Radar point cloud and echo simulation method based on spatial features

Abstract

The invention discloses a radar point cloud and echo simulation method based on spatial features. The radar point cloud and echo simulation method comprises the following steps: converting a coordinate system to determine a relative relation; acquiring discrete target edges by detection points; screening the detection points to remove shielding data; according to the distance resolution and the angle resolution, the edge of a target is dispersed into detection points, shielded parts are removed, ideal point cloud data are obtained, and finally echo signals containing the ground multipath effect are generated. Compared with actual measurement data acquired by a radar sensor development board, the problems of fixed scene setting and limited experimental environment can be solved, the spatial characteristics of radar point clouds in a multi-target complex scene are quantitatively represented through detection point acquisition and detection point screening, and the real-time performance of the radar point clouds in the multi-target complex scene is improved. Therefore, high-flexibility and low-cost simulation of continuous wave radar point cloud data and echo signals is realized.

Description

Technical Field

[0001] The present invention belongs to the field of radar signal processing, and particularly relates to a method for simulating radar point cloud and echo based on spatial features, which is applicable to the simulation of frequency modulated continuous wave (FMCW) radar point cloud data and echo signals in scenarios such as vehicle driving, autonomous parking, and UAV taxiing. Background Art

[0002] Frequency Modulated Continuous Wave (FMCW) radar, with its strong penetration ability, all-weather usability, and stable performance, is commonly used for the perception of the surrounding environment in scenarios such as vehicle driving, autonomous parking, and UAV taxiing. It can obtain and analyze the surrounding target information in real time and plays a crucial role in the military, civilian, and scientific research fields. The ground collision perception and warning technology based on FMCW radar is one of the effective ways to improve the deployment efficiency and ensure driving safety. Radar point cloud data and echo signals provide decision-making basis for the autonomous obstacle avoidance systems of driverless vehicles and UAVs.

[0003] Currently, the collision perception and warning technology based on continuous wave radar requires a large amount of point cloud data. The development and performance verification of autonomous obstacle avoidance algorithms mainly use publicly available datasets or actual measurements sampled by radar sensor development boards. However, the publicly available datasets are for fixed scenario settings and cannot match the characteristics of various autonomous obstacle avoidance algorithms, and it is difficult to achieve the best verification effect. Some methods use actual measurements sampled by radar sensor development boards to obtain radar echoes and then obtain point cloud data through signal processing. However, the scenario settings of this method are limited by the experimental environment and the cost is relatively high, which cannot meet the needs of the development and verification of various algorithms.

[0004] The scenario settings of simulated point clouds are more flexible. On the premise that the accuracy meets the requirements, it is more suitable for the development and performance verification of autonomous obstacle avoidance algorithms in various detection scenarios. In multi-target scenarios, due to the complexity of the detection scenarios, it is difficult for the accuracy and robustness of simulated point clouds to meet the actual requirements. Therefore, it is necessary to study effective methods for simulating continuous wave radar point cloud data and echo signals, consider the spatial correlation between adjacent detection points, determine an accurate method for screening line-of-sight detection points, and achieve the high-flexibility and low-cost generation of algorithm verification data. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for simulating radar point cloud and echo based on spatial features to quantitatively characterize the spatial characteristics of radar point clouds in multi-target complex scenarios and achieve the high-flexibility and low-cost generation of point cloud data.

[0006] To achieve the above task, the present invention adopts the following technical solutions:

[0007] A method for simulating radar point cloud and echo based on spatial features, comprising:

[0008] Model the detection targets in the detection scene of the frequency-modulated continuous-wave radar in the reference coordinate system, and model the detection targets as polygons composed of multiple vertices; determine the position parameters and velocity parameters of the radar, detection targets, and vertices in the reference coordinate system;

[0009] Construct a radar rectangular coordinate system and a radar polar coordinate system, and use the position parameters and velocity parameters of the radar, detection targets, and vertices determined in the reference coordinate system to determine the distance and azimuth angle of all vertices of each detection target in the radar polar coordinate system;

[0010] For each detection target, after sorting all its vertices, screen out the boundary vertices according to the azimuth angle of the vertices, and determine the vertex set as the line-of-sight vertices based on the distances between the vertices within and outside the boundary vertices and the radar, and discretize the target edge between adjacent line-of-sight vertices into a series of detection points, and then determine the position parameters, radial velocity, and azimuth angle of each detection point relative to the radar;

[0011] For each adjacent group of detection points on each detection target, regard the area between the detection points as an occlusion interval, and calculate in turn whether each detection point of other targets will be occluded by the occlusion interval and be outside the line-of-sight range, so as to screen out all the detection points within the line-of-sight range. These detection points form the ideal point cloud of the radar detection scene, denoted as real detection points;

[0012] Based on the multipath effect of radar signals in the detection scene, obtain three types of multipath detection points corresponding to each real detection point in the ideal point cloud, and determine the position parameters, radial velocity, and azimuth angle of each type of multipath detection point;

[0013] Determine the radar echo signal matrix corresponding to the real detection points and the three types of multipath detection points, so as to simulate and generate radar echo signals.

