A method for generating 2D occupancy grids from 3D point clouds based on multi-layer grid fusion
By generating and fusing multi-layer raster maps, the problems of insufficient preservation of low-lying obstacles and low computational efficiency in existing technologies are solved, and efficient generation of 2D occupancy raster maps is achieved.
Patent Information
- Application Number
- CN202311363667.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-20
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-10-20
AI Technical Summary
Existing technologies suffer from insufficient preservation of low-lying obstacles and low computational efficiency when converting 3D point cloud maps into 2D occupancy grid maps.
A multi-layer raster fusion method is adopted, which generates and merges multi-layer raster maps, updates the raster probability using ground extraction and traversal optical path method, and calculates obstacle layers by combining ground height parameters to generate the final 2D occupied raster map.
It effectively preserves low-lying obstacles and reduces computational complexity and resource consumption.
Smart Images

Figure CN117570956B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of robot SLAM, specifically relating to a method for generating 2D occupancy grids from 3D point clouds based on multi-layer grid fusion. Background Technology
[0002] SLAM (Simultaneous Localization and Mapping) is an important component for robots to complete autonomous tasks. Its purpose is to determine the robot's position in the environment and build a map as needed.
[0003] SLAM systems that use lasers as sensors are called laser SLAM. Multi-line LiDAR can acquire 3D point clouds with rich information. As the cost of multi-line LiDAR decreases, many robots with high localization requirements are beginning to use multi-line LiDAR instead of single-line LiDAR. For robots working in ground-based environments, 2D maps are crucial, as robot path planning needs to be performed on 2D maps rather than 3D maps. Therefore, it is necessary to convert 3D point cloud maps into 2D maps. The most commonly used 2D map is the 2D occupancy grid map. Currently, there are two main methods for converting 3D point cloud maps into 2D occupancy grid maps:
[0004] 1) Capture 3D point clouds within a certain range, reduce them to 2D, and use these 2D points to generate a 2D grid map;
[0005] 2) First generate a 3D occupying grid map, then convert it into a 2D occupying grid map.
[0006] Currently, there are two main methods for converting 3D point cloud maps into 2D occupies a raster map:
[0007] 1) Capturing 3D point clouds within a certain range: First, this method requires determining the point cloud capture range in advance. If the range is not selected properly, the ground may be captured, causing empty areas to become occupied. Second, this method may cause objects that are relatively short in the Z-axis (height) direction to be incorrectly filtered out, that is, grids that should be occupied are judged as empty.
[0008] 2) First generate a 3D occupying grid map, then convert it into a 2D occupying grid map. Compared to generating a 2D occupying grid, generating a 3D occupying grid adds an extra search dimension, which will greatly increase resource consumption and computation time. In addition, during the process of converting a 3D grid to a 2D grid, it is also necessary to select a certain height to delete the ground, otherwise path planning cannot be performed. Summary of the Invention
[0009] To address the shortcomings of the existing technologies, this invention proposes a method for generating 2D occupancy grids from 3D point clouds based on multi-layer grid fusion, which can simultaneously address the issues of preserving low-lying obstacles and improving computational efficiency.
[0010] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:
[0011] A method for generating 2D occupancy grids from 3D point clouds based on multi-layer grid fusion, the method comprising the following steps:
[0012] Step S1: Generate a multi-layer raster map;
[0013] Step S2: Merge the generated multi-layer raster maps.
[0014] The steps for generating a multi-layer raster map in step S1 are as follows:
[0015] Step S1-1: Extract the ground from the point cloud to obtain the ground height parameter h. min ;
[0016] Step S1-2: Extract the returned ground height h based on the ground. min The maximum height h is set. max Calculate the number of layers l based on the spacing g;
[0017] Steps S1-3: For a point p in the point cloud i =[x,y,z] T If the point is in (h min ,h max If the point is between z and z, then calculate which raster map layer the point belongs to.
[0018] Step S1-4: Update the probability of occupying a grid cell in this layer using the traversal ray path method in the grid map.
[0019] Step S1-5, repeat steps S1-3 and S1-4 until all points in the point cloud have been traversed.
[0020] Furthermore, the method for fusing the generated multi-layer raster maps in step S2 includes:
[0021] Step S2-1: For a certain two-dimensional position (x,y), determine whether at least two raster maps occupy this position. If so, the raster map will eventually occupy this position; otherwise, proceed to step S2-2.
[0022] Step S2-2: Determine whether the location has at most one grid cell occupied and at least one grid cell free. If so, the location will be free in the final grid map. If not, proceed to step S2-3.
