A vehicle high-precision positioning method and system based on laser radar point cloud data

Abstract

The application discloses a kind of vehicle high-precision positioning method and system based on laser radar point cloud data, comprising: by time stamp synchronization and coordinate system conversion, point cloud data is integrated into the fusion point cloud data formed under vehicle ontology coordinate system, and stable features such as road marking and building contour are extracted and compared with high-precision map, and position deviation vector is calculated in real time.When the deviation exceeds threshold value, dynamic correction mechanism and iterative closest point algorithm are used for pose correction and enhancement, and quantitative indicators such as transverse dislocation and heading angle deviation are calculated through geometric relationship analysis, and if the indicators are over limit, the pose is adjusted by triggering repositioning process. Finally, motion interpolation compensation is carried out by combining vehicle dynamics model, and error propagation chain is analyzed by point cloud density and matching residual, and position confidence index is obtained. The application improves the positioning stability and precision in complex environment through multi-level deviation correction and error quantization, and provides reliable support for automatic driving.

Description

Technical Field

[0001] This invention belongs to the field of autonomous driving, and in particular relates to a method and system for high-precision vehicle positioning based on lidar point cloud data. Background Technology

[0002] In the fields of intelligent driving and autonomous driving, high-precision vehicle positioning technology is a core pillar for ensuring safety and reliability, and its importance is self-evident. Accurate positioning not only concerns the vehicle's accurate perception of its surroundings but also directly impacts the correctness of path planning and decision-making control, making it a crucial link in moving autonomous driving from theory to reality. However, despite the significant attention this field has received, existing technologies still face numerous challenges and urgently need breakthroughs. Currently, many positioning methods perform poorly in complex environments, especially when facing dynamically changing road scenarios, often struggling to maintain consistent accuracy and stability. The problem lies not only in the limitations of the technology itself but also in the insufficient ability to comprehensively capture and process environmental information in real time. Particularly in urban roads or high-speed driving scenarios, the environmental information around the vehicle changes rapidly. Existing solutions are inadequate in integrating multi-source information and responding to sudden changes, leading to deviations in positioning results and even affecting the reliability of the entire system.

[0003] In-depth analysis reveals that the core technical challenge of existing technologies lies in effectively processing and fusing large amounts of environmental perception data. LiDAR, as an important sensing device, can provide rich three-dimensional spatial information, but the temporal and spatial discontinuities of its data pose a significant obstacle. This discontinuity directly leads to difficulties in accurately determining the real-time position of the vehicle when it is moving rapidly or in environments with sparse features. For example, during high-speed driving, the data collected by LiDAR may have intervals due to scanning frequency limitations, and the vehicle's positional changes between two frames of data cannot be accurately captured, resulting in jumps or deviations in the positioning trajectory.

[0004] Therefore, ensuring the continuity and consistency of environmental perception data in dynamic environments and eliminating positioning deviations through effective data fusion have become key issues in the field of high-precision vehicle positioning. Solving this problem requires not only overcoming time discontinuities in data acquisition but also achieving accurate matching and real-time correction of environmental features in complex scenarios to ensure stable vehicle positioning under various driving conditions. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method and system for high-precision vehicle positioning based on lidar point cloud data. Specifically, the method for high-precision vehicle positioning based on lidar point cloud data includes:

[0006] The point cloud data collected by multi-directional lidar is time-stamped and coordinate system transformed to obtain fused point cloud data;

[0007] Based on the fused point cloud data, stable features are extracted and compared with map reference features to obtain a position deviation vector;

[0008] The pose is adjusted by triggering a dynamic correction mechanism based on the relationship between the position deviation vector and the preset threshold, and the corrected pose estimate is obtained.

[0009] Based on the corrected pose estimation, iterative registration is performed on multiple consecutive frames of point cloud data in a feature-sparse environment to obtain the enhanced pose estimation.

[0010] Based on the enhanced pose estimation, the geometric relationship between point cloud features and map elements is analyzed to obtain a misalignment quantification index.

[0011] Based on the relationship between the misalignment quantification index and the preset threshold, a relocation process is triggered to update the pose, thereby obtaining the updated pose.

[0012] Based on the updated pose and the vehicle dynamics model, motion changes are estimated and point cloud sequence interpolation is performed to obtain a continuous point cloud sequence.

[0013] Based on the continuous point cloud sequence, the point cloud density is tracked and the residual is matched, and error propagation calculation is performed to obtain the location reliability index and its positioning result.

[0014] Preferably, the process of obtaining the fused point cloud data includes:

[0015] The acquisition time of each frame of point cloud is obtained by parsing the timestamp information of the raw point cloud data collected by the multi-source lidar.

[0016] The time offset of each LiDAR is determined based on the deviation between the acquisition time and the vehicle's master clock.

[0017] The point cloud timestamps are corrected based on the relationship between the time offset and the preset threshold to obtain a time-corrected point cloud.

[0018] Based on the vehicle's current pose and the lidar extrinsic parameter matrix, the time-corrected point cloud is transformed to the vehicle's body coordinate system to obtain a unified coordinate point cloud.

[0019] The nearest neighbor search method is used to find the spatial overlapping region of different lidars in the unified coordinate point cloud to obtain the overlapping region point cloud.

[0020] The overlapping region point cloud is spatially divided using a voxel mesh partitioning method to obtain a voxelized point cloud set.

[0021] The position averaging of points from different lidars is calculated based on the voxelized point cloud set to obtain sparse fused point cloud data.

[0022] Preferably, the process of obtaining the position deviation vector includes:

[0023] Preprocessing the sparse fused point cloud data to filter out noise and irrelevant points yields a clean point cloud dataset.

[0024] Based on the clean point cloud dataset, road markings and building outlines are extracted using a geometric segmentation method to obtain a stable feature set;

[0025] Reference feature data is retrieved from the map and initially registered with the stable feature set to obtain an initial matching relationship.

[0026] The initial matching relationship is finely adjusted based on a real-time comparison method to obtain an accurate matching result;

[0027] Calculate the positional deviation based on the precise matching result and generate deviation vector data;

[0028] The final position deviation vector is obtained by performing a final verification based on the deviation vector data and map reference features.

[0029] Preferably, the process of obtaining the corrected pose estimate includes:

[0030] The magnitude is calculated based on the position deviation vector, and the current frame is marked as having an excessive deviation state based on the relationship between the magnitude and a preset threshold.

[0031] The dynamic correction mechanism is activated based on the deviation exceeding the limit state. The historical pose estimation and the current deviation vector are fused by a Kalman filter to obtain the preliminary corrected pose.

[0032] The optimized corrected pose is obtained by performing consistency constraint optimization on the preliminary corrected pose and the pose estimation of the most recent multi-frames using the least squares method.

[0033] The optimized pose is overlaid on the original pose estimate to obtain the corrected pose estimate.

[0034] Preferably, the process of obtaining the enhanced pose estimation includes:

[0035] An initial point cloud dataset is obtained by collecting point cloud data from multiple consecutive frames.

[0036] Based on the initial point cloud dataset, the iterative nearest-point algorithm is used to perform preliminary registration of adjacent frames to obtain the point cloud relationship after preliminary registration.

