Moving obstacle identification method based on laser radar and terminal

Abstract

The invention discloses a moving obstacle identification method based on a laser radar and a terminal, the method comprises the following steps: driving the laser radar to scan a target direction, determining whether there is an obstacle in the target direction, and if yes, establishing a relative motion coordinate system by taking the laser radar as a center; the shape and size information of the obstacle and the movement information relative to the laser radar are calculated under the relative motion coordinate system; and according to the shape and size information and the movement information, controlling the mobile carrier where the laser radar is located to move so as to avoid obstacles. According to the method, the moving speed and direction of the obstacle are calculated, whether the collision risk exists or not can be judged more accurately, and compared with a linear distance, finer and more comprehensive obstacle information is provided, so that unmanned driving is safer and more intelligent.

Description

Technical Field

[0001] The present invention relates to the technical field of mobile obstacle recognition, and in particular to a mobile obstacle recognition method and terminal based on laser radar. Background Art

[0002] LiDAR is an important sensor for unmanned driving at this stage. Through LiDAR, the surrounding environment can be measured and an environmental map can be drawn. Obstacles encountered by unmanned vehicles can also be identified through LiDAR.

[0003] As the level of unmanned driving technology becomes higher and higher, the requirements for obstacle recognition are also getting higher and higher; the existing lidar can only identify the distance between the vehicle and the obstacle, and then trigger an alarm when the distance is too close. In actual application scenarios, unmanned vehicles and other obstacles, such as pedestrians and other vehicles, are all in a moving state. Only the relative distance is used to control the movement of the vehicle body. It is easy for the unmanned vehicle to respond untimely and the distance control to be inaccurate. Especially in congested roads and situations where vehicle movement is greatly restricted, it is more necessary to accurately control the vehicle's moving distance. Simply providing the distance between the vehicle and the obstacle cannot meet the usage needs of actual scenarios. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a mobile obstacle recognition method and terminal based on laser radar, provide more detailed and comprehensive obstacle information, and make unmanned driving safer and smarter.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: A method for identifying moving obstacles based on laser radar comprises the following steps: S1, drive the laser radar to scan the target direction, and determine whether there is an obstacle in the target direction. If yes, execute step S2, otherwise scan again; S2. Establishing a relative motion coordinate system with the laser radar as the center; S3, calculating the shape and size information of the obstacle and the movement information relative to the laser radar in the relative motion coordinate system; S4. Control the movement of the mobile vehicle where the laser radar is located according to the shape and size information and the movement information to avoid the obstacle.

[0006] In order to solve the above technical problems, another technical solution adopted by the present invention is: A laser radar-based mobile obstacle identification terminal includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: S1, drive the laser radar to scan the target direction, and determine whether there is an obstacle in the target direction. If yes, execute step S2, otherwise scan again; S2. Establishing a relative motion coordinate system with the laser radar as the center; S3, calculating the shape and size information of the obstacle and the movement information relative to the laser radar in the relative motion coordinate system; S4. Control the movement of the mobile vehicle where the laser radar is located according to the shape and size information and the movement information to avoid the obstacle.

[0007] The beneficial effects of the present invention are: providing a mobile obstacle recognition method and terminal based on laser radar, driving the laser radar to establish a relative motion coordinate system when discovering an obstacle in the target direction near a mobile vehicle, and calculating the shape and size information of the obstacle and the movement information relative to the mobile vehicle based on the system, so as to grasp the moving speed and direction of the obstacle, etc., and can more accurately judge whether there is a collision risk. Compared with the straight-line distance, it provides more detailed and comprehensive obstacle information, making unmanned driving safer and smarter. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A schematic diagram of the steps of a laser radar-based mobile obstacle identification method of the present invention; Figure 2 A schematic diagram of a relative motion coordinate system of a mobile obstacle identification method based on laser radar according to the present invention; Figure 3 A schematic diagram of a vertical point cloud of a mobile obstacle recognition method based on laser radar according to the present invention; Figure 4 The system block diagram of a laser radar-based mobile obstacle recognition terminal of the present invention.

[0009] Description of symbols 1. A mobile obstacle recognition terminal based on laser radar; 2. A processor; 3. A memory. DETAILED DESCRIPTION

[0010] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0011] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0012] Please refer to Figures 1 to 3 , a mobile obstacle recognition method based on laser radar, comprising the following steps: S1, drive the laser radar to scan the target direction, and determine whether there is an obstacle in the target direction. If yes, execute step S2, otherwise scan again; S2. Establishing a relative motion coordinate system with the laser radar as the center; S3, calculating the shape and size information of the obstacle and the movement information relative to the laser radar in the relative motion coordinate system; S4. Control the movement of the mobile vehicle where the laser radar is located according to the shape and size information and the movement information to avoid the obstacle.

