Inversion method of laser radar sensor

Abstract

The invention discloses an inversion method of a laser radar sensor. The method comprises the following steps: building a virtual unmanned ship collision detection system; a quaternion calculation method is adopted, measurement in a local coordinate system of a virtual laser radar in a virtual unmanned ship collision detection system is converted into a world coordinate system, and the emission direction and the starting point position of rays emitted by the virtual laser radar are determined; based on the emission direction and the starting point position of rays emitted by the virtual laser radar, the view field range of the virtual laser radar is determined through scanning coverage in the horizontal dimension and the vertical dimension, and ray tracing of the virtual laser radar is achieved; performing space division on the established virtual unmanned ship collision detection system; checking the intersection condition of the rays and the bounding box, screening out an area where collision possibly occurs, and checking whether the rays collide with objects in the area or not for detection; and when the ray detects collision, point cloud data points are generated, and inversion calculation of the laser radar sensor is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of virtual simulation, and particularly relates to an inversion method for lidar, providing key technology for the testing and evaluation of virtual scenarios. Background Art

[0002] Testing an unmanned ship in a real environment requires facing complex and changeable marine conditions, including weather changes, ocean current effects, wave interference, etc. These factors not only increase the difficulty and cost of testing, but also may bring uncontrollable risks. In addition, real-scenario testing usually requires a large amount of manpower, material resources and time investment, and it is difficult to meet the large-scale and diversified testing requirements. At the same time, testing in some extreme or dangerous environments may be difficult to carry out safely and effectively, further restricting the comprehensive verification and optimization of unmanned ship technology. To overcome the many deficiencies of real testing scenarios, virtual testing has become an important supplementary means. By constructing a highly simulated virtual environment, diverse and repetitive testing of unmanned ships can be carried out under controllable and safe conditions. This not only significantly reduces the testing cost and risk, but also improves the testing efficiency and coverage. In addition, the virtual testing platform can flexibly simulate various complex and extreme marine environments, helping to comprehensively evaluate the performance of unmanned ships under different conditions, thus accelerating the R & D and optimization process of unmanned ship technology.

[0003] In virtual testing, sensors are important components for unmanned ships to obtain environmental information, achieve autonomous navigation and decision-making. High-precision sensor simulation can truly reproduce the working principle and performance characteristics of actual sensors, ensuring that the virtual testing results have high credibility and practicality. Among them, lidar, as an efficient environmental perception sensor, is widely used in functions such as navigation, obstacle avoidance and environmental modeling of unmanned ships. However, existing lidar simulation methods still have certain deficiencies in terms of accuracy, efficiency and scalability, and it is difficult to meet the high requirements of testing in complex virtual environments.

[0004] In view of the above problems, it is particularly important to study the inversion method for lidar. Sensor inversion technology aims to accurately simulate the signal emission, propagation and reception processes of lidar in the actual environment through the optical characteristics and physical models in the virtual environment. This can not only improve the simulation accuracy of lidar in virtual testing, but also enhance the ability of the virtual environment to truly reflect the behavior of unmanned ships. Through effective sensor inversion methods, high-precision restoration of lidar data can be achieved, supporting more comprehensive and in-depth performance evaluation and optimization of unmanned ships. Summary of the Invention

[0005] In order to solve the problems of incomplete sensor coverage and insufficient authenticity in traditional virtual testing scenarios, the technical solution adopted by the present invention is: an inversion method for a lidar sensor, including the following steps:

[0006] Build a virtual unmanned ship collision detection system;

[0007] Adopt the calculation method of quaternion to convert the measurement in the local coordinate system of the virtual lidar in the virtual unmanned ship collision detection system into the world coordinate system, so as to determine the emission direction and starting position of the ray emitted by the virtual lidar;

[0008] Based on the emission direction and starting position of the ray emitted by the virtual lidar, determine the field of view range of the virtual lidar by scanning and covering in two dimensions of horizontal and vertical, so as to realize the ray tracing of the virtual lidar;

[0009] Conduct spatial division on the built virtual unmanned ship collision detection system;

[0010] Check the intersection of the ray and the bounding box, screen out the areas where collisions may occur, and for the areas where collisions may occur, check whether the ray will collide with the objects in the area for detection;

[0011] When the ray detects a collision, generate point cloud data points to realize the inversion calculation of the lidar sensor.

