Track precision evaluation method and device based on map positioning

Abstract

The invention relates to the technical field of self-moving robot positioning, and discloses a trajectory precision evaluation method and device based on map localization, and the method comprises the steps: obtaining an image sequence of self-moving operation equipment during operation and truth value trajectory data synchronized with the image sequence in time, and carrying out the repositioning of each frame of image in the image sequence and a pre-created map, calculating a relocation pose under a map coordinate system; calculating an initial transformation matrix representing a transformation relation between a map coordinate system and a true value coordinate system based on the relocation pose and the true value pose in the true value trajectory data under the same timestamp, and processing to obtain an optimized final transformation matrix; and converting the true value trajectory data to a map coordinate system by using the final transformation matrix, and comparing the true value trajectory data with a map positioning trajectory to evaluate precision. The method does not need additional alignment operation, is objective and reliable in evaluation result, adapts to various self-moving operation equipment such as mowers and sweepers, can provide data support for a positioning algorithm and map optimization, and assists in improving the product positioning precision.

Description

Technical Field

[0001] This invention relates to the field of self-moving robot positioning technology, specifically to a method and apparatus for evaluating trajectory accuracy based on map positioning. Background Technology

[0002] In the field of positioning technology for self-propelled mobile equipment (such as lawnmowers and sweepers), trajectory accuracy assessment is a core aspect of evaluating equipment positioning performance, directly impacting the equipment's operational path planning, the completeness of operational area coverage, and operational efficiency. Current mainstream trajectory accuracy assessment methods generally employ a scheme that directly compares the positioning trajectory with the RTK (Real-Time Kinematic) trajectory, using alignment operations to verify the accuracy of both. This method can only meet the accuracy assessment needs of basic positioning scenarios such as odometers.

[0003] However, traditional methods evaluate the odometry performance of the positioning system itself. But in map-based positioning scenarios, positioning accuracy depends not only on real-time sensing and matching algorithms, but also fundamentally on the accuracy and global consistency of the pre-created map. Traditional methods cannot effectively assess the contribution or constraint of map quality on the final positioning accuracy.

[0004] In map-based positioning, the positioning results have already been bound to the map coordinate system through methods such as relocation. If traditional trajectory alignment methods are continued to be used, it is essentially equivalent to using the positioning results to correct the map coordinate system. This masks the systematic errors that may exist in the map itself (such as cumulative drift during mapping), and fails to achieve the core objective of assessing the impact of the map on positioning accuracy.

[0005] For devices such as lawnmowers and robotic vacuum cleaners that need to maintain centimeter-level positioning accuracy during long-term, large-scale repetitive operations, their positioning systems are highly dependent on high-precision and consistent maps. Traditional evaluation methods cannot provide a precise transformation relationship between the map coordinate system and the global ground truth coordinate system (such as the RTK north-south coordinate system), making it difficult to conduct an effective and reliable quantitative evaluation of the long-term positioning performance based on maps. Summary of the Invention

[0006] To address the technical challenge of accurately evaluating the trajectory accuracy of a positioning system based on a pre-created map, this invention provides a method and apparatus for evaluating trajectory accuracy based on map positioning. Instead of performing traditional trajectory alignment operations, it directly calculates and optimizes the precise transformation relationship between the map coordinate system and the true coordinate system. This allows for an objective evaluation of the quality of the map itself and its actual improvement effect on positioning accuracy, providing a reliable technical tool for the research, development, testing, and optimization of high-precision positioning products.

[0007] In a first aspect, the present invention provides a trajectory accuracy evaluation method based on map positioning, comprising: Image sequences and time-synchronized true trajectory data of the mobile work equipment during operation are acquired, the true trajectory data being acquired based on a preset positioning system; Each frame of the image sequence is relocated to a pre-created map, and the relocation pose of each frame in the map coordinate system is calculated. Based on the repositioning pose and the true pose in the true trajectory data at the same timestamp, calculate the initial transformation matrix representing the transformation relationship between the map coordinate system and the true coordinate system; The initial transformation matrix is ​​sequentially subjected to outlier filtering and optimization processes to obtain the final transformation matrix between the optimized map coordinate system and the ground value coordinate system. The true trajectory data is transformed to a map coordinate system using the final transformation matrix and compared with the positioning trajectory data based on the map to evaluate the trajectory accuracy of the positioning system.

[0008] The trajectory accuracy evaluation method based on map positioning provided in this invention achieves accuracy evaluation by simultaneously acquiring image sequences and ground truth trajectory data, followed by repositioning, transformation matrix calculation, optimization, and trajectory comparison. This addresses the pain points of traditional methods, which require manual alignment and fail to reflect the map's contribution to improving positioning accuracy. Evaluation can be performed directly without additional alignment operations, accurately reflecting the true performance of map-based positioning. The process is logically coherent, forming a complete closed loop from data acquisition to accuracy quantification, and is compatible with various self-moving equipment such as lawnmowers and sweepers. By establishing a dual coordinate system association through a transformation matrix, combined with outlier filtering and optimization, it effectively resists data noise and environmental interference. The evaluation results are objective and reliable, quantifying the positioning system accuracy and indirectly reflecting the map construction quality, providing comprehensive data support for positioning algorithms and map optimization, and facilitating the development of high-precision positioning products.

