CN-120195712-B - Multi-source data fusion positioning method based on low-orbit satellite assistance

CN120195712BCN 120195712 BCN120195712 BCN 120195712BCN-120195712-B

Abstract

The invention discloses a multi-source data fusion positioning method based on low-orbit satellite assistance, which comprises the steps of obtaining three-dimensional space information and vehicle data of surrounding environments of a vehicle by using LiDAR, IMU and ODO installed on the vehicle, carrying out pre-integration calculation on the vehicle data obtained by measuring the IMU and the ODO to obtain motion estimation information of the vehicle, carrying out motion compensation on laser point cloud data by using the motion estimation information, carrying out feature extraction and matching on the compensated laser point cloud data to calculate relative pose information of the vehicle, obtaining global position information of the vehicle in an urban canyon complex scene by using a low-orbit satellite, fusing the global position information with the motion estimation information and the relative pose information, and realizing high-precision positioning of the vehicle in the urban canyon scene by using a graph optimization method.

Inventors

LI DAICHENG
GAO CHAOJUN
SUN JIANGLAN
CAI YINGFENG
WANG HAI
LIU ZE
CHEN LONG
JIANG JIN

Assignees

江苏大学

Dates

Publication Date: 20260512
Application Date: 20250415

Claims (7)

1. The multi-source data fusion positioning method based on the assistance of the low-orbit satellite is characterized by comprising the following steps of: Step 1, three-dimensional space information and vehicle data of the surrounding environment of a vehicle are obtained by using LiDAR, IMU and ODO installed on the vehicle; Step 2, pre-integral calculation is carried out on vehicle data obtained by measuring the IMU and the ODO to obtain motion estimation information of the vehicle, wherein the pre-integral calculation comprises the steps of respectively carrying out pre-integral calculation on acceleration and angular velocity measured by the IMU and displacement and velocity measured by the ODO, fusing pre-integral results of the IMU and the ODO through an EKF algorithm to obtain the motion estimation information of the vehicle, and the motion estimation information of the vehicle is recorded as Including the current position of the vehicle Speed and velocity of And a gesture ; Step 3, performing motion compensation on the laser point cloud data by utilizing the motion estimation information, performing feature extraction and matching on the compensated laser point cloud data, and calculating relative pose information of the vehicle; Step 4, acquiring global position information of the vehicle in the urban canyon complex scene by using a low-orbit satellite, fusing the global position information with motion estimation information and relative pose information, and realizing high-precision positioning of the vehicle in the urban canyon scene by using a graph optimization method; In the process of optimizing the map, global position information provided by low-orbit satellites is provided As global constraint conditions, the target function is constructed by fusing motion estimation information and relative pose information of the vehicle, and pose estimation of the vehicle is optimized by minimizing observation errors, and is recorded as follows: ; In the formula, Is a state vector of the prediction and, Is the gain of the kalman, Is an observation value of the current, Is a nonlinear observation function; the calculation formula of the observation error is expressed as follows: ; In the formula, And The position and attitude of the vehicle during the map optimization process, Is global position information provided by the low-orbit satellites, Is the pose information provided by the IMU.
2. The method for multi-source data fusion positioning based on low-orbit satellite assistance according to claim 1, wherein the IMU measures acceleration and angular velocity information of the vehicle in real time.
3. The method for multi-source data fusion positioning based on low-orbit satellite according to claim 1, wherein the ODO is mounted on a wheel, and the measured wheel rotation angle and wheel rotation speed are used to calculate the displacement and speed of the vehicle.
4. The method for multi-source data fusion positioning based on low-orbit satellite assistance according to claim 1, wherein the LiDAR scans the surrounding environment of the vehicle in real time, acquires high-resolution laser point cloud data, and provides three-dimensional space information of the surrounding environment of the vehicle.
5. The method for multi-source data fusion positioning based on low-orbit satellite based assistance according to claim 1, wherein the motion compensation is divided into translational compensation and rotational compensation, and the current position of the vehicle is used first Translational compensation is carried out on laser point cloud, and posture of vehicle is used again And performing rotation compensation on the laser point cloud.
6. The method for multi-source data fusion positioning based on low-orbit satellite assistance according to claim 5, wherein the motion compensation and rotation compensation formulas are expressed as: ; ; In the formula, Is the compensated laser point cloud coordinates, Is the original coordinates of the laser point cloud, Is the current location of the vehicle and, Is the gesture.
7. The method for multi-source data fusion positioning based on low-orbit satellite assistance according to claim 1, wherein the method for extracting and matching the characteristics of the laser point cloud data is as follows: taking one point in the laser point cloud, setting a point set in the neighborhood of the point cloud as a point set, and calculating the curvature of the point by calculating the neighborhood average distance of the point cloud; based on curvature and threshold of points in the point cloud Comparing, dividing the points into angular points, plane points and edge points, and extracting a group of characteristic points consisting of the angular points and the edge points; and calculating the rigid transformation between the two frames of laser point cloud data according to the characteristic point matching errors in the laser point clouds at the current moment and the previous moment, and obtaining the relative pose of the vehicle.