[0012] Furthermore, the process of adopting the calculation method of quaternion to convert the measurement in the local coordinate system of the virtual lidar in the virtual unmanned ship collision detection system into the world coordinate system to determine the emission direction and starting position of the ray emitted by the virtual lidar is as follows:

[0013] Each beam of laser of the virtual lidar can be regarded as a ray, and its propagation process is described by the ray equation:

[0014] P(t) = P0 + td (1)

[0015] Where P(t) represents any point on the ray, P0 is the starting point of the ray, t is the ray parameter, and d is the direction vector of the ray;

[0016] In the local coordinate system, the position vector is expressed as:

[0017] P l =(x l ,y l ,z l ) (2)

[0018] The direction quaternion is defined as:

[0019]

[0020] Where θ is the rotation angle, (u x ,u y ,u zis the unit vector of the rotation axis.

[0021] The complete coordinate transformation chain is expressed as:

[0022] p w = q v (q l p l q l -1 )q v -1 + p v (4)

[0023] Among them, p w is the position vector of the point in the world coordinate system, p l is the position vector of the point in the local coordinate system of the lidar, p v is the position of the hull in the world coordinate system, q v is the rotation quaternion of the hull relative to the world, q l is the rotation quaternion of the lidar relative to the hull.

[0024] Furthermore, based on the emission direction and starting position of the ray emitted by the virtual lidar, the process of determining the virtual lidar field of view range through scanning coverage in the horizontal and vertical dimensions and realizing the ray tracing of the virtual lidar is as follows:

[0025] The horizontal scanning angle α h is calculated by the formula:

[0026]

[0027] Among them, i is the horizontal number of the current laser beam, N h is the horizontal angular resolution;

[0028] The vertical scanning angle α v is calculated by the formula:

[0029]

[0030] Among them, j is the number in the vertical direction, α upper and α lower are the upper and lower boundary angles of the vertical field of view respectively.

[0031] Combining the horizontal and vertical angles, the direction vector of the laser beam in the local coordinate system is obtained.

[0032] d = (cosα v cosα h , cosα v sinα h , sinα v ) (7)

[0033] Further: The process of building the virtual unmanned ship collision detection system is as follows:

[0034] S11: Use 3D modeling software to build models including unmanned ships, lidars, islands, and buoys, and convert the format to FBX format and export it through 3dsMax software;

[0035] S12: Import the FBX format model into Unreal Engine software, and build a virtual test environment consistent with the real test environment, and add collision boxes to the models in the virtual test environment;

[0036] S13: Set the parameters of the lidar, including the number of channels, horizontal rotation angular velocity, field of view range, and measurement range. Through these parameters, generate the emission angles of multiple beams of laser emitted by the lidar, covering the entire field of view range.

[0037] An inversion device for a lidar sensor, comprising:

[0038] A building module: used to build a virtual unmanned ship collision detection system;

[0039] A determination module: used to adopt the calculation method of quaternions to convert the measurement in the local coordinate system of the virtual lidar in the virtual unmanned ship collision detection system to the world coordinate system, and realize the determination of the emission direction and starting position of the rays emitted by the virtual lidar;

[0040] A ray tracing module: used to determine the virtual lidar field of view range based on the emission direction and starting position of the rays emitted by the virtual lidar, and realize the ray tracing of the virtual lidar through scanning coverage in two dimensions, horizontal and vertical;

[0041] A division module: used to perform spatial division on the built virtual unmanned ship collision detection system;

[0042] A collision detection module: used to check the intersection of the rays and the bounding box, screen out the areas where collisions may occur, and for the areas where collisions may occur, check whether the rays will collide with the objects in the area;

[0043] A generation module: used to generate point cloud data points when the rays detect a collision, and realize the inversion calculation of the lidar sensor.

[0044] A computer device, comprising: a processor and a memory, the memory stores program modules, and the program modules run on the processor to implement the method described in any one of the above.

[0045] A readable storage medium stores program modules, characterized in that the program modules can implement the method described in any one of the above when running in a processor.

