular to a multi-view fusion automobile radar synchronous positioning and mapping method. Background Art

[0002] Simultaneous Localization And Mapping (SLAM) is a key technology for vehicles to accurately locate their own position and perceive the surrounding environment in unknown environments. Most existing SLAM systems are based on lidar or optical cameras, but these sensors are easily affected by external environments such as haze and darkness. In recent years, radar sensors have provided new solutions for SLAM with their all-day and all-weather working characteristics.

[0003] Multiple In Multiple Out (MIMO) radar has become the mainstream radar system for vehicle applications due to its low cost and high frame rate. There are currently three main technical routes for MIMO radar SLAM: radar SLAM based on space-Doppler information, SLAM based on radar images, and SLAM based on radar point clouds. With the rise of 4D radar in recent years, automotive radar has achieved a higher angular resolution and has the ability to form a three-dimensional point cloud similar to that of lidar. However, due to the different imaging mechanisms, compared with lidar point clouds, radar point clouds have RCS (Rich Communication Services) flickering, which makes point cloud-based radar SLAM unable to work properly.

[0004] The laser radar point cloud coordinates reflect the geometric position of the target, but the radar point cloud coordinates reflect the scattering center position of the target. For extended targets such as cars and buildings, the radar echo is a superposition of local strong scattering point echoes. With different observation angles, the echo synthesis phase will continue to change, causing the target to have scattering signals with drastic changes in intensity at different directions, which is the RCS flicker phenomenon. Due to the RCS flicker phenomenon, the two frames of radar images / point clouds continuously collected by the radar platform during the movement process will be de-correlated, making the matching-based SLAM method invalid. Summary of the invention

[0005] In view of the above analysis, an embodiment of the present invention aims to provide a multi-perspective fusion automotive radar synchronous positioning and mapping method to solve the problem of RCS flickering in the prior art.

[0006] On the one hand, an embodiment of the present invention provides a multi-view fusion automotive radar synchronous positioning and mapping method, including:

[0007] Acquire radar raw data and corresponding two-dimensional radar images acquired by various vehicle radars located at different positions of the vehicle;

[0008] Based on the radar image calculated from the original radar data, time and space synchronization is performed between multiple vehicle radars to obtain radar images in the same spatial coordinate system at the same time;

[0009] Based on the time and space synchronization results, the imaging results of each radar are integrated;

[0010] Extract the radar point cloud at the current moment based on the fused radar image;

[0011] Match the radar point cloud at the current moment with the radar point cloud at the previous moment to obtain the position transformation relationship between the vehicle at the current moment and the previous moment;

[0012] Solve the pose transformation relationship for positioning and mapping.

[0013] Optionally, performing radar imaging signal processing according to raw data obtained by each automobile radar to obtain a two-dimensional radar image includes:

[0014] Acquire the echo signals of each automobile radar in different directions, and establish the automobile radar imaging signal model according to the azimuth;

[0015] A two-dimensional radar image is obtained based on the established automobile radar imaging signal model.

[0016] Optionally, the radar image calculated based on the radar raw data is synchronized in time and space between multiple vehicle radars to obtain radar images in the same spatial coordinate system at the same time, including:

[0017] Time synchronization is performed by comparing the timestamp information of the radar raw data;

[0018] Spatial synchronization is achieved through matching relations between radar image correlations.

[0019] Optionally, the fusing imaging results of each radar based on time and space synchronization includes:

[0020] The radar image is fused according to the current moment, and the radar image fusion method is:

[0021]

[0022] Among them, I n (x, y) is the radar image to be fused obtained by the radar numbered n in the same spatial coordinate system at the same time, I(x, y) is the fused radar image, α n is the weighting factor of the nth radar sensor, and the weighting factor is negatively correlated with the radar measurement error.

[0023] Optionally, matching the radar point cloud at the current moment with the radar point cloud at the previous moment to obtain a posture transformation relationship of the vehicle between the two stations includes:

[0024] According to the fused radar image, extract the point cloud in the radar image;

[0025] Perform constant false alarm rate detection based on the extracted radar point cloud to obtain the radar image corresponding to the current vehicle at a certain moment;

[0026] According to the matching of the current radar point cloud with the radar point cloud at the previous moment, the posture transformation relationship between the vehicle at the current moment and the previous moment is obtained.

