Map generation method of driving test robot in driving test scene

By building sensor transformation matrix and generating electronic fence files, the problem of incomplete map construction in the existing technology is solved, and the generation of global and local maps of driving test robots is realized, and intelligent and secure navigation and path planning is supported, which is suitable for standard intelligence Iterative optimization of driving framework.

CN118936454APending Publication Date: 2024-11-12YIXIAN INTELLIGENCE
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411146311.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

In the prior art, there are fewer map construction methods for driving test scenarios and cannot meet the safety and intelligence requirements. They cannot generate a complete navigation map and cannot connect with the standard intelligent driving framework, which limits the sustainability of the system architecture. Iterative optimization.

Method used

By constructing a static transformation matrix and dynamic transformation matrix between sensors of the driving test robot, electronic fence files and point cloud files are generated, global and local raster maps are spliced, and global and local map generation is achieved using ROS function packages and mapserver.

Benefits of technology

The generated global and local grid maps can be used for global path planning and navigation, ensuring the intelligence and safety of the vehicle, being able to connect with the standard intelligent driving framework, and supporting the continuous iterative optimization of the system architecture.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118936454A_ABST
    Figure CN118936454A_ABST
Patent Text Reader

Abstract

The invention discloses a map generation method for a driving test robot in a driving test scene. Comprising the following steps: constructing a static conversion matrix among sensors, constructing a dynamic conversion matrix of a coordinate system gps sensor relative to a map coordinate system, making an electronic fence file of a driving test scene, analyzing and issuing an electronic fence point cloud file, constructing a global grid map of the driving test scene, and splicing local point cloud of a driving test robot. And constructing a local grid map of the driving test robot. The method disclosed by the invention is based on the electronic fence and the real-time sensor, the generated global grid map can be used for global path planning and navigation algorithms, the local grid map can be used for vehicle collision detection, speed planning, obstacle avoidance algorithms and the like, and the method has the characteristics of intelligence, safety and continuous iterative optimization based on a standard intelligent driving framework.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving, and more particularly to a map generation method for a driving test robot in a driving test scenario. Background Art

[0002] With the rise of artificial intelligence and autonomous driving technology, the application of driving test robots or robot coaches in driving test scenarios is becoming more and more popular. In this scenario, a global map is needed to achieve reasonable planning of the subjects and driving routes of the driving test robot, and a local map is needed to achieve timely braking and obstacle avoidance of the driving test robot. The global map and local map can help the driving test robot better understand the panoramic information of the driving test site and the surrounding environment of the vehicle, which plays a vital role in the intelligence and safety of the driving test robot.

[0003] However, there are few map construction methods that can be used for driving test scenarios in the prior art and they cannot meet the requirements of safety and intelligence. For example, some methods use key point collection and set the passable relationship between points to construct the topological structure of the driving test scene, but this method is relatively primitive and cannot fully describe the real environment of the scene. When the topological relationship is not constructed locally, navigation cannot be generated. Regarding safety strategies, the speed planning of driving test scenarios in the prior art makes decisions for a single sensor separately, and it is impossible to unify the description of the entire perception system. In addition, the local environment map is missing, and local path planning cannot be achieved, and only braking can be performed. In addition, the methods in the prior art cannot be connected with standard intelligent driving frameworks including ros-navigation or autoware, which greatly inhibits the sustainable iterative optimization of the system architecture. Summary of the invention

[0004] In order to solve one of the problems existing in the prior art, the present invention provides a map generation method for a driving test robot in a driving test scenario, comprising the following steps:

[0005] S1: constructing a static conversion matrix between the sensors of the driving test robot;

[0006] S2: Construct a dynamic transformation matrix of the GPS sensor coordinate system of the driving test robot relative to the map coordinate system;

[0007] S3: Creating an electronic fence file for the driving test scene where the driving test robot is located;

[0008] S4: parse and publish the electronic fence point cloud file of the driving test scene where the driving test robot is located;

[0009] S5: constructing a global grid map of the driving test scene where the driving test robot is located;

[0010] S6: stitching the local point cloud of the driving test robot;

[0011] S7: Construct a local grid map of the driving test robot.