Description

Multi-source data fusion positioning method based on low-orbit satellite assistance Technical Field The invention relates to the technical field of intelligent automobile high-precision positioning, in particular to a multi-source data fusion positioning method based on low-orbit satellite assistance. Background In recent years, with rapid development of intelligent automobiles and autopilot technologies, high-precision positioning has become one of core technologies in intelligent driving systems. The accurate positioning information of the vehicle is the basis for realizing the functions of follow-up environment sensing, path planning, decision control and the like, and is very important for guaranteeing the safety and reliability of intelligent automobile running. The existing intelligent automobile positioning technology is generally divided into a single sensor and a multi-sensor data fusion method. Single sensors include GNSS, liDAR, inertial Measurement Unit (IMU), vision cameras, and the like. GNSS can provide stable positioning information in open scenes, but in urban canyons or other signal-prone areas, positioning stability is significantly degraded. LiDAR can generate high resolution environmental maps, but is costly and performs poorly in severe weather. Although IMU can collect vehicle dynamic information at high frequency and calculate the pose in the course of travel in combination with a given initial pose, accumulated errors can also occur over time. Visual cameras can provide high accuracy positioning information in well lit environments, but can be limited by illumination variations and computer image processing capabilities. In order to solve the above problems, the multi-sensor data fusion positioning technology is becoming a research hotspot. By fusing GNSS, IMU, liDAR with visual and other multi-source data, the defects existing in the positioning of a single sensor can be effectively overcome. However, in the urban canyon semi-closed scene, the traditional multi-sensor data fusion method still has the problems of insufficient GNSS signals, difficult extraction of environmental features and the like, so that the positioning accuracy is difficult to meet the actual driving requirement of the intelligent automobile. Disclosure of Invention In order to solve the defects in the prior art, the application provides a multi-source data fusion positioning method based on low-orbit satellite assistance, which remarkably improves the high-precision positioning capability of an intelligent automobile in an urban canyon complex environment by utilizing low-orbit satellite, multi-sensor data fusion and graph optimization technology. The technical scheme adopted by the invention is as follows: A multi-source data fusion positioning method based on low orbit satellite assistance comprises the following steps: Step 1, three-dimensional space information and vehicle data of the surrounding environment of a vehicle are obtained by using LiDAR, IMU and ODO installed on the vehicle; Step 2, pre-integral calculation is carried out on the vehicle data obtained through IMU and ODO measurement, and motion estimation information of the vehicle is obtained; Step 3, performing motion compensation on the laser point cloud data by utilizing the motion estimation information, performing feature extraction and matching on the compensated laser point cloud data, and calculating relative pose information of the vehicle; And 4, acquiring global position information of the vehicle in the urban canyon complex scene by using a low-orbit satellite, fusing the global position information with motion estimation information and relative pose information, and realizing high-precision positioning of the vehicle in the urban canyon scene by using a graph optimization method. Further, the IMU measures acceleration and angular velocity information of the vehicle in real time. Further, the ODO is mounted on the wheels, and the measured wheel rotation angle and wheel rotation speed are used for calculating the displacement and speed of the vehicle. Further, the LiDAR scans the surrounding environment of the vehicle in real time, acquires high-resolution laser point cloud data, and provides three-dimensional space information of the surrounding environment of the vehicle. Further, the pre-integral calculation comprises the steps of respectively carrying out pre-integral calculation on the acceleration and the angular velocity measured by the IMU and the displacement and the velocity measured by the ODO, and fusing the pre-integral results of the IMU and the ODO through an EKF algorithm to obtain vehicle motion estimation information, wherein the vehicle motion estimation information is recorded asIncluding the current position P k, speed V k, and attitude R k of the vehicle. Further, the motion compensation is divided into translational compensation and rotational compensation, wherein the translational compensation is performed on