[0027] Optionally, obtaining a posture transformation relationship between the vehicle at the current moment and the previous moment according to matching the current radar point cloud with the radar point cloud at the previous moment includes:

[0028] According to the nearest neighbor point cloud matching algorithm, the following optimization objective function is constructed:

[0029]

[0030] Among them, P m With Q m Represent the mth point in point clouds P and Q respectively;

[0031] Based on the optimization objective function of the nearest neighbor point cloud matching algorithm, the matching objective function between the current and previous radar point clouds is obtained as follows:

[0032]

[0033] Among them, P and Q are the matching point sets in the two site clouds; (x, y, z) represents the three-dimensional space coordinates; (c) represents the point cloud intensity; w is the weight of the point cloud; the matching objective function is used to represent the posture transformation relationship.

[0034] Optionally, solving the posture transformation to perform positioning and mapping includes:

[0035] Based on the point cloud matching results, the pose transformation relationship between radar point clouds is obtained;

[0036] Solve the posterior probability according to the posture transformation relationship;

[0037] Based on the solved posterior probability, the binary Bayesian formula is used to update the radar synchronization and map information.

[0038] Optionally, obtaining the posture transformation relationship between radar point clouds based on the point cloud matching result includes:

[0039] According to the point cloud matching results, after the rotation matrix and translation matrix, the motion relationship between the radar point cloud of the vehicle at the current moment and the previous moment is obtained.

[0040] Optionally, the posterior probability is:

[0041]

[0042] Among them, m xy,t represents the occupancy rate of the map m at the (x, y) coordinate at time t, and is a 0-1 distribution random variable. The probability of the grid cell being occupied is expressed as P(m xy,t ), Z 1:t is the radar measurement information Z in discrete time 1~t 1:t , X 1:t is the position information within time 1~t, P(m xy,t |Z 1:t ,X 1:t ) represents the occupancy probability of the map m at the (x, y) coordinate obtained based on the sensor measurement information and posture information in the time period 1 to t. The operator It means to multiply each (x,y) value.

[0043] Optionally, the obtained solved posterior probability is updated by a binary Bayesian formula to obtain radar synchronization and map information, including:

[0044]

[0045] Among them, L(m xy,t ) represents the logarithmic operation of the occupation probability of the (x, y) coordinate in the map m at time t, m xy,t represents the occupancy rate of the map m at the (x, y) coordinate at time t, and is a 0-1 distribution random variable. The probability of the grid cell being occupied is expressed as P(m xy,t ), Z1: t is the radar measurement information Z in discrete time 1~t 1:t , X 1:t is the position information within time 1~t, P(m xy,t |Z 1:t ,X 1:t ) represents the occupancy probability of the map m at the (x, y) coordinate obtained based on the sensor measurement information and posture information in the time period 1 to t.

[0046] On the other hand, compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0047] 1. The present invention establishes a model for fusing the observation results of multi-angle radars through time and space synchronization between radar sensors, improves the target RCS flickering problem faced by single-angle observation, and improves the system robustness;

[0048] 2. The present invention fully considers the spatial position and intensity of radar point clouds, proposes a matching algorithm suitable for radar point clouds, improves the accuracy of point cloud matching, and thus improves the accuracy of subsequent positioning and mapping.

[0049] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0051] Figure 1 It is a schematic diagram of a method flow of a multi-view fusion automotive radar synchronous positioning and mapping method proposed in one embodiment of the present invention;

[0052] Figure 2 It is a detailed flow chart of a multi-view fusion automotive radar synchronous positioning and mapping method proposed in one embodiment of the present invention;

[0053] Figure 3 is the result of the multi-view radar image fusion method before fusion proposed by an embodiment of the present invention; wherein Figure 3 a. Figure 3 b. Figure 3 c. Figure 3 d are radar images obtained at different orientations;

[0054] Figure 4 is the fusion result of the multi-view radar image fusion method proposed in one embodiment of the present invention;