[0014] Furthermore, the reference coordinate system takes the ground as the xoy plane, the radar rectangular coordinate takes the position of the radar as the origin o′, and the normal direction of the radar antenna array as the y′ axis; the radar polar coordinate system takes o′ as the pole, the y′ axis as the polar axis, and the clockwise direction as the positive direction of the polar angle; according to the conversion relationship between the reference coordinate system and the radar rectangular coordinate system and the corresponding relationship between the radar rectangular coordinate system and the radar polar coordinate system, obtain the distance and azimuth angle of all vertices of each detection target in the radar polar coordinate system.

[0015] Further, the method for constructing a radar rectangular coordinate system and a radar polar coordinate system, and determining the position parameters and velocity parameters of the radar, the detection target, and the vertex in the reference coordinate system, and determining the distance and azimuth angle of all vertices of each detection target in the radar polar coordinate system includes:

[0016] First, arrange all vertices of each detection target in a clockwise order in the radar polar coordinate system, select two vertices corresponding to the maximum azimuth angle and the minimum azimuth angle as the boundary vertices, and the vertices between the boundary vertices and the vertices outside the boundary vertices respectively form two vertex sets; calculate the distance r between all vertices in each vertex set and the radar ij The average value of, then the vertex set with a smaller distance mean value contains the vertices that are line-of-sight vertices, and the target edge between adjacent line-of-sight vertices is the line-of-sight edge that can be detected by the radar, and it is used as the object for discrete detection points;

[0017] Secondly, according to the preset distance resolution r res And the angle resolution θ res Discretize the target edge between adjacent line-of-sight vertices on each detection target into detection points X k ;

[0018] Finally, according to the azimuth angle θ of the detection point X k And the velocity of the detection target in the radar rectangular coordinate system, determine the component of the velocity of the detection point relative to the radar on the line connecting the two, that is, the radial velocity v k k , define the radial velocity as positive when the detection point approaches the radar, and obtain the radial velocity of the detection point X k On the target edge; thus determine the distance r of all detection points X k Relative to the radar , radial velocity v k , azimuth angle θ k , and the height h from the ground k k k .

[0019] Further, the method for discretizing the target edge between adjacent line-of-sight vertices on each detection target into detection points X res According to the preset distance resolution r res And the angle resolution θ k Includes:

[0020] For any set of adjacent line-of-sight vertices on a detection target, assume that the distance of one vertex is r m , the azimuth angle is θ m , the height is h m , the distance of the other vertex is r n , the azimuth angle is θ n , and the height from the ground is h n, then the distance range Δr of the target edge between the two is Δr = |r m - r n |, and the angle range Δθ = |θ m - θ n |. The detection point X on the target edge k has a distance r relative to the radar k and an azimuth angle θ k The relationship can be expressed as:

[0021]

[0022] The method of discretizing the target edge into detection points can be specifically divided into two categories:

[0023] When |Δr / r res | > |Δθ / θ res |, the number N of detection points X on the target edge k is N 1 = Δr / r res + 1; when |Δr / r res | ≤ |Δθ / θ res |, the number of detection points X i is N 2 = Δθ / θ res + 1; the distance r of the detection point X k is r k = k·r res + min{r m , r n}, and the height h from the ground is h k = k·|h m - h n | / (N 1 - 1) + min{h m , h n}, where k represents the k-th detection point, k = 0, 1,..., (N 1 - 1), and the azimuth angle θ of the detection point is calculated according to the above formula k .

[0024] Furthermore, for all adjacent groups of detection points on each detection target, the area between the detection points is used as an occlusion interval, and it is calculated in turn whether each detection point of other targets will be occluded by the occlusion interval and be out of the line-of-sight range, including:

[0025] Among all the detection points discretized from the target edge between any adjacent line-of-sight vertices on the detection target, A and B are two adjacent detection points, and the area between these two detection points A and B is used as the occlusion interval AB; let the distances of A, B, and P relative to the radar be r a , r b , r p, with azimuth angles being θ a , θ b , θ p , and height values being h a , h b , h p , then the condition for the detection point P to be blocked by the occlusion interval AB is:

[0026] Condition 1, in the angular domain, P is within the angular range of AB, that is:

[0027] (θ a - θ p )(θ b - θ p ) ≤ 0

[0028] Condition 2, in the angle - distance domain, P is "behind" AB. Draw PP 0 ⊥AB, and the foot of the perpendicular is denoted as P 0 . Then this condition is equivalent to the distance value of the vector being greater than or equal to the distance value of the vector , where and are expressed as:

[0029]

[0030] Condition 3, in the distance - height domain, the slope coefficient k p of P is less than or equal to the slope coefficient k ab of the occlusion interval AB, that is k p ≤ k ab ; Since the distance between adjacent detection points is extremely small, the slope coefficient of the occlusion interval is approximately the mean of the slope coefficients of adjacent detection points. Then the expressions for k p and k ab are:

[0031]

[0032] If conditions 1 to 3 are simultaneously satisfied, then P is blocked by AB and is out of the line - of - sight range.