[0046] The inversion method of a lidar provided by the present invention has the following advantages:

[0047] In the virtual reality software of the present invention, ray tracing and collision detection methods are used to obtain virtual point cloud data, and the virtual lidar is configured according to the real lidar parameters to output the effect closest to the real lidar point cloud data.

[0048] Through this systematic implementation solution, the present invention can generate high-quality point cloud data, with the following technical advantages: relatively high measurement accuracy, can accurately simulate the reflection characteristics of different material surfaces, and supports real-time generation of large-scale point clouds.

[0049] Compared with the prior art, the beneficial effects of the present invention are:

[0050] The present invention constructs a virtual simulation environment that conforms to the real scene through virtual simulation. In this environment, the virtual lidar is used to simulate the working process of the actual lidar, and the collision of the laser beam with the objects in the environment is detected through ray tracing technology. It not only overcomes the limitations of traditional mathematical models but also provides measurement results closer to the real lidar. These distance data accurately reflect the spatial characteristics of the environment and are transmitted to the control system through the network, so as to realize the testing of algorithms such as automatic navigation, obstacle avoidance planning, and environmental perception. The simulation environment solves the problems in the actual lidar test limited by geographical location, weather conditions, and equipment costs, can quickly and accurately obtain environmental information, avoid the high costs and high risks in real tests, and improve the development and test efficiency of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0052] Figure 1 is the method flow chart of the present invention;

[0053] Figure 2 is the virtual scene in the UE;

[0054] Figure 3 is the visualization result of the lidar virtual point cloud. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0056] To make the objectives, 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 with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are only a part rather than all of the embodiments of the present invention. The following description of at least one exemplary embodiment is actually only illustrative and in no way limits the present invention and its application or use. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0057] Figure 1 is the method flow chart of the present invention;

[0058] Figure 2 is the virtual scene in the UE;

[0059] An inversion method for a lidar sensor includes the following steps:

[0060] S1: Build a virtual unmanned ship collision detection system;

[0061] S2: Using the calculation method of quaternions, convert the measurement in the local coordinate system of the virtual lidar in the virtual unmanned ship collision detection system to the world coordinate system to determine the emission direction and starting position of the ray emitted by the virtual lidar;

[0062] S3: Based on the emission direction and starting position of the ray emitted by the virtual lidar, determine the virtual lidar field of view by scanning and covering in two dimensions, horizontal and vertical, to achieve ray tracing of the virtual lidar;

[0063] S4: Divide the space of the built virtual unmanned ship collision detection system;

[0064] S5: Check the intersection of the ray and the bounding box, screen out the areas where collisions may occur, and for the areas where collisions may occur, check whether the ray will collide with the objects in the area for detection;

[0065] S6: When the ray detects a collision, generate point cloud data points to achieve the inversion calculation of the lidar sensor.

[0066] The steps S1 / S2 / S3 / S4 / S5 / S6 are executed in sequence;

[0067] The process of building the virtual unmanned ship collision detection system is as follows:

[0068] S11: Build models including an unmanned ship, lidar, islands, and buoys using 3D modeling software, convert the format to FBX format through 3dsMax software and export it.

[0069] S12: Import the FBX format model into the Unreal Engine simulation software, build a virtual test environment consistent with the real test environment, and add collision boxes to the models in the virtual test environment. Since the acquisition of simulated point cloud data depends on the collision detection between laser rays and the object to be measured, whether the laser rays can be detected after colliding with the object to be measured mainly depends on whether an appropriate collision box is set for the static mesh of the object to be measured.

[0070] S13: Set the parameters of the lidar, including the number of channels, horizontal rotation angular velocity, field of view range, and measurement range. Through these parameters, generate the emission angles of multiple laser beams of the lidar, covering the entire field of view range.

[0071] The process of using the quaternion calculation method to convert the measurement in the local coordinate system of the virtual lidar in the virtual unmanned ship collision detection system to the world coordinate system to determine the emission direction and starting position of the rays emitted by the virtual lidar is as follows:

[0072] Use ray tracing and collision detection methods to determine whether the laser rays collide with virtual obstacles. If there is no collision, it means there are no obstacles in that direction or the obstacles are outside the detection range of the lidar. If a collision is detected, the coordinates of the collision point will be stored, and these coordinates represent the positions of the obstacles measured by the lidar.