[0012] Specifically, the sensors of the driving test robot include a vehicle ego sensor, a forward single-line laser radar sensor, a rearward single-line laser radar sensor, a millimeter-wave radar sensor, and a coordinate system GPS sensor;

[0013] The specific steps of step S1 are:

[0014] S11: Obtain the external parameters between any two sensors of the driving test robot, wherein the external parameters include x, y, z, roll, pitch, and yaw; wherein x represents the relative translation of the two sensors in the x direction, y represents the relative translation of the two sensors in the y direction, z represents the relative translation of the two sensors in the z direction, roll represents the relative roll angle between the two sensors on the x axis, pitch represents the relative pitch angle between the two sensors on the y axis, and yaw represents the relative yaw angle between the two sensors on the z axis;

[0015] S12: Use the function package in ROS to configure the static transformation matrix between any two sensors in step S11, where the static transformation matrix includes three translations and quaternions from one sensor to another sensor.

[0016] Specifically, the step S2 specifically includes the following steps:

[0017] S21: Obtain the real-time dynamic position and posture of the coordinate system GPS sensor of the driving test robot relative to the map coordinate system, wherein the position and posture include x0, y0, z0, roll 0, pitch0, and yaw0, wherein x0 represents the relative translation of the coordinate system GPS sensor relative to the map coordinate system in the x direction, y0 represents the relative translation of the coordinate system GPS sensor relative to the map coordinate system in the y direction, and z0 represents the relative translation of the coordinate system GPS sensor relative to the map coordinate system in the z direction; roll 0 represents the relative roll angle of the coordinate system GPS sensor on the x-axis relative to the map coordinate system; pitch0 represents the relative pitch angle of the coordinate system GPS sensor on the y-axis relative to the map coordinate system; and yaw0 represents the relative yaw angle of the coordinate system GPS sensor on the z-axis relative to the map coordinate system;

[0018] S22: Use the ROS function package to configure the dynamic transformation matrix of the coordinate system GPS sensor of the driving test robot in step S21 relative to the map coordinate system, and publish the dynamic transformation matrix.

[0019] Specifically, step S3 includes the following steps:

[0020] S31: Draw a high-precision point cloud map of the driving school scene where the driving test robot is located by using a laser radar or IMU or RTK fusion mapping method;

[0021] S32: Importing the high-precision point cloud map formed in step S31 into the control system of the driving test robot;

[0022] S33: Draw the driving boundary of the driving school scene where the driving test robot is located, and generate an electronic fence file of the driving test scene where the driving test robot is located.

[0023] Specifically, step S4 includes the following steps:

[0024] S41: parsing the electronic fence file using the functional module of ROS, interpolating according to a given resolution, and generating point cloud points of the entire scene;

[0025] S42: Publish the geo-fence point cloud file according to the format set in ROS.

[0026] Specifically, step S5 includes the following steps:

[0027] S51: using the function module of ROS to convert the point cloud drawing in the electronic fence point cloud file into a point cloud file of a target format;

[0028] S52: using the function module of ROS to convert the point cloud file formed in step S51 into a raster map file;

[0029] S53: Use mapserver to load the raster map file formed in step S52 and form a global raster map of the driving test scene where the driving test robot is located.

[0030] Specifically, step S6 includes the following steps:

[0031] S61: Subscribe to the point cloud formed by the forward single-line laser radar sensor, the backward single-line laser radar sensor, and the electronic fence point cloud file formed in step S4;

[0032] S62: performing coordinate transformation on the different point clouds obtained in step S61 according to the static transformation matrix and the dynamic transformation matrix formed in step S1, and transferring them to the coordinate system of the vehicle ego sensor;

[0033] S63: Fusing and splicing the point clouds of multiple topics generated in step S62.