[0013] From the above description, it can be seen that the beneficial effect of the present invention is that when the laser radar is driven to find an obstacle in the target direction near the mobile vehicle, a relative motion coordinate system is established, and the shape and size information of the obstacle and the movement information relative to the mobile vehicle are calculated based on this, so as to grasp the movement speed and direction of the obstacle, etc., and can more accurately judge whether there is a collision risk. Compared with the straight-line distance, it provides more detailed and comprehensive obstacle information, making unmanned driving safer and smarter.

[0014] Furthermore, the step S2 is specifically as follows: S21, taking the position of the laser radar as the three-dimensional origin, tiling the point cloud data generated by the laser radar scanning to obtain a laser scanning depth map; S22. Project the laser scanning depth map onto a three-dimensional basic plane based on the three-dimensional origin to obtain the relative motion coordinate system.

[0015] From the above description, it can be seen that by combining the depth map and the projection in the three-dimensional direction, the laser point cloud data is depicted in a three-dimensional coordinate system, so as to obtain a relative motion coordinate system representing the position relationship between the mobile vehicle and the obstacle, thereby facilitating the grasp of the shape and direction of the obstacle.

[0016] Furthermore, the step S3 comprises: S31, generating a vertical point cloud map of the obstacle in the relative motion coordinate system; S32, obtaining an obstacle point cloud cluster surrounding the obstacle on the facade point cloud image by a clustering algorithm; S33, converting the obstacle point cloud cluster into shape and size information of the obstacle by using a minimum bounding box method.

[0017] From the above description, it can be seen that the elevation point cloud map is the distribution point cloud map of the obstacle. The point cloud of the obstacle and its vicinity is traversed by the clustering algorithm, and then the shape and size of the obstacle are concretized as much as possible by relying on the minimum bounding box method, so as to obtain the shape and geometric position of the obstacle more accurately and in detail.

[0018] Furthermore, the step S3 further includes: Calculating the current coordinate information of the obstacle in the relative motion coordinate system according to the shape and size information; Get the position coordinates of the laser radar relative to the world coordinate system (x t ,y t ), the current coordinate information (x, y) of the obstacle in the relative motion coordinate system is rotated and translated to the world coordinate system to obtain the world position coordinates of the obstacle relative to the world coordinate system ( , ), the rotation and translation transformation formula is as follows: ; Among them, angle θ represents the angle between the vehicle body direction and the y-axis of the relative motion coordinate system; Obtain and compare the two world position coordinates of the obstacle before and after a preset time to obtain movement information.

[0019] Furthermore, before step S32, the following steps are also included: After the preset time period, the target direction is rescanned, and the scanning result is matched with the shape and size information of the obstacle to relocate the obstacle.

[0020] From the above description, it can be seen that the shape and size information can be used to re-identify obstacles after they move, which is particularly applicable in some complex environments to avoid inconsistencies between the front and rear targets in the direction grasp and improve the reliability of the vehicle's obstacle avoidance.

[0021] Please refer to Figure 4 A mobile obstacle identification terminal 1 based on laser radar includes a processor 2, a memory 3, and a computer program stored in the memory 3 and executable on the processor 2. When executing the computer program, the processor 2 implements the following steps: S1, drive the laser radar to scan the target direction, and determine whether there is an obstacle in the target direction. If yes, execute step S2, otherwise scan again; S2. Establishing a relative motion coordinate system with the laser radar as the center; S3, calculating the shape and size information of the obstacle and the movement information relative to the laser radar in the relative motion coordinate system; S4. Control the movement of the mobile vehicle where the laser radar is located according to the shape and size information and the movement information to avoid the obstacle.

[0022] From the above description, it can be seen that the beneficial effect of the present invention is that when the laser radar is driven to find an obstacle in the target direction near the mobile vehicle, a relative motion coordinate system is established, and the shape and size information of the obstacle and the movement information relative to the mobile vehicle are calculated based on this, so as to grasp the movement speed and direction of the obstacle, etc., and can more accurately judge whether there is a collision risk. Compared with the straight-line distance, it provides more detailed and comprehensive obstacle information, making unmanned driving safer and smarter.

