CN-121977563-A - Space positioning optimization method and system based on multi-source fusion

CN121977563ACN 121977563 ACN121977563 ACN 121977563ACN-121977563-A

Abstract

The invention discloses a space positioning optimization method and a system based on multi-source fusion, which belong to the technical field of space positioning, and the multi-source positioning data are fused and complemented by collecting the multi-source positioning data and constructing a fixed world coordinate system, so that a multi-source data barrier is broken, the reliability of the data under the complex working condition environment is ensured through a multi-checking mechanism, the higher robustness is ensured on the premise of realizing global positioning, the precision of scene positioning is ensured, and finally the accumulated drift is globally eliminated through a space precision optimization mechanism, thereby dynamically compensating the error caused by scene change, ensuring the positioning reliability and accuracy, and solving the problem that the positioning precision is lower due to the difficulty in ensuring the positioning robustness under the condition of realizing global positioning in the existing space positioning method.

Inventors

ZHANG LEI
LOU XIN
BAO TINGHUA
He Qianxiao
ZHU ZHUOLING
ZHOU NING

Assignees

国网浙江省电力有限公司金华供电公司
国网浙江省电力有限公司科技创新中心

Dates

Publication Date: 20260505
Application Date: 20260122

Claims (10)

1. A space positioning optimization method based on multi-source fusion is characterized by comprising the following steps: S1, acquiring multi-source positioning data in a scene in real time and preprocessing the multi-source positioning data to obtain target positioning data; S2, constructing a world coordinate system, and mapping target positioning data into the world coordinate system based on a multi-source coordinate mapping mechanism; s3, checking the world coordinate system based on a multiple checking mechanism to obtain an original 3D scene map; And S4, extracting features of the target positioning data to obtain target positioning features, and performing global optimization on the original 3D scene map based on a spatial precision optimization mechanism and combining the target positioning features to obtain a target 3D scene map.
2. The spatial localization optimization method based on multi-source fusion according to claim 1, wherein: s1, multi-source positioning data in a scene are acquired in real time and preprocessed to obtain target positioning data, and the method comprises the following steps: acquiring environmental point cloud data, image data and pose data of each positioning target in a scene in real time; extracting point cloud data belonging to the ground in the environmental point cloud data to obtain ground point cloud data, and clustering non-ground point cloud data; Based on the clustering result and combining the image data, dividing the non-ground point cloud data into point cloud clusters of different target types, taking the point cloud clusters and the ground point cloud data as target point cloud data, and associating the target point cloud data with the corresponding image data and pose data to obtain target positioning data.
3. The spatial localization optimization method based on multi-source fusion according to claim 2, wherein: s2, constructing a world coordinate system, which comprises the following steps: Based on SLAM technology, the center of image data in the target positioning data at the initial acquisition time is taken as the origin of a coordinate system, the direction perpendicular to the ground point cloud data is taken as the Z axis, the advancing direction of a certain positioning target is taken as the X axis, and the horizontal direction perpendicular to the advancing direction is taken as the Y axis to construct a world coordinate system.
4. The spatial localization optimization method based on multi-source fusion according to claim 2, wherein: In S2, mapping the target positioning data into the world coordinate system based on the multi-source coordinate mapping mechanism includes the following steps: Constructing a first conversion matrix based on the position offset and posture rotation relation of the target point cloud data and the corresponding image data, and multiplying the first conversion matrix by the point cloud coordinates of the target point cloud data to map the target point cloud data into a two-dimensional coordinate system of the image data to obtain image homogeneous coordinates; Constructing a second conversion matrix based on the position deviation and posture rotation relation of the image data and the corresponding posture data, and updating in real time based on the posture data change; and multiplying the second conversion matrix by the image homogeneous coordinates to map the cloud data of the target point to the world coordinate system to obtain the world homogeneous coordinates.
5. A method for optimizing spatial localization based on multi-source fusion according to claim 3, wherein: in S3, the multiple checking mechanism includes the following steps: Randomly extracting n points from ground point cloud data in a world coordinate system to fit into a plane to be verified, and calculating the deviation distance from the plane to be verified to the origin of the coordinate system and the deviation included angle between the normal vector of the plane to be verified and the Z axis; taking point cloud coordinates of point cloud clusters with different static time duration t as reference point clouds, taking world homogeneous coordinates of corresponding point cloud clusters in a world coordinate system as check point clouds, and calculating root mean square errors of the reference point clouds and the check point clouds; and if the deviation distance is greater than or equal to the distance deviation threshold value or the deviation included angle is greater than or equal to the included angle deviation threshold value or the root mean square error is greater than or equal to the error threshold value, remapping is carried out.
6. The spatial localization optimization method based on multi-source fusion according to claim 2, wherein: S4, extracting features of the target positioning data to obtain target positioning features, wherein the method comprises the following steps: performing scale invariant feature transformation and internal shape description on target point cloud data in target positioning data to obtain a position with geometric significance in the point cloud as a key point; constructing a fast point characteristic histogram for the cloud data of the target point, dividing a spherical neighborhood of the key point into N space sectors, and counting normal vector distribution of each space sector as a characteristic descriptor; And taking the key points and the feature descriptors as target positioning features.
7. The spatial localization optimization method based on multi-source fusion according to claim 6, wherein: in S4, the spatial precision optimization mechanism includes the following steps: when the moving distance of the positioning target exceeds s, acquiring pose data of each frame in the moving process of the positioning target by the distance s; acquiring key points of the same physical point in each frame, calculating the theoretical position of a positioning target in the current frame based on a feature descriptor corresponding to the key points, comparing the actual position in the pose data of the positioning target with the theoretical position to acquire a deviation degree, and globally adjusting all coordinates of an original 3D scene map based on the deviation degree; If the positioning target repeatedly passes through a physical point, calculating the closed-loop deviation degree of the theoretical position and the actual position of the positioning target when passing through based on the key point of the physical point and the corresponding feature descriptor, and performing global adjustment on all coordinates on the path of the positioning target based on the closed-loop deviation degree.
8. The space positioning optimization system based on the multi-source fusion is suitable for the space positioning optimization method based on the multi-source fusion, and is characterized by comprising a data acquisition module, a data mapping module, a verification module, a feature extraction module and a global optimization module; The data acquisition module acquires multi-source positioning data in a scene in real time and performs preprocessing to obtain target positioning data; The data mapping module maps the target positioning data into a world coordinate system based on a multi-source coordinate mapping mechanism; The verification module verifies the world coordinate system based on a multiple verification mechanism to obtain an original 3D scene map; The feature extraction module performs feature extraction on the target positioning data to obtain target positioning features; The global optimization module performs global optimization on the original 3D scene map based on a spatial precision optimization mechanism and combining with target positioning features to obtain a target 3D scene map.
9. The computer equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are in communication with each other through the communication bus, the memory is used for storing a computer program, and the processor is used for realizing the steps of the space positioning optimization method based on the multi-source fusion according to any one of claims 1-7 when the program stored on the memory is executed.
10. A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the computer program realizes the steps of a spatial localization optimization method based on multi-source fusion as set forth in any one of claims 1 to 7.