[0034] Specifically, step S7 includes the following steps:

[0035] S71: Generate a blank grid map based on the coordinate system of the vehicle ego sensor, and set the length, width, resolution and original coordinates of the blank grid map;

[0036] S72: traversing the fused and spliced ​​point cloud formed in step S63, and calculating the index value of each point cloud point in the grid map formed in step S71;

[0037] S73: Assign the index value of each point cloud point formed in step S72 to a set value, so that there is an obstacle in the grid.

[0038] S74: Publishing the local grid map of the driving test robot.

[0039] The map generation method of the driving test robot in the driving test scenario proposed in the present invention is based on electronic fences and real-time sensors, and can generate a global grid map of the driving test scenario and a local grid map near the driving test robot. The global grid map generated by this method can be used in global path planning and navigation algorithms, and the local grid map can be used in vehicle collision detection, speed planning and acc, and obstacle avoidance algorithms. It is intelligent, safe, and can be continuously iterated and optimized based on a standard intelligent driving framework. At present, all of them have been applied to robot coaches or driving test robots developed by the company where the patent applicant is located. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the specific implementation manner of the present invention, the drawings required for use in the specific implementation manner or the description of the prior art will be briefly introduced below.

[0041] Figure 1 It is a flow chart of the method disclosed in the embodiment of the first aspect of the present invention.

[0042] Figure 2 It is a specific flow chart of the method disclosed in the embodiment of the first aspect of the present invention. DETAILED DESCRIPTION

[0043] The structure and operation of the present invention will be further described in detail below in conjunction with the accompanying drawings. Obviously, the accompanying drawings are provided only for a better understanding of the present invention and should not be construed as limiting the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] In the description of the present invention, it should be noted that, if the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. appear, the orientation or position relationship indicated is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, if the terms "first" and "second" appear, they are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0045] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0046] Example 1

[0047] Combination Figure 1 to Figure 2 As shown, this embodiment discloses a map generation method of a driving test robot in a driving test scenario, which includes the following steps:

[0048] S1: constructing a static conversion matrix between the sensors of the driving test robot;

[0049] S2: Construct a dynamic transformation matrix of the GPS sensor coordinate system of the driving test robot relative to the map coordinate system;

[0050] S3: Creating an electronic fence file for the driving test scene where the driving test robot is located;

[0051] S4: parse and publish the electronic fence point cloud file of the driving test scene where the driving test robot is located;

[0052] S5: constructing a global grid map of the driving test scene where the driving test robot is located;

[0053] S6: stitching the local point cloud of the driving test robot;

[0054] S7: Construct a local grid map of the driving test robot.

[0055] In a specific embodiment, the sensors of the driving test robot include a vehicle ego sensor, a forward single-line laser radar sensor, a backward single-line laser radar sensor, a millimeter-wave radar sensor, and a coordinate system GPS sensor.

[0056] The specific steps of step S1 are:

[0057] S11: Obtain the external parameters between any two sensors of the driving test robot, wherein the external parameters include x, y, z, roll, pitch, and yaw; wherein x represents the relative translation of the two sensors in the x direction, y represents the relative translation of the two sensors in the y direction, z represents the relative translation of the two sensors in the z direction, roll represents the relative roll angle between the two sensors on the x axis, pitch represents the relative pitch angle between the two sensors on the y axis, and yaw represents the relative yaw angle between the two sensors on the z axis;

[0058] S12: Use the function package in ROS to configure the static transformation matrix between any two sensors in step S11, where the static transformation matrix includes three translations and quaternions from one sensor to another sensor.

[0059] ROS (Robot Operating System) is an open source robot operating system framework that can provide support for various functional modules, software algorithms, etc. In a specific embodiment, the tf2_ros function package of ROS can be used to configure the static transformation matrix, and the quaternion in the static transformation matrix is ​​converted from roll, pitch, yaw, etc.

