A Method for Removing Dynamic Obstacles from Laser Point Clouds Combining with a Map Model

Abstract

The present invention belongs to the technical field of vehicle positioning systems, and particularly relates to a method for dynamically removing obstacles from laser point clouds in combination with a map model. It mainly includes processes such as ground segmentation, extracting non-ground obstacle point clouds, querying road segments according to positioning, and removing obstacle point clouds in combination with structured road map model parameters. The present invention quickly and efficiently removes dynamic interference obstacle point clouds on the road by combining a map road simplification model, reduces the interference of other traffic participants on point cloud matching positioning, and improves the positioning accuracy and stability; the present invention combines road segmentation with map model parameters, the method is simple, the algorithm is fast, and there are no requirements for the speed of the host vehicle and the speed of other objects.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle positioning systems, and particularly relates to a method for removing dynamic obstacles from laser point clouds in combination with a map model. Background Art

[0002] Point cloud matching positioning based on high-precision maps and lidar is an important non-GNSS (Global Navigation Satellite System) positioning method in intelligent driving vehicles. However, in an open urban structured road scenario, the performance of this method is easily interfered by other dynamic participants. In particular, large traffic participants (such as large trucks, freight vehicles, buses, etc.) will have a great impact on the point cloud matching positioning result, and even cause positioning failure. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method for removing dynamic obstacles from laser point clouds in combination with a map model, which can efficiently and quickly remove the dynamic obstacle point clouds in structured roads, has no requirement for the speed of other traffic participants, and has good real-time performance.

[0004] The technical solution adopted by the present invention to solve the above technical problem is as follows:

[0005] A method for removing dynamic obstacles from laser point clouds in combination with a map model, which is used in a vehicle positioning system, specifically includes the following steps:

[0006] S1. After the vehicle positioning system receives the current frame of point cloud data, preprocess the point cloud to remove noise abnormal data;

[0007] S2. Perform ground segmentation on the preprocessed point cloud, divide the point cloud into ground point cloud PC1 and non-ground point cloud PC2, perform subsequent processing on the non-ground point cloud PC2, and keep the ground point cloud;

[0008] S3. According to the position of the vehicle itself, query the high-precision map to obtain the current road section to which the vehicle belongs;

[0009] S4. Obtain the map model parameters of the current road section to which the vehicle belongs, and calculate the coordinates in the Lidar point cloud coordinate system of the vehicle through Heron's formula;

[0010] S5. Calculate the filtering range of dynamic obstacles in the current frame of point cloud, and filter it to obtain the filtered point cloud PC3;

[0011] S6. Merge the ground point cloud PC1 and the filtered point cloud PC3, and output the point cloud result PC4.

[0012] Further, in step S2, the non-ground point cloud PC2 includes road edges, road signs, street lights, trees, isolation belts, bridge piers, as well as pedestrians and vehicles.

[0013] Further, in step S3, the urban structured roads are divided into multiple road segments based on the map road simplification model.

[0014] Further, in step S4, the map model parameters of the current road segment to which the vehicle belongs include the starting point P start and the ending point P end of the road segment; the coordinates of the vehicle in the Lidar point cloud coordinate system include the distance L left between the vehicle and the left side of the road segment, the distance L right between the vehicle and the right side of the road segment, the distance L fr o nt between the vehicle and the starting point of the road segment, and the distance L rear between the vehicle and the ending point of the road segment.

[0015] Further, the calculation process of the coordinates of the vehicle in the Lidar point cloud coordinate system is as follows:

[0016] l 0 = |P start - P end |

[0017] l 1 = |P car - P start |

[0018] l 2 = |P car - P end |

[0019]

[0020] L right = width - l left

[0021]

[0022] L rear = l 0 - L front

[0023] where P car is the vehicle position.

[0024] Further, in step S5, the specific process of obtaining the filtered point cloud PC3 is as follows: According to the current vehicle heading angle yaw, first convert the vehicle Lidar point cloud coordinate system to the local Gaussian coordinate system; then, in the local Gaussian coordinate system, use L left , L right , L front and Lrear Directly remove the obstacle points within the road section to obtain the filtered point cloud PC3.