[0037] Based on the comparison results between the pre-registered point cloud relationship and the preset threshold, the point pairs in the local area are adjusted to obtain the adjusted point cloud registration result;

[0038] Based on the adjusted point cloud registration results, obtain the relative pose transformation matrix between multiple frames of point clouds, and determine the continuity of the pose transformation.

[0039] Based on the continuity of the pose transformation, a cumulative transformation method is used for global registration to obtain globally consistent point cloud registration data.

[0040] The final pose estimate is calculated based on the globally consistent point cloud registration data to obtain the enhanced pose estimate result.

[0041] Preferably, the process of obtaining the misalignment quantification index includes:

[0042] Key feature points are extracted from point cloud data and geometrically matched with map elements to obtain initial pose estimation results.

[0043] Based on the initial pose estimation results, the geometric relationship difference between point cloud features and map elements is calculated to obtain a preliminary value of lateral misalignment.

[0044] The matching error is optimized using the least squares method based on the preliminary value of the lateral misalignment to obtain the lateral misalignment data;

[0045] Based on the analysis of the lateral misalignment data, the distribution of the heading angle deviation is obtained, and the heading angle deviation parameters are obtained.

[0046] The point cloud features are locally adjusted based on the comparison result between the heading angle deviation parameter and the preset threshold to obtain the corrected deviation value;

[0047] Based on the corrected deviation value and the lateral misalignment data, the misalignment quantification parameters are derived to obtain the misalignment quantification index.

[0048] Preferably, the process of obtaining the updated pose includes:

[0049] The misalignment quantization error is calculated based on the current pose and the map matching, and the misalignment quantization error value is obtained.

[0050] A trigger signal is generated based on the comparison result between the misalignment quantization error value and the preset threshold.

[0051] The relocation process is initiated based on the trigger signal, loading global map data and current sensor observation data;

[0052] The current sensor observation data is distributed to the global map data according to the particle filtering algorithm, forming a particle set and corresponding weights.

[0053] A resampling operation is performed based on the particle set and its corresponding weights, retaining high-weight particles to obtain a convergent particle set.

[0054] Based on the convergent particle set, select the pose corresponding to the particle with the highest weight to determine the updated pose.

[0055] Preferably, the process of obtaining a continuous point cloud sequence includes:

[0056] Based on the vehicle dynamics model, the original pose data is obtained from the acquisition interval and combined with the historical motion trajectory for preliminary analysis to obtain the initial estimation result after pose update.

[0057] Based on the initial estimation results after pose update, the Kalman filter method is used to smooth the motion changes and determine the trend of motion trajectory changes within the acquisition interval.

[0058] Based on the trend of motion trajectory changes and the time span of the acquisition interval, the missing spatiotemporal points in the point cloud sequence are predicted to obtain preliminary point cloud supplementary data.

[0059] Based on the preliminary point cloud supplementary data, the point cloud sequence is filled using interpolation compensation technology to generate smoothly transitioning intermediate point cloud data.

[0060] Based on the intermediate point cloud data, a sequence optimization method is used to perform consistency correction on the overall point cloud sequence, resulting in a point cloud sequence with higher continuity.

[0061] Based on the point cloud sequence results, the local non-smooth areas are dynamically adjusted and optimized to determine the final continuous point cloud sequence.

[0062] Preferably, the process of obtaining the location reliability index and its location results includes:

[0063] The initial values ​​of point cloud density and matching residuals are extracted based on the acquisition and preprocessing of continuous point cloud sequences to obtain a preliminary dataset.

[0064] Based on the preliminary dataset, a continuous analysis method is used to perform time-series comparison of point cloud density, identify density change patterns, and determine significant intervals of density change.

[0065] Based on the significant intervals of the density changes, and combined with the distribution of the matching residuals, the influence weight of the residual factors on the positioning is calculated to determine the main sources of error.

[0066] Based on the main sources of error, a propagation path is constructed using an error propagation model, resulting in a structure diagram of the propagation path.

[0067] Based on the structure diagram of the propagation link, analyze the degree of impact of error propagation on the positioning signal, deduce the fluctuation range of the positioning signal, and determine the preliminary assessment value of the confidence level;

[0068] The final value of the confidence index is calculated based on the preliminary assessment value of the confidence level and the results of the sequence analysis, thus obtaining the quantitative result of the location confidence index.

[0069] The parameters of the error propagation model are adjusted based on the comparison between the quantitative results of the location confidence and the preset threshold to obtain the optimized confidence index value.

[0070] The predicted vehicle positioning result is output based on the optimized confidence index value.

[0071] The present invention also provides a high-precision vehicle positioning system based on lidar point cloud data, comprising:

[0072] The data fusion module is used to synchronize the timestamps and transform the coordinate system of the point cloud data collected by multi-directional lidar to obtain fused point cloud data.

[0073] The feature comparison module is used to extract stable features from the fused point cloud data and compare them with map reference features to obtain a position deviation vector;

[0074] The dynamic correction module is used to trigger a dynamic correction mechanism to adjust the pose based on the relationship between the position deviation vector and the preset threshold, so as to obtain the corrected pose estimate.

[0075] The clustering and registration module is used to iteratively register multiple consecutive frames of point cloud data in a feature-sparse environment based on the corrected pose estimation to obtain the enhanced pose estimation.

[0076] The misalignment quantization module is used to analyze the geometric relationship between point cloud features and map elements based on the enhanced pose estimation to obtain misalignment quantization index.

[0077] The repositioning module is used to trigger the repositioning process to update the pose based on the relationship between the misalignment quantification index and the preset threshold, so as to obtain the updated pose.

[0078] The interpolation compensation module is used to estimate motion changes based on the updated pose and the vehicle dynamics model, and to perform point cloud sequence interpolation to obtain a continuous point cloud sequence.

[0079] The positioning evaluation module is used to track the point cloud density and match the residuals based on the continuous point cloud sequence and perform error propagation calculations to obtain the positioning reliability index and its positioning results.

[0080] Compared with the prior art, the present invention has the following advantages and technical effects:

[0081] This invention achieves a comprehensive improvement in vehicle positioning accuracy and stability by constructing a closed-loop processing chain of "deviation detection—dynamic correction—enhanced registration—quantitative evaluation—intelligent relocation—continuous compensation—confidence propagation". Specifically, by introducing least-squares consistency constraints for optimization, the dynamically corrected pose maintains a smooth transition across multiple frames, avoiding the abrupt changes in traditional methods. In feature-sparse environments, multi-frame clustering registration and pose continuity judgment are employed, and global consistency optimization is achieved using cumulative transformation, significantly reducing the cumulative effect of registration errors. By performing least-squares optimization and local feedback adjustment on lateral misalignment and heading angle deviation, more accurate and reliable misalignment quantification indicators are obtained. Based on the misalignment quantification indicators, the relocation process is intelligently triggered, and particle filtering and resampling mechanisms are combined to quickly converge to the optimal pose, improving relocation efficiency and accuracy. The trajectory change trend is estimated using a vehicle dynamics model, and missing spatiotemporal points are filled through Kalman filtering smoothing and interpolation compensation techniques. After sequence optimization and local dynamic adjustment, a highly smooth and spatiotemporally consistent continuous point cloud sequence is generated. By tracking point cloud density and matching residuals and constructing an error propagation chain model, the error coupling relationship and accumulation path are systematically analyzed, obtaining more objective and comprehensive positioning reliability quantification results, thus effectively supporting high-precision and reliable vehicle positioning in complex dynamic scenarios. Attached Figure Description

[0082] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0083] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0084] Figure 2 This is a schematic diagram of the system structure according to an embodiment of the present invention. Detailed Implementation

[0085] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0086] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0087] Example 1

[0088] like Figure 1 As shown, this embodiment provides a high-precision vehicle positioning method based on lidar point cloud data, including:

[0089] The point cloud data collected by multi-directional lidar is time-stamped and coordinate system transformed to obtain fused point cloud data;

[0090] Stable features are extracted from the fused point cloud data and compared with map reference features to obtain the position deviation vector;

[0091] The pose is adjusted by triggering a dynamic correction mechanism based on the relationship between the position deviation vector and a preset threshold, thus obtaining the corrected pose estimate.