[0055] Figure 5 is the processing result of the radar point cloud matching method after multi-view fusion proposed in one embodiment of the present invention; wherein, Figure 5 a is the radar point cloud image at the previous moment;

[0056] Figure 5 b is the radar point cloud image at the current moment;

[0057] Figure 5 c is the point cloud matching result image;

[0058] Figure 6 It is the processing result of executing the traditional point cloud matching method without multi-view fusion;

[0059] in, Figure 6 a is the radar point cloud image at the previous moment;

[0060] Figure 6 b is the radar point cloud image at the current moment;

[0061] Figure 6 c is the point cloud matching result image;

[0062] Figure 7 This is a graph of the simultaneous positioning and mapping results of a car radar proposed in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0064] A specific embodiment of the present invention discloses a multi-view fusion automotive radar synchronous positioning and mapping method, such as Figure 1 shown.

[0065] Step S1: acquiring radar raw data and corresponding two-dimensional radar images acquired by various vehicle radars located at different positions of the vehicle;

[0066] Specifically, the data received by each automobile radar at the current moment is obtained as the original data, and radar imaging signal processing is performed on the original data obtained by each automobile radar to obtain a two-dimensional radar image; the two-dimensional radar image refers to an image obtained after echo signal processing in azimuth and range directions; the original data is the signal data returned by the radar when the radar detects an obstacle.

[0067] Among them, each automobile radar in the embodiment of the present invention refers to multiple automobile radars installed at different positions of a car respectively, which are used to obtain radar echo signals from different directions, thereby making the radar echo signal more accurate and the obtained image more precise.

[0068] After acquiring the raw data, each automotive radar first performs range pulse compression. In order to perform azimuth imaging, the automotive radar imaging signal model is established as follows:

[0069]

[0070] Among them, s(t) is the source signal, γ k and θ krepresent the amplitude and angular position of the kth scatterer, u(t) is Gaussian noise, a R (θ k ) and a T (θ k ) represent the steering vectors of the receiving and transmitting arrays respectively and can be expressed as:

[0071]

[0072]

[0073] Where j is the imaginary unit, λ is the operating wavelength, and d T With d R They represent the transmitting array element spacing and the receiving array element spacing respectively.

[0074] Based on orthogonality, each transmit and receive channel can be separated to obtain N T ×N R virtual array elements, and sample the signals of each virtual array element at L points, thereby obtaining a size of (N T ×N R )×L data matrix:

[0075]

[0076] Among them, x 11 It indicates the radar echo signal received by the virtual array element composed of the first transmitting antenna and the first receiving antenna.

[0077] Virtual array elements can form a signal at the nth range gate Among them, x 11 (n) represents the time sampling of the radar echo signal received by the virtual array element composed of the first transmitting antenna and the first receiving antenna at the current moment.

[0078] In order to estimate the radar signal s in the azimuth direction, it is necessary to construct the array flow matrix A according to the spatial layout of the radar virtual array elements, so the following set of equations is constructed and solved:

[0079] y=A·s (5)

[0080] Among them, s is the azimuth signal obtained at the lth range gate, and A is called the array flow pattern:

[0081]

[0082] Among them, the operator Represents the Kronecker product. By solving the above formula using the least squares method, we can get the range and azimuth two-dimensional radar image.

[0083] Step S2: Based on the radar image calculated from the radar raw data, time and space synchronization is performed between multiple vehicle radars to obtain radar images in the same spatial coordinate system at the same time;

[0084] Specifically, after obtaining the radar images of each radar, it is necessary to perform time and space synchronization between multiple radars based on the radar raw data and radar images. Time synchronization is mainly achieved by comparing the timestamp information of each radar raw data. When the difference between two timestamps is less than a fixed threshold, the two sets of radar data can be considered to be time synchronized. The fixed threshold is a timestamp error range set based on experience. Spatial synchronization is mainly achieved by the correlation between two-dimensional radar images. Specifically, the spatial synchronization method is to solve the phase relationship of each radar image in the frequency domain:

[0085]

[0086] in, It represents the maximum value of the phase relationship function modulus between the main radar image and the reference radar image in the frequency domain at the current moment, the spatial offset of the main radar image (x0, y0), and the main radar image is I n (x,y), the reference radar image is I ref (x,y), represents the Fourier transform, represents the inverse Fourier transform, {·} * represents the complex conjugate operation, and || represents the modulo operation.

