Efficient obstacle recognition method and terminal based on laser radar

Abstract

The invention discloses a laser radar-based efficient obstacle recognition method and terminal, and the method comprises the steps: carrying out the scanning through a laser radar with a preset number of wire harnesses, and obtaining the laser point cloud data in a preset range in front of a vehicle; filtering ground points in the laser point cloud data, and only retaining the ground points; with the laser radar as the center and the preset angle as the unit, a preset range in front of the vehicle is divided into fan-shaped areas; for each fan-shaped region, calculating a spatial distance between a point cloud point in the fan-shaped region and the laser radar, classifying the point cloud point according to the spatial distance, and identifying an obstacle according to a classification result; according to the method, operation is simplified, the obstacle recognition efficiency is improved, the hardware requirement is reduced, and the refresh rate stability of obstacle recognition is improved, so that the requirement for rapidness of obstacle emergency processing is met.

Description

Technical Field

[0001] The present invention relates to the field of unmanned driving technology, and in particular to a laser radar-based efficient obstacle recognition method and terminal. Background Art

[0002] LiDAR is undoubtedly an important sensor for unmanned driving at this stage. Through LiDAR, the surrounding environment can be measured. Fixed obstacles can be mapped, and real-time obstacle information can be generated for mobile or temporary obstacles. Existing technologies often use complex classification and clustering algorithms to try to describe and classify specific details such as the outline of obstacles. The result is a huge amount of calculation. In the case of insufficient hardware conditions, it often leads to information lag, affecting the control of the wire-controlled chassis. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a laser radar-based efficient obstacle recognition method and terminal, which can more efficiently recognize obstacles.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is: An efficient obstacle recognition method based on laser radar comprises the following steps: S1. Scanning with a laser radar with a preset number of beams, and obtaining laser point cloud data within a preset range in front of the vehicle; S2, filtering the ground points in the laser point cloud data and retaining only the ground points; S3, dividing the preset range in front of the vehicle into fan-shaped areas with the laser radar as the center and the preset angle as the unit; S4. For each sector-shaped area, calculate the spatial distance between the point cloud points and the laser radar, classify the point cloud points according to the spatial distance, and identify obstacles according to the classification results.

[0005] In order to solve the above technical problems, another technical solution adopted by the present invention is: A laser radar-based efficient obstacle recognition terminal comprises a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above laser radar-based efficient obstacle recognition method are implemented.

[0006] The beneficial effects of the present invention are as follows: the present invention provides an efficient obstacle recognition method and terminal based on laser radar, which performs two-layer filtering on the laser point cloud data obtained by the laser radar scanning, and only retains the ground points within a preset range in front of the vehicle, thereby greatly reducing the number of point cloud points that need to be identified subsequently, and avoiding the use of existing complex clustering algorithms. Through fan-shaped area division and distance calculation, obstacles can be integrated and segmented, the number of generated obstacles can be controlled, calculations can be simplified, the efficiency of obstacle recognition can be improved, the hardware requirements can be reduced, and the refresh rate stability of obstacle recognition can be improved to ensure the rapidity of emergency handling of obstacles. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 This is a flowchart of an efficient obstacle recognition method based on laser radar according to an embodiment of the present invention; Figure 2 This is a structural diagram of an efficient obstacle recognition terminal based on laser radar according to an embodiment of the present invention; Figure 3 An example diagram of ground points and above-ground points according to an efficient obstacle recognition method based on laser radar according to an embodiment of the present invention; Figure 4 A schematic diagram of sector area division based on a laser radar efficient obstacle recognition method according to an embodiment of the present invention; Description of labels: 1. A highly efficient obstacle recognition terminal based on laser radar; 2. Processor; 3. Memory. DETAILED DESCRIPTION

[0008] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in conjunction with the implementation modes and the accompanying drawings.

[0009] Please refer to Figure 1 , Figure 3 and Figure 4 , an efficient obstacle recognition method based on laser radar, comprising the steps of: S1. Scanning with a laser radar with a preset number of beams, and obtaining laser point cloud data within a preset range in front of the vehicle; S2, filtering the ground points in the laser point cloud data and retaining only the ground points; S3, dividing the preset range in front of the vehicle into fan-shaped areas with the laser radar as the center and the preset angle as the unit; S4. For each sector-shaped area, calculate the spatial distance between the point cloud points and the laser radar, classify the point cloud points according to the spatial distance, and identify obstacles according to the classification results.

