Elevation map construction method of quadruped robot

Abstract

The invention discloses an elevation map construction method of a quadruped robot, and relates to the technical field of elevation map construction. Complete data coverage is ensured during data collection, real-time data quality monitoring is carried out, the accuracy of data collection can be ensured, the accuracy of data collection can be further improved by carrying out processing, classification, calibration and target segmentation on data, and therefore the accuracy of a map can be ensured; map construction is achieved based on the quadruped robot, the working efficiency can be improved, the intelligent degree is high, a three-dimensional map and a two-dimensional map can be generated, and therefore the applicability can be improved, in addition, through the arranged historical database and updating unit, the accuracy of the map can be guaranteed, the timeliness of the map can be prolonged, and the quality of the map can be guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of elevation map construction, and specifically relates to a method for constructing an elevation map of a quadruped robot. Background Art

[0002] An elevation map is a graph used to represent the height of terrain, which shows the undulations of the terrain through specific symbols and lines. Elevation maps play an important role in the fields of Geographic Information System (GIS), navigation, planning, and engineering. An elevation map usually includes the following elements: Viewpoint and viewing direction: Determine the position and orientation of the observer to correctly interpret the map. Scale and elevation interval: The scale is used to represent the ratio of the distance on the map to the actual ground distance, and the elevation interval is used to represent the change in terrain height. Contour lines: Contour lines are lines connecting points of the same height, and each contour line represents a specific height value. The interval between contour lines (contour interval) is clearly marked on the map. Terrain features: Include features such as buildings, rivers, and roads, which are usually represented by symbols and lines. Elevation annotation: Numerical annotations on elevation points or contour lines, used to clearly indicate the elevation information of specific points.

[0003] Traditional elevation map measurement methods are to use GPS or other measurement tools to record the coordinates and height information of each point, then draw these data into a plan view, form contour lines by connecting the points, and finally add other information such as buildings, rivers, and roads, and perform typesetting and decoration. This method is limited by the experience of the staff, has low work efficiency, and has great limitations. Therefore, we propose a method for constructing an elevation map of a quadruped robot. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for constructing an elevation map of a quadruped robot to solve the problems mentioned in the above background art.

[0005] The present invention specifically adopts the following technical solutions to achieve the above purpose:

[0006] A method for constructing an elevation map of a quadruped robot includes:

[0007] Step 1, positioning information: Locate the initial location according to the acquisition device and use this as a reference point;

[0008] Step 2, data acquisition: Conduct ground measurement, record and save the coordinates and height of each point;

[0009] Step 3, data processing: Conduct preliminary processing according to the data;

[0010] Step 4, data classification: Use machine learning algorithms to classify point clouds, such as buildings, vegetation, roads, and rivers;

[0011] Step 5, Data Calibration: Correct and modify the data manually;

[0012] Step 6, Target Segmentation: Segment the collected point cloud data according to the target object;

[0013] Step 7, 3D Modeling: Construct a 3D model based on the above results;

[0014] Step 8, Map Generation: Based on the completed 3D model, construct a local elevation map region by region, eliminate local errors, and construct a global elevation map on the basis of the local elevation map;

[0015] Step 9, Data Update: Periodically repeat the above steps multiple times, determine the coincidence degree of the data collected each time with the previous data. If they coincide, end the update. If not, overwrite the previous data for effective data update.

[0016] Furthermore, both the positioning information and the collected point cloud data are realized based on a quadruped robot, and the quadruped robot is equipped with a measurement unit, a positioning unit, a data conduction unit, and a pose calculation unit. Among them, the measurement unit is used to collect image data, the positioning unit is used to locate the position information of the measurement location, the data conduction unit is used to transmit the data of the measurement unit and the positioning unit, and the pose calculation unit is used to assist the quadruped robot in gait planning.

[0017] Furthermore, the measurement unit is based on a lidar and a camera. Among them, the lidar is used to accurately measure the position, motion state, and shape of the target, and the camera is used to define a 3D view. The viewing angle and magnification ratio of the camera can be adjusted through parameters such as position, target, and focal length; the positioning unit selects one of the global positioning system and the Beidou satellite positioning system, and the data conduction unit is built-in with a data transceiver module.

[0018] Furthermore, the quadruped robot further includes: a historical database unit, a map generation unit, and an update unit. Among them, the historical database unit is used to store the data collected multiple times, the map generation unit is used to generate a 3D map and a 2D map, and the update unit is used to perform periodic data updates.

[0019] Furthermore, the map generation unit includes a 3D map generation module and a 2D map generation module. Among them, the 3D map generation module is used to generate a 3D map, and the 2D map generation module converts the 3D map into a 2D map based on CAD software.

