A combined positioning method, device and medium in a denial environment

Abstract

The application discloses a combined positioning method and device in a denial environment and a medium, and belongs to the positioning field. The application obtains local coordinate system pose trajectory by tightly coupling the inertial data of an inertial measurement unit and the point cloud data of a laser radar and performing pose calculation on the inertial data and the point cloud data, filters global positioning data based on a quality parameter, constructs a trajectory point pair set, solves the transformation parameters between the local and global coordinate systems by using a sliding window optimization algorithm, converts the local coordinate system pose trajectory to the global coordinate system according to the transformation parameters, and finally generates a global positioning result. The application effectively solves the problem that the prior art cannot accurately and efficiently perform global positioning in a GNSS denial environment.

Description

Technical Field

[0001] This invention relates to the field of positioning, and more particularly to a combined positioning method, apparatus and medium in a denied environment. Background Technology

[0002] In the fields of mobile mapping, autonomous driving, embodied intelligence, and indoor / outdoor robot positioning, there is an extremely high demand for continuous, high-precision global navigation and positioning capabilities. Centimeter-level positioning accuracy is the core foundation for stable operation and precise navigation of various devices. Currently, real-time differential positioning technology based on the Global Navigation Satellite System (GNSS) is the mainstream high-precision positioning method in open outdoor environments. This technology can achieve three-dimensional global positioning of longitude, latitude, and elevation based on satellite signals, with positioning accuracy reaching the centimeter level. It has the advantages of convenient operation and continuous positioning, and has become the preferred solution for outdoor positioning in various fields.

[0003] However, the positioning accuracy of real-time differential positioning technology is highly dependent on the signal quality of the Global Navigation Satellite System (GNSS). In GNSS-denied environments such as densely built-up areas, canyons, indoors, woodlands, and garages, where satellite signals are blocked or interfered with, its positioning accuracy drops sharply to the decimeter or even meter level. In some scenarios, positioning failure may even occur. This deficiency has become a core bottleneck restricting the cross-scenario application of real-time differential positioning technology. To solve this problem, multi-sensor fusion navigation and positioning schemes have become the main research direction of existing technologies. Among them, the fusion scheme of real-time differential positioning and inertial measurement unit (INS) is widely used, attempting to compensate for the accuracy loss of real-time differential positioning in GNSS-denied environments by leveraging the short-time, high-precision pose estimation capability of the INS.

[0004] However, inertial measurement units (IMUs) can only calculate pose using motion parameters collected by their own sensors and lack the ability to perceive the surrounding environment. Their positioning errors accumulate over time and distance. Even in high-precision IMU applications, positioning errors can reach meter-level after only tens of seconds of operation. The error accumulation problem is even more pronounced in low-cost IMUs, making it impossible to maintain high-precision positioning in environments where GNSS is denied for extended periods. In contrast, laser-synchronized positioning and mapping (LSMT) technology, based on lidar-inertial odometry, combines the environmental point cloud perception capabilities of lidar with the motion parameters of the IMU to achieve centimeter-level relative pose constraints within a local area. Furthermore, its positioning drift is extremely slow, making it an effective technique for local positioning in GNSS-denied environments.

[0005] However, laser-synchronized positioning and mapping (LSPM) technology still has significant technical shortcomings. Relying on its own constructed local world coordinate system, it cannot independently provide global positioning information and lacks global coordinate consistency. Furthermore, drift errors such as scale and rotation in local positioning accumulate over time, leading to a decline in positioning accuracy under long-term operation. In summary, all existing navigation and positioning methods have insurmountable technical limitations. Using them alone cannot achieve consistent high-precision global positioning both indoors and outdoors. Simple sensor fusion schemes have not solved core issues such as data filtering, precise coordinate system alignment, and dynamic error suppression. An effective fusion scheme that can comprehensively integrate global positioning and high-precision local pose estimation has not yet been formed. These deficiencies prevent existing technologies from accurately and efficiently performing global positioning in GNSS-denied environments. Summary of the Invention

[0006] This invention provides a combined positioning method, apparatus, and medium in a GNSS denied environment to solve the problem that existing technologies cannot accurately and efficiently perform global positioning in a GNSS denied environment.

[0007] Firstly, this application provides a combined localization method under denied conditions, comprising: Acquire environmental point cloud data collected by lidar, inertial data collected by inertial measurement unit, and global positioning data and mass parameters collected by real-time differential positioning device; The inertial data and the environmental point cloud data are tightly coupled to calculate the pose, and the pose trajectory in the local coordinate system is obtained. Based on the quality parameters, the global positioning data is filtered, and the filtered global positioning data and the pose trajectory in the local coordinate system are used to construct a set of trajectory point pairs. Based on the sliding window optimization algorithm, the set of trajectory point pairs is solved to obtain the transformation parameters between the local coordinate system and the global coordinate system; Based on the transformation parameters, the pose trajectory in the local coordinate system is transformed to the global coordinate system to generate a global positioning result.

[0008] This application acquires environmental point cloud data from a lidar system, inertial data from an inertial measurement unit (IMU), and global positioning data and mass parameters from a real-time differential positioning (RTD) device. It then performs tight-coupled pose estimation using the inertial data and environmental point cloud data to obtain the pose trajectory in a local coordinate system. Based on the mass parameters, it filters the global positioning data to construct a set of trajectory point pairs. A sliding window optimization algorithm is then used to solve the trajectory point pair set to obtain the transformation parameters between the local and global coordinate systems. Finally, based on the transformation parameters, the pose trajectory in the local coordinate system is transformed to the global coordinate system to generate a global positioning result. Since the lidar inertial odometry can... While providing high-precision relative pose constraints with slow drift, real-time differential positioning (RTD) offers global positioning information but lacks pose determination capabilities and its accuracy decreases when the high-precision attenuation factor or coordinate standard deviation is large. Therefore, this application filters out low-quality global positioning data through quality parameters and uses a sliding window optimization algorithm to calculate the transformation parameters between the two coordinate systems in real time. This ensures that the local high-precision pose of the lidar inertial odometry achieves global consistency, while avoiding the impact of global positioning data quality degradation or long-term cumulative drift of the lidar inertial odometry on positioning accuracy. This allows for the continuation of high-precision global positioning capabilities even in GNSS-denied environments, achieving seamless high-precision positioning both indoors and outdoors. This application effectively solves the problem that existing technologies cannot accurately and efficiently perform global positioning in GNSS-denied environments.

[0009] Furthermore, the tight-coupled pose estimation of the inertial data and the environmental point cloud data specifically involves: The inertial data is integrated to obtain the initial pose prediction value; Based on the initial pose prediction value, motion distortion correction is performed on the environmental point cloud data to obtain the corrected point cloud data; The corrected point cloud data is then registered to obtain the registered pose. Based on the registered pose and the inertial data, a joint optimization function is constructed; The joint optimization function is optimized and solved to obtain the optimized pose as the pose trajectory in the local coordinate system.