[0010] From the above description, it can be seen that the beneficial effects of the present invention are: the present invention is an efficient obstacle recognition method and terminal based on laser radar, which performs two layers of filtering on the laser point cloud data obtained by the laser radar scanning, and only retains the ground points within the preset range in front of the vehicle, thereby greatly reducing the number of point cloud points that need to be identified subsequently, and avoiding the use of existing complex clustering algorithms. Through fan-shaped area division and distance calculation, obstacles can be integrated and segmented, the number of generated obstacles can be controlled, calculations can be simplified, the efficiency of obstacle recognition can be improved, the hardware requirements can be reduced, and the refresh rate stability of obstacle recognition can be improved to ensure the rapidity of emergency handling of obstacles.

[0011] Furthermore, the step S1 is specifically as follows: Scanning with a laser radar with a preset number of beams to obtain laser point cloud data; According to the coordinate information of the point cloud points, the laser point cloud data is segmented into regions, and only the laser point cloud data in the rectangular region in front of the vehicle is retained; The rectangular area is specifically a rectangular area formed by a first distance in front of the laser radar and a second distance on the left and right sides of the vehicle body.

[0012] From the above description, it can be seen that only the laser point cloud data of the rectangular area formed by the preset distance in front of the vehicle and the second distance on the left and right sides of the vehicle body is obtained, which effectively improves the effectiveness of data acquisition and avoids the load on the calculation caused by a large amount of potentially useless data.

[0013] Furthermore, the laser radar with the preset number of beams is specifically a 16-beam laser radar.

[0014] From the above description, it can be seen that the laser radar uses a 16-line or 32-line laser radar. The more beams, the stronger the obstacle height resolution capability is and the greater the amount of calculation. The present invention uses a 16-line laser radar, which effectively reduces the required amount of calculation, thereby reducing hardware requirements.

[0015] Furthermore, the first distance is 10 m, the second distance is 1.5 m, and the preset angle is 12 degrees.

[0016] From the above description, it can be seen that the first distance is 10m, the second distance is 1.5m, and the preset angle is 15 degrees as an embodiment of the present invention.

[0017] Furthermore, the step S2 is specifically as follows: According to the preset lidar installation height h, deviation threshold d and height coordinate z of the point cloud point, the point cloud points are identified and divided, and the point cloud points with z values ​​in the range of [-hd,-h+d] are ground points, and the point cloud points with z values ​​greater than -h+d are ground points.

[0018] From the above description, it can be seen that the installation height and deviation threshold of the laser radar are used as the data basis for ground point recognition, so as to effectively distinguish between bottom points and ground points. The deviation threshold can ensure that the point cloud points due to possible deviations of the laser radar and some low obstacles that do not need to be avoided, such as speed bumps, are filtered out.

[0019] Furthermore, the step S4 is specifically as follows: For each sector area, the spatial distance between the point cloud points and the laser radar is calculated, and all point cloud points whose difference in spatial distance with the laser radar is less than a preset distance threshold are classified into a cluster; For each cluster of point cloud points, determine whether the number of point cloud points with the same height coordinates exceeds the first number and whether the number of point cloud points with the same horizontal coordinates exceeds the second number. If so, identify the corresponding point cloud cluster as an obstacle, otherwise, filter out the noise.

[0020] From the above description, it can be seen that obstacle recognition is performed based on divided fan-shaped areas, that is, obstacles are divided by spatial areas rather than actual obstacles. The actual number of obstacles is uncontrollable, but regional division is controllable. By dividing the areas, small obstacles in the same area are merged, and large obstacles across different areas are divided. This can achieve the integration and segmentation of obstacles, thereby controlling the number of obstacles. In addition, obstacle recognition also takes into account the number of point cloud points in the horizontal and numerical directions of the point cloud points divided into clusters, which can effectively avoid the possibility of misidentifying scattered point cloud points caused by possible interference as obstacles.

[0021] Furthermore, the distance threshold is 1.5m.

[0022] Furthermore, the first number is 10, and the second number is 3.

[0023] From the above description, it can be seen that the distance threshold is 1.5m, the first number is 10, and the second number is 3 as an embodiment of the present invention.

[0024] Furthermore, the method further comprises the steps of: S5. For the identified obstacle, obtain and establish a rectangular frame of the obstacle according to the maximum x coordinate, maximum y coordinate, maximum z coordinate, minimum x coordinate, minimum y coordinate and minimum z coordinate of each point cloud point in the corresponding point cloud point cluster.

[0025] It can be seen from the above description that, for the identified obstacles, the present invention ignores the specific details of the obstacles and directly establishes a rectangular frame that wraps the obstacles. The rectangular frame can contain all the point cloud points of the obstacles while ensuring the recognition efficiency.

[0026] Please refer to Figure 2A laser radar-based efficient obstacle recognition terminal includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above laser radar-based efficient obstacle recognition method are implemented.