[0092] Based on the corrected pose estimation, iterative registration is performed on multiple consecutive frames of point cloud data in a feature-sparse environment to obtain the enhanced pose estimation.

[0093] Based on the analysis of the geometric relationship between point cloud features and map elements using enhanced pose estimation, a misalignment quantification index is obtained.

[0094] The repositioning process is triggered based on the relationship between the misalignment quantification index and the preset threshold to update the pose and obtain the updated pose.

[0095] Based on the updated pose and the vehicle dynamics model, motion changes are estimated and point cloud sequence interpolation is performed to obtain a continuous point cloud sequence.

[0096] By tracking point cloud density and matching residuals based on continuous point cloud sequences and performing error propagation calculations, the location reliability index and its positioning results are obtained.

[0097] Furthermore, the process of obtaining the fused point cloud data includes:

[0098] The acquisition time of each frame of point cloud is obtained by parsing the timestamp information of the raw point cloud data collected by the multi-source lidar.

[0099] The time offset of each LiDAR is determined based on the deviation between the acquisition time and the vehicle's master clock.

[0100] The point cloud timestamps are corrected based on the relationship between the time offset and a preset threshold to obtain a time-corrected point cloud;

[0101] Based on the vehicle's current pose and the lidar extrinsic matrix, the time-corrected point cloud is transformed to the vehicle's body coordinate system to obtain a unified coordinate point cloud.

[0102] The spatial overlapping region of different lidars is found in the coordinate unified point cloud using the nearest neighbor search method, and the overlapping region point cloud is obtained.

[0103] The overlapping point cloud is spatially partitioned using a voxel mesh method to obtain a set of voxelized point clouds.

[0104] The position average of points from different lidars is calculated based on the voxelized point cloud set to obtain sparse fused point cloud data.

[0105] Furthermore, for example, in this embodiment, timestamp parsing is the primary step when processing multi-source LiDAR point cloud data. Assume there are three LiDAR devices, denoted as L1, L2, and L3, each acquiring point cloud data with a timestamp. Through standardized protocol parsing, the point cloud acquisition time per frame is found to be 10.2 seconds for L1, 10.3 seconds for L2, and 10.5 seconds for L3, while the vehicle's main clock displays 10.0 seconds. After comparing the discrepancies, L1 is offset by 0.2 seconds, L2 by 0.3 seconds, and L3 by 0.5 seconds. If the preset threshold is 0.4 seconds, then L3 needs correction by uniformly adding a 0.5-second offset to its point cloud timestamp, resulting in a time-corrected point cloud. This ensures that the multi-source data is aligned in the time dimension, laying the foundation for subsequent fusion.

[0106] In one possible implementation, the coordinate transformation stage relies on the vehicle pose and the extrinsic parameter matrix of the LiDAR. Assuming L1 is mounted on the top of the vehicle, L2 on the left, and L3 on the right, the extrinsic parameter matrix describes their position and orientation relative to the vehicle's body coordinate system. Through matrix transformation, the coordinates of a point in the L1 point cloud are converted from the sensor coordinate system to a new value in the vehicle coordinate system. The same transformation is performed on the L2 and L3 point clouds, ultimately forming a point cloud with unified coordinates. This step ensures spatial data consistency and avoids fusion errors caused by different coordinate systems.

[0107] For example, this embodiment uses a nearest neighbor search method to determine the overlapping area point cloud. Assuming that the L1 and L2 point clouds spatially overlap 5 meters in front of the vehicle, the closest pair of points among the point groups collected by the two devices is found within a search radius of 0.1 meters, and this is confirmed as the overlapping area point cloud. This helps identify redundant parts of multi-source data, providing a basis for subsequent optimization.

[0108] In one possible implementation, this embodiment divides the coordinate-uniform point cloud into cubic grids with sides of 0.2 meters using voxel meshing. Overlapping point clouds are assigned to the same voxels, forming a voxelized point cloud set. This processing effectively reduces the amount of data, improves computational efficiency, and preserves spatial features.

[0109] For example, in this embodiment, when calculating the positional average within each voxel, assuming a voxel contains 3 points from L1 and 2 points from L2, a single representative point is obtained by calculating the spatial positional average of these points. After all voxels are processed, sparsely fused point cloud data is formed. This method reduces redundant points, improves data consistency, and facilitates the efficient execution of subsequent perception tasks such as object detection and path planning.

[0110] In one possible implementation, this embodiment reduces the amount of point cloud data from millions of points to hundreds of thousands by sparsely fusing the point cloud, significantly reducing computational resource consumption. Simultaneously, due to temporal and spatial alignment, perception accuracy is improved by approximately 10%. This provides more reliable environmental information for the autonomous driving system, enhancing safety.

[0111] Furthermore, the process of obtaining the position deviation vector includes:

[0112] Preprocessing the sparse fused point cloud data to filter out noise and irrelevant points yields a clean point cloud dataset.

[0113] Based on the clean point cloud dataset, road markings and building outlines are extracted using a geometric segmentation method to obtain a stable feature set;

[0114] Reference feature data is retrieved from the map and initially registered with a stable feature set to obtain an initial matching relationship.

[0115] The initial matching relationship is finely adjusted based on the real-time comparison method to obtain accurate matching results;

[0116] Calculate the positional deviation based on the precise matching results and generate deviation vector data;

[0117] The final position deviation vector is obtained by comparing the deviation vector data with map reference features.

[0118] Furthermore, for example, in the autonomous driving positioning system of this embodiment, after acquiring point cloud data, the raw data first needs to be preprocessed to remove outliers caused by ground reflection and rain / fog interference.

[0119] Specifically, a statistical filtering method can be used, which requires each point to have at least 30 neighboring points within a radius of 5 meters; otherwise, they are considered noise and removed, thus obtaining a clean point cloud dataset. This step significantly reduces the false alarm rate of subsequent matching.

[0120] In one possible implementation, when extracting stable features from a clean point cloud using a geometric segmentation method, this embodiment prioritizes identifying road markings and building facades.