[0060] In a specific embodiment, the step S2 specifically includes the following steps:

[0061] S21: Obtain the real-time dynamic position and posture of the coordinate system GPS sensor of the driving test robot relative to the map coordinate system, wherein the position and posture include x0, y0, z0, roll 0, pitch0, and yaw0, wherein x0 represents the relative translation of the coordinate system GPS sensor relative to the map coordinate system in the x direction, y0 represents the relative translation of the coordinate system GPS sensor relative to the map coordinate system in the y direction, and z0 represents the relative translation of the coordinate system GPS sensor relative to the map coordinate system in the z direction; roll 0 represents the relative roll angle of the coordinate system GPS sensor on the x-axis relative to the map coordinate system; pitch0 represents the relative pitch angle of the coordinate system GPS sensor on the y-axis relative to the map coordinate system; and yaw0 represents the relative yaw angle of the coordinate system GPS sensor on the z-axis relative to the map coordinate system;

[0062] S22: Use the ROS function package to configure the dynamic transformation matrix of the coordinate system GPS sensor of the driving test robot in step S21 relative to the map coordinate system, and publish the dynamic transformation matrix.

[0063] Among them, the map coordinate system refers to the map coordinate system, which is a reference frame that describes the position and direction of various elements in the map (such as roads, landmarks, obstacles, etc.). The map coordinate system in this embodiment is a geographic coordinate system based on longitude and latitude or a customized local coordinate system. The coordinate system GPS sensor of the driving test robot can provide the position (x0, y0, z0, roll 0, pitch 0, yaw0) of the driving test robot in the map coordinate system in real time based on RTK technology or laser radar odometer.

[0064] Specifically, you can use the TransfromBroadcaster tool of ROS to publish a real-time dynamic transformation matrix.

[0065] In a specific embodiment, step S3 specifically includes the following steps:

[0066] S31: Draw a high-precision point cloud map of the driving school scene where the driving test robot is located by using a laser radar or IMU or RTK fusion mapping method;

[0067] S32: Importing the high-precision point cloud map formed in step S31 into the control system of the driving test robot;

[0068] S33: Draw the driving boundary of the driving school scene where the driving test robot is located, and generate an electronic fence file of the driving test scene where the driving test robot is located.

[0069] The specific steps of the fusion mapping method used in step S31 are as follows:

[0070] 1: Subscribe to the point cloud topic and use the first frame of point cloud as the initial point cloud map;

[0071] 2: Perform voxel filtering on each frame of the subscribed point cloud to achieve downsampling;

[0072] 3: For each frame of point cloud, set the NDT operator, use the NDT algorithm and point cloud map for registration, calculate the movement pose of the point cloud relative to the point cloud map, and solve the transformation matrix;

[0073] 4: If the moving distance of the frame point cloud relative to the map is greater than a certain value, the frame point cloud is transformed using the transformation matrix and spliced ​​with the point cloud map to generate a new point cloud map.

[0074] The method used in step S32 may use the secondary developed autoware adasmap or other prior art methods.

[0075] The electronic fence file generated in step S33 may be in a custom format.

[0076] In a specific embodiment, step S4 specifically includes the following steps:

[0077] S41: parsing the electronic fence file using the functional module of ROS, interpolating according to a given resolution, and generating point cloud points of the entire scene;

[0078] S42: Publish the geo-fence point cloud file according to the format set in ROS.

[0079] Specifically, in step S41, the tinyxml module of ROS can be used to parse the electronic fence file, and in step S42, the electronic fence point cloud file can be published according to the format of sensor_msgs::Pointl oud2 of ROS.

[0080] In a specific embodiment, step S5 specifically includes the following steps:

[0081] S51: using the function module of ROS to convert the point cloud drawing in the electronic fence point cloud file into a point cloud file of a target format;

[0082] S52: using the function module of ROS to convert the point cloud file formed in step S51 into a raster map file;

[0083] S53: Use mapserver to load the raster map file formed in step S52 and form a global raster map of the driving test scene where the driving test robot is located.