[0087] The phase relationship C(x,y) is used as the evaluation index to evaluate the main radar image I n The spatial coordinates in (x, y) are translated to find the maximum value of C(x, y). n The spatial coordinate translation (x0, y0) of (x, y) is the spatial synchronization relationship between radars. That is, through the correlation matching between the two radar images, the optimal solution of the spatial translation between the radar images is found according to the maximum value of the correlation matching, thereby reflecting the spatial synchronization relationship between radars.

[0088] Step S3: Based on the time and space synchronization results, the imaging results of each radar are integrated;

[0089] Specifically, after executing the time synchronization and the space synchronization, the radar image fusion method for the current multi-view radar image is as follows:

[0090]

[0091] Among them, I n(x, y) is the radar image to be fused obtained by the radar numbered n in the same spatial coordinate system at the same time, I(x, y) is the fused radar image, α n is the weighting factor of the nth radar sensor, and the weighting factor is negatively correlated with the radar measurement error.

[0092] Step S4: extracting the radar point cloud at the current moment according to the fused radar image;

[0093] After obtaining the fused radar image through the above formula, it is necessary to extract the point cloud in the radar image. Before extracting the point cloud, in order to improve the influence of side lobes and noise in the radar image, it is necessary to perform constant false alarm rate detection. The spatial coordinates corresponding to the radar image pixel points retained by the constant false alarm rate detection are the radar point clouds extracted at a certain moment of the current vehicle position. The said certain moment refers to the moment corresponding to the acquisition of this radar image.

[0094] Step S5: Matching the radar point cloud at the current moment with the radar point cloud at the previous moment to obtain the position transformation relationship between the vehicle at the current moment and the previous moment;

[0095] Specifically, by matching the radar point cloud of the current station with the radar point cloud of the previous station, the position transformation relationship of the vehicle between the two stations can be obtained, including rotation R and translation T. Based on the nearest neighbor point cloud matching algorithm, the following optimization objective function is constructed:

[0096]

[0097] Among them, P m With Q m Represent the mth point in point clouds P and Q respectively.

[0098] However, when searching for corresponding point sets, the nearest neighbor point cloud matching algorithm uses the shortest distance as the objective function. But in the radar point cloud, the intensity of the radar point cloud contains the RCS information of the target. By installing car radars at different positions on a car and fusing multiple radar echo signals, the purpose of eliminating RCS flicker is achieved. After eliminating RCS flicker, the matching accuracy of the radar point cloud can be improved by increasing the matching in the point cloud intensity dimension. Therefore, in order to obtain the match between the radar point cloud at the current moment and the previous moment, the objective function of the radar point cloud matching algorithm is obtained as follows:

[0099]

[0100] Among them, P and Q are the matching point sets in the clouds of the two stations; (x, y, z) represents the three-dimensional spatial coordinates; (c) represents the point cloud intensity; the denominator mainly plays the role of eliminating the dimension, so that the spatial information and the intensity information can be directly added; w is the weight of the point cloud, which reflects the frequency of the point in multiple radar observations. Point cloud matching uses the current station radar point cloud and the average radar point cloud of multiple historical stations to match to improve the matching stability.

[0101] Step S6: Solve the posture transformation relationship to perform positioning and mapping.

[0102] Specifically, through the point cloud matching results, we can obtain the position transformation relationship between the radar point clouds, that is, the rotation matrix R and the translation matrix T, and then we can get the motion relationship between the vehicle's radar point clouds at the current moment and the previous moment. In order to build an environmental map, the map construction steps are transformed into solving the following posterior probability:

[0103]

[0104] Among them, m xy,t represents the occupancy rate of the map m at the (x, y) coordinate at time t, and is a 0-1 distribution random variable. The probability of the grid cell being occupied is expressed as P(m xy,t ), Z1: t is the radar measurement information Z in discrete time 1~t 1:t , X 1:t is the position information within time 1~t, P(m xy,t |Z 1:t ,X 1:t ) represents the occupancy probability of the map m at the (x, y) coordinate obtained based on the sensor measurement information and posture information in the time period 1 to t. The operator It means to multiply each (x,y) value.