[0121] For example, the ground is extracted by clustering points whose point cloud normal vectors make an angle of less than 15 degrees with the vertical direction, and then white markings are segmented by reflectivity regions with an intensity threshold greater than 200. At the same time, Euclidean clustering is performed on vertical surfaces with height abrupt changes exceeding 3 meters to extract building outlines, forming a stable feature set containing only road marking points and building edge points. This choice is made because these features remain unchanged in high-precision maps over a long period of time and are least affected by occlusion from high-speed trains.

[0122] It should be noted that, in this embodiment, when retrieving reference features from a high-precision map, a pre-stored vector markings and building outline point cloud within a 50-meter radius are loaded, centered on the vehicle's approximate GNSS position. The extracted stable feature set is then preliminarily registered with these features using an intensity-weighted ICP algorithm with an initial tolerance error set of 0.8 meters. This rapidly establishes the correspondence between feature points, laying the foundation for subsequent refinement.

[0123] For example, during the real-time comparison and fine-tuning stage, if the mean residual value after initial registration still reaches 0.45 meters, exceeding the preset 0.3-meter threshold, a re-registration mechanism is triggered. At this time, the system increases feature weights and prioritizes using stable calibrated corner points from three or more consecutive frames for secondary iteration, bringing the final matching error to 0.12 meters and significantly improving positioning robustness.

[0124] In one possible implementation, when calculating the positional deviation based on the accurate matching result, this embodiment treats the map reference features as true values. The real-time features are rotated and translated, and the difference between them and the true values ​​is calculated to generate deviation vectors with three components: eastward, northward, and azimuth. For example, the measured eastward deviation is -0.28 meters and the northward deviation is 0.15 meters. Through vector synthesis, a total deviation of 0.32 meters and an azimuth angle of approximately 151 degrees are obtained.

[0125] It should be noted that during the final verification stage, if the deviation vector direction is found to be more than 30 degrees opposite to the cumulative drift trend within the past 10 seconds, it is determined that there may be temporary occlusion interference. At this time, the threshold is automatically relaxed and iterative adjustment is initiated. By gradually reducing the rotation matrix step size by 0.005 degrees until the deviation vector direction is consistent with the expectation, a stable and reliable position deviation vector is finally output for vehicle pose correction. This complete process enables the vehicle to maintain positioning accuracy within 0.15 meters in complex urban environments, effectively supporting the safe operation of high-level autonomous driving.

[0126] Furthermore, the process of obtaining the corrected pose estimate includes:

[0127] The magnitude is calculated based on the position deviation vector, and the current frame is marked as having exceeded the deviation limit based on the relationship between the magnitude and a preset threshold.

[0128] The dynamic correction mechanism is activated based on the deviation exceeding the limit. The historical pose estimation and the current deviation vector are fused by a Kalman filter to obtain the preliminary corrected pose.

[0129] The optimized corrected pose is obtained by applying the least squares method to the consistency constraint optimization between the preliminary corrected pose and the pose estimation of the most recent multi-frames.

[0130] The optimized and corrected pose is overlaid on the original pose estimate to obtain the corrected pose estimate.

[0131] Furthermore, for example, in this embodiment, during the acquisition of the real-time position deviation vector, the current frame data collected by the sensors is compared with reference data from a high-precision map to determine the vehicle's position offset on the road. Assuming a city road environment, the vehicle acquires point cloud data via LiDAR, and the system calculates the magnitude of the position deviation vector to be 2.5 meters, while the preset threshold is 1.0 meter, clearly exceeding the acceptable range. In this case, the system marks the current frame as exceeding the deviation limit. This marking mechanism helps to promptly detect positioning anomalies, providing a basis for subsequent corrections.

[0132] For example, in this embodiment, when the dynamic correction mechanism is activated in a deviation exceeding the limit state, the system dynamically adjusts the positioning parameters based on the current environment and historical data. Suppose the vehicle is in an area with many tall buildings, and signal obstruction causes positioning drift. The dynamic correction mechanism will prioritize calling historical trajectory data and combine it with the current deviation vector for preliminary correction. This approach can effectively address positioning challenges in complex environments.

[0133] For example, in this embodiment, fusing historical pose estimates with the current deviation vector using a Kalman filter can be considered a smoothing process aimed at reducing abrupt errors. Suppose that the vehicle's pose estimates over five consecutive frames show a gradually increasing positional offset, while the current deviation vector indicates a large jump in offset, the Kalman filter will combine the weights of both to generate a smoother initial corrected pose. The advantage of this method is that it avoids drastic fluctuations in the localization results, improving system stability.

[0134] For example, in this embodiment, when the least squares method is used to optimize the consistency constraint between the preliminary corrected pose and the pose estimation of the most recent multiple frames, it can be viewed as a process of overall adjustment of multi-source data. Suppose the system takes the pose data of the most recent 10 frames and finds that 2 frames are significantly abnormal due to sensor noise. The least squares method will optimize the calculation to reduce the impact of the abnormal data, obtaining a corrected pose that better conforms to the overall trend. The benefit of this optimization is that it improves the reliability of the positioning results.

[0135] For example, in this embodiment, when outputting the optimized and corrected pose and overriding the original pose estimate, the system directly applies the final correction result to the navigation module. Suppose that at an intersection, the original pose estimate deviates from the actual position by 10 meters due to blurred road markings. The optimized and corrected pose reduces this deviation to 0.5 meters, and after overriding, the navigation system can guide the vehicle more accurately. The advantage of this overriding mechanism is that it ensures that the real-time output pose data is closer to the actual position.

[0136] For example, this embodiment logically progresses from the core solution to the extended solution for all the above-mentioned steps. The core solution quickly identifies problems based on deviation vector calculation and out-of-limit markers; the extended solution gradually improves positioning accuracy through filtering and optimization. Assuming long-distance high-speed driving, the core solution can quickly capture deviations, while the extended solution optimizes through multi-frame data to ensure stability during extended driving. The advantage of this layered design is that it balances real-time performance and accuracy, adapting to different scenario requirements.

[0137] Furthermore, the process of obtaining the enhanced pose estimation includes:

[0138] An initial point cloud dataset is obtained by collecting point cloud data from multiple consecutive frames.

[0139] Based on the initial point cloud dataset, the iterative nearest point algorithm is used to perform preliminary registration of adjacent frames to obtain the point cloud relationship after preliminary registration.

[0140] Based on the comparison between the point cloud relationship after preliminary registration and the preset threshold, the point pairs in the local area are adjusted to obtain the adjusted point cloud registration result.

[0141] Based on the adjusted point cloud registration results, obtain the relative pose transformation matrix between multiple frames of point clouds and determine the continuity of pose transformation.

[0142] Based on the continuity of pose transformation, a cumulative transformation method is used for global registration to obtain globally consistent point cloud registration data.

[0143] The final pose estimate is calculated based on globally consistent point cloud registration data, resulting in an enhanced pose estimate.

[0144] Furthermore, for example, this embodiment extracts specific implementation methods from real-world application scenarios when processing the acquisition and registration of continuous multi-frame point cloud data. For acquiring the initial point cloud dataset, a LiDAR device continuously scans the environment in a fixed scene, such as during the operation of an autonomous vehicle. The LiDAR acquires 10 frames of point cloud data per second, with each frame containing 100,000 points, forming the initial dataset. This method ensures the continuity and integrity of the data, laying the foundation for subsequent processing.