[0033] Furthermore, according to whether the transmitted and scattered radar signals pass through ground reflection, the multipath signal propagation paths are divided into three categories: "radar - ground - target - ground - radar", "radar - target - ground - radar", "radar - ground - target - radar"; Therefore, each real detection point T in the ideal point cloud corresponds to three types of multipath detection points, namely the first - type multipath detection point T 1 , the second - type multipath detection point T 2 and the third - type multipath detection point T 3 ;

[0034] Record the distance r of the real detection point T relative to the radar 0 , the radial velocity v 0 , the azimuth angle θ 0 and the height h from the ground 0 . Establish a plane rectangular coordinate system XOY in the plane perpendicular to the ground. Denote the projection point R of the radar R on the ground 0 as the origin O of the plane rectangular coordinate system, and denote the projection point of the real detection point T on the ground as T 0 , with the projection height being h R . Then the X-axis points from R 0 to T 0 , and the Y-axis points from R 0 to R; Therefore, the position coordinates of the radar R are (0, h R ), the position coordinates of the real detection point T are (r 0 , h 0 ), and the position coordinates of the reflection point G of the radar signal on the ground are (r 0 h R / (h 0 + h R ), 0); Within a short time Δt of the detection target's movement, it can be approximately considered that the movement state of the detection target remains unchanged, that is, the real detection point moves in a uniform straight line, and the coordinates of the position T' reached are (r 0 - v 0 Δt, h 0 ). Denote the positions reached by T 1 , T 2 and T 3 as T' 1 , T' 2 and T' 3 ;

[0035] For the first type of multipath detection point T 1 , since T 1 is symmetric to T, T' 1 is symmetric to T' about the X-axis, the distance r 1 of T 1 relative to the radar is |RT 1 | = |RG| + |TG|, the radial velocity v 1 of T 1 is equal to the radial velocity v 0 of the real detection point T, and the azimuth angle θ 1 of T 1 is determined according to the distance r 1 ;

[0036] For the second type of multipath detection point T 2 , T 2 is on the extension line of RG and |RT 2| = 1 / 2(|RG| + |TG| + |RT|), according to the position parameter of T 2 the position parameter of T 2 ' and the azimuth angle θ 2 then the distance r of T 2 is |RT 2 |, and the radial velocity v 2 is v 2 = v 2 = v 0 ·|T 2 T' 2 | / |TT'|;

[0037] For the third type of multipath detection point, T 3 is on the extension line of RT and |RT 3 | = 1 / 2(|RG| + |TG| + |RT|), according to the position parameter of T 3 the position parameter of T 3 ' and the azimuth angle θ 3 then the distance r of T 3 is |RT 3 |, and the radial velocity v 3 is v 3 = v 3 = v 0 ·|T 3 T' 3 | / |TT'|.

[0038] Furthermore, denote the detection point Q as a real detection point or a certain multipath detection point, then the distance of the detection point Q relative to the radar is r, the radial velocity is v, and the azimuth angle is θ; for the m q th array element receiving, the n q th echo pulse, and the i q th signal sampling point, the expression of the signal time delay τ is:

[0039]

[0040] where the speed of light is c, the array element spacing is d, the signal time width is T r and the signal sampling frequency is f s , the fast time t = (i q -1) / f s , then the expression of any element S q (m q , n q , i q ) in the echo signal matrix is:

[0041]

[0042] In the above formula, the imaginary unit is j, the signal bandwidth is B, and the signal carrier frequency is f c ;

[0043] After obtaining the echo signal matrix S corresponding to all real detection points and the corresponding three types of multipath detection points q Then, the echo signal matrix S of all real detection points and the corresponding three types of multipath detection points within the radar beam illumination range q is accumulated to obtain the frequency-modulated continuous-wave radar echo signal S of the detection scenario.

[0044] A terminal device includes a processor, a memory, and a computer program stored in the memory; when the processor executes the computer program, the radar point cloud and echo simulation method based on spatial features is implemented.

[0045] A computer-readable storage medium stores a computer program; when the computer program is executed by a processor, the radar point cloud and echo simulation method based on spatial features is implemented.

[0046] Compared with the prior art, the present invention has the following technical characteristics:

[0047] Compared with the actual data collected by the radar sensor development board, the present invention can overcome the problems of fixed scenario setting and limited experimental environment, and quantify and characterize the spatial characteristics of the radar point cloud in a multi-target complex scenario through detection point acquisition and detection point screening, so as to realize the highly flexible and low-cost simulation of continuous-wave radar point cloud data and echo signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a schematic flow chart of the radar point cloud data and echo signal simulation method in an embodiment of the present invention;

[0049] Figure 2 is a schematic diagram of the detection scenario and the radar beam direction;

[0050] Figure 3 is a schematic diagram of the coordinate system conversion result;

[0051] Figure 4 is a schematic diagram of the detection point acquisition result;

[0052] Figure 5 is a schematic diagram of the ideal point cloud simulation result;

[0053] Figure 6 is a constant false alarm rate detection result diagram of the echo signal range-Doppler feature map. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] The present invention provides a radar point cloud and echo simulation method based on spatial features to generate an ideal point cloud and a frequency modulated continuous wave radar echo in a multi-target detection scene, which mainly includes the following steps:

[0055] Step 1: Modeling the detection target in the FMCW radar detection scene in a reference coordinate system, and modeling the detection target as a polygon composed of multiple vertices; determining the position parameters and speed parameters of the radar, the detection target, and the vertices in the reference coordinate system.