[0020] Furthermore, the data collection further includes collecting the surrounding obstacles and their 3D position information in the scene.

[0021] Furthermore, the data processing includes the following steps:

[0022] Step 31, denoising: removing the noise points generated during the scanning process;

[0023] Step 32, registration: using the iterative closest point algorithm to unify the multi-point scanning data into the same coordinate system;

[0024] Step 33, filtering: separating the ground points from the non-ground points and extracting the point cloud of the target object.

[0025] Furthermore, the 3D modeling includes the following steps:

[0026] Step 71, surface reconstruction: generating a 3D model through the triangular meshing algorithm;

[0027] Step 72, texture mapping: mapping the image data onto the model surface;

[0028] Step 73, model optimization: simplifying the mesh and repairing the holes.

[0029] Furthermore, the map generation includes the following steps:

[0030] Step 81, format conversion: exporting the point cloud data or the model into a common format;

[0031] Step 82, visual display: using professional software to display the results;

[0032] Step 83, data analysis: extracting the geometric information in the model, such as length, area, volume, etc.

[0033] The beneficial effects of the present invention are as follows:

[0034] 1. When collecting data in the present invention, it ensures complete data coverage and monitors the real-time data quality, which can guarantee the accuracy of data collection. Further improving the accuracy of data collection through data processing, classification, calibration, and target segmentation can thus ensure the accuracy of the map.

[0035] 2. The present invention realizes the construction of the map based on a quadruped robot, which can improve work efficiency, has a high degree of intelligence, and can generate 3D maps and 2D maps, thereby improving applicability. Moreover, by setting up the historical database and the update unit, it can not only ensure the accuracy of the map but also extend the timeliness of the map and guarantee the quality of the map. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 is the workflow diagram of the present invention;

[0037] Figure 2It is the workflow diagram of data processing in the present invention;

[0038] Figure 3 It is the workflow diagram of 3D modeling in the present invention;

[0039] Figure 4 It is the workflow diagram of generating a map in the present invention;

[0040] Figure 5 It is the block diagram of four groups of robots in the present invention. Specific implementation mode

[0041] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0042] Please refer to Figure 1 - Figure 5 , the present invention provides a method for constructing an elevation map of a quadruped robot, including:

[0043] Step 1, positioning information: Locate the initial location according to the acquisition device and use it as a reference point;

[0044] Step 2, data acquisition: Conduct ground measurement, record and save the coordinates and heights of each point; Ensure that the data coverage is complete and monitor the quality of real-time data to ensure the accuracy of data collection.

[0045] Step 3, data processing: Conduct preliminary processing according to the data;

[0046] Step 4, data classification: Use machine learning algorithms to classify point clouds, such as buildings, vegetation, roads and rivers;

[0047] Step 5, data calibration: Manually correct and modify the data to ensure the accuracy and effectiveness of the data;

[0048] Step 6, target segmentation: Segment the collected point cloud data according to the target objects; Facilitate subsequent retrieval and modeling;

[0049] Step 7, 3D modeling: Construct a 3D model according to the above results;

[0050] Step 8, map generation: Based on the completed 3D model, construct a local elevation map region by region, eliminate local errors, and construct a global elevation map on the basis of the local elevation map; Improve the accuracy of overall mapping

[0051] Step 9, Data Update: Repeat the above steps periodically multiple times, determine the coincidence degree between the data collected each time and the previous data. If they coincide, end the update; if not, overwrite the previous data to perform effective data update.

[0052] In this embodiment, preferably, both the positioning information and the collected point cloud data are implemented based on a quadruped robot, and the quadruped robot is equipped with a measurement unit, a positioning unit, a data conduction unit, and a pose calculation unit. Among them, the measurement unit is used to collect image data, the positioning unit is used to locate the position information of the measurement location, the data conduction unit is used to transmit the data of the measurement unit and the positioning unit, and the pose calculation unit is used to assist the quadruped robot in gait planning.

[0053] In this embodiment, preferably, the measurement unit is based on a lidar and a camera. Among them, the lidar is used to accurately measure the position (distance and angle), motion state (speed, vibration, and pose) and shape of the target, and the camera is used to define a three-dimensional view, and the viewing angle and magnification ratio of the camera can be adjusted through parameters such as position, target, and focal length; the positioning unit selects one of the global positioning system and the Beidou satellite positioning system, and the data conduction unit is built-in with a data transceiver module.

[0054] In this embodiment, preferably, the quadruped robot further includes: a historical database unit, a map generation unit, and an update unit. Among them, the historical database unit is used to store the data collected multiple times, the map generation unit is used to generate a three-dimensional map and a planar map, and the update unit is used to perform periodic data update.