[0145] For example, this embodiment uses an iterative nearest-point algorithm for initial registration, which can be understood as finding the closest point pairs and calculating the transformation relationship by comparing the geometric features of point clouds in adjacent frames. Suppose that in a certain registration, the initial transformation matrix between two point clouds shows a translation error of 0.5 meters and a rotation error of 2 degrees. If the preset thresholds are 0.3 meters for translation and 1 degree for rotation, further adjustments are needed. This initial registration provides a starting point for subsequent optimization.

[0146] For example, in this embodiment, after detecting that the registration error exceeds the threshold, the adjustment of local point pairs can be specifically implemented by selecting areas with larger errors, such as the overlapping area of ​​point clouds when a vehicle is turning, and recalculating the matching relationship of local point pairs to reduce the error to within 0.2 meters. This local adjustment can effectively improve registration accuracy and ensure the accuracy of point cloud relationships.

[0147] For example, in this embodiment, when acquiring the relative pose transformation matrix between multiple frames of point cloud data, the transformation matrix between every two frames is calculated using five consecutive frames of point cloud data to determine whether the translation and rotation changes are smooth. If it is found that the translation amount of the transformation matrix in a certain frame suddenly increases to 1 meter, while in other frames it is only 0.1 meters, it can be marked as an anomaly and further verification is required. This continuity judgment helps to discover potential problems.

[0148] For example, for the cumulative transformation of global registration, this embodiment progressively superimposes the point cloud data from the first 10 frames based on a relative transformation matrix to form a point cloud distribution in a unified coordinate system. This global consistency processing can avoid cumulative errors and ensure the spatial consistency of the overall point cloud data.

[0149] For example, in calculating the final pose estimate, this embodiment combines global registration data and uses a weighted average method to determine the vehicle's pose at a certain moment, such as a final translation of 5.2 meters and a rotation angle of 10 degrees. This enhanced pose estimation result provides a reliable basis for subsequent navigation.

[0150] For example, in this embodiment, the stability index of pose estimation is determined by analyzing the pose change amplitude over 10 consecutive frames. If the translational change is within 0.05 meters and the rotational change is within 0.5 degrees, the stability is considered high. This reliability assessment provides an important reference for system decision-making and helps improve the overall operational safety and stability.

[0151] Furthermore, the process of obtaining the misalignment quantification index includes:

[0152] Key feature points are extracted from point cloud data and geometrically matched with map elements to obtain initial pose estimation results.

[0153] The geometric relationship difference between point cloud features and map features is calculated based on the initial pose estimation results to obtain the preliminary value of lateral misalignment;

[0154] The matching error is optimized using the least squares method based on the initial value of lateral misalignment to obtain lateral misalignment data;

[0155] The distribution of heading angle deviation is analyzed based on the lateral misalignment data to obtain the heading angle deviation parameters;

[0156] The point cloud features are locally adjusted based on the comparison between the heading angle deviation parameter and the preset threshold to obtain the corrected deviation value.

[0157] Based on the corrected deviation value and the lateral misalignment data, the misalignment quantification parameters are derived to obtain the misalignment quantification index.

[0158] Furthermore, for example, in this embodiment, when extracting key feature points from point cloud data, significant geometric shapes, such as edges or corners, are identified in the point cloud as the basis for subsequent matching. Assuming that in an autonomous driving scenario, the point cloud data collected by LiDAR includes guardrails and road signs on both sides of the road, these structures have obvious linear or planar characteristics and can serve as key feature points. This extraction method can effectively reduce the amount of data processing while retaining important geometric information, laying the foundation for subsequent alignment.

[0159] For example, this embodiment uses a geometric matching method to perform preliminary alignment with map features, which can be understood as comparing point cloud feature points with road features in a pre-loaded high-precision map.

[0160] In one possible implementation, assuming the guardrail lines in the point cloud data are matched with the guardrail lines in the map, the initial calculation shows that the vehicle's translational deviation is 0.8 meters and the rotational deviation is 3 degrees. This initial alignment can quickly locate the approximate position of the vehicle, providing a starting point for subsequent optimization.

[0161] For example, in this embodiment, when calculating the geometric differences between point cloud features and map elements, a preliminary value for lateral misalignment is determined by comparing the distance and angle between the two. Suppose that in a certain calculation, the location of a landmark in the point cloud is laterally offset by 0.6 meters from its corresponding location on the map; this value can be used as preliminary misalignment data. This difference analysis helps to clarify the direction and magnitude of the adjustment that needs to be made.

[0162] For example, in this embodiment, when optimizing the matching error using the least squares method, fine adjustments can be made to the lateral misalignment data. Assuming the initial lateral misalignment is 0.6 meters, after multiple iterations of optimization, the misalignment value is reduced to 0.2 meters. This optimization method can significantly improve the accuracy of the matching, providing more reliable data support for subsequent analysis.

[0163] For example, in this embodiment, when analyzing the distribution of heading angle deviation, the specific deviation parameters can be derived by statistically analyzing the optimized point cloud and map matching results. Assuming that the average heading angle deviation is 2 degrees and the maximum is 4 degrees in multiple consecutive frames of data, this distribution analysis can help determine whether the deviation is within an acceptable range.

[0164] For example, if the heading angle deviation exceeds a preset threshold, such as 1 degree, then local adjustments to the point cloud features are needed. Suppose that in a certain frame of data the deviation reaches 3 degrees, then by re-matching feature points in the local area, such as road signs or lane lines, the deviation can be corrected to 0.5 degrees. This local adjustment can effectively reduce the accumulation of deviations and improve overall consistency.

[0165] For example, in this embodiment, when deriving the misalignment quantification parameters by combining the corrected deviation value and lateral misalignment data, translation and rotation data can be integrated to form the final quantification result. Assuming the final lateral misalignment is 0.2 meters and the heading angle deviation is 0.5 degrees, these parameters can serve as a reference for vehicle positioning. This comprehensive quantification method can provide more accurate input for subsequent path planning.

[0166] Furthermore, the process of obtaining the updated pose includes:

[0167] The misalignment quantization error is calculated based on the current pose and the map matching, and the misalignment quantization error value is obtained.

[0168] A trigger signal is generated based on the comparison result between the misalignment quantization error value and the preset threshold.

[0169] The relocation process is initiated based on the trigger signal, loading global map data and current sensor observation data;

[0170] The current sensor observation data is distributed to the global map data according to the particle filtering algorithm, forming a particle set and corresponding weights.

[0171] A resampling operation is performed based on the particle set and its corresponding weights, retaining high-weight particles to obtain a convergent particle set;

[0172] The pose corresponding to the particle with the highest weight is selected from the convergent particle set to determine the updated pose.

[0173] Furthermore, for example, in this embodiment, when processing the misalignment quantization error in pose and map matching, the calculation of the misalignment quantization error is usually based on the spatial difference between the current pose and the map data. Suppose that in a certain scenario, a vehicle is driving on an urban road, and the point cloud data collected by the sensor has a certain offset from the pre-loaded high-precision map. By comparing the geometric features of the two, the misalignment quantization error value is found to be 0.5 meters, while the preset threshold is 0.3 meters, which clearly exceeds the limit. In this case, the system will generate a trigger signal, indicating that further pose adjustment is needed to ensure positioning accuracy.