[0056] For the detection scene with a height h above the ground R Frequency modulated continuous wave radar, with N T To generate the radar point cloud and echo data of the detection scene, a reference coordinate system is established with the ground as the xoy plane. The position coordinates of the x and y axes of the radar in the reference coordinate system are marked as (x 0 ,y 0 ), speed is recorded as (v x0 ,v y0 ), the angle of the normal of the radar antenna array relative to the y-axis clockwise rotation is denoted as α; let the speed of the i-th detected target be (v xi ,v yi ), where i = 1, 2, ..., N T , each type of detection target is modeled as a different polygon composed of multiple vertices, and the position coordinates of the jth vertex of the i-th detection target are (x ij ,y ij ), the height of the vertex from the ground is h ij , where j = 1, 2, ..., N Pi , N Pi Indicates the number of vertices contained in the polygon corresponding to the i-th detection target.

[0057] Step 2: Construct a radar rectangular coordinate system and a radar polar coordinate system, and use the position parameters and speed parameters of the radar, the detection target, and the vertex in the reference coordinate system to determine the distance and azimuth of all the vertices of each detection target in the radar polar coordinate system.

[0058] With the radar location as the origin o′ and the normal of the radar antenna array as the y′ axis, a radar rectangular coordinate system x′o′y′ is established. The clockwise rotation angle of the radar rectangular coordinate system relative to the reference coordinate system is α; the position coordinates (x ij ,y ij ) is converted into the position coordinate (x′) in the radar rectangular coordinate system ij ,y′ ij ), the velocity of all detected targets in the reference coordinate system (v xi ,vyi ) Convert to the velocity (v′ xi , v′ yi ) in the radar rectangular coordinate system, that is

[0059]

[0060] Establish a radar polar coordinate system with o′ as the pole, the y′ axis as the polar axis, and the clockwise direction as the positive direction of the polar angle. Then, for all vertices of each detected target, the distance r ij and the azimuth angle θ ij in the radar polar coordinate system are expressed as follows:

[0061]

[0062] Step 3: For each detected target, after sorting all its vertices, screen out the boundary vertices according to the azimuth angle of the vertices, and determine the set of vertices as the line-of-sight vertices based on the distances between the vertices within and outside the boundary vertices and the radar. Discretize the target edge between adjacent line-of-sight vertices into a series of detection points, and then determine the position parameters (including the distance from the radar and the height from the ground), radial velocity, and azimuth angle of each detection point relative to the radar.

[0063] First, arrange all the vertices of each detected target in clockwise order in the radar polar coordinate system. Select the two vertices corresponding to the maximum azimuth angle and the minimum azimuth angle as the boundary vertices. The vertices between the boundary vertices and the vertices outside the boundary vertices form two sets of vertices respectively; calculate the average value of the distances r ij from all the vertices in each set of vertices to the radar. Then, the set of vertices with a smaller average distance contains the line-of-sight vertices. The target edge between adjacent line-of-sight vertices is the line-of-sight edge that can be detected by the radar, and it is used as the object to be discretized into detection points.

[0064] Secondly, according to the preset distance resolution r res and angle resolution θ res , discretize the target edge between adjacent line-of-sight vertices on each detected target into detection points X k . Specifically:

[0065] For any set of adjacent line-of-sight vertices on a detected target, assume the distance of one vertex is r m , the azimuth angle is θ m , and the height is h m . The distance of the other vertex is r n , the azimuth angle is θ n , and the height from the ground is h n . Then, the distance range Δr of the target edge between the two is Δr = |r m - r n|, the angular range Δθ = |θ m -θ n |, the detection point X on the target edge k The distance r relative to the radar k And the azimuth angle θ k The relationship can be expressed as:

[0066]

[0067] The method of discretizing the target edge into detection points can be specifically divided into two categories:

[0068] When |Δr / r res | > |Δθ / θ res |, the number N of detection points X on the target edge k = Δr / r 1 + 1; when |Δr / r res | ≤ |Δθ / θ res |, the number of detection points X res is N i = Δθ / θ 2 + 1; the distance r of the detection point X res = k·r k + min{r k , r res}, the height from the ground is h m , r n} = k·|h k - h m | / (N n - 1)+ min{h 1 , h m , h n}, where k represents the k-th detection point, k = 0, 1, …, (N 1 - 1), and calculate the azimuth angle θ of the detection point according to formula (3) k .