[0010] This application obtains an initial pose prediction value by integrating inertial data. Based on this initial pose prediction value, motion distortion correction is performed on the environmental point cloud data to obtain corrected point cloud data. Then, point cloud registration is performed on the corrected point cloud data to obtain the registered pose. Based on the registered pose and inertial data, a joint optimization function is constructed and optimized to obtain the pose trajectory in the local coordinate system. Since inertial data integration can provide high-frequency pose prediction but has accumulated errors, while point cloud registration can provide relatively accurate pose but is affected by motion distortion, the pose prediction is obtained by integrating inertial data to correct the motion distortion of the point cloud. Then, the corrected point cloud registration result is used to jointly optimize with the inertial data. This allows the accumulated error of the inertial data to be constrained by the point cloud registration result, and the accuracy of the point cloud registration is also supported by the high-frequency prediction of the inertial data. This improves the accuracy and stability of the pose trajectory calculation in the local coordinate system, providing a high-quality local pose foundation for subsequent fusion and alignment with global positioning data.

[0011] Furthermore, the filtering of the global positioning data based on the quality parameters specifically involves: Extract the horizontal accuracy attenuation factor and coordinate standard deviation from the quality parameters; Determine whether the horizontal accuracy attenuation factor is greater than a first preset threshold, or determine whether the coordinate standard deviation is greater than a second preset threshold; If the horizontal accuracy attenuation factor is greater than the first preset threshold, or the coordinate standard deviation is greater than the second preset threshold, then the corresponding global positioning data is discarded.

[0012] This application extracts the horizontal accuracy attenuation factor and coordinate standard deviation from the quality parameters, and determines whether the horizontal accuracy attenuation factor is greater than a first preset threshold or whether the coordinate standard deviation is greater than a second preset threshold. If the horizontal accuracy attenuation factor is greater than the first preset threshold or the coordinate standard deviation is greater than the second preset threshold, the corresponding global positioning data is removed. Since the horizontal accuracy attenuation factor can reflect the influence of satellite geometric distribution on horizontal positioning error in advance, and the coordinate standard deviation can directly characterize the accuracy level of positioning coordinates, by setting thresholds to jointly judge these two parameters, low-quality data can be identified and removed in a timely manner when the accuracy of global positioning data degrades or satellite signals are blocked. This avoids global positioning data with large errors from participating in subsequent trajectory alignment calculations, thereby improving the calculation accuracy and stability of transformation parameters between the local coordinate system and the global coordinate system, and ensuring the reliability of global positioning results under denied conditions.

[0013] Furthermore, the step of constructing a set of trajectory point pairs from the filtered global positioning data and the pose trajectory in the local coordinate system specifically involves: Obtain the timestamp corresponding to the filtered global location data; Based on the timestamp, pose interpolation is performed on the pose trajectory in the local coordinate system to obtain the local pose corresponding to the filtered global positioning data. The filtered global positioning data is combined with the corresponding local pose to form trajectory point pairs; Based on the trajectory point pairs, a set of trajectory point pairs corresponding to the filtered global positioning data and the pose trajectory in the local coordinate system is constructed.

[0014] This application obtains the timestamps corresponding to the filtered global positioning data, performs pose interpolation on the pose trajectory in the local coordinate system based on the timestamps to obtain the local poses corresponding to the filtered global positioning data, and then forms trajectory point pairs with the filtered global positioning data and the corresponding local poses to construct a trajectory point pair set. Since there are differences in the data acquisition frequency and transmission delay between the real-time differential positioning device and the lidar inertial odometry, it is difficult for the data of the two to correspond directly at the same time. Therefore, pose interpolation is performed using timestamps to enable the high-frequency local pose trajectory to be accurately aligned with the low-frequency global positioning data in the time dimension, thereby ensuring the temporal consistency and spatial correspondence between the global positioning data and the local poses, and providing an accurate and reliable trajectory point pair foundation for subsequent trajectory alignment calculation based on the sliding window optimization algorithm.

[0015] Furthermore, the sliding window optimization algorithm is used to solve the set of trajectory point pairs to obtain the transformation parameters between the local coordinate system and the global coordinate system, specifically: Based on the relative pose changes between adjacent local poses in the set of trajectory point pairs, a pose factor is constructed. An alignment factor is constructed based on the coordinate difference between the filtered global positioning data and the corresponding local pose in the trajectory point pair set. Construct a factor graph based on the pose factor and the alignment factor; Incremental smoothing optimization is performed on the factor graph to obtain the rotation transformation parameters and translation transformation parameters between the local coordinate system and the global coordinate system, which are then used as the transformation parameters.

[0016] This application constructs a pose factor based on the relative pose change between adjacent local poses in a set of trajectory point pairs, and an alignment factor based on the coordinate difference between the global positioning data and the corresponding local poses after filtering the global positioning data in the set of trajectory point pairs. Then, a factor graph is constructed based on the pose factor and the alignment factor, and incremental smoothing optimization is performed to obtain the rotation transformation parameters and translation transformation parameters between the local coordinate system and the global coordinate system as transformation parameters. Since the pose factor can constrain the relative geometric relationship between adjacent poses in the local coordinate system to ensure the continuity of the local trajectory, and the alignment factor can constrain the consistency between the local pose and the global positioning data after transformation, the joint optimization of the two types of factors enables the pose trajectory in the local coordinate system to achieve accurate alignment with the global coordinate system while maintaining relative geometric consistency. At the same time, the incremental smoothing optimization algorithm can efficiently process newly added trajectory point pairs and update the transformation parameters in real time, thereby realizing the real-time dynamic calculation of the transformation parameters between the local coordinate system and the global coordinate system. This effectively suppresses the cumulative drift caused by the long-term operation of the lidar inertial odometry and ensures the accuracy and timeliness of the global positioning results in the denied environment.

[0017] Furthermore, the incremental smoothing optimization of the factor graph specifically involves: Add the alignment factor corresponding to the most recently added trajectory point pair in the set of trajectory point pairs to the factor graph; Determine if the number of aligned factors in the current factor graph exceeds the preset window size; If the number of alignment factors in the current factor graph exceeds the preset window size, then remove the alignment factor with the earliest timestamp from the factor graph; The updated factor graph is optimized based on incremental smoothing and graph building algorithms to obtain optimized rotation transformation parameters and translation transformation parameters; Calculate the heading angle error based on the optimized rotation transformation parameters; Determine whether the heading angle error is greater than a preset angle threshold; If the heading angle error is greater than a preset angle threshold, the corresponding alignment factor is marked as an abnormal factor, and the abnormal factor is removed from the factor graph before the next optimization.