[0174] For example, the discussion on initiating the relocation process based on the trigger signal is based on the loading of global map data and the fusion of sensor observation data.

[0175] In one possible implementation, the system loads global map data covering the current area while simultaneously acquiring local environmental information collected in real time by sensors. Assuming the vehicle is at a complex intersection with multiple similar buildings nearby, the sensor data may be ambiguous. In this case, by performing a preliminary comparison between the sensor data and the global map, the system can identify the approximate location range, providing a basis for subsequent particle filtering.

[0176] For example, this embodiment focuses on the application of a particle filtering algorithm, including particle distribution and weight calculation. The system transforms sensor observation data into multiple particles, scattered across the global map, with each particle representing a possible pose hypothesis. Assuming 1000 particles are distributed near an intersection, the system calculates the corresponding weight by comparing the matching degree of each particle with map features. Some particles may be concentrated in the center of the road, receiving higher weights, while particles deviating from the road have lower weights. This method can progressively filter out pose hypotheses that better reflect reality.

[0177] For example, this embodiment analyzes the resampling operation and the generation of the convergent particle set from the perspective of particle selection. During the resampling process, the system retains particles with higher weights and discards particles with lower weights. Assume that after one resampling, 1000 particles are reduced to 200, and these particles are concentrated within a 5-meter radius of the road centerline, forming a convergent particle set. This selection mechanism can effectively reduce computational load while improving the reliability of pose estimation.

[0178] For example, in this embodiment, the pose corresponding to the particle with the highest weight is selected based on the final pose update. Assuming that the pose corresponding to the particle with the highest weight in the converged particle set is 0.2 meters east of the road centerline, the system will use this pose as the updated current pose. This update method can ensure that the vehicle maintains high positioning accuracy in complex environments, providing a reliable basis for subsequent path planning.

[0179] In summary, the various stages described above, from misalignment error assessment to final pose update, form a complete closed-loop process. Whether it's error comparison, trigger signal generation, particle filtering, or resampling operations, each stage is closely interconnected, working together to improve the stability of the positioning system. Especially in complex urban environments, this process effectively addresses subtle differences between the map and the actual environment, ensuring the continuity and reliability of system operation.

[0180] Furthermore, the process of obtaining a continuous point cloud sequence includes:

[0181] Based on the vehicle dynamics model, the original pose data is obtained from the acquisition interval and combined with the historical motion trajectory for preliminary analysis to obtain the initial estimation result after pose update.

[0182] Based on the initial estimation results after pose update, the Kalman filter method is used to smooth the motion changes and determine the trend of motion trajectory changes within the acquisition interval.

[0183] Based on the trend of motion trajectory changes and the time span of the acquisition interval, the missing spatiotemporal points in the point cloud sequence are predicted to obtain preliminary point cloud supplementary data.

[0184] Based on the preliminary point cloud supplementary data, the point cloud sequence is filled using interpolation compensation technology to generate intermediate point cloud data with a smooth transition.

[0185] Based on the intermediate point cloud data, a sequence optimization method is used to perform consistency correction on the overall point cloud sequence, resulting in a point cloud sequence with higher continuity.

[0186] Based on the point cloud sequence results, the local non-smooth areas are dynamically adjusted and optimized to determine the final continuous point cloud sequence.

[0187] Furthermore, in this embodiment, the original pose data is obtained from the acquisition interval of adjacent laser frames through the vehicle dynamics model in the high-precision vehicle positioning system. Combined with the historical motion trajectory, a preliminary analysis is performed to obtain the initial estimation result after pose update.

[0188] Specifically, if the vehicle travels straight at a constant speed of 30 km / h within a 0.1-second data acquisition interval, and the pose change calculated jointly by the wheel speed sensor and the IMU is 2.5 meters forward and 0.8 degrees yaw, then this will be used as the initial estimate to avoid abrupt changes caused by directly using a single sensor.

[0189] In one embodiment, the motion changes are smoothed by using an extended Kalman filter on the initial estimation result, and the observations of the accelerometer and gyroscope are fused to significantly suppress high-frequency noise, so that the trend of motion trajectory changes within the acquisition interval is more consistent with the actual vehicle kinematic characteristics.

[0190] It should be noted that after filtering, the yaw rate fluctuation can be reduced from ±5 deg / s to within ±0.3 deg / s, providing a stable basis for subsequent point cloud prediction.

[0191] Understandably, based on the trend of the smoothed motion trajectory and combined with a time span of 0.1 seconds, the missing spatiotemporal points in the point cloud sequence caused by motion distortion are predicted.

[0192] For example, a 360-degree single-frame scan by a lidar takes 0.1 seconds. The point cloud at the front of the vehicle has moved forward by about 0.42 meters, while the point cloud at the rear of the vehicle remains in the same position. At this point, a uniform speed model is used to interpolate all points to the end of the frame to obtain preliminary supplementary point cloud data and eliminate obvious stretching deformation.

[0193] For example, the sparse areas are further filled by spline interpolation compensation technology for the preliminary supplementary data. When the turning radius is 50 meters and the vehicle speed is 36 km / h, the density of the outer point cloud is reduced by about 30%. By inserting the intermediate point along the motion trajectory with cubic splines, the distance between adjacent points is restored from 15 cm to about 5 cm, generating intermediate point cloud data with a smooth transition.

[0194] In one embodiment, a sequence optimization method, such as sliding window bundle adjustment, is used to perform consistency correction on the overall point cloud sequence. The point cloud of 10 consecutive frames is jointly optimized, so that the pose residual between adjacent frames is reduced from 0.15 meters to less than 0.03 meters, resulting in a point cloud sequence with significantly improved continuity.

[0195] Specifically, if there are local uneven areas on slopes or bumpy roads, the point cloud in these areas is optimized a second time through dynamic adjustment technology. For example, the local point set with z-axis fluctuations exceeding 8 cm is re-weighted and iterated to improve the ground flatness of the final continuous point cloud sequence to within 3 cm.

[0196] Preferably, the final continuous point cloud sequence is combined with the vehicle dynamics model for verification updates. For example, the error between the predicted displacement and the actual wheel speed integral is less than 0.5%, thereby generating a complete data sequence that is uniform in time and continuous in space. This provides high-quality input for subsequent map construction and real-time positioning, significantly reduces cumulative drift, and improves system robustness.

[0197] Furthermore, the process of obtaining the location reliability index and its location results includes:

[0198] The initial values ​​of point cloud density and matching residuals are extracted based on the acquisition and preprocessing of continuous point cloud sequences to obtain a preliminary dataset.

[0199] Based on the preliminary dataset, a continuous analysis method was used to perform time-series comparison of point cloud density, identify density change patterns, and determine significant intervals of density change.

[0200] Based on the significant intervals of density changes, and combined with the distribution of matching residuals, the influence weight of residual factors on positioning is calculated to determine the main sources of error.

[0201] Based on the main sources of error, a propagation path is constructed using an error propagation model, resulting in a structure diagram of the propagation path.

