Roadside laser radar beam distribution optimization method and device based on vehicle shielding

Abstract

The invention provides a roadside laser radar beam distribution optimization method and device based on vehicle shielding, and relates to the technical field of laser radar optimizing.The method comprises the steps that the expected shielding effect of a target vehicle is constructed based on the shielding relation between a roadside laser radar and a dynamic vehicle; quantifying the perception capability of the laser radar, and calculating a point cloud number expected value which can be identified by the laser radar based on an expected shielding effect and energy attenuation of the laser radar; an objective function is constructed based on the point cloud number expected value, optimization parameters are set, and the optimization parameters are vertical angle distribution functions of the laser radar; and optimizing the optimization parameters through an objective function and an interval search algorithm to obtain an optimal vertical angle distribution function, and determining laser radar beam distribution through the optimal vertical angle distribution function. According to the invention, the problem of unreasonable light beam distribution in the existing roadside laser radar is solved.

Description

Technical Field

[0001] The present invention relates to the field of laser radar optimization technology, and in particular to a method and device for optimizing roadside laser radar beam distribution based on vehicle occlusion. Background Art

[0002] With the rapid development of autonomous driving technology, perception systems play a vital role in ensuring vehicle safety and reliability. Roadside LiDAR, as a supplement to on-board sensors, can provide a wider field of view, reduce blind spots, and enable multi-vehicle collaborative perception in complex traffic scenarios such as intersections and highways.

[0003] However, the perception performance of roadside lidar is affected by many factors, especially traffic density and the configuration of the lidar itself. In high-density traffic flow, the dynamic occlusion effect between vehicles will significantly reduce the detection performance of the lidar. In addition, the beam distribution of the lidar has an important impact on its perception performance. If the beam distribution is unreasonable, it will lead to sparse coverage of the target point cloud, further limiting its perception ability. Existing research on lidar beam distribution optimization mainly focuses on static occlusion models, which cannot accurately reflect the occlusion effects in dynamic traffic flows. In addition, existing optimization methods usually require a lot of computing resources, and it is difficult to complete the optimal configuration of lidars with a high number of beams in a short time. These problems limit the application effect of lidar in actual traffic environments and make it difficult to meet the needs of autonomous driving and intelligent transportation systems. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and device for optimizing roadside LiDAR beam distribution based on vehicle occlusion to improve the above-mentioned problem. To achieve the above-mentioned purpose, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present application provides a method for optimizing roadside lidar beam distribution based on vehicle occlusion, comprising:

[0006] Based on the occlusion relationship between the road test LiDAR and the dynamic vehicle, the expected occlusion effect of the target vehicle is constructed;

[0007] Quantify the perception capability of the lidar and calculate the expected number of point clouds that the lidar can recognize based on the expected occlusion effect and the energy attenuation of the lidar;

[0008] Constructing an objective function based on the expected value of the point cloud number and setting an optimization parameter, wherein the optimization parameter is the vertical angle distribution function of the laser radar;

[0009] The optimization parameters are optimized through the objective function and interval search algorithm to obtain the optimal vertical angle distribution function, and the lidar beam distribution is determined by the optimal vertical angle distribution function.

[0010] In a second aspect, the present application also provides a roadside lidar beam distribution optimization device based on vehicle occlusion, comprising:

[0011] The first construction module is used to construct the expected occlusion effect of the target vehicle based on the occlusion relationship between the road test laser radar and the dynamic vehicle;

[0012] The calculation module is used to quantify the perception capability of the lidar and calculate the expected number of point clouds that can be recognized by the lidar based on the expected occlusion effect and the energy attenuation of the lidar;

[0013] The second construction module is used to construct an objective function based on the expected value of the point cloud number and set an optimization parameter, wherein the optimization parameter is the vertical angle distribution function of the laser radar;

[0014] The optimization module is used to optimize the optimization parameters through the objective function and interval search algorithm to obtain the optimal vertical angle distribution function, and determine the laser radar beam distribution through the optimal vertical angle distribution function.

[0015] The beneficial effects of the present invention are as follows: the present invention uses a geometric method to calculate the occlusion area between the target vehicle and the LiDAR, and converts the three-dimensional occlusion problem into two dimensions by constructing a projection plane and calculating the normal vector, thereby achieving accurate quantification of the occlusion effect and calculable area of ​​the occlusion region. The number of point clouds obtained from the target vehicle is used to evaluate the LiDAR perception capability, and the introduction of distance energy attenuation and occlusion effects makes the evaluation more comprehensive. An interval search algorithm is designed to optimize the angular distribution of the laser beam as a whole, avoiding the complex operation of optimizing vertical angles one by one, significantly improving the optimization efficiency, and finding the optimal solution within a relatively small number of iterations. The optimal vertical angle distribution function is obtained to determine the laser radar beam distribution.

[0016] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the embodiments of the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 Schematic diagram of the process of optimizing the roadside lidar beam distribution based on vehicle occlusion according to an embodiment of the present invention;

[0019] Figure 2 Schematic diagram of vehicle occlusion relationship in traffic flow according to an embodiment of the present invention;

[0020] Figure 3 Schematic diagram of vehicle occlusion relationship based on projection in an embodiment of the present invention;

[0021] Figure 4 Schematic diagram showing the position of an obstructing vehicle in the blocked area according to an embodiment of the present invention;

[0022] Figure 5 Schematic diagram of the beam distribution of the laser radar in an embodiment of the present invention;

[0023] Figure 6 Schematic diagram of the field of view of a laser radar on a vehicle in an embodiment of the present invention;

[0024] Figure 7 Schematic diagram of the sensing capability of the laser radar in an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0026] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are used only to distinguish the description and should not be understood as indicating or implying relative importance.