[0055] In this embodiment, preferably, the map generation unit includes a three-dimensional map generation module and a planar map generation module. Among them, the three-dimensional map generation module is used to generate a three-dimensional map, and the planar map generation module converts the three-dimensional map into a two-dimensional map based on CAD software.

[0056] In this embodiment, preferably, the data collection further includes collecting the surrounding obstacles and their three-dimensional position information in the scene, which can further improve the accuracy of the map.

[0057] In this embodiment, preferably, the data processing includes the following steps:

[0058] Step 31, Denoising: Remove the noise points generated during the scanning process;

[0059] Step 32, Registration: Use the iterative closest point algorithm to unify the multi-point scan data into the same coordinate system;

[0060] Step 33, Filtering: Separate the ground points from the non-ground points and extract the point cloud of the target object.

[0061] In this embodiment, preferably, the three-dimensional modeling includes the following steps:

[0062] Step 71, surface reconstruction: Generate a 3D model through a triangular meshing algorithm (such as Delaunay triangulation);

[0063] Step 72, texture mapping: Map the image data onto the model surface; enhance the realism of the model.

[0064] Step 73, model optimization: Simplify the mesh and repair holes; improve the model quality.

[0065] In this embodiment, preferably, generating a map includes the following steps:

[0066] Step 81, format conversion: Export the point cloud data or model to a common format, such as LAS, OBJ, FBX, etc.;

[0067] Step 82, visualization display: Use professional software (such as CloudCompare, MeshLab) to display the results;

[0068] Step 83, data analysis: Extract geometric information from the model, such as length, area, volume, etc.

[0069] The working principle and usage process of the present invention:

[0070] Step 1, positioning information: Locate the initial location according to the acquisition device and use this as a reference point.

[0071] Step 2, data acquisition: Conduct ground measurements, record and save the coordinates and heights of each point to ensure complete data coverage, and monitor the quality of real-time data to ensure the accuracy of data collection.

[0072] Both the positioning information and the acquired point cloud data are realized based on a quadruped robot, and the quadruped robot is equipped with a measurement unit, a positioning unit, a data conduction unit, and a pose calculation unit. Among them, the measurement unit is used to collect image data, the positioning unit is used to locate the position information of the measurement location, the data conduction unit is used to transmit the data of the measurement unit and the positioning unit, and the pose calculation unit is used to assist the quadruped robot in gait planning.

[0073] The measurement unit is based on a lidar and a camera. Among them, the lidar is used to accurately measure the position (distance and angle), motion state (speed, vibration, and pose) and shape of the target, and the camera is used to define a three-dimensional view, and the viewing angle and magnification ratio of the camera can be adjusted through parameters such as position, target, and focal length; the positioning unit selects one of the global positioning system and the Beidou satellite positioning system, and the data conduction unit is built-in with a data transceiver module.

[0074] The quadruped robot further includes: a historical database unit, a map generation unit, and an update unit. Among them, the historical database unit is used to store the data collected multiple times, the map generation unit is used to generate a three-dimensional map and a planar map, and the update unit is used to perform periodic updates of the data.

[0075] The map generation unit includes a three-dimensional map generation module and a planar map generation module. Among them, the three-dimensional map generation module is used to generate a three-dimensional map, and the planar map generation module converts the three-dimensional map into a two-dimensional map based on CAD software.

[0076] The data collection further includes collecting the surrounding obstacles and their three-dimensional position information in the scene, which can further improve the accuracy of the map.

[0077] Step 3: Data processing: Perform preliminary processing according to the data; including the following steps:

[0078] Step 31: Denoising: Remove the noise points generated during the scanning process;

[0079] Step 32: Registration: Use the iterative closest point algorithm to unify the multi-point scanning data into the same coordinate system;

[0080] Step 33: Filtering: Separate the ground points from the non-ground points and extract the point cloud of the target object.

[0081] Step 4: Data classification: Use machine learning algorithms to classify the point cloud, such as buildings, vegetation, roads, and rivers;

[0082] Step 5: Data calibration: Perform correction and modification of the data manually; ensure the accuracy and effectiveness of the data.

[0083] Step 6: Target segmentation: Segment the collected point cloud data according to the target object; facilitate subsequent retrieval and modeling.

[0084] Ensure that the data coverage is complete during data collection and monitor the real-time data quality, which can guarantee the accuracy of data collection. Further improving the accuracy of data collection through data processing, classification, calibration, and target segmentation can ensure the accuracy of the map.

[0085] Step 7: Three-dimensional modeling: Construct a three-dimensional model according to the above results; including the following steps:

[0086] Step 71: Surface reconstruction: Generate a three-dimensional model through a triangular meshing algorithm (such as Delaunay triangulation);

[0087] Step 72: Texture mapping: Map the image data onto the model surface; enhance the realism of the model.