[0018] This application adds the alignment factor corresponding to the most recently added trajectory point pair in the trajectory point pair set to the factor graph. It then determines whether the number of alignment factors in the current factor graph exceeds the preset window size. If it does, the alignment factor with the earliest timestamp is removed from the factor graph. Subsequently, the updated factor graph is optimized based on incremental smoothing and mapping algorithms to obtain optimized rotation transformation parameters and translation transformation parameters. The heading angle error is calculated based on the optimized rotation transformation parameters. If the heading angle error is greater than a preset angle threshold, the corresponding alignment factor is marked as an abnormal factor and removed before the next optimization. Since the pose trajectory of the LiDAR inertial odometry drifts slowly with time and distance, and the global positioning data may contain outliers that have not been completely removed, a sliding window mechanism is used to retain only the most recently added trajectory point pairs for optimization. This allows the transformation parameters to be updated in real time as the device moves to adapt to the drift changes of the local coordinate system. At the same time, the heading angle error detection identifies and removes alignment factors that cause abnormal optimization results, avoiding the impact of low-quality global positioning data or local pose anomalies on the transformation parameter calculation. This achieves real-time dynamic optimization of transformation parameters and robust handling of anomalies, ensuring the long-term accuracy and stability of global positioning results in the denied environment.

[0019] Furthermore, the step of transforming the pose trajectory in the local coordinate system to the global coordinate system according to the transformation parameters to generate a global positioning result specifically involves: Determine whether the quality parameters meet the preset GNSS signal normal determination conditions; If the quality parameters meet the GNSS signal normal determination conditions, the global positioning data collected by the real-time differential positioning device will be output as the global positioning result. If the quality parameters do not meet the GNSS signal normal determination conditions, the pose trajectory in the local coordinate system is transformed to the global coordinate system according to the transformation parameters, and the transformed coordinate data is output as the global positioning result.

[0020] This application determines whether the quality parameters meet the preset GNSS signal normality judgment conditions. If they do, the global positioning data collected by the real-time differential positioning device is output as the global positioning result. If not, the pose trajectory in the local coordinate system is transformed to the global coordinate system according to the transformation parameters, and the transformed coordinate data is output as the global positioning result. Since the real-time differential positioning device can provide direct high-precision global positioning when the GNSS signal is good, but the signal quality degrades in the GNSS rejection environment, leading to a decrease or failure in positioning accuracy, the quality parameters are used for real-time judgment. When the signal is normal, the real-time differential positioning result is directly used to ensure positioning accuracy and efficiency. When the signal is abnormal, it automatically switches to a combined positioning mode based on lidar inertial odometry and transformation parameters to continue the global positioning capability. This achieves seamless switching and adaptive positioning between indoor and outdoor environments, ensuring the continuity and reliability of global positioning results in complex environments.

[0021] Furthermore, after generating the global localization result, the process also includes: Obtain the rigid body structural parameters of the device in the lidar coordinate system, and determine the local coordinates of the two reference points of the centering rod in the lidar coordinate system; Based on the local coordinates of the two reference points, the direction vector of the centering rod tip is calculated, and the direction vector is normalized to obtain the unit direction vector of the centering rod tip. Obtain the preset centering rod height, and extend the local coordinates of the preset reference point among the two reference points towards the rod tip along the unit direction vector to calculate the local coordinates of the centering rod tip in the lidar coordinate system. The local coordinates of the centering rod tip are transformed to the global coordinate system according to the transformation parameters to obtain the global positioning coordinates of the centering rod tip.

[0022] This application determines the local coordinates of two reference points of the centering rod in the lidar coordinate system by obtaining the rigid body structural parameters of the device after generating the global positioning result. Then, it calculates the direction vector of the centering rod tip based on the local coordinates of the two reference points and normalizes it to obtain the unit direction vector. Next, it obtains the preset centering rod height and extends the unit direction vector from the local coordinates of the preset reference point to the rod tip to calculate the local coordinates of the centering rod tip in the lidar coordinate system. Finally, it transforms the local coordinates of the centering rod tip to the global coordinate system according to the transformation parameters to obtain the global positioning coordinates of the centering rod tip. Since the target point that the user needs to measure in actual surveying operations is the tip of the centering rod, rather than the center of the lidar or the phase center of the GNSS antenna, the direction of the centering rod is determined by the rigid body structural parameters of the device and tilt measurement correction is performed in combination with the preset rod height. This allows the positioning result to be corrected from the center of the lidar to the actual measurement target point, i.e., the tip of the centering rod. This achieves high-precision global positioning of the centering rod tip, meets the need for direct positioning of actual target points in mobile surveying in complex environments, and improves the convenience and accuracy of surveying operations.

[0023] Secondly, this application provides a combined positioning device for a denied environment. The combined positioning device for a denied environment includes: The acquisition module is used to acquire environmental point cloud data collected by the lidar, inertial data collected by the inertial measurement unit, and global positioning data and mass parameters collected by the real-time differential positioning device. The estimation module is used to perform tightly coupled pose estimation on the inertial data and the environmental point cloud data to obtain the pose trajectory in the local coordinate system. The filtering module is used to filter the global positioning data based on the quality parameters, and construct a set of trajectory point pairs by combining the filtered global positioning data with the pose trajectory in the local coordinate system. The solution module is used to solve the set of trajectory point pairs based on the sliding window optimization algorithm to obtain the transformation parameters between the local coordinate system and the global coordinate system; The transformation module is used to transform the pose trajectory in the local coordinate system to the global coordinate system according to the transformation parameters, and generate a global positioning result.

[0024] This application acquires environmental point cloud data from a lidar sensor, inertial data from an inertial measurement unit (IMU), and global positioning data and mass parameters from a real-time differential positioning (RTD) device via an acquisition module. An estimation module tightly couples the inertial data and environmental point cloud data to calculate the pose trajectory in the local coordinate system. A filtering module filters the global positioning data based on the mass parameters and constructs a set of trajectory point pairs with the filtered global positioning data and the pose trajectory in the local coordinate system. A solution module uses a sliding window optimization algorithm to solve the trajectory point pair set to obtain the transformation parameters between the local and global coordinate systems. Finally, a transformation module... The module transforms the pose trajectory in the local coordinate system to the global coordinate system based on the transformation parameters to generate a global positioning result. Since each functional module corresponds to a key processing link in the method and the data flow relationship is clear, the modular design enables the efficient execution and collaborative operation of the combined positioning method in the denied environment. This allows the high-precision local relative positioning capability of the lidar inertial odometry to be organically integrated with the global positioning capability of real-time differential positioning. At the same time, the real-time dynamic processing of the filtering module and the solution module ensures the system's adaptability and robustness in complex environments, thereby realizing the automated and engineering application of high-precision global positioning in the denied environment.