[0025] The present invention has the following main advantages compared with the prior art:

[0026] 1. The present invention combines the map road simplification model to quickly and efficiently remove the dynamic interference obstacle point clouds on the road, reduce the interference of other traffic participants on the point cloud matching and positioning, and improve the positioning accuracy and stability.

[0027] 2. The present invention combines road segmentation with map model parameters, with a simple method and a fast algorithm, and has no requirements for the speed of the host vehicle and the speeds of other objects. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a flowchart of a method for removing dynamic obstacles from laser point clouds in combination with a map model according to the present invention;

[0029] Figure 2 It is a schematic diagram of road segmentation of the map model in an embodiment of the present invention;

[0030] Figure 3 It is a schematic diagram of road section attributes in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0032] It should be noted that according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the objectives of the present invention.

[0033] Actual roads can generally be divided into two categories: structured roads and unstructured roads. Structured roads generally refer to highways, urban arterial roads and other roads with good structure. Such roads have clear road marking lines, a relatively simple road background environment, and obvious geometric features of the road.

[0034] A method for removing dynamic obstacles from laser point clouds in combination with a map model provided by the present invention mainly includes processes such as ground segmentation, extracting non-ground obstacle point clouds, querying road sections according to positioning, and removing obstacle point clouds in combination with structured road map model parameters.

[0035] Such asFigure 1 As shown in the figure, it specifically includes the following steps:

[0036] S1. After the system receives the current frame of point cloud data, it first preprocesses the point cloud to remove noisy and abnormal data.

[0037] S2. Perform ground segmentation on the preprocessed point cloud to divide the point cloud into ground point cloud PC1 and non-ground point cloud PC2.

[0038] Mainly process the non-ground point cloud PC2, while keeping the ground point cloud; the non-ground point cloud mainly includes inherent road features such as curbs, road signs, street lights, trees, isolation belts, and bridge piers near the road, as well as road traffic participants such as pedestrians and vehicles.

[0039] S3. Query the high-precision map according to the position of the vehicle itself to obtain the current section.

[0040] S4. Obtain the map model parameters (starting point, ending point, road width, etc.) of the current section.

[0041] S5. Calculate the filtering range of dynamic obstacles in the current frame of point cloud and filter it to obtain the filtered point cloud PC3.

[0042] S6. Merge the ground point cloud PC1 and the filtered point cloud PC3 and output the point cloud result PC4.

[0043] Furthermore, this embodiment mainly aims at urban structured roads, such as Figure 2 As shown in the figure, the urban structured road is described by a map road simplified model, and the map road simplified model divides the road into multiple sections;

[0044] Such as Figure 3 As shown in the figure, each section has attributes such as starting point, ending point, number of lanes, and width.

[0045] Furthermore, in step S3, according to the current position P car of the vehicle itself, query the current section of the vehicle itself in the high-precision map.

[0046] Furthermore, in step S4, obtain the starting point P start and the ending point P end of the current section. When P start , P end and P car are known, calculate L left (the distance between the vehicle itself and the left side of the section), and further obtain L right (the distance between the vehicle itself and the right side of the section), L front (the distance between the vehicle itself and the starting point of the section) and L rear(Distance between the host vehicle and the end point of the road segment), the specific calculation process is as follows:

[0047] l 0 = P start - P end |

[0048] l 1 = |P car - P start |

[0049] l 2 = |P car - P end |

[0050]

[0051] L right = width - l left

[0052]

[0053] L rear = l 0 - L front

[0054] Further, in step S5, according to the current host vehicle heading angle yaw, first convert the host vehicle Lidar point cloud coordinate system to the local Gaussian coordinate system; then, in the local Gaussian coordinate system, use L calculated in step S4 left ,L right ,L front and L rear to directly remove the obstacle points within the road segment range to obtain the filtered point cloud PC3.

[0055] Further, in step S6, merge the ground point cloud PC1 with the filtered point cloud PC3 and output the point cloud result PC4.

[0056] In summary, the present invention provides a method for dynamically removing obstacles from laser point clouds in combination with a map model:

[0057] 1. The present invention combines the map road segment simplification model to quickly and efficiently remove the dynamic interference obstacle point clouds on the road segment, reduce the interference of other traffic participants on the point cloud matching and positioning, and improve the positioning accuracy and stability.