[0073] Ray tracing is a technology for simulating the propagation of rays in three-dimensional space. It calculates the intersection points of rays with the surface of objects through a collision detection system to obtain spatial information.

[0074] Each laser beam of the virtual lidar can be regarded as a ray, and its propagation process is described by the ray equation:

[0075] P(t) = P0 + td (1)

[0076] Among them, P(t) represents any point on the ray, P0 is the starting point of the ray, t is the ray parameter, and d is the direction vector of the ray; in the application scenario of marine lidar, the starting point P0 represents the position of the lidar in the world coordinate system, and the direction vector d is determined by the scanning angle of the lidar.

[0077] However, in an actual marine lidar system, the situation is more complex. Since the lidar is installed on the hull, as the ship moves, the position and attitude of the lidar will continuously change. This requires us to establish an accurate coordinate transformation system to convert the measurements in the lidar local coordinate system to the world coordinate system. This transformation process needs to consider two aspects: one is the installation position and attitude of the lidar relative to the hull, and the other is the position and attitude of the hull in the world coordinate system.

[0078] To achieve accurate coordinate transformation, we adopt a calculation method based on quaternions. Compared with Euler angles, quaternions can avoid the gimbal lock problem and provide better numerical stability. First, define the basic coordinate transformation relationships:

[0079] In the local coordinate system, the position vector is expressed as:

[0080] P l =(x l ,y l ,z l ) (2)

[0081] The direction quaternion is defined as:

[0082]

[0083] where θ is the rotation angle, and (u x ,u y ,u z ) is the unit vector of the rotation axis.

[0084] The complete coordinate transformation chain is expressed as:

[0085] p w =q v (q l p l q l -1 )q v -1 +p v (4)

[0086] where p w is the position vector of the point in the world coordinate system, p l is the position vector of the point in the lidar local coordinate system, p v is the position of the hull in the world coordinate system, q v is the rotation quaternion of the hull relative to the world, and q l is the rotation quaternion of the lidar relative to the hull.

[0087] This transformation chain ensures that the position information is first correctly oriented in the local coordinate system of the lidar, then transformed to the hull coordinate system, and finally mapped to the world coordinate system.

[0088] After determining the starting point of the laser beam, based on the emission direction and starting point position of the rays emitted by the virtual lidar, the field of view of the virtual lidar is determined by scanning coverage in two dimensions, horizontal and vertical. The process of implementing ray tracing for the virtual lidar is as follows:

[0089] Horizontal scan angle α h The calculation formula is:

[0090]

[0091] where i is the horizontal number of the current laser beam, and N h is the horizontal angular resolution;

[0092] Vertical scan angle α v The calculation formula is:

[0093]

[0094] where j is the number in the vertical direction, and α upper and α lower are the upper and lower boundary angles of the vertical field of view respectively.

[0095] Combining the horizontal and vertical angles, the direction vector of the laser beam in the local coordinate system is obtained.

[0096] d = (cosα v cosα h , cosα v sinα h , sinα v ) (7)

[0097] This direction vector also needs to be transformed to the world coordinate system through the coordinate transformation chain introduced earlier before it can be used for ray tracing calculations.

[0098] Through this complete mathematical framework, we can accurately calculate the starting point and direction of each laser beam in the world coordinate system, providing a basis for subsequent collision detection. This quaternion-based method not only ensures the stability of numerical calculations but also provides a reliable mathematical tool for handling complex attitude changes caused by the movement of the hull.

[0099] After determining the starting point and direction of the ray, the next step is to implement collision detection between the ray and the environment. This process involves complex spatial geometric operations and physical simulations and needs to be designed from both theoretical and engineering implementation levels.

[0100] In three-dimensional space, the propagation of a ray can be represented by a parametric equation:

[0101] P(t) = P0 + td (8)

[0102] When the ray intersects the surface of an object in the scene, the intersection point needs to satisfy both the ray equation and the object surface equation. Taking the most basic plane collision as an example, the plane equation can be expressed as:

[0103] ax + by + cz + d = 0 (9)

[0104] where (a, b, c) is the normal vector of the plane, and d is the signed distance from the plane to the origin of the coordinate system.