[0027] Example 1:

[0028] This embodiment provides a method for optimizing roadside lidar beam distribution based on vehicle occlusion.

[0029] See also Figure 1 , the figure shows that the method includes step S1, step S2, step S3 and step S4.

[0030] Step S1: Based on the occlusion relationship between the road test LiDAR and the dynamic vehicle, the expected occlusion effect of the target vehicle is constructed;

[0031] In this embodiment, the vehicle occlusion relationship in the dynamic traffic flow is first modeled, such as Figure 2 As shown in the top view, with LiDAR as the origin, there is a triangular occlusion area between the target vehicle and LiDAR, namely the occlusion area. In the figure, Ω q represents the occlusion area between the qth target vehicle and LiDAR, q = 1, 2, ... n, n represents the total number of target vehicles, LiDAR represents laser radar, X, Y and Z represent the established x-axis, y-axis and z-axis respectively, l veh and w veh Represent the length and width of the target vehicle, x t and y t Represent the horizontal and vertical coordinates of the center point of the target vehicle respectively.

[0032] In this embodiment, the position of the corresponding vehicle is represented by the coordinates of the vehicle center point.

[0033] In step S1, constructing the desired occlusion effect of the target vehicle includes:

[0034] Step S11: determining the coordinates of the diagonal vertices of the bounding box of the target vehicle according to the position and size of the target vehicle;

[0035] In this embodiment, Figure 2 As shown, located at (x t ,y t ) The coordinates of the diagonal vertices of the bounding box of the target vehicle are and The specific calculation formula is:

[0036]

[0037] Where x t and y t Indicates the horizontal and vertical coordinates of the center point of the target vehicle, l veh and w veh Represent the length and width of the target vehicle respectively.

[0038] Step S12: Calculate the occluded area of ​​the target vehicle based on the coordinates of the diagonal vertices of the bounding box and the position of the laser radar;

[0039] In this embodiment, the calculation formula for the area of ​​the target vehicle's blocked area is:

[0040]

[0041] Where τ represents the occluded area of ​​the target vehicle, and Represents the horizontal coordinates of the two diagonal vertices of the bounding box of the target vehicle, and They represent the ordinates of the two diagonal vertices of the bounding box of the target vehicle, and |·| means taking the absolute value.

[0042] Step S13: Calculate the probability of the obstructing vehicle appearing in the blocked area based on the Poisson point process;

[0043] In this embodiment, when other vehicles are in the obstruction area, they may obstruct the target vehicle. Since the arrival of vehicles in the traffic flow follows a Poisson distribution, a Poisson point process is used to model the probability of an obstructing vehicle appearing in the obstruction area. The formula is:

[0044]

[0045] Where, represents the probability that the obstacle vehicle appears in the occlusion area, λ represents the traffic density, and τ represents the area of ​​the occlusion area of ​​the target vehicle.

[0046] Step S14: constructing a projection plane, and calculating the occlusion effect of the target vehicle based on the projection plane and the obstacle vehicle;

[0047] In this embodiment, Figure 3 As shown in Figure 3, the occlusion effect is modeled as the overlapping area between the obstacle vehicle and the target vehicle on the projection plane (the diagonal section of the target vehicle's bounding box).

[0048] First, construct a general expression for the projection plane:

[0049]

[0050] Where, and Both represent plane normal vectors The weight, Represents the constant that determines the position of the plane, and x, y, and z represent the horizontal, vertical, and vertical coordinates of the point on the projection plane, respectively.

[0051] In step S14, the calculation of the occlusion effect of the target vehicle based on the projection plane and the obstacle vehicle includes:

[0052] Step S141: Calculate the normal vector of the projection plane according to the coordinates of the center point and two diagonal vertices of the target vehicle;

[0053] In this embodiment, the normal vector of the projection plane is to be solved You need to select three points on the plane, such as Figure 3As shown, the center point (p1) and two diagonal vertices (p2 and p3) of the target vehicle are selected, which are specifically:

[0054] p1=(x t ,y t ,z t )

[0055]

[0056] Where x t 、y t and z t Represent the horizontal coordinate, vertical coordinate and vertical coordinate of the center point of the target vehicle, respectively. veh 、w veh and h veh Represent the length, width and height of the target vehicle respectively.

[0057] Based on p1, p2, and p3, the normal vector of the projection plane can be expressed as:

[0058]

[0059] in, and Both represent two vectors on the plane, and Solve by substituting p1 into the general expression for the projective plane.

[0060] Step S142: Using the laser radar as a light source, construct a projection equation of the obstacle vehicle on the projection plane;

[0061] In this embodiment, the projection equation of the obstacle vehicle on the projection plane is:

[0062]

[0063] Where, and Respectively represent the coordinates before and after the obstacle vehicle projection, represents the position of the lidar, ρ represents the scaling factor, and Both represent the projection plane normal vector The weight, represents the constant that determines the position of the projection plane, x l 、y l and z l Represents the horizontal coordinate, vertical coordinate and vertical coordinate of the laser radar, x o 、y o and z o They represent the horizontal, vertical and vertical coordinates of a diagonal vertex of the obstacle vehicle respectively.