[0058] 2. The present invention combines the road segment segmentation with the map model parameters, the method is simple, the algorithm is fast, and there is no requirement for the host vehicle speed and the speeds of other objects.

[0059] Based on the above method, the present invention also provides:

[0060] An automotive positioning system, comprising a memory, a processor, and a program stored on the memory and executable on the processor, wherein when the processor executes the program, it implements a method for dynamically removing obstacles from laser point clouds in combination with a map model as described above.

[0061] A non-transitory readable storage medium, on which a program is stored, and when the program is executed by an automotive positioning system, it implements a method for dynamically removing obstacles from laser point clouds in combination with a map model as described above.

[0062] An automobile, comprising the automotive positioning system as described above.

[0063] The above embodiments are only used to illustrate the design concept and characteristics of the present invention, and the purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made based on the principles and design concepts disclosed by the present invention are within the protection scope of the present invention.

Claims

1. A laser point cloud dynamic obstacle removal method combined with a map model for use in a vehicle positioning system. It is characterized in that Combine road segmentation with map model parameters, query road segments according to positioning, and remove obstacle point clouds in combination with structured road map model parameters, specifically including the following steps: S1, after the vehicle positioning system receives the point cloud data of the current frame, it pre-processes the point cloud and removes abnormal noise data; S2, perform ground segmentation on the preprocessed point cloud, divide the point cloud into ground point cloud PC1 and non-ground point cloud PC2, perform subsequent processing on the non-ground point cloud PC2, and keep the ground point cloud; S3, based on the map road simplification model, the urban structured roads are divided into multiple sections, and according to the location of the vehicle, the high-precision map is queried to obtain the current section of the vehicle; S4, obtaining the map model parameters of the road section to which the ego vehicle currently belongs, and calculating the coordinates of the ego vehicle in the Lidar point cloud coordinate system by using the Heron formula; The map model parameters of the current road section to which the host vehicle belongs include the starting point P of the road section start , the ending point P of the road section end and the road section width width; The coordinates of the host vehicle in the Lidar point cloud coordinate system include the distance L between the host vehicle and the left side of the road section left , the distance L between the host vehicle and the right side of the road section right , the distance L between the host vehicle and the starting point of the road section front , the distance L between the host vehicle and the ending point of the road section rear ; S5, calculating the filtering range of the dynamic obstacles in the point cloud of the current frame, and filtering it to obtain the filtered point cloud PC3; The specific process of obtaining the filtered point cloud PC3 is as follows: According to the current vehicle heading angle yaw, first convert the vehicle Lidar point cloud coordinate system to the local Gaussian coordinate system; then, in the local Gaussian coordinate system, use L left , L right , L front and L rear to directly remove the obstacle points within the road section range to obtain the filtered point cloud PC3; S6, merging the ground point cloud PC1 and the filtered point cloud PC3, and outputting a point cloud result PC4.

2. According to the laser point cloud dynamic obstacle removal method combined with the map model according to claim 1, Features In step S2, the non-ground point cloud PC2 includes curbs, road signs, street lights, trees, isolation belts, bridge piers, pedestrians and vehicles.

3. According to the method for removing dynamic obstacles from laser point cloud combined with map model in claim 1, It is characterized in that The coordinate calculation process in the ego-vehicle Lidar point cloud coordinate system is as follows: l 0 = |P start - P end | l 1 = |P car - P start | l 2 = |P car - P end | L right = width - l left L rear = l 0 - L front Among them, P car is the position of the vehicle itself, l 0 , l 1 , l 2 are all intermediate parameters, and width is the width of the road section.

4. A vehicle positioning system, comprising a memory, a processor, and a program stored in the memory and executable on the processor, It is characterized in that When the processor executes the program, the method according to any one of claims 1 to 3 is implemented.

5. A non-transitory readable storage medium having a program stored thereon, It is characterized in that When the program is executed by a vehicle positioning system, the method according to any one of claims 1 to 3 is implemented.

6. A car, Features: The invention comprises the automobile positioning system as claimed in claim 4.

Images