[0202] Based on the structure diagram of the propagation link, analyze the degree of impact of error propagation on the positioning signal, deduce the fluctuation range of the positioning signal, and determine the preliminary assessment value of the confidence level;

[0203] The final value of the confidence index is calculated based on the preliminary confidence level assessment value and the sequence analysis results to obtain the location confidence quantification result;

[0204] The parameters of the error propagation model are adjusted based on the comparison between the location confidence quantification results and the preset threshold to obtain the optimized confidence index value.

[0205] The predicted vehicle positioning result is output based on the optimized confidence index value.

[0206] For example, in the scenario of building high-precision maps for autonomous driving, the acquisition of point cloud sequences is often limited by the high-speed movement of vehicles and the frequency of sensors, resulting in uneven density distribution, which directly affects the accuracy of subsequent positioning and mapping. This embodiment, through the acquisition and preprocessing of point cloud sequences, first performs voxel filtering and ground segmentation on the original LiDAR data to extract effective non-ground point clouds, and simultaneously calculates the point cloud density of each scan frame and the initial residual value after feature matching to form a preliminary dataset.

[0207] Specifically, point cloud density is defined as the number of points within a unit spherical neighborhood. For example, within a radius of 0.5 meters, the normal frame density in an urban road scene is typically between 150 and 200 points, while it may drop below 80 points when a vehicle accelerates rapidly or turns. Using a continuous analysis method, the density within 10 adjacent frames is compared temporally. It is found that the interval where the density drops sharply from 180 points to 90 points often corresponds to a vehicle entering a tunnel or a severely occluded area. At this time, the density change pattern shows obvious step characteristics and is marked as a significant interval.

[0208] In one possible implementation, the matching residual distribution is calculated simultaneously. For example, under normal circumstances, the mean NDT matching residual is about 0.12 meters, while in the interval where the density decreases significantly, the mean residual rises to 0.35 meters. At this time, by calculating the influence weight of the residual factor on positioning through weighted analysis, it is found that the decrease in density contributes about 70% of the error increment, thus determining that the main source of error is indeed highly correlated with the density change.

[0209] It should be noted that once the primary error is identified as originating from density unevenness, an error propagation model can be constructed. For example, treating density variation as an input disturbance, a propagation path can be established in a state-space format. It is observed that the error initially accumulates in the yaw angle estimation, causing the lateral positioning deviation to gradually amplify along the vehicle's direction of travel. Within 20 consecutive frames, this deviation can accumulate from 0.1 meters to 0.8 meters, forming a clear propagation path structure diagram. Further analysis of this structure diagram allows for the derivation of the positioning confidence range. For instance, the confidence level is stable above 0.95 during the normal segment, while it drops to 0.63 during the error propagation peak segment, yielding a preliminary assessment value. Subsequently, combined with the overall sequence matching score, the final positioning confidence quantification result is calculated, resulting in a weighted average of 0.78.

[0210] Preferably, when the quantization result is lower than a preset threshold of 0.85, the system automatically triggers parameter readjustment, such as reducing the process noise coefficient in the error propagation model to 60% of its original value, or increasing the density compensation weight to regenerate the propagation link, thereby raising the optimized confidence index to 0.89 to meet the requirements of subsequent fusion positioning. This confidence assessment method based on joint analysis of density and residuals effectively avoids positioning failures caused by fluctuations in point cloud quality, and improves the robustness and continuous positioning accuracy of the system in complex urban scenarios.

[0211] Example 2

[0212] like Figure 2 As shown, based on the same inventive concept, this embodiment also provides a high-precision vehicle positioning system based on lidar point cloud data, including:

[0213] The data fusion module is used to synchronize the timestamps and transform the coordinate system of the point cloud data collected by multi-directional lidar to obtain fused point cloud data.

[0214] The feature comparison module is used to extract stable features from the fused point cloud data and compare them with map reference features to obtain the position deviation vector;

[0215] The dynamic correction module is used to adjust the pose by triggering a dynamic correction mechanism based on the relationship between the position deviation vector and a preset threshold, so as to obtain the corrected pose estimate.

[0216] The cluster registration module is used to iteratively register multiple consecutive frames of point cloud data in a feature-sparse environment based on the corrected pose estimation to obtain the enhanced pose estimation.

[0217] The misalignment quantization module is used to analyze the geometric relationship between point cloud features and map elements based on enhanced pose estimation to obtain misalignment quantization indicators.

[0218] The repositioning module is used to trigger the repositioning process based on the relationship between the misalignment quantification index and the preset threshold to update the pose and obtain the updated pose.

[0219] The interpolation compensation module is used to estimate motion changes based on the updated pose and vehicle dynamics model, and to interpolate the point cloud sequence to obtain a continuous point cloud sequence.

[0220] The positioning evaluation module is used to track point cloud density and match residuals based on a continuous point cloud sequence and to perform error propagation calculations to obtain the positioning reliability index and its positioning results.

[0221] The vehicle high-precision positioning system based on lidar point cloud data provided in this embodiment has all the advantages of the vehicle high-precision positioning method based on lidar point cloud data provided in Embodiment 1.

[0222] Example 3

[0223] This embodiment also discloses a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in Embodiment 1.

[0224] Example 4

[0225] This embodiment also discloses a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0226] Example 5

[0227] This embodiment also discloses a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0228] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A high-precision vehicle positioning method based on lidar point cloud data, characterized in that, include: The point cloud data collected by multi-directional lidar is time-stamped and coordinate system transformed to obtain fused point cloud data; Based on the fused point cloud data, stable features are extracted and compared with map reference features to obtain a position deviation vector; The pose is adjusted by triggering a dynamic correction mechanism based on the relationship between the position deviation vector and the preset threshold, and the corrected pose estimate is obtained. Based on the corrected pose estimation, iterative registration is performed on multiple consecutive frames of point cloud data in a feature-sparse environment to obtain the enhanced pose estimation. Based on the enhanced pose estimation, the geometric relationship between point cloud features and map elements is analyzed to obtain a misalignment quantification index. Based on the relationship between the misalignment quantification index and the preset threshold, a relocation process is triggered to update the pose, thereby obtaining the updated pose. Based on the updated pose and the vehicle dynamics model, motion changes are estimated and point cloud sequence interpolation is performed to obtain a continuous point cloud sequence. Based on the continuous point cloud sequence, the point cloud density is tracked and the residual is matched, and error propagation calculation is performed to obtain the location reliability index and its positioning result.

2. The method according to claim 1, characterized in that, The process of obtaining fused point cloud data includes: The acquisition time of each frame of point cloud is obtained by parsing the timestamp information of the raw point cloud data collected by the multi-source lidar. The time offset of each LiDAR is determined based on the deviation between the acquisition time and the vehicle's master clock. The point cloud timestamps are corrected based on the relationship between the time offset and the preset threshold to obtain a time-corrected point cloud. Based on the vehicle's current pose and the lidar extrinsic parameter matrix, the time-corrected point cloud is transformed to the vehicle's body coordinate system to obtain a unified coordinate point cloud. The nearest neighbor search method is used to find the spatial overlapping region of different lidars in the unified coordinate point cloud to obtain the overlapping region point cloud. The overlapping region point cloud is spatially divided using a voxel mesh partitioning method to obtain a voxelized point cloud set. The position averaging of points from different lidars is calculated based on the voxelized point cloud set to obtain sparse fused point cloud data.