[0064] Step S143: Calculating the projection coordinates of two diagonal vertices of the obstacle vehicle according to the projection equation;

[0065] In this embodiment, Figure 3 As shown, and are the coordinates of the two diagonal vertices of the obstacle vehicle, and The coordinates of the two diagonal vertices of the target vehicle are obtained by the projection equation, that is, and And (x l ,y l ,z l ) represents the coordinates of the lidar.

[0066] Step S144: Calculating the overlapping area between the obstacle vehicle and the target vehicle based on the coordinates of the two diagonal vertices of the target vehicle and the projection coordinates of the two diagonal vertices of the obstacle vehicle;

[0067] In this embodiment, the overlapping area between the obstacle vehicle and the target vehicle is approximated as a rectangle, and the area of ​​the overlapping area can be calculated as:

[0068] S lap =w lap ×l lap

[0069]

[0070] Where S lap represents the overlapping area between the obstacle vehicle and the target vehicle, w lap Indicates the width of the overlapping area, l lap Indicates the length of the overlapping area, min(·) indicates the minimum value, max(·) indicates the maximum value, h veh Indicates the height of the target vehicle, and Represent the horizontal coordinates of the two diagonal vertices of the target vehicle, and Represent the ordinates of the two diagonal vertices of the target vehicle, and Represent the projected horizontal coordinates of the two diagonal vertices of the obstacle vehicle, and Respectively represent the projected ordinates of the two diagonal vertices of the obstacle vehicle, Represents the projected vertical coordinate of a diagonal vertex of the obstacle vehicle.

[0071] Step S145: Calculate the occlusion ratio of the target vehicle according to the area of ​​the overlapping region, and define the occlusion effect of the target vehicle according to the occlusion ratio.

[0072] In this embodiment, in order to quantify the occlusion effect, an occlusion ratio is defined to represent the ratio of the occluded area of ​​the target vehicle to the total area on the projection plane.

[0073] The formula for the occlusion ratio is:

[0074]

[0075] Where S lap represents the overlapping area between the obstacle vehicle and the target vehicle, w lap Indicates the width of the overlapping area, l lap Indicates the length of the overlapping area, l veh 、w veh and h veh Represent the length, width and height of the target vehicle respectively.

[0076] Since the occlusion effect of obstacle vehicles at different positions affects the perception of the target vehicle, the occlusion effect is expressed as a conditional function as follows:

[0077]

[0078] Where, represents the occlusion effect of the target vehicle, represents the occlusion penalty function, and They represent the positions of obstacle vehicles, target vehicles and lidar respectively, and λ represents the traffic density.

[0079] Step S15: discretize the occlusion area, and calculate the expected occlusion effect of the target vehicle through the probability density function of the obstacle vehicle at different positions in the occlusion area, the probability of the obstacle vehicle appearing in the occlusion area, and the occlusion effect of the target vehicle.

[0080] In this embodiment, The occlusion effect of the obstacle vehicle on the target vehicle under static position relationship is quantified, and its influence is quantified based on the geometric occlusion ratio. The occlusion effect under static position relationship is further expanded and a probability density function is introduced to characterize the spatial distribution of the obstacle vehicle in the occlusion area. Specifically, using To represent the probability density function of the obstacle vehicle at different positions in the occlusion area Ω. Assume that the probability of the obstacle vehicle appearing in the occlusion area is Then the expected occlusion effect of the target vehicle can be expressed as:

[0081]

[0082] Where, represents the expected occlusion effect of the target vehicle, represents the probability that the obstacle vehicle appears in the occluded area, represents the occlusion penalty function, represents the probability density function of the obstacle vehicle at different positions in the occlusion area Ω, x o and y o Represent the horizontal and vertical coordinates of the center point of the obstacle vehicle, and They represent the positions of obstacle vehicles, target vehicles and lidar respectively, and λ represents the traffic density.

[0083] Therefore, through Can describe the The target vehicle at the traffic flow density λ, from the perspective of LiDAR The expected occlusion effect caused by the obstructing vehicle in the occlusion area Ω is observed. However, in the above formula, the expected occlusion effect of the target vehicle is difficult to solve directly because It is a complex function derived by projection, and its exact form cannot be explicitly defined, making analytical integration difficult to implement.

[0084] To overcome this difficulty, this step simplifies the calculation by discretizing the occluded area into a series of candidate obstacle vehicle positions. Figure 4 As shown in , the position of the obstacle vehicle is generated at fixed intervals (e.g., 1 meter) along each lane in the occlusion area to calculate the expected occlusion effect of the target vehicle, where k represents the total number of lanes.

[0085] The expected occlusion effect of the discretized target vehicle is approximately expressed as:

[0086]

[0087] Where, represents the expected occlusion effect of the target vehicle, N represents the number of discretized obstacle vehicle positions within the occlusion area Ω, represents the probability that the obstacle vehicle appears in the occluded area, represents the occlusion penalty function, represents the position of the ath obstacle vehicle after discretization, represent the positions of the target vehicle and lidar respectively, λ represents the traffic density, Indicates that the obstacle vehicle is located in the occlusion area Ω The probability density function of ΔΩ represents the discretized unit area.

[0088] Since the probability of an obstacle vehicle appearing at different positions on the lane is the same, this means that the position of the obstacle vehicle obeys a uniform distribution. Therefore, Therefore, the expected occlusion effect of the target vehicle can be further simplified as:

[0089]

[0090] Where, represents the expected occlusion effect of the target vehicle, N represents the number of discretized obstacle vehicle positions within the occlusion area Ω, represents the probability that the obstacle vehicle appears in the occluded area, represents the occlusion penalty function, represents the position of the ath obstacle vehicle after discretization, and denote the positions of the target vehicle and lidar respectively, and λ denotes the traffic density.