[0069] Finally, according to the azimuth angle θ of the detection point X k and the velocity (v′ k , v′ xi , v′ yi ) of the detected target in the radar rectangular coordinate system, determine the component of the velocity of the detection point relative to the radar on the line connecting the two, that is, the radial velocity v k . Define the radial velocity as positive when the detection point approaches the radar, then the radial velocity v of the detection point X on the target edge k = -(v′ k sinθ xi + v′ k cosθ yi k ​); So far, the distance rk and radial velocity v of all detection points Xk relative to the radar k 、Azimuth angle θ k and height from the ground h k All have been obtained.

[0070] Step 4: For all adjacent detection points on each detection target, the area between the detection points is taken as the occlusion interval, and each detection point of other targets is calculated in turn to see whether it will be blocked by the occlusion interval and be outside the line of sight range, thereby screening out all detection points within the line of sight range. These detection points constitute the ideal point cloud of the radar detection scene and are recorded as real detection points.

[0071] Note that among all the detection points on the target edge between any group of adjacent sight-range vertices on the detection target, A and B are two adjacent detection points, and the area between these two detection points A and B is taken as the occlusion interval AB; for all detection points P of other detection targets, determine whether they will be blocked by the occlusion interval AB and be outside the sight-range.

[0072] Assume that the distances of A, B and P relative to the radar are r a 、r b 、r p , the azimuth angles are θ a ,θ b ,θ p , the height values ​​are h a 、h b 、h p , then the condition for the detection point P to be blocked by the blocking interval AB is:

[0073] Condition 1: In the angle domain, P is within the angle range of AB, that is:

[0074] (θ a -θ p )(θ b -θ p )≤0 (4)

[0075] Condition 2: In the angle-distance domain, P is located "behind" AB, and PP is drawn through point P. 0 ⊥AB, the foot of the perpendicular is denoted by P 0 , then this condition is equivalent to the vector The distance value is greater than or equal to the vector The distance value of and The expression is:

[0076]

[0077] Condition 3: In the distance-height domain, the slope coefficient k of P is pLess than or equal to the slope coefficient k of the occlusion interval AB ab , that is, k p ≤k ab ; Since the distance between adjacent detection points is extremely small, the slope coefficient of the occlusion interval is approximately the mean of the slope coefficients of adjacent detection points. Then the expressions of k p and k ab are as follows:

[0078]

[0079] If conditions 1 to 3 are simultaneously satisfied, then P is occluded by AB and is outside the line-of-sight range.

[0080] Successively use all occlusion intervals of each detection target to perform the above line-of-sight range determination process on all detection points of other detection targets, and screen out all detection points within the line-of-sight range. These detection points constitute the ideal point cloud of the radar detection scene, denoted as real detection points; each real detection point among them includes the distance, radial velocity, azimuth angle, and height from the ground relative to the radar.

[0081] Step 5: Based on the multipath effect of radar signals in the detection scene, obtain three types of multipath detection points corresponding to each real detection point in the ideal point cloud, and determine the position parameters, radial velocity, and azimuth angle of each type of multipath detection point.

[0082] In the detection scene, the frequency-modulated continuous-wave radar emits frequency-modulated continuous waves, and the directly scattered echo signals received by the detection target correspond to the real detection points in the ideal point cloud; however, due to the certain reflection characteristics of the ground in the detection scene, the radar receiver also receives multipath echo signals scattered by the target and reflected by the ground, and equivalently models them as radar echo signals of multipath detection points; According to whether the transmitted and scattered radar signals pass through ground reflection, the multipath signal propagation paths are divided into three categories: "radar-ground-target-ground-radar", "radar-target-ground-radar", "radar-ground-target-radar"; Therefore, each real detection point T in the ideal point cloud obtained in step 4 corresponds to three types of multipath detection points, that is, the first type of multipath detection point T 1 , the second type of multipath detection point T 2 and the third type of multipath detection point T 3 .

[0083] Denote the distance r 0 , radial velocity v 0 , azimuth angle θ 0 and height h 0 from the real detection point T to the radar, and establish a plane rectangular coordinate system XOY in the plane perpendicular to the ground. Project the radar R onto the ground at the point R 0Taking the origin O of the rectangular coordinate system as the reference, the projection point of the actual detection point T on the ground is denoted as T 0 , and the projection height is h R . Then the X-axis points from R 0 to T 0 , and the Y-axis points from R 0 to R. Therefore, the position coordinates of the radar R are (0, h R ), the position coordinates of the actual detection point T are (r 0 , h 0 ), and the position coordinates of the reflection point G of the radar signal on the ground are (r 0 h R / (h 0 + h R ), 0); within a short time Δt of the detection target's movement, it can be approximately considered that the movement state of the detection target remains unchanged, that is, the actual detection point moves in a uniform straight line, and the coordinates of the position T' reached are (r 0 - v 0 Δt, h 0 ). Denote the positions reached by T 1 , T 2 and T 3 as T' 1 , T' 2 and T' 3 .

[0084] For the first type of multipath detection point T 1 , since T 1 is symmetric to T and T', T' 1 is symmetric to T' about the X-axis, the distance r 1 of T 1 relative to the radar is |RT 1 | = |RG| + |TG|, and the radial velocity v 1 of T 1 is equal to the radial velocity v 0 of the actual detection point T, and the azimuth angle θ 1 of T 1 is determined according to the distance r 1 .