3. The method according to claim 1, characterized in that, The process of obtaining the position deviation vector includes: Preprocessing the sparse fused point cloud data to filter out noise and irrelevant points yields a clean point cloud dataset. Based on the clean point cloud dataset, road markings and building outlines are extracted using a geometric segmentation method to obtain a stable feature set; Reference feature data is retrieved from the map and initially registered with the stable feature set to obtain an initial matching relationship. The initial matching relationship is finely adjusted based on a real-time comparison method to obtain an accurate matching result; Calculate the positional deviation based on the precise matching result and generate deviation vector data; The final position deviation vector is obtained by performing a final verification based on the deviation vector data and map reference features.

4. The method according to claim 1, characterized in that, The process of obtaining the corrected pose estimate includes: The magnitude is calculated based on the position deviation vector, and the current frame is marked as having an excessive deviation state based on the relationship between the magnitude and a preset threshold. The dynamic correction mechanism is activated based on the deviation exceeding the limit state. The historical pose estimation and the current deviation vector are fused by a Kalman filter to obtain the preliminary corrected pose. The optimized corrected pose is obtained by performing consistency constraint optimization on the preliminary corrected pose and the pose estimation of the most recent multi-frames using the least squares method. The optimized pose is overlaid on the original pose estimate to obtain the corrected pose estimate.

5. The method according to claim 1, characterized in that, The process of obtaining enhanced pose estimation includes: An initial point cloud dataset is obtained by collecting point cloud data from multiple consecutive frames. Based on the initial point cloud dataset, the iterative nearest-point algorithm is used to perform preliminary registration of adjacent frames to obtain the point cloud relationship after preliminary registration. Based on the comparison results between the pre-registered point cloud relationship and the preset threshold, the point pairs in the local area are adjusted to obtain the adjusted point cloud registration result; Based on the adjusted point cloud registration results, obtain the relative pose transformation matrix between multiple frames of point clouds, and determine the continuity of the pose transformation. Based on the continuity of the pose transformation, a cumulative transformation method is used for global registration to obtain globally consistent point cloud registration data. The final pose estimate is calculated based on the globally consistent point cloud registration data to obtain the enhanced pose estimate result.

6. The method according to claim 1, characterized in that, The process of obtaining the misalignment quantification index includes: Key feature points are extracted from point cloud data and geometrically matched with map elements to obtain initial pose estimation results. Based on the initial pose estimation results, the geometric relationship difference between point cloud features and map elements is calculated to obtain a preliminary value of lateral misalignment. The matching error is optimized using the least squares method based on the preliminary value of the lateral misalignment to obtain the lateral misalignment data; Based on the analysis of the lateral misalignment data, the distribution of the heading angle deviation is obtained, and the heading angle deviation parameters are obtained. The point cloud features are locally adjusted based on the comparison result between the heading angle deviation parameter and the preset threshold to obtain the corrected deviation value; Based on the corrected deviation value and the lateral misalignment data, the misalignment quantification parameters are derived to obtain the misalignment quantification index.

7. The method according to claim 1, characterized in that, The process of obtaining the updated pose includes: The misalignment quantization error is calculated based on the current pose and the map matching, and the misalignment quantization error value is obtained. A trigger signal is generated based on the comparison result between the misalignment quantization error value and the preset threshold. The relocation process is initiated based on the trigger signal, loading global map data and current sensor observation data; The current sensor observation data is distributed to the global map data according to the particle filtering algorithm, forming a particle set and corresponding weights. A resampling operation is performed based on the particle set and its corresponding weights, retaining high-weight particles to obtain a convergent particle set. Based on the convergent particle set, select the pose corresponding to the particle with the highest weight to determine the updated pose.

8. The method according to claim 1, characterized in that, The process of obtaining a continuous point cloud sequence includes: Based on the vehicle dynamics model, the original pose data is obtained from the acquisition interval and combined with the historical motion trajectory for preliminary analysis to obtain the initial estimation result after pose update. Based on the initial estimation results after pose update, the Kalman filter method is used to smooth the motion changes and determine the trend of motion trajectory changes within the acquisition interval. Based on the trend of motion trajectory changes and the time span of the acquisition interval, the missing spatiotemporal points in the point cloud sequence are predicted to obtain preliminary point cloud supplementary data. Based on the preliminary point cloud supplementary data, the point cloud sequence is filled using interpolation compensation technology to generate smoothly transitioning intermediate point cloud data. Based on the intermediate point cloud data, a sequence optimization method is used to perform consistency correction on the overall point cloud sequence, resulting in a point cloud sequence with higher continuity. Based on the point cloud sequence results, the local non-smooth areas are dynamically adjusted and optimized to determine the final continuous point cloud sequence.

9. The method according to claim 1, characterized in that, The process of obtaining the location reliability index and its location results includes: The initial values ​​of point cloud density and matching residuals are extracted based on the acquisition and preprocessing of continuous point cloud sequences to obtain a preliminary dataset. Based on the preliminary dataset, a continuous analysis method is used to perform time-series comparison of point cloud density, identify density change patterns, and determine significant intervals of density change. Based on the significant intervals of the density changes, and combined with the distribution of the matching residuals, the influence weight of the residual factors on the positioning is calculated to determine the main sources of error. Based on the main sources of error, a propagation path is constructed using an error propagation model, resulting in a structure diagram of the propagation path. Based on the structure diagram of the propagation link, analyze the degree of impact of error propagation on the positioning signal, deduce the fluctuation range of the positioning signal, and determine the preliminary assessment value of the confidence level; The final value of the confidence index is calculated based on the preliminary assessment value of the confidence level and the results of the sequence analysis, thus obtaining the quantitative result of the location confidence index. The parameters of the error propagation model are adjusted based on the comparison between the quantitative results of the location confidence and the preset threshold to obtain the optimized confidence index value. The predicted vehicle positioning result is output based on the optimized confidence index value.

10. A high-precision vehicle positioning system based on lidar point cloud data, characterized in that, include: The data fusion module is used to synchronize the timestamps and transform the coordinate system of the point cloud data collected by multi-directional lidar to obtain fused point cloud data. The feature comparison module is used to extract stable features from the fused point cloud data and compare them with map reference features to obtain a position deviation vector; The dynamic correction module is used to trigger a dynamic correction mechanism to adjust the pose based on the relationship between the position deviation vector and the preset threshold, so as to obtain the corrected pose estimate. The clustering and registration module is used to iteratively register multiple consecutive frames of point cloud data in a feature-sparse environment based on the corrected pose estimation to obtain the enhanced pose estimation. The misalignment quantization module is used to analyze the geometric relationship between point cloud features and map elements based on the enhanced pose estimation to obtain misalignment quantization index. The repositioning module is used to trigger the repositioning process to update the pose based on the relationship between the misalignment quantification index and the preset threshold, so as to obtain the updated pose. The interpolation compensation module is used to estimate motion changes based on the updated pose and the vehicle dynamics model, and to perform point cloud sequence interpolation to obtain a continuous point cloud sequence. The positioning evaluation module is used to track the point cloud density and match the residuals based on the continuous point cloud sequence and perform error propagation calculations to obtain the positioning reliability index and its positioning results.

Images