[0084] Specifically, in step S51, the point cloud drawing in the electronic fence point cloud file can be converted into a pcd point cloud file using the point cloud_to_pcd function of pc l_ros; in step S52, the pcd point cloud file is converted into a raster map file using the pcd2pgm function module of ROS; the raster map is similar to the pixel format of an image, and is projected onto the raster map according to the point cloud coordinates. If the corresponding grid contains a point cloud, the value of the grid is set to 100, otherwise it is set to 0, thereby obtaining the obstacle occupancy status of the map.

[0085] In a specific embodiment, step S6 specifically includes the following steps:

[0086] S61: Subscribe to the point cloud formed by the forward single-line laser radar sensor, the backward single-line laser radar sensor, and the electronic fence point cloud file formed in step S4;

[0087] S62: performing coordinate transformation on the different point clouds obtained in step S61 according to the static transformation matrix and the dynamic transformation matrix formed in step S1, and transferring them to the coordinate system of the vehicle ego sensor;

[0088] S63: The point clouds of multiple topics generated in step S62 are merged and spliced. After the coordinates of the electronic fence point cloud are converted, the point cloud is filtered using the direct filtering of the PCL point cloud library, and only the point cloud within a certain range of the vehicle is retained for splicing.

[0089] In a specific embodiment, step S7 specifically includes the following steps:

[0090] S71: Generate a blank grid map based on the coordinate system of the vehicle ego sensor, and set the length, width, resolution and original coordinates of the blank grid map;

[0091] S72: traversing the fused and spliced ​​point cloud formed in step S63, and calculating the index value of each point cloud point in the grid map formed in step S71;

[0092] S73: Assign the index value of each point cloud point formed in step S72 to a set value, so that there is an obstacle in the grid.

[0093] S74: Publish the local grid map of the driving test robot.

[0094] Specifically, in step S73, the index value is assigned 100, and it is considered that the grid has an obstacle.

[0095] The present invention can achieve the following technical effects:

[0096] The map generation method of the driving test robot in the driving test scenario proposed in the present invention is based on electronic fences and real-time sensors, and can generate a global grid map of the driving test scenario and a local grid map near the driving test robot. The global grid map generated by this method can be used in global path planning and navigation algorithms, and the local grid map can be used in vehicle collision detection, speed planning and acc, and obstacle avoidance algorithms. It is intelligent, safe, and can be continuously iterated and optimized based on a standard intelligent driving framework. At present, all of them have been applied to robot coaches or driving test robots developed by the company where the patent applicant is located.

[0097] The above embodiments are only used to illustrate the present invention, wherein the structure, connection method and manufacturing process of each component may be changed. Any equivalent transformations and improvements based on this technical solution should not be excluded from the protection scope of the present invention.

Claims

1. A method for generating a map of a driving test robot in a driving test scenario, characterized in that: The steps include: S1: constructing a static conversion matrix between the sensors of the driving test robot; S2: Construct a dynamic transformation matrix of the GPS sensor coordinate system of the driving test robot relative to the map coordinate system; S3: Creating an electronic fence file for the driving test scene where the driving test robot is located; S4: parse and publish the electronic fence point cloud file of the driving test scene where the driving test robot is located; S5: constructing a global grid map of the driving test scene where the driving test robot is located; S6: stitching the local point cloud of the driving test robot; S7: Construct a local grid map of the driving test robot.