[0025] Thirdly, this application provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the combined positioning method under a denied environment as described above. Its beneficial effects are the same as those of the combined positioning method under a denied environment provided in the first aspect of this application. Attached Figure Description

[0026] Figure 1 : A schematic flowchart of an embodiment of the combined positioning method under denied conditions provided in this application; Figure 2 : A schematic diagram of an embodiment of the RTK-SLAM hardware structure provided in this application; Figure 3 : A schematic diagram of an embodiment of STD / HDOP that undergoes mutation due to environmental influences, provided in this application; Figure 4 : A schematic diagram of an embodiment of STD / HDOP-assisted removal of pseudo-fixed RTK points under eaves provided in this application; Figure 5 : A schematic diagram of an embodiment of the factor graph structure provided in this application; Figure 6 : A schematic diagram of an embodiment of the RTK trajectory and LIO trajectory before and after alignment provided in this application; Figure 7 : A schematic diagram of an embodiment of the alignment angle error provided in this application; Figure 8 : A schematic diagram of an embodiment of the RTK-LIO sliding window alignment process provided in this application; Figure 9 : A schematic diagram of an embodiment of the RTK-LIO combined measurement mode provided in this application; Figure 10 : A schematic diagram of an embodiment of RTK-LIO tilt measurement provided in this application; Figure 11 : A schematic diagram of an embodiment of the test scenario (underground parking garage, building, indoor) provided in this application; Figure 12 : A schematic diagram of an embodiment of the combined positioning device in a denied environment provided in this application. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Example 1 Please refer to Figure 1 In order to solve the problem that existing technologies cannot accurately and efficiently perform global positioning in GNSS denied environments, this invention provides a combined positioning method in denied environments, including steps S01-S05.

[0029] This application is based on Figure 2 The multi-sensor integrated device shown includes a South Surveying and Mapping RTK unit on top, containing a GNSS antenna; and a MID-360 semi-solid-state lidar unit (with a built-in ICM40609 IMU), left and right binocular panoramic cameras, a removable handle battery, a base bracket, and a reserved centering rod interface at the bottom. The above hardware device will be referred to below as an RTK-SLAM device.

[0030] RTK measurement data includes coordinates in three dimensions: longitude, latitude, and altitude, as well as the standard deviation (STD) of the horizontal and vertical coordinates, and the horizontal dilution of precision (HDOP), which quantifies the impact of satellite geometry on horizontal positioning error. A smaller STD / HDOP indicates higher positioning accuracy. Figure 3 As shown, after timestamp 1763715836.73s, when the GNSS antenna equipment moved from outdoors to indoors, the STD value increased dramatically. The STD in the three dimensions increased from nearly 0 meters (actual order of magnitude 1e-2) to approximately 5, 6, and 10 meters, respectively. At the same time, when the equipment was close to a building, due to the obstruction of some GNSS signals in the air direction, the HDOP value fluctuated drastically and increased rapidly from a relatively stable 0.7 (or below 0.7). Moreover, HDOP increased before the standard deviation, that is, HDOP detected the GNSS denial environment before the standard deviation. These weak GNSS signals will greatly affect the accuracy and stability of the multi-source sensor data combination solution, so they need to be screened and removed. Figure 4 These are pseudo-fixed RTK points under the eaves that were removed with the help of STD / HDOP in the actual test scenario.

[0031] S01: Acquire environmental point cloud data collected by lidar, inertial data collected by inertial measurement unit, and global positioning data and mass parameters collected by real-time differential positioning device.

[0032] In a preferred embodiment of this invention, the acquisition of environmental point cloud data collected by the lidar, inertial data collected by the inertial measurement unit, and global positioning data and mass parameters collected by the real-time differential positioning device specifically includes: When performing combined positioning operations in denied environments, raw data from multiple sensors are first collected simultaneously. Among them, the lidar collects environmental point cloud data within its detection range. This lidar is a semi-solid-state lidar with a horizontal detection range of 360° and a vertical detection range of -7° to 52°, which can accurately acquire three-dimensional point cloud information of the surrounding environment. The inertial measurement unit (hereinafter referred to as "IMU") integrated with the lidar simultaneously collects three-axis acceleration data and three-axis angular velocity data as inertial data, providing kinematic basic parameters for pose estimation. The real-time differential positioning device (hereinafter referred to as "RTK device") simultaneously collects global positioning data and quality parameters. The global positioning data is the longitude, latitude, and elevation three-dimensional coordinate data obtained by the device based on the Global Navigation Satellite System (hereinafter referred to as "GNSS"). The quality parameters include the standard deviation (hereinafter referred to as "STD") of each dimension of the positioning coordinates and the horizontal accuracy attenuation factor (hereinafter referred to as "HDOP") used to quantify the influence of satellite geometric distribution on horizontal positioning error. The aforementioned STDs specifically refer to longitude STD, latitude STD, and elevation STD. These quality parameters directly reflect the positioning accuracy and satellite signal quality of the RTK device. The smaller the values ​​of HDOP and STDs in each dimension, the higher the positioning accuracy of the RTK device.

[0033] In practical applications, when a device moves from an open outdoor environment to an indoor GNSS-denied environment, the STD (Standard Distance) in each dimension increases dramatically. For example, after a specific timestamp, the longitude STD, latitude STD, and elevation STD increase from nearly 0 meters (actually on the order of 10⁻² meters) to approximately 5 meters, 6 meters, and 10 meters, respectively. When the device approaches obstructions such as buildings, the HDOP (Highest Distance Position) fluctuates drastically and rises rapidly from a relatively stable state of 0.7 (or below 0.7) due to partial obstruction of GNSS signals in the air direction. Furthermore, the change in HDOP precedes the increase in STD, meaning it can detect the arrival of GNSS-denied environments earlier than STD. These abnormal quality parameter data caused by satellite signal obstruction or interference significantly affect the accuracy and stability of multi-source sensor data combination and calculation. Therefore, it is necessary to filter and eliminate the corresponding global positioning data. For example, in scenarios where satellite signals are partially obstructed, such as under eaves, pseudo-fixed RTK positioning data can be effectively eliminated through dual judgment of HDOP and STD.

[0034] During the data acquisition process, the data collected by the lidar, IMU and RTK devices are strictly time-stamped and synchronized to ensure the consistency of data from each sensor in the time dimension. This avoids deviations in subsequent data processing due to time asynchrony and provides a multi-source raw data foundation with a unified time dimension and complete data dimension for subsequent operations such as tightly coupled pose estimation, data filtering and trajectory alignment.

[0035] S02: Perform tight-coupled pose estimation on the inertial data and the environmental point cloud data to obtain the pose trajectory in the local coordinate system.