[0105] Substituting the ray equation into the plane equation, the parameter t can be solved. This parameter t represents the distance from the starting point of the ray to the collision point.

[0106] However, this simplified model based on analytic geometry has obvious limitations: First, it cannot handle complex three-dimensional objects because objects in the actual environment are rarely describable by simple plane equations. Second, the computational efficiency is low, as it is necessary to solve equations for each object in the scene. Moreover, it is difficult to simulate the reflection characteristics of materials on the laser.

[0107] To overcome these problems, the present invention adopts a ray tracing method and a collision detection method to obtain point cloud data.

[0108] A hierarchical structure based on an axis-aligned bounding box (AABB) is constructed for the virtual unmanned ship built, and the scene space is recursively divided into multiple sub-regions from top to bottom. First, a compact basic bounding box is constructed for all objects, and then the space is divided into a hierarchical structure through an octree algorithm, and each node stores the index of the objects it contains. When the number of objects in a node reaches a threshold or the recursion reaches a preset depth, a leaf node is formed. In this way, the system can quickly determine which objects the ray may collide with, avoiding the computational overhead of performing collision detection with all scene objects.

[0109] For each ray, the system recursively detects the intersection with the AABB hierarchical structure starting from the root node. When the ray does not intersect the bounding box of a certain node, the node and all its child nodes are directly skipped; if there is an intersection, its child nodes are continuously detected until a leaf node is reached. In the leaf node, the system only performs an accurate ray-triangle intersection algorithm on the small number of objects contained to calculate the specific collision point. This top-down hierarchical detection strategy significantly reduces the number of objects that need to perform accurate collision calculations, and significantly improves the efficiency of point cloud generation.

[0110] This method further includes:

[0111] When performing collision detection, a Hit Record data structure is used to store and process collision information, record the precise three-dimensional coordinates of the collision point, calculate the normal vector of the collision point, and obtain the physical material properties of the collision surface;

[0112] When the ray detects a collision, a point cloud data point will be generated, including the precise position of the collision point in the world coordinate system, the precise distance from the lidar emission point to the collision point, and the reflection intensity calculated based on the physical material properties and the incident angle.

[0113] This method also includes:

[0114] For each generated point cloud data point, the distance is converted into the specific measurement unit of the lidar, timestamp information is added, and the data is organized into the standard point cloud data format. The lidar modifies the generated simulated point cloud data according to the corresponding real data format, and the data formats of each model of radar are different.

[0115] Before calculating the position and angle of the laser emission, all laser beams are traversed, and the distance from the emission starting point of each laser beam to the collision point is calculated, that is, the distance between the virtual lidar and the objects in the virtual environment. During each scan process, the position information of the current hull is obtained and used in real time as the starting point position of the laser beam to ensure the accuracy of the starting point position for the next scan. In the real scenario, the lidar is fixed on the hull, and as the hull moves, the position of the lidar will be updated in real time. Therefore, in the virtual simulation environment, the position of the virtual lidar will also be updated accordingly. Through this simulation method of the lidar, the distance information between the virtual lidar and the virtual objects can be obtained in real time. When the hull or the environment in the scene changes, the lidar can perceive it in real time and provide accurate simulation data support for the navigation and obstacle avoidance of the hull.

[0116] Point cloud data visualization is to communicate the collision point data (i.e., point cloud data) obtained from collision detection with the visualization software through the UDP protocol, and the specific scanning effect can be seen in the visualization software;

[0117] Using the basic process of running the simulation software, the scanning of the lidar also updates the data every frame. During the process of the ship moving forward, the distance to the obstacles will also change, which serves the purpose of perceiving the surrounding environment in real time.

[0118] Figure 3 For the visualization result of the lidar virtual point cloud;

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention 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 described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An inversion method for a lidar sensor, characterized in that: It includes the following steps: Build a virtual unmanned ship collision detection system; Adopt the quaternion calculation method to convert the measurements in the local coordinate system of the virtual lidar in the virtual unmanned ship collision detection system to the world coordinate system, so as to determine the emission direction and starting position of the rays emitted by the virtual lidar; Based on the emission direction and starting position of the rays emitted by the virtual lidar, determine the virtual lidar field of view through scanning coverage in two dimensions, horizontal and vertical, to achieve ray tracing of the virtual lidar; Perform spatial partitioning on the built virtual unmanned ship collision detection system; Check the intersection of the rays and the bounding box, filter out the areas where collisions may occur, and for the areas where collisions may occur, check whether the rays will collide with the objects in the area for detection; When the rays detect a collision, generate point cloud data points to achieve the inversion calculation of the lidar sensor.