[0085] For the second type of multipath detection point T 2 , T 2 is on the extension line of RG and RT 2 | = 1 / 2(|RG| + |TG| + |RT|), and according to the position parameters of T 2 , the position parameters and azimuth angle θ 2 of T 2 ' can be calculated. Then the distance r 2 of T 2 is |RT 2 |, and the radial velocity v2 For v 2 = v 0 ·|T 2 T′ 2 | / |TT′|.

[0086] For the third type of multipath detection points, T 3 is on the extension line of RT and |RT 3 | = 1 / 2(|RG| + |TG| + |RT|). According to the position parameter of T 3 , the position parameter of T 3 ′ and the azimuth angle θ 3 can be calculated. Then the distance r 3 of T 3 is |RT 3 |, and the radial velocity v 3 is v 3 = v 0 ·|T 3 T′ 3 | / |TT′|.

[0087] Step 6: Determine the radar echo signal matrices corresponding to the true detection points and the three types of multipath detection points, so as to simulate and generate radar echo signals.

[0088] During the radar detection process, each detection point Q will generate an echo signal matrix S q . The three dimensions of the echo signal matrix S q are defined as the receiving array element dimension, slow time dimension, and fast time dimension in sequence; assume the simulated time length is T a , the number of array elements of the radar receiving antenna is N e , the array element spacing is d, the speed of light is c, the signal carrier frequency is f c , the signal time width is T r , the signal bandwidth is B, the signal sampling frequency is f s . Then the number of slow time samples of the received signal is N a = T a / T r , and the number of in-pulse signal sampling points N p = T r f s .

[0089] According to the true detection points obtained in Step 4 and the three types of multipath detection points obtained in Step 5, determine the radar echo signals;

[0090] Denote the detection point Q as a true detection point or a certain multipath detection point. Then the distance of the detection point Q relative to the radar is r, the radial velocity is v, and the azimuth angle is θ; for the m q th array element reception, the n q th echo pulse, the i qA signal sampling point, the expression of the signal time delay τ is:

[0091]

[0092] where the fast time t = (i q - 1) / f s , then for any element S q (m q , n q , i q ) in the echo signal matrix, the expression is:

[0093]

[0094] In the above formula, j is the imaginary unit.

[0095] According to formulas (7) and (8), that is, after obtaining the echo signal matrix S corresponding to all real detection points and the corresponding three types of multipath detection points q ; accumulate the echo signal matrices S of all real detection points and the corresponding three types of multipath detection points within the radar beam irradiation range to obtain the frequency-modulated continuous-wave radar echo signal S of this detection scenario, that is, S = ∑S q q .

[0096] Embodiment:

[0097] In an embodiment of the present invention, in the detection scenario as Figure 2 shown, it is assumed that there is a frequency-modulated continuous-wave radar with a height h R of 5 m from the ground in the detection scenario, and N T = 3 detection targets are distributed around it. In order to generate the radar point cloud and echo data of this detection scenario, the ground is used as the xoy plane to establish a reference coordinate system. The position (x 0 , y 0 ) of the radar in the reference coordinate system is (100 m, 200 m), the speed (v x0 , v y0 ) is (0 m / s, 0 m / s), and the angle α by which the normal of the radar antenna array rotates clockwise relative to the y-axis is 0.1 rad. Let the speed of the i-th target be (v xi , v yi ), where i = 1, 2,..., N T , and the speeds of the targets are (0 m / s, 0 m / s), (-5 m / s, -5 m / s) and (3 m / s, 6 m / s). The different types of targets are modeled as different polygons determined by multiple vertices. The position of the j-th vertex of the i-th target is (x ij , y ij ), where j = 1, 2,..., N PiThe central position coordinates of the three targets in the reference coordinate system are (115m, 230m), (105m, 225m), and (85m, 210m). The third target is not within the radar beam coverage, and all target polygons are rectangles with 4 vertices. The height of all vertices from the ground is 6m.

[0098] The distance resolution r set in this embodiment res = 0.15m, and the angular resolution θ res = 1.5°. The detection points distributed along the target edge obtained are as Figure 4 shown, and the ideal point cloud simulation result is as Figure 5 shown; finally, the echo components corresponding to all real detection points and multipath detection points within the radar beam illumination range are accumulated to obtain the frequency-modulated continuous-wave radar echo signal of this detection scenario. Constant false alarm rate detection is performed on the range-Doppler feature map of the echo signal, as Figure 6 shown, and the range and velocity characteristics of the two targets within the radar beam range are consistent with the scenario settings.