[0036] In a preferred embodiment of this invention, the step of performing tightly coupled pose estimation on the inertial data and the environmental point cloud data to obtain the pose trajectory in the local coordinate system specifically involves: After completing the acquisition and timestamp synchronization of multi-source sensor data, tightly coupled pose estimation is performed based on inertial data and environmental point cloud data to obtain the pose trajectory in the local coordinate system. The specific estimation process is as follows: First, the inertial data such as three-axis acceleration and three-axis angular velocity acquired by the IMU are integrated and combined with the inertial navigation solution method to obtain the initial pose prediction value of the carrier. This initial pose prediction value can reflect the real-time pose change trend of the carrier during the motion process. Based on this initial pose prediction value, motion distortion correction is performed on the environmental point cloud data acquired by the lidar to eliminate the point cloud distortion problem caused by the lidar's point-by-point scanning during the carrier's motion, and the corrected point cloud data is obtained to ensure the spatial position accuracy of the point cloud data. Subsequently, point cloud registration processing is performed on the corrected point cloud data. Through feature extraction and matching, the current frame point cloud and key frame point clouds are registered to obtain the registered pose data. Then, a joint optimization function is constructed based on the registered pose and the original inertial data. Using the IMU's zero bias and the carrier's pose as optimization variables, and combining the constraint information of the lidar point cloud and the motion constraint information of the inertial data, the joint optimization function is solved nonlinearly. During the optimization process, the IMU's acceleration zero bias and angular velocity zero bias are corrected simultaneously to eliminate the influence of the inertial device's systematic errors on the calculation results. Furthermore, key frames are selected based on the carrier's motion distance and rotation angle characteristics, and a factor map is built based on the key frames. The stable triangle description algorithm is used to realize loop closure detection, further correcting the accumulated errors in the pose calculation process. Finally, the continuous and high-precision optimized pose obtained through multiple rounds of optimization is the pose trajectory in the local coordinate system.

[0037] It should be noted that pure laser inertial odometry (LIO) does not possess global positioning information. Its corresponding Simultaneous Localization and Mapping (SLAM) architecture automatically constructs a local world coordinate system as the reference coordinate system for measurement during the initialization phase. When the device is stationary, the acceleration measured by the IMU is equal in magnitude and opposite in direction to the gravitational acceleration. Therefore, in the LIO-based SLAM architecture, the IMU coordinate system in the initial stationary phase (i.e., a three-dimensional Cartesian coordinate system with the IMU center as the origin and the IMU's axes as coordinate axes) is transformed so that the z-axis of the transformed coordinate system is opposite to the direction of gravitational acceleration. The resulting coordinate system is the local world coordinate system. Although this local world coordinate system is aligned with the absolute world system in the direction of gravity, its x-axis and y-axis are not aligned with the global world coordinate system. As a result, LIO still lacks global positioning capability; that is, there is a yaw angle deviation between the local world coordinate system and the global world coordinate system. The integrated RTK device on the equipment provides global positioning information, effectively compensating for absolute errors. However, this device lacks attitude determination capabilities, thus the two complement each other. By jointly optimizing the pose trajectory of the LIO and the positioning trajectory of the RTK device within a given window, accurate estimation of yaw and translation deviations can be achieved, thereby aligning the local world coordinate system with the global world coordinate system, giving the equipment complete global positioning and attitude determination capabilities. Furthermore, while the aforementioned trajectory alignment method can achieve good global positioning accuracy locally, the positioning accuracy will still decrease after a certain time or distance. The error mainly comes from two aspects: first, the optimized calculation of alignment parameters is not absolutely accurate, resulting in errors in global pose, and these errors become more pronounced with increasing distance; second, the pose of the LIO drifts slowly with time and distance. Therefore, this solution adopts the latest real-time calculation and alignment method within a sliding window to achieve high-precision real-time global absolute positioning.

[0038] S03: Based on the quality parameters, the global positioning data is filtered, and the filtered global positioning data and the pose trajectory in the local coordinate system are used to construct a set of trajectory point pairs.

[0039] In a preferred embodiment of this invention, the step of filtering the global positioning data based on the quality parameters and constructing a set of trajectory point pairs with the filtered global positioning data and the pose trajectory in the local coordinate system is specifically as follows: After obtaining the pose trajectory in the local coordinate system, the global positioning data is first precisely filtered based on the quality parameters collected by the RTK device. Then, the filtered valid global positioning data and the local pose trajectory are used to construct a set of trajectory point pairs. The specific operation process is as follows: First, extract HDOP and the coordinate STDs of longitude, latitude, and elevation from the quality parameters. Pre-set the first preset threshold corresponding to HDOP and the second preset threshold corresponding to the coordinate STD (the first and second preset thresholds are fixed values ​​pre-calibrated according to the actual mapping scenario requirements; for example, the first preset threshold can be set to 1.5, and the second preset threshold can be set to 1.5). The value can be set to 0.5 meters to define the validity of RTK positioning data. The real-time collected HDOP is compared with the first preset threshold, and the STD of each dimension coordinate is compared with the second preset threshold. If the HDOP exceeds the first preset threshold, or the STD of any dimension coordinate exceeds the second preset threshold, the set of global positioning data is determined to be abnormal data and is removed. Only the valid global positioning data in which both HDOP and STD of each dimension coordinate are within the corresponding preset threshold range are retained. This is to exclude pseudo fixed positioning data in environments where GNSS signal obstruction or interference is denied, and to avoid abnormal data causing error interference to subsequent trajectory alignment.

[0040] After filtering the global positioning data, for each set of valid global positioning data, the timestamp corresponding to its acquisition time is extracted. Linear interpolation is then performed on the pose trajectory in the calculated local coordinate system based on this timestamp to calculate the local pose data of the LiDAR that perfectly matches the timestamp, achieving a precise correspondence between the global positioning data and the local pose data in the time dimension. Subsequently, each set of filtered global positioning data with matched timestamps is paired with its corresponding local pose data to form a trajectory point pair. All valid trajectory point pairs are then sequentially aggregated, ultimately constructing a set of trajectory point pairs corresponding to the filtered global positioning data and the pose trajectory in the local coordinate system. This provides a time-synchronized and data-valid foundation dataset for subsequent coordinate system transformation parameter calculation based on the sliding window optimization algorithm.

[0041] S04: Based on the sliding window optimization algorithm, the set of trajectory point pairs is solved to obtain the transformation parameters between the local coordinate system and the global coordinate system.

[0042] In a preferred embodiment of this invention, the step of solving the set of trajectory point pairs using the sliding window optimization algorithm to obtain the transformation parameters between the local and global coordinate systems is specifically as follows: The sliding window optimization algorithm is used to solve the set of trajectory point pairs to obtain the transformation parameters between the local and global coordinate systems. The specific solution process revolves around factor graph construction and incremental smoothing optimization, relying on the sliding window mechanism throughout to ensure the real-time performance and accuracy of the solution. Factor graphs are a mainstream graph model, obtaining a graph structure containing variable nodes and function nodes by factoring a function. This embodiment uses GTSAM (Graph Optimization Toolkit) to construct the factor graph. This toolkit provides core algorithms for factor graph construction and nonlinear optimization such as Levenberg-Marquardt. The iSAM2 incremental smoothing and graph construction optimization algorithms are selected to complete the nonlinear optimization solution of the factor graph.

[0043] like Figure 5 As shown in the constructed factor graph, blue circles represent variable nodes, i.e., the variables to be optimized, where Align is the trajectory alignment parameter that includes rotation and translation. In order to be in The pose of the lidar at any given moment, and To perform at a given time The LiDAR pose obtained by linear interpolation has an interpolation time determined by the input RTK positioning data. The squares represent factors that connect one or more variables, i.e., function nodes. The red squares are pose relative change factors (referred to as "pose factors"), which are used to constrain adjacent pose variable nodes. The green squares are RTK measurement factors defined by the design, which are used to constrain the relationship between trajectory alignment parameters and interpolated pose. Therefore, they are also called trajectory alignment factors (referred to as "alignment factors").