[0027] From the above description, it can be seen that the beneficial effects of the present invention are: the present invention is an efficient obstacle recognition method and terminal based on laser radar, which performs two layers of filtering on the laser point cloud data obtained by the laser radar scanning, and only retains the ground points within the preset range in front of the vehicle, thereby greatly reducing the number of point cloud points that need to be identified subsequently, and avoiding the use of existing complex clustering algorithms. Through fan-shaped area division and distance calculation, obstacles can be integrated and segmented, the number of generated obstacles can be controlled, calculations can be simplified, the efficiency of obstacle recognition can be improved, the hardware requirements can be reduced, and the refresh rate stability of obstacle recognition can be improved to ensure the rapidity of emergency handling of obstacles.

[0028] The invention discloses a laser radar-based efficient obstacle recognition method and terminal, which are suitable for obstacle recognition of vehicles equipped with laser radar.

[0029] Please refer to Figure 1 , Figure 3 and Figure 4 , Embodiment 1 of the present invention is: An efficient obstacle recognition method based on laser radar comprises the following steps: S1. Scanning with a laser radar with a preset number of beams, and obtaining laser point cloud data within a preset range in front of the vehicle; The step S1 is specifically as follows: Scanning with a laser radar with a preset number of beams to obtain laser point cloud data; According to the coordinate information of the point cloud points, the laser point cloud data is segmented into regions, and only the laser point cloud data in the rectangular region in front of the vehicle is retained; The rectangular area is specifically a rectangular area formed by a first distance in front of the laser radar and a second distance on the left and right sides of the vehicle body.

[0030] The laser radar uses a 16-line or 32-line laser radar. The more beams, the stronger the obstacle height resolution capability, and the greater the amount of calculation. In this embodiment, the laser radar uses a 16-line laser radar. The data processing uses an NVIDIA processor, which receives the laser radar point cloud data through the network interface and sends the processed obstacle information through the CAN bus. The laser radar point cloud frame rate can be 20Hz, and 10Hz is usually used to meet the output frequency requirements. Data is received through the network port.

[0031] In other equivalent embodiments, the hardware used can be adjusted according to actual needs.

[0032] In this embodiment, the laser radar is used for scanning, and then the received laser radar point cloud data is segmented into regions, and only the laser point cloud data in the area to be detected (preset range) is retained. In this embodiment, the preset area is a rectangular area consisting of 10m in front of the laser radar and 1.5m on the left and right sides of the vehicle body.

[0033] Specifically, each point cloud point contains an area x*y relative to the coordinate (x, y, z) detection area of ​​the laser radar, which can realize regional filtering of the point cloud and filter out points outside the area.

[0034] S2, filtering the ground points in the laser point cloud data and retaining only the ground points; The step S2 is specifically as follows: According to the preset lidar installation height h, deviation threshold d and height coordinate z of the point cloud point, the point cloud points are identified and divided, and the point cloud points with z values ​​in the range of [-hd,-h+d] are ground points, and the point cloud points with z values ​​greater than -h+d are ground points.

[0035] In this embodiment, the ground points in the detection area are filtered and the ground points in the area are retained. Figure 4 As shown, specifically, the installation height of the laser radar is determined, and the ground point is the point within the height range below the laser radar. For example, if the laser radar is installed at 80 cm, the ground point can be -80 ± 10 cm, that is, the z-axis is -80 ± 10 cm, which is the ground point. After exclusion, the ground point is the obstacle point cloud.

[0036] S3. Divide the preset range in front of the vehicle into a fan-shaped area with the laser radar as the center and the preset angle as the unit.

[0037] In this embodiment, Figure 4 As shown, the detection area is divided into fan-shaped areas radiating outward from the laser radar as the center, with one area divided every 12 degrees. A total of 15 fan-shaped areas are divided in front of the vehicle, and the spatial distance between the point cloud point and the laser radar is calculated.

[0038] S4. For each sector area, calculate the spatial distance between the point cloud points and the laser radar, classify the point cloud points according to the spatial distance, and identify obstacles according to the classification results; The step S4 is specifically as follows: For each sector area, the spatial distance between the point cloud points and the laser radar is calculated, and all point cloud points whose difference in spatial distance with the laser radar is less than a preset distance threshold are classified into a cluster; For each cluster of point cloud points, determine whether the number of point cloud points with the same height coordinates exceeds the first number and whether the number of point cloud points with the same horizontal coordinates exceeds the second number. If so, identify the corresponding point cloud cluster as an obstacle, otherwise, filter out the noise.

[0039] In this embodiment, the point cloud points in each sector area are compared with the spatial distance of the laser radar, and all point cloud points whose spatial distance difference is less than the distance threshold, and whose number of points is large, whose horizontal number (same height coordinate) is greater than the first number, and whose vertical number (same horizontal coordinate, i.e., the same x and y coordinate) is greater than the second number are classified into one category to form an obstacle, thereby obtaining a series of obstacle point cloud clusters.