2. The inversion method of a lidar sensor according to claim 1, characterized in that: The process of adopting the quaternion calculation method to convert the measurements in the local coordinate system of the virtual lidar in the virtual unmanned ship collision detection system to the world coordinate system, so as to determine the emission direction and starting position of the rays emitted by the virtual lidar is as follows: Each laser beam of the virtual lidar can be regarded as a ray, and its propagation process is described by the ray equation: P(t) = P0 + td (1) where P(t) represents any point on the ray, P0 is the starting point of the ray, t is the ray parameter, and d is the direction vector of the ray; In the local coordinate system, the position vector is expressed as: P l = (x l , y l , z l ) (2) The direction quaternion is defined as: where θ is the rotation angle, (u x , u y , u z ) is the unit vector of the rotation axis. The complete coordinate transformation chain is expressed as: p w = q v (q l p l q l -1 )q v -1 + p v (4) Among them, p w is the position vector of the point in the world coordinate system, p l is the position vector of the point in the local coordinate system of the lidar, p v is the position of the hull in the world coordinate system, q v is the rotation quaternion of the hull relative to the world, q l is the rotation quaternion of the lidar relative to the hull.

3. The inversion method of a lidar sensor according to claim 1, wherein: The process of determining the virtual lidar field of view through scanning coverage in two dimensions, horizontal and vertical, based on the emission direction and starting position of the rays emitted by the virtual lidar to achieve ray tracing of the virtual lidar is as follows: Horizontal scanning angle α h The calculation formula is as follows: where i is the horizontal number of the current laser beam, and N h is the horizontal angular resolution; Vertical scanning angle α v The calculation formula is as follows: where j is the number in the vertical direction, α upper and α lower are the upper and lower boundary angles of the vertical field of view, respectively. Combine the horizontal and vertical angles to obtain the direction vector of the laser beam in the local coordinate system. d = (cosα v cosα h , cosα v sinα h , sinα v ) (7).

4. The inversion method of a lidar sensor according to claim 1, characterized in that: The process of building the virtual unmanned ship collision detection system is as follows: S11: Use 3D modeling software to build a model including an unmanned ship, lidar, island and buoy, and convert the format to FBX format and export it through 3ds Max software; S12: Import the FBX format model into Unreal Engine software, and build a virtual test environment consistent with the real test environment, and add collision boxes to the models in the virtual test environment; S13: Set the parameters of the lidar, including the number of channels, horizontal rotation angular velocity, field of view, measurement range, and generate the emission angles of multiple laser beams of the lidar through these parameters to cover the entire field of view.

5. An inversion device for a lidar sensor, characterized in that: It includes: A building module: used to build a virtual unmanned ship collision detection system; A determination module: used to adopt the quaternion calculation method to convert the measurements in the local coordinate system of the virtual lidar in the virtual unmanned ship collision detection system to the world coordinate system, so as to determine the emission direction and starting position of the rays emitted by the virtual lidar; Ray tracing module: It is used to determine the virtual lidar field of view range based on the emission direction and starting position of the rays emitted by the virtual lidar, and realize the ray tracing of the virtual lidar through the scanning coverage in two dimensions, horizontal and vertical; Partitioning module: It is used to partition the space of the built virtual unmanned ship collision detection system; Collision detection module: It is used to check the intersection of the rays and the bounding box, screen out the areas where collisions may occur, and for the areas where collisions may occur, check whether the rays will collide with the objects in the area; Generation module: It is used to generate point cloud data points when the rays detect a collision, and realize the inversion calculation of the lidar sensor.

6. A computer device, comprising: A processor and a memory, the memory stores program modules, and is characterized in that the program modules run on the processor to implement the method according to any one of claims 1-4.

7. A readable storage medium stores program modules, characterized in that, The program modules can run on the processor to implement the method according to any one of claims 1-4.

Images