[0044] Adding appropriate alignment factors is key to achieving good alignment between the local and global coordinate systems. As the measuring device moves forward, new pose and alignment factors are continuously added to the factor graph. To ensure the real-time accuracy of trajectory alignment parameter calculation, older alignment factors within the window need to be removed. For example... Figure 6 As shown, in this embodiment, the latest α filtered RTK-LIO point pairs with a certain spacing (α is the threshold for the number of point pairs in the preset window, for example, α can be set to 20, calibrated according to the device's movement speed and positioning accuracy requirements) are selected to calculate the transformation parameters (R,t). There is a significant difference in the trajectory effect before and after alignment: before alignment, the RTK positioning trajectory (red) and the LIO trajectory (blue, which has been converted to the GNSS phase center) are not aligned. After adding β alignment factors (β is the number of alignment factors required for the first alignment calculation, and β < α, for example, β can be set to 5 to ensure the stability of the first alignment) and successfully calculating, the two sets of trajectories can be accurately aligned. Then, the calculated transformation parameters can be used to convert the LIO pose, point cloud map, etc. from local coordinates to global coordinates.

[0045] This embodiment, based on the initial screening using the STD of RTK measurements, proposes a robust RTK factor re-screening strategy, designing two additional refined screening processes to further improve the effectiveness of the RTK data participating in the calculation: First, spatial distance screening, with a preset distance threshold d (d is a fixed value preset according to the device's measurement accuracy, for example, d can be set to 0.3 meters). If the spatial distance between the RTK measurement value and the corresponding LIO interpolation point is greater than the threshold d, the measurement value is directly discarded and not added to the factor graph for calculation; Second, angle error screening, with a preset angle threshold φ (φ is a fixed value preset according to the positioning accuracy requirements, for example, φ can be set to 2°). After adding the alignment factor to the factor graph and completing the optimization calculation, the angle errors in the horizontal and vertical directions are calculated based on the optimized transformation parameters. If the error exceeds the preset threshold φ, the alignment factor is marked and removed from the factor graph before the next optimization. The specific calculation method for the angle error is as follows: Taking the first point within the sliding window as the reference point, calculate the corresponding vector by subtracting the reference point from other LIO trajectory points and RTK trajectory points within the window. Then, based on the obtained vectors, calculate the angle difference between the corresponding vectors of the two trajectories in the vertical and horizontal directions. This is the required alignment angle error. Figure 7 As shown, the blue solid circles represent LIO trajectory points, and the red solid circles represent RTK trajectory points. The above angle error calculation and RTK robustness factor screening process are described in reference [reference needed]. Figure 8 .

[0046] The entire sliding window alignment algorithm process is as follows: The specific calculation operation is as follows: First, the above factor graph model is constructed based on the data in the trajectory point pair set. The pose factor in the function node is constructed according to the relative pose change between adjacent local poses in the trajectory point pair set. The alignment factor is the trajectory alignment factor, which is the residual function node. Its residual equation consists of the difference between the global coordinate measurement value of real-time differential positioning and the local coordinate estimation value of laser inertial odometry. Its goal is to optimize the transformation parameters from local coordinates to global coordinates. The residual equation is: in, This represents the transformation parameters to be optimized. and This indicates that the center of the LiDAR laser head, as calculated by the odometer, is located at... i Position at any given moment express i The ENU coordinates (northeast celestial coordinates, obtained by geodetic coordinate transformation) of the GNSS antenna phase center measured by RTK at that time. It is the vector from the center of the laser head to the phase center in the LiDAR system (i.e., the lever between these two points), a fixed structural parameter used to correct the odometry pose representation from the center of the laser head to the phase center, so as to be consistent with the RTK measurement value.

[0047] Since pose variables and affine transformation parameters (especially rotational components) belong to non-Euclidean space, modern SLAM optimization libraries such as GTSAM typically use manifolds to represent rotation (SO(3)) and pose (SE(3)) variables to avoid singularities (such as gimbal lock) during parameterization and to ensure that iterative updates always reside within a valid geometric space. In SE(3), direct addition and subtraction of states are not allowed; they must be transformed to tangent space (SO(3)). In this process, the state value is modified based on the incremental perturbation, and the result is obtained through an exponential mapping. ) and logarithmic mapping ( The transformation is performed between the elements on the manifold and the elements on the right perturbation model. Therefore, this application derives the above residual equations relative to the right perturbation model. Partial derivative of the right perturbation: The first three columns represent the partial derivatives with respect to the rotational perturbation, and the last three columns represent the partial derivatives with respect to the translational perturbation. To address the issue of antisymmetric matrix operations, where the unobservable nature of one direction during linear motion can lead to large optimization errors or even failure, this embodiment adds... and set the first two columns as Only the yaw angle is optimized and updated. This avoids calculation errors caused by the unobservable nature of a single direction during linear motion.

[0048] After constructing the factor graph, the iSAM2 incremental smoothing and mapping algorithm is used to perform nonlinear optimization calculations on the factor graph. Simultaneously, a sliding window optimization mechanism is introduced to dynamically manage the trajectory point pair set, selecting the latest multiple sets of trajectory point pairs for calculation. When a new trajectory point pair is added to the window, if the number of trajectory point pairs in the window exceeds the preset window size (i.e., threshold α), the trajectory point pair with the earliest timestamp in the window is removed, ensuring that the transformation parameters are always calculated using the latest valid data. During the optimization calculation process, the aforementioned RTK robust factor multiple screening strategy is used to complete a secondary robust screening of the RTK measurement data, achieving precise control over the global positioning data. Through a series of operations including factor graph construction, dynamic updating of the sliding window, incremental smoothing optimization, and robustness screening, the trajectory point pair set is iteratively calculated multiple times. The final converged rotation and translation parameters are the optimal transformation parameters between the local and global coordinate systems. These parameters include the rotation matrix and translation vector required to achieve the transformation from local to global coordinates, enabling precise alignment of the two coordinate systems.

[0049] S05: Based on the transformation parameters, the pose trajectory in the local coordinate system is transformed to the global coordinate system to generate a global positioning result.