[0091] Step S2: quantify the perception capability of the lidar, and calculate the expected number of point clouds that can be recognized by the lidar based on the expected occlusion effect and the energy attenuation of the lidar;

[0092] In this embodiment, a quantitative relationship between the laser radar beam distribution and the perception capability is first established, and the number of point clouds obtained from the target vehicle is used as a key indicator for evaluating the laser radar perception capability. The laser radar is modeled as a set of rays defined by horizontal emission angles and vertical emission angles, where the horizontal emission angle is represented by the horizontal angle and the vertical emission angle is represented by the vertical angle, as shown in the figure. Figure 5 As shown, its mathematical expression is as follows:

[0093] α={α i ∣α i ∈[0,2π),i=1,2,…,N α}

[0094] β={β j ∣β j ∈[β min ,β max ],j=1,2,…,N β}

[0095] Where α and β represent the horizontal angle and vertical angle of the LiDAR laser, respectively, i represents the i-th horizontal angle of the LiDAR laser, β j represents the jth vertical angle of the LiDAR laser, β min and β max Represents the minimum vertical angle and the maximum vertical angle, N ɑ and N β denote the number of horizontal and vertical angles respectively, Δɑ represents the horizontal angular resolution.

[0096] like Figure 6As shown, for the t ,y t ,z t ) target vehicle, the effective sensing range of the horizontal and vertical angle ranges of the LiDAR beam are as follows:

[0097]

[0098] Where Δα t and Δβ t They represent the horizontal angle sensing range and vertical angle sensing range of the laser radar beam to the target vehicle, respectively, and x t and y t Indicates the horizontal and vertical coordinates of the center point of the target vehicle, and Represent the horizontal coordinates of the two diagonal vertices of the target vehicle, and Represents the ordinates of the two diagonal vertices of the target vehicle, h veh Indicates the height of the target vehicle, Z l represents the vertical coordinate of the laser radar, arctan(·) represents the inverse tangent function, α t and β t represent the horizontal and vertical angles of the lidar beam to the target vehicle, respectively.

[0099] In step S2, the calculation steps of the expected value of the point cloud number are:

[0100] Step S21: Calculating the theoretical number of point clouds on the target vehicle based on first information and a vertical angle distribution function, wherein the first information is the horizontal angle sensing range and the vertical angle sensing range of the laser radar beam to the target vehicle;

[0101] In this embodiment, we focus on a mechanical rotating LiDAR, where the horizontal angular resolution Δα is fixed, while the vertical angular distribution is determined by internal design and characterized by a distribution function or model. Therefore, the theoretical number of point clouds of a target vehicle can be expressed as:

[0102]

[0103] Where, represents the theoretical point cloud number of the target vehicle, Δα t and Δβ t They represent the horizontal and vertical angle sensing ranges of the laser radar beam to the target vehicle, Δα represents the horizontal angular resolution, represents the vertical angle distribution function, and β represents the vertical angle of the LiDAR laser.

[0104] Step S22: Calculating the first point cloud loss ratio on the target vehicle based on the energy attenuation of the laser radar;

[0105] In this embodiment, Figure 7 As shown in Figure 2, LiDAR's ability to perceive a vehicle point cloud is affected by laser reflectivity and occlusion. Regarding reflectivity, laser energy attenuates with increasing distance, significantly increasing the probability that distant targets will not be detected. Therefore, this energy attenuation is modeled based on the probability of loss over distance, as expressed as follows:

[0106]

[0107] Where, represents the loss ratio of the first point cloud, which is actually the loss ratio of the point cloud on the target vehicle caused by energy attenuation. γ represents the attenuation rate, which characterizes the growth rate of the loss probability with increasing distance. t 、y t and z t Represent the horizontal, vertical and vertical coordinates of the center point of the target vehicle respectively.

[0108] Step S23: Calculating a second point cloud loss ratio on the target vehicle based on the expected occlusion effect;

[0109] In this embodiment, regarding the occlusion problem, obstructing vehicles within the occlusion area will reduce the effective sensing area of ​​the target vehicle. Specifically, the occlusion area on the projection plane represents the portion of the target vehicle that the LiDAR cannot directly observe due to occlusion. Therefore, the reduction in the number of LiDAR detectable point clouds caused by occlusion can be quantified by the occlusion effect of the target vehicle, which is expressed as follows:

[0110]

[0111] Where, Indicates the loss ratio of the second point cloud, which is actually the loss ratio of the point cloud on the target vehicle caused by the occlusion effect. represents the expected occlusion effect of the target vehicle.

[0112] Step S24: calculating second information based on the theoretical point cloud quantity, the first point cloud loss ratio, and the second point cloud loss ratio, wherein the second information is the number of valid point clouds detectable on the target vehicle;

[0113] In this embodiment, in order to further quantify this point cloud loss relationship, a function is established to relate the reduction in the number of LiDAR detectable point clouds on the target vehicle to the energy attenuation and occlusion effects. The expression is as follows:

[0114]

[0115] Where, Indicates the horizontal angular resolution Δα and vertical angular distribution function Next, the target vehicle v t The number of valid point clouds that can be detected, represents the theoretical number of point clouds of the target vehicle, Indicates the loss ratio of the first point cloud, Indicates the loss ratio of the second point cloud.