[0040] In this embodiment, the distance threshold is 1.5m, the first number is 10, and the second number is 3. In other equivalent embodiments, the distance threshold, the first number, and the second number can be adjusted according to actual needs. S5. For the identified obstacle, obtain and establish a rectangular frame of the obstacle according to the maximum x coordinate, maximum y coordinate, maximum z coordinate, minimum x coordinate, minimum y coordinate and minimum z coordinate of each point cloud point in the corresponding point cloud point cluster.

[0041] In this embodiment, the maximum and minimum x, y, and z points in each cluster of point clouds are taken to form a rectangular frame to realize the distribution of the obstacle point cloud cluster in the lidar coordinate system, and then the obstacle information is transmitted to the control unit for judgment through the CAN bus or other means.

[0042] Please refer to Figure 2 , Embodiment 2 of the present invention is: A laser radar-based efficient obstacle recognition terminal 1 includes a processor 2, a memory 3, and a computer program stored in the memory 3 and executable on the processor 2. When the processor 2 executes the computer program, the steps of the laser radar-based efficient obstacle recognition method in the above embodiment 1 are implemented.

[0043] In summary, the present invention provides an efficient obstacle recognition method and terminal based on laser radar, which performs two-layer filtering on the laser point cloud data obtained by laser radar scanning, and only retains the ground points within a preset range in front of the vehicle, thereby greatly reducing the number of point cloud points that need to be identified subsequently, and avoiding the use of existing complex clustering algorithms. Through fan-shaped area division and distance calculation, obstacles can be integrated and segmented, the number of generated obstacles can be controlled, calculations can be simplified, the efficiency of obstacle recognition can be improved, the hardware requirements can be reduced, and the refresh rate stability of obstacle recognition can be improved to ensure the rapidity of emergency handling of obstacles.

[0044] 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. An efficient obstacle recognition method based on laser radar, characterized in that: Includes steps: S1. Scanning with a laser radar with a preset number of beams, and obtaining laser point cloud data within a preset range in front of the vehicle; S2, filtering the ground points in the laser point cloud data and retaining only the ground points; S3, dividing the preset range in front of the vehicle into fan-shaped areas with the laser radar as the center and the preset angle as the unit; S4. For each sector area, calculate the spatial distance between the point cloud points and the laser radar, classify the point cloud points according to the spatial distance, and identify obstacles according to the classification results; The step S4 is specifically as follows: For each sector area, the spatial distance between the point cloud points and the laser radar is calculated, and all point cloud points whose difference in spatial distance with the laser radar is less than a preset distance threshold are classified into a cluster; For each cluster of point cloud points, determine whether the number of point cloud points with the same height coordinates exceeds the first number and whether the number of point cloud points with the same horizontal coordinates exceeds the second number. If so, identify the corresponding point cloud cluster as an obstacle, otherwise, filter out the noise.

2. The efficient obstacle recognition method based on laser radar according to claim 1 is characterized in that: The step S1 is specifically as follows: Scanning with a laser radar with a preset number of beams to obtain laser point cloud data; According to the coordinate information of the point cloud points, the laser point cloud data is segmented into regions, and only the laser point cloud data in the rectangular region in front of the vehicle is retained; The rectangular area is specifically a rectangular area formed by a first distance in front of the laser radar and a second distance on the left and right sides of the vehicle body.

3. The efficient obstacle recognition method based on laser radar according to claim 1 or 2, characterized in that: The laser radar with the preset number of beams is specifically a 16-beam laser radar.

4. The efficient obstacle recognition method based on laser radar according to claim 2 is characterized in that: The first distance is 10m, the second distance is 1.5m, and the preset angle is 12 degrees.

5. The efficient obstacle recognition method based on laser radar according to claim 1 is characterized in that: The step S2 is specifically as follows: According to the preset lidar installation height h, deviation threshold d and height coordinate z of the point cloud point, the point cloud points are identified and divided, and the point cloud points with z values ​​in the range of [-hd,-h+d] are ground points, and the point cloud points with z values ​​greater than -h+d are ground points.

6. The efficient obstacle recognition method based on laser radar according to claim 1 is characterized in that: The distance threshold is 1.5m.

7. The efficient obstacle recognition method based on laser radar according to claim 1 is characterized in that: The first number is 10, and the second number is 3.

8. The efficient obstacle recognition method based on laser radar according to claim 1 is characterized in that: Also includes the steps: S5. For the identified obstacle, obtain and establish a rectangular frame of the obstacle according to the maximum x coordinate, maximum y coordinate, maximum z coordinate, minimum x coordinate, minimum y coordinate and minimum z coordinate of each point cloud point in the corresponding point cloud point cluster.

9. A laser radar-based efficient obstacle recognition terminal, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of any one of the above laser radar-based efficient obstacle recognition methods in claims 1-8 are implemented.

Images