[0050] In a preferred embodiment of this invention, the step of transforming the pose trajectory in the local coordinate system to the global coordinate system according to the transformation parameters to generate a global positioning result specifically involves: After calculating the optimal transformation parameters between the local and global coordinate systems, the pose trajectory in the local coordinate system is transformed to the global coordinate system based on these parameters. Combined with the GNSS signal quality assessment results, adaptive switching of the positioning mode is achieved, ultimately generating continuous and high-precision global positioning results. Simultaneously, the RTK-SLAM equipment in this embodiment can fully realize high-precision combined positioning for indoor and outdoor collaboration, and can complete the coordinate correction of the pole tip during surveying operations. The specific operation process is as follows: First, the positioning mode is seamlessly switched based on the real-time assessment results of GNSS signal quality: such as Figure 9 As shown, when the device is in an open outdoor environment, it directly uses RTK measurements as the global high-precision positioning result. At the same time, the SLAM algorithm receives the latest high-precision RTK measurements in real time, continuously calculates and updates the alignment transformation parameters between the local coordinate system and the global coordinate system, ensuring that the device always has the ability to achieve high-precision global positioning outdoors. When the device enters an environment where GNSS is degraded and denied, such as when it is obstructed by buildings or indoors, the device will switch to the RTK-SLAM combined measurement mode without interruption. It no longer relies on direct RTK measurements, but instead relies on the SLAM+GNSS fusion alignment algorithm and the optimal transformation parameters calculated to continue to maintain high-precision global positioning results, achieving continuous positioning without breaks both indoors and outdoors.

[0051] Secondly, the GNSS signal quality is assessed in real time, verifying whether its HDOP and STD of each dimension meet the preset GNSS signal normal judgment conditions (i.e., HDOP ≤ first preset threshold, STD of each dimension ≤ second preset threshold). This determines whether the current satellite signal environment is a normal open environment or a rejected environment. If the quality parameters meet the GNSS signal normal judgment conditions, it indicates that the current satellite signal is good, and the RTK device can provide high-precision global positioning data. At this time, the longitude, latitude, and elevation three-dimensional global positioning data collected by the RTK device are directly output as the global positioning result. At the same time, the transformation parameters of the local and global coordinate systems are continuously optimized and updated using this high-precision global positioning data to ensure the accuracy of the parameters. If the quality parameters do not meet the GNSS signal normal judgment conditions, it indicates that the current environment is GNSS denied, and the positioning accuracy of the RTK device cannot meet the requirements. At this time, based on the calculated rotation and translation transformation parameters, the pose trajectory of the LIO in the local coordinate system is completely transformed to the global coordinate system. During the transformation process, the pose estimation results of the tightly coupled lidar and IMU are relied upon to make up for the lack of global positioning data in the GNSS denied environment. The transformed global coordinate data is continuously output as the global positioning result.

[0052] During actual positioning operations, the system monitors and judges quality parameters in real time at the millisecond level. During positioning mode switching, it relies on the real-time update of transformation parameters and the continuous calculation of local pose trajectories to ensure the continuity and accuracy consistency of global positioning results, and there will be no positioning interruption or jump. It achieves seamless global positioning in indoor and outdoor, open environments and GNSS denied environments.

[0053] Meanwhile, in actual surveying scenarios, the target point that users actually need to measure is the tip of the centering rod, not the center of the LiDAR laser head and the phase center of the GNSS antenna. Therefore, if... Figure 10 As shown, this embodiment uses tilt measurement correction to accurately correct the coordinate measurement values ​​to the tip position of the centering rod: with the laser head center as the reference point. In a LiDAR coordinate system with the origin at the known center of the host bottom, Center of the bottom of the handle The "tip direction" of the centering rod is determined by multiple reference rigid body points, and then this direction is extended from the reference points to the rod tip to calculate the rod tip coordinates. First, the direction vector of the centering rod in the Lidar coordinate system is calculated: Let the height of the rod be the vector extending along that direction to the tip. Then, the coordinates of the tip in the Lidar system are: The pose of the LiDAR in the world coordinate system is and Then the absolute coordinates of the rod tip in the world system are: Furthermore, throughout the entire process of generating global positioning results, the transformation parameters are continuously updated and optimized using a sliding window optimization algorithm. The latest trajectory points are used to correct minor errors in the transformation parameters and offset the slow drift of LIO local pose estimation. This ensures that even under long-term GNSS rejection conditions, the converted global positioning results and the absolute coordinates of the centering pole tip can still maintain centimeter-level high accuracy, meeting the actual needs of mobile surveying, engineering surveying, and other scenarios for continuous high-precision global positioning and accurate measurement of target points.

[0054] In summary, as Figure 11 As shown, this application can effectively achieve high-precision global positioning under GNSS denial conditions. It has good test results in GNSS denial environments such as between buildings, indoor office buildings, garages, and eaves, and meets the difficulties and pain points of traditional mobile mapping in GNSS denial environments where it cannot be measured or cannot be measured accurately.

[0055] The specific evaluation results are shown in Tables 1 and 2. The checkpoints were measured using the NT10 high-precision surveying robot from Southern Surveying and Mapping.

[0056] Table 1. Accuracy Assessment of Indoor and Outdoor Combined Solution in the Southern Surveying and Mapping Building Park Table 2 Outdoor track test (with extensive tree obstruction) In summary, this application acquires environmental point cloud data from a lidar system, inertial data from an inertial measurement unit (IMU), and global positioning data and mass parameters from a real-time differential positioning device. It then tightly couples the inertial data with the environmental point cloud data to calculate the pose trajectory in a local coordinate system. Based on the mass parameters, it filters the global positioning data to construct a set of trajectory point pairs. Finally, it uses a sliding window optimization algorithm to solve the trajectory point pair set to obtain the transformation parameters between the local and global coordinate systems. Based on these transformation parameters, it converts the pose trajectory in the local coordinate system to the global coordinate system to generate a global positioning result. Since the lidar inertial odometry is limited to a local range... While existing technologies can provide high-precision relative pose constraints with slow drift, real-time differential positioning (RTD) provides global positioning information but lacks pose determination capabilities and its accuracy decreases when the high-precision attenuation factor or coordinate standard deviation is large. Therefore, this application filters out low-quality global positioning data by using quality parameters and calculates the transformation parameters between the two coordinate systems in real time using a sliding window optimization algorithm. This ensures that the local high-precision pose of the lidar inertial odometry achieves global consistency, while avoiding the impact of degraded global positioning data quality or long-term cumulative drift of the lidar inertial odometry on positioning accuracy. This allows for the continuation of high-precision global positioning capabilities even in GNSS-denied environments, achieving seamless high-precision positioning both indoors and outdoors. This application effectively solves the problem that existing technologies cannot accurately and efficiently perform global positioning in GNSS-denied environments.

[0057] Example 2 Please refer to Figure 12 This is a combined positioning device for a denied environment provided in the embodiments of this application.

[0058] In this embodiment, the combined positioning device in a denied environment includes: The acquisition module 10 is used to acquire environmental point cloud data collected by the lidar, inertial data collected by the inertial measurement unit, and global positioning data and mass parameters collected by the real-time differential positioning device. The estimation module 20 is used to perform tightly coupled pose estimation on the inertial data and the environmental point cloud data to obtain the pose trajectory in the local coordinate system. The filtering module 30 is used to filter the global positioning data based on the quality parameters, and construct a set of trajectory point pairs by combining the filtered global positioning data with the pose trajectory in the local coordinate system. The solution module 40 is used to solve the set of trajectory point pairs based on the sliding window optimization algorithm to obtain the transformation parameters between the local coordinate system and the global coordinate system. The transformation module 50 is used to transform the pose trajectory in the local coordinate system to the global coordinate system according to the transformation parameters, and generate a global positioning result.