[0088] Step 73, Model Optimization: Simplify the grid and fix the loopholes to improve the model quality.

[0089] Step 8, Generate Map: Based on the completed 3D model, construct local elevation maps region by region, eliminate local errors, and construct a global elevation map based on the local elevation maps to improve the accuracy of the overall mapping. The steps are as follows:

[0090] Step 81, Format Conversion: Export the point cloud data or model into common formats such as LAS, OBJ, FBX, etc.

[0091] Step 82, Visualization Display: Use professional software (such as CloudCompare, MeshLab) to display the results.

[0092] Step 83, Data Analysis: Extract geometric information in the model, such as length, area, volume, etc.

[0093] Step 9, Data Update: Periodically repeat the above steps multiple times, determine the coincidence degree of the data collected each time with the previous data. If they coincide, end the update. If not, overwrite the previous data to perform effective data update.

[0094] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing an elevation map of a quadruped robot, characterized in that, Including: Step 1, Location Information: Locate the initial location based on the acquisition device and use this as the reference point; Step 2, Data Acquisition: Conduct ground measurements, record and save the coordinates and heights of each point; Step 3, Data Processing: Conduct preliminary processing based on the data; Step 4, Data Classification: Use machine learning algorithms to classify the point cloud, such as buildings, vegetation, roads, and rivers; Step 5, Data Calibration: Manually correct and revise the data; Step 6, Target Segmentation: Segment the collected point cloud data according to the target objects; Step 7, 3D Modeling: Construct a 3D model based on the above results; Step 8, Map Generation: Based on the completed 3D model, construct a local elevation map for each area, eliminate local errors, and construct a global elevation map on the basis of the local elevation map; Step 9, Data Update: Periodically repeat the above steps multiple times, determine the coincidence degree of the data collected each time with the previous data. If they coincide, end the update. If not, overwrite the previous data for effective data update.

2. The method for constructing an elevation map of a quadruped robot according to claim 1, wherein: The above location information and the collected point cloud data are both realized based on a quadruped robot, and the quadruped robot is equipped with a measurement unit, a positioning unit, a data transmission unit, and a pose calculation unit. Among them, the measurement unit is used to collect image data, the positioning unit is used to locate the position information of the measurement location, the data transmission unit is used to transmit the data of the measurement unit and the positioning unit, and the pose calculation unit is used to assist the quadruped robot in gait planning.

3. A method for constructing an elevation map of a quadruped robot according to claim 2, wherein: The measurement unit is based on a lidar and a camera. Among them, the lidar is used to accurately measure the position, motion state, and shape of the target, and the camera is used to define a 3D view. The viewing angle and magnification ratio of the camera can be adjusted through parameters such as position, target, and focal length; the positioning unit selects one of the global positioning system and the Beidou satellite positioning system, and the data transmission unit is built-in with a data transceiver module.

4. A method for constructing an elevation map of a quadruped robot according to claim 2, characterized in that, The quadruped robot also includes: a historical database unit, a map generation unit, and an update unit. Among them, the historical database unit is used to store the data collected multiple times, the map generation unit is used to generate a 3D map and a 2D map, and the update unit is used to perform periodic updates of the data.

5. A method for constructing an elevation map of a quadruped robot according to claim 4, characterized in that, The map generation unit includes a 3D map generation module and a 2D map generation module. Among them, the 3D map generation module is used to generate a 3D map, and the 2D map generation module converts the 3D map into a 2D map based on CAD software.

6. The method for constructing an elevation map of a quadruped robot according to claim 1, wherein: The data acquisition also includes collecting the surrounding obstacles and their 3D position information in the scene.

7. A method for constructing an elevation map of a quadruped robot according to claim 1, characterized in that, The data processing includes the following steps: Step 31, Denoising: Remove the noise points generated during the scanning process; Step 32, Registration: Use the iterative closest point algorithm to unify the multi-point scan data into the same coordinate system; Step 33, Filtering: Separate the ground points from the non-ground points and extract the point cloud of the target object.

8. A method for constructing an elevation map of a quadruped robot according to claim 1, characterized in that The 3D modeling includes the following steps: Step 71, Surface Reconstruction: Generate a 3D model through the triangular meshing algorithm; Step 72, Texture Mapping: Map the image data onto the model surface; Step 73, Model Optimization: Simplify the mesh and repair the loopholes.

9. A method for constructing an elevation map of a quadruped robot according to claim 1, characterized in that, The generation of the map includes the following steps: Step 81, Format Conversion: Export the point cloud data or model into a common format; Step 82, Visualization Display: Use professional software to display the results; Step 83, Data Analysis: Extract geometric information in the model, such as length, area, volume, etc.

Images