[0116] Step S25: Calculate the expected number of point clouds that can be identified by the laser radar using the second information.

[0117] In this embodiment, in order to expand the analysis to cover all possible target vehicle positions within the LiDAR detection range, the spatial distribution of target vehicles is considered. The expected value of spatial probability density function Describes the target vehicle in the detection area Any position within (x t ,y t ). Therefore, considering the change in the position of the target vehicle, the expected value of the number of point clouds that can be identified by LiDAR can be expressed as:

[0118]

[0119] Where, represents the expected number of point clouds that can be identified by the lidar, Indicates the detection area, Indicates the horizontal angular resolution Δα and vertical angular distribution function Below, located at (x t ,y t ) of the target vehicle, represents the spatial probability density function.

[0120] in, In fact, it describes the expected number of point clouds that can be identified by the LiDAR when considering all possible target vehicle positions within the LiDAR detection range. The detection area is calculated by using the discretization method. Divide into a finite number of discrete points. Specifically, the LiDAR detection range is gridded based on lanes and a specific length interval (e.g., 5 meters). The formula after discretization is as follows:

[0121]

[0122] Where, represents the expected number of point clouds that can be identified by the lidar, M represents the total number of possible positions of the target vehicle, Indicates the horizontal angular resolution Δα and vertical angular distribution function Below, located The number of valid point clouds that can be detected on the target vehicle, represents the coordinates of the bth discrete position, Represents the area corresponding to each discrete point, represents the probability density function.

[0123] Since the target vehicle is located at each point in the same lane, the probability is uniform, that is, Therefore, we can further simplify and obtain the expected occlusion effect of the target vehicle as follows:

[0124]

[0125] In the formula represents the expected number of point clouds that can be identified by the lidar, M represents the total number of possible positions of the target vehicle, Indicates the horizontal angular resolution Δα and vertical angular distribution function Below, located The number of valid point clouds that can be detected on the target vehicle, Represents the coordinates of the bth discrete position.

[0126] Step S3: constructing an objective function based on the expected value of the point cloud number and setting an optimization parameter, wherein the optimization parameter is the vertical angle distribution function of the laser radar;

[0127] In step S3, constructing an objective function based on the expected value of the point cloud number includes:

[0128] Step S31: constructing an initial objective function through the expected value of the point cloud number;

[0129] In this embodiment, in order to optimize the laser beam distribution of LiDAR to maximize its perception capability, and focusing on the vertical angle distribution function, an optimization model is constructed with the expected value of the point cloud number as the initial objective function. The optimization model is expressed as:

[0130]

[0131] Where M represents the total number of possible locations of the target vehicle, Indicates the horizontal angular resolution Δα and vertical angular distribution function Below, located The number of valid point clouds that can be detected on the target vehicle, represents the coordinates of the bth discrete position, Indicates that the objective function reaches its maximum value The value of represents the optimal vertical angle distribution function.

[0132] Step S32: constructing a weighted function for distance, and modifying the initial objective function using the weighted function to obtain an objective function;

[0133] In this embodiment, the optimization model aims to maximize the expected perception capability of the vehicle point cloud. However, optimizing only for overall perception may cause the LiDAR laser beam distribution to be biased toward closer targets, as occlusion and laser attenuation reduce the point cloud gain of distant vehicles. To address this issue, a distance-based weighting function is introduced to adjust the contribution weight of the target vehicle position according to its distance, resulting in the objective function. The modified optimization model based on the objective function is expressed as follows:

[0134]

[0135] Where M represents the total number of possible locations of the target vehicle, Indicates the horizontal angular resolution Δɑ and the vertical angular distribution function Below, located The number of valid point clouds that can be detected on the target vehicle, represents the coordinates of the bth discrete position, Indicates that the objective function reaches its maximum value The value of represents the optimal vertical angle distribution function, Express about The weighting function of k represents the slope of weight growth, which determines the intensity of the influence of distance growth on the weight.

[0136] Step S33: setting constraints of the objective function, wherein the constraints include a beam angle constraint and a minimum angular resolution constraint. The beam angle constraint requires that the vertical angle of each laser beam is not less than the minimum vertical angle and not greater than the maximum vertical angle.

[0137] In this embodiment, the beam angle restriction constraint limits the field of view of the laser, ensuring that β min allows detection of the bottom of the nearest vehicle, while β max The top of the farthest vehicle can be observed. The specific formula is as follows:

[0138] β min ≤β j ≤β max

[0139]

[0140] Where, β min and β max denote the minimum vertical angle and the maximum vertical angle respectively, arctan(·) denotes the inverse tangent function, z l Indicates the vertical coordinate of the lidar, hveh Indicates the height of the target vehicle, d min and d max Represent the minimum detection distance and maximum detection distance of LiDAR, β j represents the j-th vertical angle of the LiDAR laser.

[0141] The formula for the minimum angular resolution constraint is:

[0142] β j+1 -β j ≥Δ

[0143] Where, β j and β j+1 They represent the j-th and j+1-th vertical angles of the LiDAR laser, respectively, and Δ represents the preset resolution.

[0144] Step S4: Optimize the optimization parameters through the objective function and interval search algorithm to obtain the optimal vertical angle distribution function, and determine the laser radar beam distribution through the optimal vertical angle distribution function.

[0145] In this embodiment, since the revised optimization model involves multiple variables, i.e., multiple vertical angles, the interdependence of the multiple vertical angles makes it difficult to find the global optimal solution. Therefore, an interval search algorithm is designed to solve the problem based on the objective function to obtain the optimal vertical angle distribution function.