[0099] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A radar point cloud and echo simulation method based on spatial features, characterized in that: include: Modeling the detection target in the FMCW radar detection scene in a reference coordinate system, and modeling the detection target as a polygon composed of multiple vertices; Determine the position parameters and velocity parameters of the radar, the detected target and the vertex in the reference coordinate system; Construct a radar rectangular coordinate system and a radar polar coordinate system, use the position parameters and speed parameters of the radar, the detection target and the vertex in the reference coordinate system to determine the distance and azimuth of all the vertices of each detection target in the radar polar coordinate system; For each detection target, after sorting all its vertices, the boundary vertices are screened out according to the azimuth angles of the vertices, and the vertex set as the line-of-sight vertex is determined based on the distance between the vertex sets formed by the vertices between and outside the boundary vertices and the radar, and the target edges between adjacent line-of-sight vertices are discretized into a series of detection points, and then the position parameters, radial velocity and azimuth angle of each detection point relative to the radar are determined; For all adjacent detection points on each detection target, the area between the detection points is taken as the occlusion interval, and each detection point of other targets is calculated in turn to see whether it will be blocked by the occlusion interval and be out of the line of sight range, so as to screen out all the detection points within the line of sight range. These detection points constitute the ideal point cloud of the radar detection scene and are recorded as real detection points. Based on the multipath effect of the radar signal in the detection scene, three types of multipath detection points corresponding to each real detection point in the ideal point cloud are obtained, and the position parameters, radial velocity and azimuth of each type of multipath detection point are determined; The radar echo signal matrix corresponding to the real detection point and three types of multipath detection points is determined, so as to simulate and generate the radar echo signal.

2. The radar point cloud and echo simulation method based on spatial features according to claim 1 is characterized in that: The reference coordinate system uses the ground as the xoy plane, the radar rectangular coordinate uses the radar location as the origin o′, and the normal of the radar antenna array is the y′ axis; the radar polar coordinate system uses o′ as the pole, the y′ axis as the polar axis, and the clockwise direction as the positive polar angle direction; according to the conversion relationship between the reference coordinate system and the radar rectangular coordinate system, and the corresponding relationship between the radar rectangular coordinate system and the radar polar coordinate system, the distance and azimuth of all vertices of each detection target in the radar polar coordinate system are obtained.

3. The radar point cloud and echo simulation method based on spatial features according to claim 1 is characterized in that: The construction of the radar rectangular coordinate system and the radar polar coordinate system, using the position parameters and speed parameters of the radar, the detection target and the vertex in the reference coordinate system to determine the distance and azimuth of all the vertices of each detection target in the radar polar coordinate system, includes: First, all vertices of each detected target are arranged in clockwise order in the radar polar coordinate system, and the two vertices corresponding to the maximum azimuth and the minimum azimuth are selected as the boundary vertices. The vertices between the boundary vertices and the vertices outside the boundary vertices constitute two vertex sets respectively; the distance r between all vertices in each vertex set and the radar is calculated. ij The average value of , then the vertices contained in the vertex set with smaller distance mean are the line-of-sight vertices, and the target edge between adjacent line-of-sight vertices is the line-of-sight edge detectable by the radar, which is used as the object of discrete detection point; Secondly, according to the preset distance resolution r res and angular resolution θ res Discretize the target edge between adjacent sight-range vertices on each detection target into detection points X k ; Finally, according to the detection point X k The azimuth angle θ k As well as the velocity of the detected target in the radar rectangular coordinate system, determine the component of the velocity of the detection point relative to the radar on the line connecting the two, that is, the radial velocity v k , define the radial velocity as positive when the detection point is close to the radar, and get the detection point X on the edge of the target k The radial velocity of all detection points X is determined k Distance r relative to the radar k , radial velocity v k 、Azimuth angle θ k and height from the ground h k .

4. The radar point cloud and echo simulation method based on spatial features according to claim 3 is characterized in that: The preset distance resolution r res and angular resolution θ res Discretize the target edge between adjacent sight-range vertices on each detection target into detection points X k ,include: For any set of adjacent line-of-sight vertices on a detection target, let the distance of one of the vertices be r m , azimuth angle is θ m , height h m , the distance to the other vertex is r n , azimuth angle is θ n , height from the ground is h n , then the distance range of the target edge between the two is Δr=|r m -r n |, angle range Δθ=|θ m -θ n |, the detection point X on the edge of the target k Distance r relative to the radar k With azimuth θ k The relationship can be expressed as: The methods of discretizing target edges into detection points can be divided into two categories: When |Δr / r res |>|Δθ / θ res |, the detection point X on the edge of the target k The number of N1 = Δr / r res +1; when |Δr / r res |≤|Δθ / θ res |, the detection point X i The number of is N2 = Δθ / θ res +1; detection point X k The distance r k = k·r res +min{r m ,r n }, the height from the ground is h k =k·|h m -h n | / (N1-1)+min{h m ,h n }, where k represents the kth detection point, k = 0, 1, ..., (N1-1), and the azimuth angle θ of the detection point is calculated according to the above formula k .