[0023] Furthermore, the step S2 is specifically as follows: S21, taking the position of the laser radar as the three-dimensional origin, tiling the point cloud data generated by the laser radar scanning to obtain a laser scanning depth map; S22. Project the laser scanning depth map onto a three-dimensional basic plane based on the three-dimensional origin to obtain the relative motion coordinate system.

[0024] From the above description, it can be seen that by combining the depth map and the projection in the three-dimensional direction, the laser point cloud data is depicted in a three-dimensional coordinate system, so as to obtain a relative motion coordinate system representing the position relationship between the mobile vehicle and the obstacle, thereby facilitating the grasp of the shape and direction of the obstacle.

[0025] Furthermore, the step S3 comprises: S31, generating a vertical point cloud map of the obstacle in the relative motion coordinate system; S32, obtaining an obstacle point cloud cluster surrounding the obstacle on the facade point cloud image by a clustering algorithm; S33, converting the obstacle point cloud cluster into shape and size information of the obstacle by using a minimum bounding box method.

[0026] From the above description, it can be seen that the elevation point cloud map is the distribution point cloud map of the obstacle. The point cloud of the obstacle and its vicinity is traversed by the clustering algorithm, and then the shape and size of the obstacle are concretized as much as possible by relying on the minimum bounding box method, so as to obtain the shape and geometric position of the obstacle more accurately and in detail.

[0027] Furthermore, the step S3 further includes: Calculating the current coordinate information of the obstacle in the relative motion coordinate system according to the shape and size information; Get the position coordinates of the laser radar relative to the world coordinate system (x t ,y t ), the current coordinate information (x, y) of the obstacle in the relative motion coordinate system is rotated and translated to the world coordinate system to obtain the world position coordinates of the obstacle relative to the world coordinate system ( , ), the rotation and translation transformation formula is as follows: ; Among them, angle θ represents the angle between the vehicle body direction and the y-axis of the relative motion coordinate system; Obtain and compare the two world position coordinates of the obstacle before and after a preset time to obtain movement information.

[0028] Furthermore, before step S32, the following steps are also included: After the preset time period, the target direction is rescanned, and the scanning result is matched with the shape and size information of the obstacle to relocate the obstacle.

[0029] From the above description, it can be seen that the shape and size information can be used to re-identify obstacles after they move, which is particularly applicable in some complex environments to avoid inconsistencies between the front and rear targets in the direction grasp and improve the reliability of the vehicle's obstacle avoidance.

[0030] Please refer to Figures 1 to 3 , Embodiment 1 of the present invention is: A method for identifying moving obstacles based on laser radar comprises the following steps: S1, drive the laser radar to scan the target direction to determine whether there is an obstacle in the target direction. If so, execute step S2, otherwise scan again; In this embodiment, the laser radar can adopt a 32-line laser radar with many beams and strong obstacle height resolution capability.

[0031] S2. Establish a relative motion coordinate system with the laser radar as the center; Wherein, step S2 is specifically as follows: S21, taking the position of the laser radar as the three-dimensional origin, tiling the point cloud data generated by the laser radar scanning to obtain a laser scanning depth map; In this embodiment, if Figure 2 As shown in the figure, after receiving a frame of point cloud data from the laser radar, the point cloud data is processed, and each point data is used as the (x, y, z) coordinate relative to the laser radar, and the laser point cloud can be tiled into a depth map.

[0032] S22. Project the laser scanning depth map onto a three-dimensional basic plane based on the three-dimensional origin to obtain a relative motion coordinate system, as shown in the figure. By projecting the xoy plane, the angle between the line connecting the (x, y) point and the origin and the positive direction of the y-axis is the laser scanning angle. Taking the laser scanning resolution of 0.02 degrees as an example, the laser scanning image is unfolded, and the image scanned from 0 degrees to 360 degrees has a horizontal coordinate of 0 to 360 degrees, with a total of 1800 data columns, and the vertical coordinate is the laser scanning beam, xoz plane, the angle between the (x, z) point and the x-axis, and then divided by the laser radar column resolution, for example 1 degree, the vertical coordinate is 0~31 rows, with a total of 32 data rows, and the depth is the distance between the (x, y, z) point and the origin.