[0146] In step S4, obtaining the optimal vertical angle distribution function includes:

[0147] Step S41: calculating a first vertical angle and a second vertical angle of each laser beam based on the uniform angle distribution and the uniform distance distribution;

[0148] In this embodiment, the LiDAR has multiple vertical angles, each corresponding to a laser beam. The LiDAR's vehicle perception capabilities differ under two different laser beam distribution modes: uniform angular distribution and uniform distance distribution. When the LiDAR's laser beams are evenly distributed at equal angular intervals within the field of view, it has a strong ability to perceive close-range vehicles but has difficulty detecting distant targets. In contrast, a LiDAR laser beam with a uniform distance distribution across the detection range effectively detects distant vehicles but has reduced perception of close-range targets.

[0149] The calculation formulas for the first vertical angle and the second vertical angle are:

[0150]

[0151] Where, and denote the first vertical angle and the second vertical angle of the j′th laser beam, respectively, and Nβ represents the number of vertical angles, β min and β max denote the minimum vertical angle and the maximum vertical angle respectively, arctan(·) denotes the inverse tangent function, z l Indicates the vertical coordinate of the laser radar, d min and d max Respectively represent the minimum detection distance and maximum detection distance of LiDAR, d j′ represents the detection distance of the j′th laser beam.

[0152] Step S42: calculating the lower limit and the upper limit of the vertical angle of each laser beam by using the first vertical angle and the second vertical angle;

[0153] In this embodiment, the calculation formulas for the lower and upper limits of the vertical angle of each laser beam are:

[0154]

[0155] Where, and denote the lower and upper bounds of the vertical angle of the j′th laser beam, respectively, and They represent the first vertical angle and the second vertical angle of the j′th laser beam respectively, min(·) represents the minimum value, and max(·) represents the maximum value.

[0156] Step S43: defining the lower bound and upper bound of the initial vertical angle distribution function by the lower bound and upper bound of the vertical angle of each laser beam;

[0157] In this embodiment, the boundaries of the initial vertical angle distribution function are as follows:

[0158]

[0159] Where, and denote the lower and upper bounds of the initial vertical angle distribution function, respectively. and They represent the lower and upper bounds of the vertical angle of the j′th laser beam, respectively, and N β Indicates the number of vertical angles.

[0160] Step S44: using the lower bound and upper bound of the initial vertical angle distribution function as a search interval, and iteratively optimizing the search interval based on the segmentation ratio to obtain an optimal search interval;

[0161] In this embodiment, the search interval is in, and represent the lower and upper bounds of the initial vertical angle distribution function, respectively.

[0162] In step S44, the optimal search interval is obtained, including:

[0163] Step S441: generating a first laser beam distribution and a second laser beam distribution for each laser beam according to the segmentation ratio and the search space;

[0164] In this embodiment, the segmentation ratio is pre-set. In each iteration, two new laser beams are obtained according to the segmentation ratio, specifically:

[0165]

[0166] j′=1,2,…,N β

[0167] Where, and denote the first laser beam distribution and the second laser beam distribution of the j′th laser beam, respectively, and They represent the lower and upper bounds of the j′th laser beam, φ represents the division ratio, and N β Indicates the number of vertical angles.

[0168] Step S442: combining all first laser beam distributions into a first tentative distribution, and combining all second laser beam distributions into a second tentative distribution;

[0169] Step S443: For the first trial distribution and the second trial distribution, calculate the corresponding first function value and second function value using the objective function;

[0170] In this embodiment, for the first trial distribution and the second trial distribution Calculate the corresponding first function value through the objective function and the second function value

[0171] Step S444: updating the search space by updating the strategy, the first function value, and the second function value;

[0172] In this embodiment, if Then the second trial distribution As Otherwise, the first trial distribution As Through such an update strategy, the boundaries of the search space are continuously updated.

[0173] Step S445: Determine whether the termination threshold is reached. If so, use the updated search interval as the optimal search interval. Otherwise, perform the next iteration based on the updated search interval.

[0174] Step S45: Calculate the optimal vertical angle distribution function according to the lower bound and the upper bound of the optimal search interval.

[0175] In this embodiment, the calculation formula of the optimal vertical angle distribution function is:

[0176]

[0177] Where, represents the optimal vertical angle distribution function.

[0178] In this embodiment, the interval search algorithm optimizes the LiDAR laser beam angles as a whole distribution, rather than optimizing each laser beam angle individually. This simplifies the inherently complex multidimensional optimization problem and significantly improves the efficiency of the optimization process. Compared to the traditional method of adjusting each laser beam angle individually, this overall optimization method can simultaneously consider the interrelationships between multiple angles in a single search. As a result, the algorithm can quickly find the optimal laser beam angle distribution in a relatively small number of iterations, significantly improving optimization efficiency and maintaining stable convergence on large-scale problems.

[0179] In summary, this method uses geometric methods to calculate the occlusion area between the target vehicle and the LiDAR, accurately quantifying the occlusion effect. By constructing a projection plane and calculating the normal vector, the occlusion problem in three-dimensional space can be transformed into a geometric problem on a two-dimensional plane, simplifying the analysis of complex occlusion relationships and making the area calculation of the occlusion area feasible.

[0180] Second, evaluating LiDAR perception capabilities by the number of point clouds captured by the target vehicle provides an intuitive and quantifiable evaluation metric. Incorporating distance-based energy attenuation and occlusion effects makes the evaluation of LiDAR perception capabilities more comprehensive, reflecting the complex influencing factors in the real environment.