[0009] Secondly, the present invention provides a trajectory accuracy evaluation device based on map positioning, the device comprising: The data acquisition module is used to acquire image sequences from the mobile work equipment during operation and real-value trajectory data synchronized with its time, wherein the real-value trajectory data is acquired based on a preset positioning system; The relocation calculation module is used to relocate each frame of the image sequence with a pre-created map and calculate the relocation pose of each frame in the map coordinate system. The initial matrix calculation module is used to calculate an initial transformation matrix representing the transformation relationship between the map coordinate system and the true coordinate system based on the repositioning pose and the true pose in the true trajectory data at the same timestamp. The final transformation matrix acquisition module is used to sequentially perform outlier filtering and optimization processing on the initial transformation matrix to obtain the final transformation matrix between the optimized map coordinate system and the true coordinate system. The accuracy evaluation module is used to transform the true trajectory data to a map coordinate system using the final transformation matrix, and compare it with the positioning trajectory data based on the map to evaluate the trajectory accuracy of the positioning system.

[0010] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the trajectory accuracy evaluation method based on map positioning described in the first aspect or any corresponding embodiment thereof.

[0011] Fourthly, the present invention provides a self-moving working device, comprising: a moving component, a working component, a vision component, a positioning component, a radar component, and a controller, wherein: The mobile component is used to provide the device with mobility and motion capability; The task component is used to execute specific task actions according to the task instructions; The vision component is used to acquire images and depth information of the working environment in real time; The positioning component is used to determine the position, attitude and movement trajectory of the device in the working environment in real time; The radar component is used for environmental perception and positioning assistance; The controller executes the trajectory accuracy evaluation method based on map positioning as described in the first aspect or any corresponding embodiment.

[0012] The self-moving work device provided in this invention integrates a positioning component, a vision component, and a radar component, which complement each other to eliminate the measurement blind spots and errors of a single sensor. The controller dynamically corrects positioning deviations through trajectory evaluation, significantly improving the device's positioning and trajectory tracking accuracy and avoiding missed or repetitive tasks. The moving component executes motion commands according to a precise trajectory, and the actions of the work component are highly matched with the path, effectively improving work quality and efficiency while reducing ineffective energy consumption. The device has a high degree of integration and a compact structure, does not affect the original work functions, and is suitable for various work scenarios such as lawn mowing and sweeping. Through its autonomous evaluation capabilities, the device can continuously optimize its positioning performance, improve the accuracy of the work path and work efficiency, provide users with a more reliable and efficient work experience, and enhance the product's market competitiveness.

[0013] Fifthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the map-based trajectory accuracy evaluation method of the first aspect or any corresponding embodiment thereof.

[0014] In a sixth aspect, the present invention provides a computer program product, including computer instructions for causing a computer to execute the map-based positioning trajectory accuracy evaluation method described in the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a trajectory accuracy evaluation method based on map positioning according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating another trajectory accuracy evaluation method based on map positioning according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating another trajectory accuracy evaluation method based on map positioning according to an embodiment of the present invention; Figure 4 This is a flowchart illustrating another trajectory accuracy evaluation method based on map positioning according to an embodiment of the present invention; Figure 5 This is a structural block diagram of a trajectory accuracy evaluation device based on map positioning according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention; Figure 7 This is a schematic diagram of a self-moving operation device according to an embodiment of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0018] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0019] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] As an optional application scenario of this invention, this embodiment applies to autonomous mobile robots, which can be devices such as lawnmowers, sweeping robots, warehouse handling robots, and park patrol robots that require high-precision positioning based on maps. Specifically, the autonomous mobile robot is equipped with a visual sensor and a positioning module, enabling it to move autonomously and perform tasks within its work area. Before performing a task, the autonomous mobile robot creates and stores an environmental map using SLAM technology; during task execution, it performs real-time positioning and navigation based on the map.

[0021] This invention provides an embodiment of a trajectory accuracy evaluation method based on map positioning. Instead of traditional trajectory alignment, it directly calculates and optimizes the precise transformation relationship between the map coordinate system and the true coordinate system. This allows for an objective evaluation of the map's quality and its actual improvement in positioning accuracy, providing a reliable technical tool for the research, testing, and optimization of high-precision positioning products. It should be noted that the steps shown in the flowcharts can be executed in a computer system, such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowcharts, in some cases, the steps shown or described may be performed in a different order than that presented here. Figure 1 This is a flowchart of a trajectory accuracy evaluation method based on map positioning according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps: Step S101: Obtain the image sequence of the mobile work equipment during operation and the real trajectory data synchronized with its time. The real trajectory data is obtained based on a preset positioning system.