Description

Space positioning optimization method and system based on multi-source fusion Technical Field The invention relates to the technical field of space positioning, in particular to a space positioning optimization method and system based on multi-source fusion. Background Along with the rapid development of technologies such as automatic driving, intelligent robots, digital twinning and the like, the requirements on high-precision positioning and three-dimensional modeling of environmental space are increasingly urgent. The technology core is that environmental data is acquired through a sensor, and accurate space coordinate information and a three-dimensional model are generated after the environmental data is processed, so that a data basis is provided for autonomous decision-making of equipment. At present, the mainstream space positioning and modeling schemes mainly comprise a vision/inertia fusion scheme based on 3D SLAM (synchronous positioning and map construction), a scheme of realizing positioning and map construction through vision feature matching and motion estimation by relying on a camera, an IMU (inertial measurement unit) and other sensors, and a scheme of obtaining environment three-dimensional point cloud data by utilizing a laser ranging principle based on a laser radar, and directly generating space contour information. However, the prior art has obvious defects and shortcomings that the pure 3D SLAM scheme is easily influenced by environmental illumination change and texture missing scene, the success rate of feature matching is reduced, positioning drift is caused, the precision of image construction is reduced, and the robustness is insufficient. The laser radar single-point ranging scheme lacks global positioning capability, is easy to generate accumulated errors when being used alone, and has weak distinguishing capability on dynamic obstacles and high redundancy of point cloud data. Most of the existing fusion schemes are simple data superposition, depth coordination is not realized, the problem of poor data synchronism and inconsistent space coordinate mapping exists, and complementary advantages of the two technologies are difficult to develop. The space precision optimizing means of the partial scheme is single, only depends on the upgrade of sensor hardware or the correction of a simple algorithm, does not form a full-flow precision control system, and is difficult to meet the use requirement of a high-precision scene. According to the method, a device, computer equipment and a storage medium for positioning the laser radar SLAM are disclosed in China patent publication No. 117949968A, publication No. 2024 and No. 6 and 21, a target scene is scanned through a laser radar and a vision sensor to obtain a laser SLAM point cloud containing a vision image, identification points are arranged in the target scene, under the condition that the scene degradation degree of the target scene meets preset conditions, semantic identification processing is conducted on the laser SLAM point cloud to identify the identification points in the laser SLAM point cloud, three-dimensional coordinates of the identification points are extracted, priori characteristics of the identification points are determined, information of the characteristic points and the characteristic points of the laser SLAM point cloud is obtained, and on the basis of the information of the characteristic points and the characteristic points, the three-dimensional coordinates and the priori characteristics of the identification points are combined, pose optimization is conducted to obtain positioning data, and the technical problems that in the scenes with fewer textures such as tunnels and long corridor are difficult to match, positioning is unstable are effectively solved, but the laser SLAM algorithm characteristics are only fused with the technology of the laser SLAM point cloud to a certain extent, and the dynamic position change of the scenes with fewer textures in the corridor is difficult to adapt to the dynamic scene positioning of the scenes with fewer and the corridor. Disclosure of Invention Aiming at the problem that the existing space positioning method is difficult to ensure the positioning accuracy is lower due to the robustness of positioning under the condition of realizing global positioning, the invention provides a space positioning optimization method and a system based on multi-source fusion, which break the multi-source data barriers by collecting multi-source positioning data and constructing a fixed world coordinate system to fuse and complement the multi-source positioning data, the data reliability under the complex working condition environment is guaranteed through a multiple verification mechanism, the higher robustness can be guaranteed on the premise of realizing global positioning, the precision of scene positioning is guaranteed, and finally the accumulated drift is globally eliminated through a