[0181] Finally, the laser beam angle distribution is optimized as a whole through the interval search algorithm. When obtaining the optimal lidar beam distribution, the complexity of optimizing each laser beam angle one by one is avoided, the optimization efficiency is significantly improved, and the algorithm can find the optimal solution within a smaller number of iterations.

[0182] Example 2:

[0183] This embodiment provides a device for optimizing roadside laser radar beam distribution based on vehicle occlusion, the device comprising:

[0184] The first construction module is used to construct the expected occlusion effect of the target vehicle based on the occlusion relationship between the road test laser radar and the dynamic vehicle;

[0185] The calculation module is used to quantify the perception capability of the lidar and calculate the expected number of point clouds that can be recognized by the lidar based on the expected occlusion effect and the energy attenuation of the lidar;

[0186] The second construction module is used to construct an objective function based on the expected value of the point cloud number and set an optimization parameter, wherein the optimization parameter is the vertical angle distribution function of the laser radar;

[0187] The optimization module is used to optimize the optimization parameters through the objective function and interval search algorithm to obtain the optimal vertical angle distribution function, and determine the laser radar beam distribution through the optimal vertical angle distribution function.

[0188] The first building block includes:

[0189] A determination unit, configured to determine the coordinates of diagonal vertices of a bounding box of the target vehicle according to the position and size of the target vehicle;

[0190] A first calculation unit is used to calculate the area of ​​the occlusion region of the target vehicle according to the coordinates of the diagonal vertices of the bounding box and the position of the laser radar;

[0191] The second calculation unit is used to calculate the probability of the obstructing vehicle appearing in the blocked area based on the Poisson point process;

[0192] A first construction unit is configured to construct a projection plane and calculate an occlusion effect of the target vehicle based on the projection plane and the obstacle vehicle;

[0193] The third calculation unit is used to discretize the occlusion area and calculate the expected occlusion effect of the target vehicle through the probability density function of the obstacle vehicle at different positions in the occlusion area, the probability of the obstacle vehicle appearing in the occlusion area and the occlusion effect of the target vehicle.

[0194] The calculation module includes:

[0195] a fourth calculation unit, configured to calculate a theoretical number of point clouds on the target vehicle based on the first information and the vertical angle distribution function, wherein the first information is a horizontal angle perception range and a vertical angle perception range of the laser radar beam to the target vehicle;

[0196] a fifth calculation unit, configured to calculate a first point cloud loss ratio on the target vehicle based on energy attenuation of the laser radar;

[0197] a sixth calculation unit, configured to calculate a second point cloud loss ratio on the target vehicle based on an expected occlusion effect;

[0198] a seventh calculation unit, configured to calculate second information according to the theoretical point cloud quantity, the first point cloud loss ratio, and the second point cloud loss ratio, wherein the second information is the number of valid point clouds detectable on the target vehicle;

[0199] The eighth calculation unit is used to calculate the expected value of the number of point clouds that can be recognized by the laser radar based on the second information.

[0200] It should be noted that, regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated on here.

[0201] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

[0202] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A roadside lidar beam distribution optimization method based on vehicle occlusion, characterized in that: include: Based on the occlusion relationship between the road test LiDAR and the dynamic vehicle, the expected occlusion effect of the target vehicle is constructed; Quantify the perception capability of the lidar and calculate the expected number of point clouds that the lidar can recognize based on the expected occlusion effect and the energy attenuation of the lidar; Constructing an objective function based on the expected value of the point cloud number and setting an optimization parameter, wherein the optimization parameter is the vertical angle distribution function of the laser radar; The optimization parameters are optimized through the objective function and interval search algorithm to obtain the optimal vertical angle distribution function, and the lidar beam distribution is determined by the optimal vertical angle distribution function.

2. The method for optimizing roadside laser radar beam distribution based on vehicle occlusion according to claim 1 is characterized in that ,The desired occlusion effect of the target vehicle is constructed including: Determine the coordinates of the diagonal vertices of the bounding box of the target vehicle according to the position and size of the target vehicle; Calculate the occluded area of ​​the target vehicle based on the coordinates of the diagonal vertices of the bounding box and the position of the lidar; Based on the Poisson point process, the probability of the obstacle vehicle appearing in the blocked area is calculated; Construct a projection plane, and calculate the occlusion effect of the target vehicle based on the projection plane and the obstacle vehicle; The occlusion area is discretized, and the expected occlusion effect of the target vehicle is calculated through the probability density function of the obstacle vehicle at different positions in the occlusion area, the probability of the obstacle vehicle appearing in the occlusion area, and the occlusion effect of the target vehicle.

3. The method for optimizing roadside laser radar beam distribution based on vehicle occlusion according to claim 2 is characterized in that ,The occlusion effect of the target vehicle is calculated based on the ,projection plane and the obstacle vehicle, including: Calculate the normal vector of the projection plane according to the coordinates of the center point and two diagonal vertices of the target vehicle; Using the laser radar as the light source, the projection equation of the obstacle vehicle on the projection plane is constructed; Calculate the projection coordinates of the two diagonal vertices of the obstacle vehicle according to the projection equation; Calculate the overlapping area between the obstacle vehicle and the target vehicle based on the coordinates of the two diagonal vertices of the target vehicle and the projection coordinates of the two diagonal vertices of the obstacle vehicle; The occlusion ratio of the target vehicle is calculated according to the area of ​​the overlapping region, and the occlusion effect of the target vehicle is defined according to the occlusion ratio.