2. The method according to claim 1, characterized in that: The sensors of the driving test robot include a vehicle ego sensor, a forward single-line laser radar sensor, a rearward single-line laser radar sensor, a millimeter-wave radar sensor, and a coordinate system GPS sensor; The specific steps of step S1 are: S11: Obtain the external parameters between any two sensors of the driving test robot, wherein the external parameters include x, y, z, roll, pitch, and yaw; wherein x represents the relative translation of the two sensors in the x direction, y represents the relative translation of the two sensors in the y direction, z represents the relative translation of the two sensors in the z direction, roll represents the relative roll angle between the two sensors on the x axis, pitch represents the relative pitch angle between the two sensors on the y axis, and yaw represents the relative yaw angle between the two sensors on the z axis; S12: Use the function package in ROS to configure the static transformation matrix between any two sensors in step S11, where the static transformation matrix includes three translations and quaternions from one sensor to another sensor.

3. The method according to claim 1, characterized in that: The step S2 specifically includes the following steps: S21: Obtain the real-time dynamic position and posture of the coordinate system GPS sensor of the driving test robot relative to the map coordinate system, the position and posture include x0, y0, z0, roll 0, pitch0, yaw0, the x0 represents the relative translation of the coordinate system GPS sensor relative to the map coordinate system in the x direction, the y0 represents the relative translation of the coordinate system GPS sensor relative to the map coordinate system in the y direction, the z0 represents the relative translation of the coordinate system GPS sensor relative to the map coordinate system in the z direction; the roll0 represents the relative roll angle of the coordinate system GPS sensor on the x-axis relative to the map coordinate system; the pitch0 represents the relative pitch angle of the coordinate system GPS sensor on the y-axis relative to the map coordinate system; the yaw0 represents the relative yaw angle of the coordinate system GPS sensor on the z-axis relative to the map coordinate system; S22: Use the ROS function package to configure the dynamic transformation matrix of the coordinate system GPS sensor of the driving test robot in step S21 relative to the map coordinate system, and publish the dynamic transformation matrix.

4. The method according to claim 1, characterized in that: The step S3 specifically includes the following steps: S31: Draw a high-precision point cloud map of the driving school scene where the driving test robot is located by using a laser radar or IMU or RTK fusion mapping method; S32: Importing the high-precision point cloud map formed in step S31 into the control system of the driving test robot; S33: Draw the driving boundary of the driving school scene where the driving test robot is located, and generate an electronic fence file of the driving test scene where the driving test robot is located.

5. The method according to claim 1, characterized in that: The step S4 specifically includes the following steps: S41: parsing the electronic fence file using the functional module of ROS, interpolating according to a given resolution, and generating point cloud points of the entire scene; S42: Publish the geo-fence point cloud file according to the format set in ROS.

6. The method according to claim 1, characterized in that: The step S5 specifically includes the following steps: S51: using the function module of ROS to convert the point cloud drawing in the electronic fence point cloud file into a point cloud file of a target format; S52: using the function module of ROS to convert the point cloud file formed in step S51 into a raster map file; S53: Use mapserver to load the raster map file formed in step S52 and form a global raster map of the driving test scene where the driving test robot is located.

7. The method according to claim 1, characterized in that: The step S6 specifically includes the following steps: S61: Subscribe to the point cloud formed by the forward single-line laser radar sensor, the backward single-line laser radar sensor, and the electronic fence point cloud file formed in step S4; S62: performing coordinate transformation on the different point clouds obtained in step S61 according to the static transformation matrix and the dynamic transformation matrix formed in step S1, and transferring them to the coordinate system of the vehicle ego sensor; S63: Fusing and splicing the point clouds of multiple topics generated in step S62.

8. The method according to claim 1, characterized in that: The step S7 specifically includes the following steps: S71: Generate a blank grid map based on the coordinate system of the vehicle ego sensor, and set the length, width, resolution and original coordinates of the blank grid map; S72: traversing the fused and spliced ​​point cloud formed in step S63, and calculating the index value of each point cloud point in the grid map formed in step S71; S73: Assign the index value of each point cloud point formed in step S72 to a set value, so that there is an obstacle in the grid. S74: Publish the local grid map of the driving test robot.