[0059] For ease of description and brevity, the embodiments of the device of the present invention include all the implementation methods in the above-described embodiments of the combined positioning method under the denied environment, and will not be repeated here.

[0060] Example 3 This application provides a computer-readable storage medium including a stored computer program, wherein the computer program, when running, controls the device where the computer-readable storage medium is located to execute the combined positioning method under a denial environment. The combined positioning method under denied conditions, if implemented as a software functional unit and used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0061] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A combined localization method under a denied environment, characterized in that, include: Acquire environmental point cloud data collected by lidar, inertial data collected by inertial measurement unit, and global positioning data and mass parameters collected by real-time differential positioning device; The inertial data and the environmental point cloud data are tightly coupled to calculate the pose, and the pose trajectory in the local coordinate system is obtained. Based on the quality parameters, the global positioning data is filtered, and the filtered global positioning data and the pose trajectory in the local coordinate system are used to construct a set of trajectory point pairs. Based on the sliding window optimization algorithm, the set of trajectory point pairs is solved to obtain the transformation parameters between the local coordinate system and the global coordinate system; Based on the transformation parameters, the pose trajectory in the local coordinate system is transformed to the global coordinate system to generate a global positioning result.

2. The combined positioning method under denied conditions according to claim 1, characterized in that, The specific steps for tightly coupled pose estimation of the inertial data and the environmental point cloud data are as follows: The inertial data is integrated to obtain the initial pose prediction value; Based on the initial pose prediction value, motion distortion correction is performed on the environmental point cloud data to obtain the corrected point cloud data; The corrected point cloud data is then registered to obtain the registered pose. Based on the registered pose and the inertial data, a joint optimization function is constructed; The joint optimization function is optimized and solved to obtain the optimized pose as the pose trajectory in the local coordinate system.

3. The combined positioning method under denied conditions according to claim 1, characterized in that, The filtering of the global positioning data based on the quality parameters specifically involves: Extract the horizontal accuracy attenuation factor and coordinate standard deviation from the quality parameters; Determine whether the horizontal accuracy attenuation factor is greater than a first preset threshold, or determine whether the coordinate standard deviation is greater than a second preset threshold; If the horizontal accuracy attenuation factor is greater than the first preset threshold, or the coordinate standard deviation is greater than the second preset threshold, then the corresponding global positioning data is discarded.

4. The combined positioning method under denied conditions according to claim 1, characterized in that, The step of constructing a set of trajectory point pairs from the filtered global positioning data and the pose trajectory in the local coordinate system is specifically as follows: Obtain the timestamp corresponding to the filtered global location data; Based on the timestamp, pose interpolation is performed on the pose trajectory in the local coordinate system to obtain the local pose corresponding to the filtered global positioning data. The filtered global positioning data is combined with the corresponding local pose to form trajectory point pairs; Based on the trajectory point pairs, a set of trajectory point pairs corresponding to the filtered global positioning data and the pose trajectory in the local coordinate system is constructed.

5. The combined positioning method under denied conditions according to claim 1, characterized in that, The sliding window optimization algorithm solves the set of trajectory point pairs to obtain the transformation parameters between the local and global coordinate systems, specifically: Based on the relative pose changes between adjacent local poses in the set of trajectory point pairs, a pose factor is constructed. An alignment factor is constructed based on the coordinate difference between the filtered global positioning data and the corresponding local pose in the trajectory point pair set. Construct a factor graph based on the pose factor and the alignment factor; Incremental smoothing optimization is performed on the factor graph to obtain the rotation transformation parameters and translation transformation parameters between the local coordinate system and the global coordinate system, which are then used as the transformation parameters.

6. The combined positioning method under denied conditions according to claim 5, characterized in that, The incremental smoothing optimization of the factor graph specifically involves: Add the alignment factor corresponding to the most recently added trajectory point pair in the set of trajectory point pairs to the factor graph; Determine if the number of aligned factors in the current factor graph exceeds the preset window size; If the number of alignment factors in the current factor graph exceeds the preset window size, then remove the alignment factor with the earliest timestamp from the factor graph; The updated factor graph is optimized based on incremental smoothing and graph building algorithms to obtain optimized rotation transformation parameters and translation transformation parameters; Calculate the heading angle error based on the optimized rotation transformation parameters; Determine whether the heading angle error is greater than a preset angle threshold; If the heading angle error is greater than a preset angle threshold, the corresponding alignment factor is marked as an abnormal factor, and the abnormal factor is removed from the factor graph before the next optimization.

7. The combined positioning method under denied conditions according to claim 1, characterized in that, The step of transforming the pose trajectory in the local coordinate system to the global coordinate system according to the transformation parameters to generate a global positioning result is as follows: Determine whether the quality parameters meet the preset GNSS signal normal determination conditions; If the quality parameters meet the GNSS signal normal determination conditions, the global positioning data collected by the real-time differential positioning device will be output as the global positioning result. If the quality parameters do not meet the GNSS signal normal determination conditions, the pose trajectory in the local coordinate system is transformed to the global coordinate system according to the transformation parameters, and the transformed coordinate data is output as the global positioning result.

8. The combined positioning method under denied conditions according to claim 7, characterized in that, After generating the global localization result, the process also includes: Obtain the rigid body structural parameters of the device in the lidar coordinate system, and determine the local coordinates of the two reference points of the centering rod in the lidar coordinate system; Based on the local coordinates of the two reference points, the direction vector of the centering rod tip is calculated, and the direction vector is normalized to obtain the unit direction vector of the centering rod tip. Obtain the preset centering rod height, and extend the local coordinates of the preset reference point among the two reference points towards the rod tip along the unit direction vector to calculate the local coordinates of the centering rod tip in the lidar coordinate system. The local coordinates of the centering rod tip are transformed to the global coordinate system according to the transformation parameters to obtain the global positioning coordinates of the centering rod tip.

9. A combined positioning device for a denied environment, characterized in that, include: The acquisition module is used to acquire environmental point cloud data collected by the lidar, inertial data collected by the inertial measurement unit, and global positioning data and mass parameters collected by the real-time differential positioning device. The estimation module is used to perform tightly coupled pose estimation on the inertial data and the environmental point cloud data to obtain the pose trajectory in the local coordinate system. The filtering module is used to filter the global positioning data based on the quality parameters, and construct a set of trajectory point pairs by combining the filtered global positioning data with the pose trajectory in the local coordinate system. The solution module is used to solve the set of trajectory point pairs based on the sliding window optimization algorithm to obtain the transformation parameters between the local coordinate system and the global coordinate system; The transformation module is used to transform the pose trajectory in the local coordinate system to the global coordinate system according to the transformation parameters, and generate a global positioning result.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the combined positioning method under a denial environment as described in any one of claims 1 to 8.

Images