4. The method for optimizing roadside laser radar beam distribution based on vehicle occlusion according to claim 1 is characterized in that ,The calculation steps of the expected value of the point cloud number are: Calculating the theoretical number of point clouds on the target vehicle based on first information and a vertical angle distribution function, wherein the first information is the horizontal angle perception range and the vertical angle perception range of the laser radar beam to the target vehicle; Calculate the first point cloud loss ratio on the target vehicle based on the energy attenuation of the laser radar; Calculate the second point cloud loss ratio on the target vehicle based on the expected occlusion effect; Calculating second information based on the theoretical point cloud quantity, the first point cloud loss ratio, and the second point cloud loss ratio, wherein the second information is the number of valid point clouds detectable on the target vehicle; The expected number of point clouds recognizable by the lidar is calculated using the second information.

5. The method for optimizing roadside laser radar beam distribution based on vehicle occlusion according to claim 1 is characterized in that ,The objective function is constructed based on the expected value of the point cloud number, including: Construct the initial objective function through the expected value of the point cloud number; Construct a weighted function about distance, and modify the initial objective function by the weighted function to obtain the objective function; The constraint conditions of the objective function are set, wherein the constraint conditions include a beam angle constraint and a minimum angular resolution constraint. The beam angle constraint is that the vertical angle of each laser beam is not less than the minimum vertical angle and not greater than the maximum vertical angle.

6. The method for optimizing roadside laser radar beam distribution based on vehicle occlusion according to claim 1 is characterized in that , the obtaining of the optimal vertical angle distribution function includes: respectively calculating a first vertical angle and a second vertical angle of each laser beam based on the uniform angle distribution and the uniform distance distribution; Calculating a lower bound and an upper bound of a vertical angle of each laser beam using the first vertical angle and the second vertical angle; defining lower and upper bounds of the initial vertical angle distribution function by lower and upper bounds of the vertical angle of each laser beam; The lower and upper bounds of the initial vertical angle distribution function are used as the search interval, and the search interval is iteratively optimized based on the segmentation ratio to obtain the optimal search interval; The optimal vertical angle distribution function is calculated based on the lower and upper bounds of the optimal search interval.

7. The method for optimizing roadside laser radar beam distribution based on vehicle occlusion according to claim 6 is characterized in that , get the optimal search interval, including: generating a first laser beam distribution and a second laser beam distribution for each laser beam according to the division ratio and the search space; combining all first laser beam profiles into a first tentative profile, and combining all second laser beam profiles into a second tentative profile; For the first trial distribution and the second trial distribution, the corresponding first function value and the second function value are calculated by the objective function; updating the search space by updating the strategy, the first function value, and the second function value; Determine whether the termination threshold is reached. If so, use the updated search interval as the optimal search interval. Otherwise, perform the next iteration based on the updated search interval.

8. A roadside laser radar beam distribution optimization device based on vehicle occlusion, characterized in that: include: The first construction module is used to construct the expected occlusion effect of the target vehicle based on the occlusion relationship between the road test laser radar and the dynamic vehicle; The calculation module is used to quantify the perception capability of the lidar and calculate the expected number of point clouds that can be recognized by the lidar based on the expected occlusion effect and the energy attenuation of the lidar; The second construction module is used to construct an objective function based on the expected value of the point cloud number and set an optimization parameter, wherein the optimization parameter is the vertical angle distribution function of the laser radar; The optimization module is used to optimize the optimization parameters through the objective function and interval search algorithm to obtain the optimal vertical angle distribution function, and determine the laser radar beam distribution through the optimal vertical angle distribution function.

9. The device for optimizing roadside laser radar beam distribution based on vehicle occlusion according to claim 8, characterized in that: The first building block includes: A determination unit, configured to determine the coordinates of diagonal vertices of a bounding box of the target vehicle according to the position and size of the target vehicle; A first calculation unit is used to calculate the area of ​​the occlusion region of the target vehicle according to the coordinates of the diagonal vertices of the bounding box and the position of the laser radar; The second calculation unit is used to calculate the probability of the obstructing vehicle appearing in the blocked area based on the Poisson point process; A first construction unit is configured to construct a projection plane and calculate an occlusion effect of the target vehicle based on the projection plane and the obstacle vehicle; The third calculation unit is used to discretize the occlusion area and calculate the expected occlusion effect of the target vehicle through the probability density function of the obstacle vehicle at different positions in the occlusion area, the probability of the obstacle vehicle appearing in the occlusion area and the occlusion effect of the target vehicle.

10. The device for optimizing roadside laser radar beam distribution based on vehicle occlusion according to claim 8, characterized in that: The calculation module includes: a fourth calculation unit, configured to calculate a theoretical number of point clouds on the target vehicle based on the first information and the vertical angle distribution function, wherein the first information is a horizontal angle perception range and a vertical angle perception range of the laser radar beam to the target vehicle; a fifth calculation unit, configured to calculate a first point cloud loss ratio on the target vehicle based on energy attenuation of the laser radar; a sixth calculation unit, configured to calculate a second point cloud loss ratio on the target vehicle based on an expected occlusion effect; a seventh calculation unit, configured to calculate second information according to the theoretical point cloud quantity, the first point cloud loss ratio, and the second point cloud loss ratio, wherein the second information is the number of valid point clouds detectable on the target vehicle; The eighth calculation unit is used to calculate the expected value of the number of point clouds that can be recognized by the laser radar based on the second information.

Images