[0033] S3, calculating the shape and size information of the obstacle and the movement information relative to the moving vehicle in the relative motion coordinate system; In this embodiment, step S3 includes: S31, generating a vertical point cloud image of the obstacle in a relative motion coordinate system; In this embodiment, if Figure 2 and Figure 3 As shown, the horizontal distance and height difference between two points in the same column form a triangle. Through the relationship of trigonometric functions, the height of the opposite side divided by the horizontal distance gives the tangent value of the angle θ, and then θ can be calculated through the inverse trigonometric function: ; The smaller the angle θ is, the closer the two points are to the plane, and the larger the angle is, the closer they are to the facade. By calculating column by column and marking the plane point cloud and non-plane point cloud, the depth map can be decomposed into two images, one is the plane point cloud map, and the other is the facade point cloud map. The facade point cloud map is the obstacle distribution point cloud map.

[0034] S32. Obtain the obstacle point cloud cluster that surrounds the obstacle on the facade point cloud map through the clustering algorithm. The operation process of the clustering algorithm is as follows: traverse the facade point cloud map, select unmarked points as the starting point of the traversal queue, traverse the coordinate points around the obstacle, add the untraversed points to the queue to be traversed, and skip the marked points. In this way, gradually expand outward from one point until there are no points around or the points have been marked, that is, the entire obstacle, the queue traversal is completed, and a cluster of obstacle point clouds is obtained. Then re-select unmarked points in the obstacle distribution point cloud as the starting point of the second queue to continue traversal. In this way, continuous points in the same area are marked as an obstacle point cloud cluster. After traversing all points, multiple obstacle point cloud clusters are obtained, that is, the obstacle distribution of the current frame is obtained.

[0035] S33. The obstacle point cloud cluster is converted into the shape and size information of the obstacle through the minimum bounding box method.

[0036] The obstacle point cloud cluster can be obtained by the minimum bounding box method, which can be concretized into rectangular size information and rectangular geometric center position information. Thus, the relative vehicle body position (obstacle lidar) and obstacle position information are obtained. Among them, the bounding box is also called the minimum external rectangle, which is an algorithm for solving the optimal enclosing space of a discrete point set. The basic idea is to use a slightly larger and simpler geometric body (called a bounding box) to approximately replace complex geometric objects.

[0037] S34, calculating the current coordinate information of the obstacle in the relative motion coordinate system according to the shape and size information; In this embodiment, the position coordinates (x t ,y t ), the world coordinate system can usually be taken from the satellite positioning coordinate system or the initial position coordinate system relative to the laser radar mapping. The current coordinate information (x, y) of the obstacle in the relative motion coordinate system is rotated and translated to the world coordinate system to obtain the world position coordinates (x', y') of the obstacle relative to the world coordinate system. The rotation and translation transformation formula is as follows: .

[0038] The angle θ represents the angle between the vehicle body direction and the y-axis of the relative motion coordinate system.

[0039] Obtain and compare the two world position coordinates of the obstacle before and after the preset time to obtain movement information.

[0040] Combined with the above content, when the same obstacle is matched, the obstacle movement speed and direction are calculated according to the geometric center positions of the two obstacles in front and behind, that is, the difference in world position coordinates and the preset duration, so as to obtain the movement information of the obstacle. Specifically, the continuous multi-frame method can be used to improve data stability and multiple comparisons to judge the movement of the obstacle.

[0041] S4. Control the movement of the mobile vehicle where the laser radar is located according to the shape and size information and the movement information to avoid obstacles.

[0042] After understanding the movement of obstacles, the mobile vehicle can more clearly judge how to move itself and better avoid obstacles.

[0043] Please refer to Figure 4 , Embodiment 2 of the present invention is: A laser radar-based mobile obstacle identification terminal 1, based on the above-mentioned embodiment one, includes a processor 2, a memory 3, and a computer program stored in the memory 3 and executable on the processor 2. When executing the computer program, the processor 2 implements a laser radar-based mobile obstacle identification method of embodiment one.

[0044] In summary, the present invention discloses a mobile obstacle recognition method and terminal based on laser radar. When the laser radar is driven to find an obstacle in the target direction near a mobile vehicle, a relative motion coordinate system is established. Based on this, a clustering algorithm is used to traverse the point cloud of the obstacle and its vicinity, and then the minimum bounding box method is used to concretize the outer dimensions of the obstacle as much as possible. The position change of the obstacle is accurately grasped through the coordinate change after a preset time, so that the moving direction and speed of the obstacle can be obtained, thereby mastering the moving speed and direction of the obstacle, etc., and can more accurately judge whether there is a collision risk. Compared with the straight-line distance, more detailed and comprehensive obstacle information is provided, making unmanned driving safer and smarter.