5. The radar point cloud and echo simulation method based on spatial features according to claim 1, characterized in that: For all adjacent detection points on each detection target, the area between the detection points is used as the occlusion interval, and each detection point of other targets is calculated in turn to determine whether it will be blocked by the occlusion interval and be outside the visual range, including: Note that among all the detection points on the target edge between any set of adjacent sight vertices on the detection target, A and B are two adjacent detection points, and the area between these two detection points A and B is taken as the occlusion interval AB; let the distances of A, B and P relative to the radar be r a 、r b 、r p , the azimuth angles are θ a ,θ b ,θ p , the height values ​​are h a 、h b 、h p , then the condition for the detection point P to be blocked by the blocking interval AB is: Condition 1: In the angle domain, P is within the angle range of AB, that is: (i a -θ p )(θ b -θ p )≤0 Condition 2: In the angle-distance domain, P is located "behind" AB. Draw PP0⊥AB through point P, and record the foot of the perpendicular as P0. Then this condition is equivalent to the vector The distance value is greater than or equal to the vector The distance value of and The expression is: Condition 3: In the distance-height domain, the slope coefficient k of P is p The slope coefficient k is less than or equal to the occlusion interval AB ab , that is, k p ≤k ab ; Since the distance between adjacent detection points is extremely small, the slope coefficient of the occlusion interval is approximately the mean of the slope coefficients of adjacent detection points, so k p and k ab The expression is: If conditions one to three are met at the same time, P is blocked by AB and is out of sight.

6. The radar point cloud and echo simulation method based on spatial features according to claim 1, characterized in that: According to whether the transmitted and scattered radar signals are reflected by the ground, the multipath signal propagation paths are divided into three categories: "radar-ground-target-ground-radar", "radar-target-ground-radar", and "radar-ground-target-radar". Therefore, each real detection point T in the ideal point cloud corresponds to three types of multipath detection points, namely, the first type of multipath detection point T1, the second type of multipath detection point T2, and the third type of multipath detection point T3. Record the distance r0, radial velocity v0, azimuth θ0 and height h0 of the real detection point T relative to the radar, establish a plane rectangular coordinate system XOY in a plane perpendicular to the ground, take the projection point R0 of the radar R on the ground as the origin O of the plane rectangular coordinate system, record the projection point of the real detection point T on the ground as T0, and the projection height as h R , then the X-axis points from R0 to T0, and the Y-axis points from R0 to R; therefore, the position coordinates of radar R are (0, h R ), the position coordinates of the actual detection point T are (r0,h0), and the position coordinates of the reflection point G of the radar signal on the ground are (r0h R / (h0+h R ),0); within the short time Δt of the detection target movement, it can be approximately considered that the detection target movement state remains unchanged, that is, the real detection point moves in a uniform straight line, and the coordinates of the position T′ are (r0-v0Δt,h0), and the positions reached by T1, T2 and T3 are recorded as T1′, T2′ and T3′; For the first type of multipath detection point T1, since T1 and T, T1′ and T′ are symmetric about the X-axis, the distance r1 of T1 relative to the radar is |RT1|=|RG|+|TG|, the radial velocity v1 of T1 is equal to the radial velocity v0 of the real detection point T, and the azimuth θ1 of T1 is determined according to the distance r1; For the second type of multipath detection point T2, T2 is on the extension line of RG and |RT2|=1 / 2(|RG|+|TG|+|RT|). According to the position parameters of T2, the position parameters of T2′ and the azimuth angle θ2 can be calculated. Then the distance r2 of T2 is |RT2|, and the radial velocity v2 is v2=v0·|T2T2′| / |TT′|. For the third type of multipath detection point, T3 is on the extension line of RT and |RT3|=1 / 2(|RG|+|TG|+|RT|). The position parameters of T3′ and the azimuth angle θ3 can be calculated based on the position parameters of T3. Then the distance r3 of T3 is |RT3|, and the radial velocity v3 is v3=v0·|T3T3′| / |TT′|.

7. The radar point cloud and echo simulation method based on spatial features according to claim 1, characterized in that: The detection point Q represents a real detection point or a multipath detection point. The distance of the detection point Q relative to the radar is r, the radial velocity is v, and the azimuth is θ. q The array element receives the nth q The i-th echo pulse q signal sampling points, the expression of signal delay τ is: Among them, the speed of light is c, the array element spacing is d, and the signal time width is T r , the signal sampling frequency is f s , fast time t=(i q -1) / f s , then any element S in the echo signal matrix q (m q ,n q ,i q ) is: In the above formula, the imaginary unit is j, the signal bandwidth is B, and the signal carrier frequency is f c ; After obtaining the echo signal matrix S corresponding to all real detection points and the corresponding three types of multipath detection points q Then, the echo signal matrix S of all real detection points and the corresponding three types of multipath detection points within the radar beam illumination range is calculated. q Accumulate and obtain the frequency modulated continuous wave radar echo signal S of the detection scene.

8. A terminal device comprising a processor, a memory and a computer program stored in the memory; characterized in that: When the processor executes the computer program, the radar point cloud and echo simulation method based on spatial features according to any one of claims 1 to 7 is implemented.

9. A computer-readable storage medium, wherein a computer program is stored in the medium; characterized in that: When the computer program is executed by a processor, the radar point cloud and echo simulation method based on spatial features according to any one of claims 1 to 7 is implemented.

Images