[0105] In addition, in order to avoid numerical instability caused by very small and very large probabilities, based on the above posterior probability, the logarithmic representation method is used to obtain the following map update formula through the binary Bayesian filter:

[0106]

[0107] Among them, L(m xy,t ) represents the logarithmic operation of the occupation probability of the (x, y) coordinate in the map m at time t.

[0108] A specific embodiment of the present invention is:

[0109] The detailed flow chart of this method is as follows Figure 2 As shown:

[0110] First, according to the raw data obtained by each vehicle radar, the radar imaging operation is implemented by solving formula (5) to obtain N (for example, N = 4) radar images, as shown in Figure 4 As shown;

[0111] However, in order to fuse images from different radar sensors, each radar needs to be synchronized in time and space. During the data acquisition phase of each radar, the radar will record the timestamp of each frame of data, so the time synchronization between multiple radars can be achieved by aligning the timestamps. Spatial synchronization is divided into two parts: coarse synchronization and fine synchronization. When the radar equipment is installed, the installation position and installation angle of the radar can be measured by measuring instruments, and these parameters can be used to achieve coarse synchronization of the radar spatial position. In order to improve the accuracy of subsequent radar point cloud extraction and point cloud matching, fine synchronization is required on the basis of coarse synchronization of the radar spatial position. Fine synchronization of the radar spatial position is achieved by formula (7).

[0112] After synchronizing the radar images of each radar sensor in time and space, since each radar sensor may have different measurement errors, the radar images can be weighted fused using formula (8) to obtain: Figure 4 The fused radar image shown.

[0113] For the fused radar image, the radar point cloud can be obtained after constant false alarm rate detection. After obtaining the multi-station radar point cloud, point cloud matching is performed using formula (10). Figure 6 In the example, due to the lack of multi-view fusion, the RCS flickering phenomenon between the radar point clouds of the two stations is serious. Figure 5 is the radar point cloud matching result obtained after processing in this embodiment. By accumulating the required posture transformation relationship, the simultaneous positioning and map construction of the vehicle can be achieved, wherein the map construction can be achieved by formulas (11) to (12).

[0114] Figure 5 This is the radar synchronous positioning and mapping result obtained in this embodiment. This embodiment can achieve a relative positioning error of about 1%. Therefore, the method of this embodiment can obtain more accurate positioning results in a larger-scale scenario and reconstruct a map corresponding to the real environment.

[0115] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0116] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0117] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A multi-view fusion automotive radar synchronous positioning and mapping method, characterized in that: include: Acquire radar raw data and corresponding two-dimensional radar images acquired by various vehicle radars located at different positions of the vehicle; Based on the radar image calculated from the original radar data, time and space synchronization is performed between multiple vehicle radars to obtain radar images in the same spatial coordinate system at the same time; Based on the time and space synchronization results, the imaging results of each radar are integrated; Extract the radar point cloud at the current moment based on the fused radar image; Match the radar point cloud at the current moment with the radar point cloud at the previous moment to obtain the position transformation relationship between the vehicle at the current moment and the previous moment; Solve the posture transformation relationship for positioning and mapping; The imaging results of each radar are integrated based on the time and space synchronization results, including: The radar image is fused according to the current moment, and the radar image fusion method is: Among them, I n (x, y) is the radar image to be fused obtained by the radar numbered n in the same spatial coordinate system at the same time, I(x, y) is the fused radar image, α n is the weighting factor of the nth radar sensor, and the weighting factor is negatively correlated with the radar measurement error; The matching of the radar point cloud at the current moment with the radar point cloud at the previous moment to obtain the position transformation relationship between the vehicle at the current moment and the previous moment includes: According to the nearest neighbor point cloud matching algorithm, the following optimization objective function is constructed: Among them, P m With Q m Represent the mth point in point clouds P and Q respectively; Based on the optimization objective function of the nearest neighbor point cloud matching algorithm, the matching objective function between the current and previous radar point clouds is obtained as follows: in, P and Q are the matching point sets in the two site clouds; (x, y, z) represent the three-dimensional space coordinates; (c) represents the point cloud intensity; w is the weight of the point cloud; the matching objective function is used to solve the posture transformation relationship.