[0045] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for identifying mobile obstacles based on laser radar, characterized in that: The steps include: S1, drive the laser radar to scan the target direction, and determine whether there is an obstacle in the target direction. If yes, execute step S2, otherwise scan again; S2. Establishing a relative motion coordinate system with the laser radar as the center; S3, calculating the shape and size information of the obstacle and the movement information relative to the laser radar in the relative motion coordinate system; S4. Control the movement of the mobile vehicle where the laser radar is located according to the shape and size information and the movement information to avoid the obstacle.

2. The method for identifying mobile obstacles based on laser radar according to claim 1, characterized in that: The step S2 is specifically as follows: S21, taking the position of the laser radar as the three-dimensional origin, tiling the point cloud data generated by the laser radar scanning to obtain a laser scanning depth map; S22. Project the laser scanning depth map onto a three-dimensional basic plane based on the three-dimensional origin to obtain the relative motion coordinate system.

3. The method for identifying mobile obstacles based on laser radar according to claim 2, characterized in that: The step S3 comprises: S31, generating a vertical point cloud map of the obstacle in the relative motion coordinate system; S32, obtaining an obstacle point cloud cluster surrounding the obstacle on the facade point cloud image by a clustering algorithm; S33, converting the obstacle point cloud cluster into shape and size information of the obstacle by using a minimum bounding box method.

4. The method for identifying mobile obstacles based on laser radar according to claim 2, characterized in that: The step S3 further comprises: Calculating the current coordinate information of the obstacle in the relative motion coordinate system according to the shape and size information; Get the position coordinates of the laser radar relative to the world coordinate system (x t ,y t ), the current coordinate information (x, y) of the obstacle in the relative motion coordinate system is rotated and translated to the world coordinate system to obtain the world position coordinates of the obstacle relative to the world coordinate system ( , ), the rotation and translation transformation formula is as follows: ; Among them, the angle θ represents the angle between the vehicle body direction and the y-axis of the relative motion coordinate system; Obtain and compare the two world position coordinates of the obstacle before and after a preset time to obtain movement information.

5. The method for identifying moving obstacles based on laser radar according to claim 4, characterized in that: Before step S32, the following steps are also included: After the preset time period, the target direction is rescanned, and the scanning result is matched with the shape and size information of the obstacle to relocate the obstacle.

6. A mobile obstacle recognition terminal based on laser radar, characterized in that: The invention comprises a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: S1, drive the laser radar to scan the target direction, and determine whether there is an obstacle in the target direction. If yes, execute step S2, otherwise scan again; S2. Establishing a relative motion coordinate system with the laser radar as the center; S3, calculating the shape and size information of the obstacle and the movement information relative to the laser radar in the relative motion coordinate system; S4. Control the movement of the mobile vehicle where the laser radar is located according to the shape and size information and the movement information to avoid the obstacle.

7. The laser radar-based mobile obstacle identification terminal according to claim 6, characterized in that: The step S2 is specifically as follows: S21, taking the position of the laser radar as the three-dimensional origin, tiling the point cloud data generated by the laser radar scanning to obtain a laser scanning depth map; S22. Project the laser scanning depth map onto a three-dimensional basic plane based on the three-dimensional origin to obtain the relative motion coordinate system.

8. The laser radar-based mobile obstacle identification terminal according to claim 7, characterized in that: The step S3 comprises: S31, generating a vertical point cloud map of the obstacle in the relative motion coordinate system; S32, obtaining an obstacle point cloud cluster surrounding the obstacle on the facade point cloud image by a clustering algorithm; S33, converting the obstacle point cloud cluster into shape and size information of the obstacle by using a minimum bounding box method.

9. The laser radar-based mobile obstacle identification terminal according to claim 7, characterized in that: The step S3 further comprises: Calculating the current coordinate information of the obstacle in the relative motion coordinate system according to the shape and size information; Get the position coordinates of the laser radar relative to the world coordinate system (x t ,y t ), the current coordinate information (x, y) of the obstacle in the relative motion coordinate system is rotated and translated to the world coordinate system to obtain the world position coordinates of the obstacle relative to the world coordinate system ( , ), the rotation and translation transformation formula is as follows: ; Among them, the angle θ represents the angle between the vehicle body direction and the y-axis of the relative motion coordinate system; Obtain and compare the two world position coordinates of the obstacle before and after a preset time to obtain movement information.

10. The laser radar-based mobile obstacle identification terminal according to claim 9, characterized in that: Before step S32, the following steps are also included: After the preset time period, the target direction is rescanned, and the scanning result is matched with the shape and size information of the obstacle to relocate the obstacle.

Images