[0023] Step S2-3: Determine that the position of the final grid is unknown;
[0024] Step S2-4: Repeat steps S2-1 to S2-3 until all positions have been traversed.
[0025] The present invention has the following beneficial effects:
[0026] (1) Compared with the method of extracting 3D point cloud within a certain range, reducing it to 2D, and using these 2D points to generate a 2D occupied grid map, the method proposed in this invention can retain more low-lying obstacles.
[0027] (2) The method proposed in this invention involves a small amount of computation: first, a 3D occupying grid map is generated, and then it is converted into a 2D occupying grid map. Attached Figure Description
[0028] Figure 1 This is a flowchart illustrating the generation of a multi-layer raster map according to the present invention.
[0029] Figure 2 This is a diagram of the fusion strategy proposed in this invention. Detailed Implementation
[0030] The present invention is described below based on embodiments, but the invention is not limited to these embodiments. In the detailed description of the invention below, certain specific details are described in detail. Those skilled in the art will fully understand the invention even without these details. To avoid obscuring the essence of the invention, well-known methods, processes, flows, elements, and circuits are not described in detail.
[0031] Unless the context explicitly requires it, the words "comprising," "including," and similar terms throughout the specification and claims should be interpreted as encompassing rather than being exclusive or exhaustive; that is, meaning "including but not limited to."
[0032] In the description of this invention, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means two or more.
[0033] A method for generating 2D occupancy grids from 3D point clouds based on multi-layer grid fusion, the method comprising the following steps:
[0034] Step S1: Generate a multi-layer raster map;
[0035] Step S2: Merge the generated multi-layer raster maps.
[0036] The steps for generating a multi-layer raster map in step S1 are as follows:
[0037] Step S1-1: Extract the ground from the point cloud to obtain the ground height parameter h. min ;
[0038] Step S1-2: Extract the returned ground height h based on the ground. min The maximum height h is set. max Calculate the number of layers l based on the spacing g;
[0039] Steps S1-3: For a point p in the point cloud i =[x,y,z] T If the point is in (h min ,h max If the point is between z and z, then calculate which raster map layer the point belongs to.
[0040] Step S1-4: Update the probability of occupying a grid cell in this layer using the traversal ray path method in the grid map.
[0041] Step S1-5, repeat steps S1-3 and S1-4 until all points in the point cloud have been traversed.
[0042] The method for fusing the generated multi-layer raster maps in step S2 includes:
[0043] Step S2-1: For a certain two-dimensional position (x,y), determine whether at least two raster maps occupy this position. If so, the raster map will eventually occupy this position; otherwise, proceed to step S2-2.
[0044] Step S2-2: Determine whether the location has at most one grid cell occupied and at least one grid cell free. If so, the location will be free in the final grid map. If not, proceed to step S2-3.
[0045] Step S2-3: Determine that the position of the final grid is unknown;
[0046] Step S2-4: Repeat steps S2-1 to S2-3 until all positions have been traversed.
[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for generating 2D occupancy grids from 3D point clouds based on multi-layer grid fusion, characterized in that, The method includes the following steps: Step S1: Generate a multi-layer raster map; Step S2: Merge the generated multi-layer raster maps; The steps for generating a multi-layer raster map in step S1 are as follows: Step S1-1, ground extraction is performed on the point cloud to obtain a ground height parameter h min ; Step S1-2, calculating the ground height h according to the ground extraction return min , the maximum height h max and the interval g to calculate the number of layers l; Steps S1-3: For a point p in the point cloud i =[x, y, z] T If the point is in (h min , h max If the point is between z and z, then calculate which raster map layer the point belongs to. Step S1-4: Update the probability of occupying a grid cell in this layer using the traversal ray path method in the grid map. Step S1-5: Repeat steps S1-3 and S1-4 until all points in the point cloud have been traversed; The method for fusing the generated multi-layer raster maps in step S2 includes: Step S2-1: For a given two-dimensional location (x, y), determine whether at least two raster maps occupy that location. If yes, the raster map will eventually occupy that location; otherwise, proceed to step S2-2. Step S2-2: Determine whether the location has at most one grid cell occupied and at least one grid cell free. If so, the location will be free in the final grid map. If not, proceed to step S2-3. Step S2-3: Determine that the position of the final grid is unknown; Step S2-4: Repeat steps S2-1 to S2-3 until all positions have been traversed.
Citation Information
Patent Citations
Fusion method of multi-robot three-dimensional geometrical map
CN101944240A
Preparation method of indoor occupation grid map based on RGB-D information
CN111598916A