2. The method for synchronous positioning and mapping of multi-view fusion automotive radar according to claim 1, characterized in that: The radar imaging signal processing according to the raw data obtained by each automobile radar to obtain a two-dimensional radar image includes: Acquire the echo signals of each automobile radar in different directions, and establish the automobile radar imaging signal model according to the azimuth; A two-dimensional radar image is obtained based on the established automobile radar imaging signal model.

3. The method for synchronous positioning and mapping of multi-view fusion automotive radar according to claim 1, characterized in that: The radar image calculated based on the radar raw data is synchronized in time and space between multiple vehicle radars to obtain radar images in the same spatial coordinate system at the same time, including: Time synchronization is performed by comparing the timestamp information of the radar raw data; Spatial synchronization is achieved through matching relations between radar image correlations.

4. The method for synchronous positioning and mapping of multi-view fusion automotive radar according to claim 1, characterized in that: The matching of the radar point cloud at the current moment with the radar point cloud at the previous moment to obtain the position transformation relationship of the vehicle between the two stations includes: According to the fused radar image, extract the point cloud in the radar image; Perform constant false alarm rate detection based on the extracted radar point cloud to obtain the radar image corresponding to the current vehicle at a certain moment; According to the matching of the current radar point cloud with the radar point cloud at the previous moment, the posture transformation relationship between the vehicle at the current moment and the previous moment is obtained.

5. The method for synchronous positioning and mapping of multi-view fusion automotive radar according to claim 1, characterized in that: Solving the posture transformation relationship for positioning and mapping includes: Based on the point cloud matching results, the pose transformation relationship between radar point clouds is obtained; Solve the posterior probability according to the posture transformation relationship; Based on the solved posterior probability, the binary Bayesian formula is used to update the radar synchronization and map information.

6. The method for synchronous positioning and mapping of multi-view fusion automotive radars according to claim 5, characterized in that: The obtaining of the pose transformation relationship between radar point clouds based on the point cloud matching result includes: According to the point cloud matching results, after the rotation matrix and translation matrix, the motion relationship between the radar point cloud of the vehicle at the current moment and the previous moment is obtained.

7. The method for synchronous positioning and mapping of multi-view fusion automotive radars according to claim 5, characterized in that: The posterior probability is: Among them, m xy,t represents the occupancy rate of the map m at the (x, y) coordinate at time t, and is a 0-1 distribution random variable. The probability of the grid cell being occupied is expressed as P(m xy,t ), Z 1:t is the radar measurement information Z in discrete time 1~t 1:t , X 1:t is the position information within time 1~t, P(m xy,t |Z 1:t ,X 1:t ) represents the occupancy probability of the map m at the (x, y) coordinate obtained based on the sensor measurement information and posture information in the time period 1 to t. The operator It means to multiply each (x,y) value.

8. The method for synchronous positioning and mapping of multi-view fusion automotive radars according to claim 5, characterized in that: The obtained posterior probability after solution is updated by binary Bayesian formula to obtain radar synchronization and map information including: Among them, L(m xy,t ) represents the logarithmic operation of the occupation probability of the (x, y) coordinate in the map m at time t, m xy,t represents the occupancy rate of the map m at the (x, y) coordinate at time t, and is a 0-1 distribution random variable. The probability of the grid cell being occupied is expressed as P(m xy,t ), Z 1:t is the radar measurement information Z in discrete time 1~t 1:t , X 1:t is the position information within time 1~t, P(m xy,t |Z 1:t ,X 1:t ) represents the occupancy probability of the map m at the (x, y) coordinate obtained based on the sensor measurement information and posture information in the time period 1 to t.

Images