[0022] Specifically, this step is the basic data acquisition stage for trajectory accuracy assessment in this invention, and its core is to simultaneously acquire two types of key data: 1. Image sequence: A series of images captured and stored at preset intervals (or at preset distance intervals) by the vision sensors (cameras, etc.) mounted on self-moving operating equipment (such as lawnmowers, sweepers) during operation, recording real-time environmental visual information during equipment operation.

[0023] 2. Time-synchronized true trajectory data: Real-time pose data (including position coordinates and attitude angles) of the device during operation is collected through a preset high-precision positioning system, such as RTK (Real-Time Kinematic) positioning module. This data strictly corresponds to the same timestamp as the above image sequence, ensuring that each frame of the image can match the actual motion state of the device at the same moment. The true trajectory data provides an objective benchmark for subsequent accuracy evaluation.

[0024] This invention utilizes ground truth trajectory data from high-precision positioning systems such as RTK, which possesses high reliability and accuracy, serving as an objective standard to measure the quality of map-based positioning trajectories. The image sequences contain rich environmental visual features, and when combined with synchronized ground truth data, they can be adapted to different operational scenarios such as grasslands and indoor environments, providing ample information support for subsequent multi-feature extraction and relocation calculations, ensuring the effectiveness of the evaluation method in complex environments. Through the time synchronization design between the image sequences and ground truth trajectory data, evaluation errors caused by data temporal misalignment are avoided, providing accurately matched basic data for subsequent relocation pose calculations and coordinate system transformation matrix solutions, ensuring the rigor of the evaluation logic.

[0025] Step S102: Relocate each frame of the image sequence to the pre-created map and calculate the relocation pose of each frame in the map coordinate system.

[0026] This invention, based on continuous image sequences acquired during the operation of a self-moving work device, employs a combination strategy of multi-feature extraction and multi-matching algorithms to perform correlation calculations with a pre-constructed scene map. Ultimately, it determines the precise pose (i.e., repositioning pose) of each image frame in the map coordinate system. The repositioning pose directly correlates the image sequence with the map coordinate system, providing crucial data support for subsequent calculations of the transformation matrix between the map coordinate system and the ground truth coordinate system. This is a prerequisite for achieving accurate evaluation without additional alignment.

[0027] Step S103: Based on the repositioning pose and the true pose in the true trajectory data at the same timestamp, calculate the initial transformation matrix representing the transformation relationship between the map coordinate system and the true coordinate system.

[0028] This invention utilizes two sets of key data at the same timestamp: the repositioning pose in the map coordinate system (from the repositioning calculation between the image and the pre-created map) and the ground truth pose in the ground truth coordinate system (from high-precision positioning systems such as RTK). Through coordinate transformation operations, a correspondence is established between the two, ultimately yielding an initial transformation matrix representing the translation and rotation relationship between the map coordinate system and the ground truth coordinate system. This matrix essentially represents a preliminary quantification of the spatial mapping relationship between two different coordinate systems, providing a basic model for subsequent optimization processing.

[0029] Step S104: The initial transformation matrix is ​​subjected to outlier filtering and optimization processes in sequence to obtain the final transformation matrix between the optimized map coordinate system and the true coordinate system.

[0030] This invention addresses the issue of rough accuracy in the initial transformation matrix caused by repositioning pose errors by sequentially filtering out and optimizing outliers in pose rotation and translation. This effectively eliminates errors caused by repositioning mismatches and environmental interference, thereby obtaining a high-precision final transformation matrix.

[0031] Step S105: The true trajectory data is transformed into the map coordinate system using the final transformation matrix and compared with the map-based positioning trajectory data to evaluate the trajectory accuracy of the positioning system.

[0032] Taking a lawnmower robot operation scenario as an example: Assume the robot has obtained the final transformation matrix between the map coordinate system and the RTK ground truth coordinate system using the method of this invention. This matrix precisely contains the translation and rotation mapping relationship between the two coordinate systems. The RTK ground truth trajectory data collected during robot operation (including the actual position coordinates and attitude angles at each timestamp) is input into this final transformation matrix to complete the transformation of the ground truth trajectory from the RTK coordinate system to the map coordinate system, resulting in the transformed ground truth trajectory in the same coordinate system as the map positioning trajectory. Subsequently, the pose data corresponding to the same timestamp in the transformed ground truth trajectory and the robot's map-based positioning trajectory are extracted. The numerical differences between the two in the position dimension (such as the deviation of X, Y, and Z axis coordinates) and the attitude dimension (such as the deviation of roll angle, pitch angle, and yaw angle) are calculated. By statistically analyzing indicators such as the mean deviation, maximum deviation, and standard deviation, the trajectory accuracy of the lawnmower robot's positioning system can be quantitatively evaluated. If the mean position deviation is ≤0.5cm and the mean angle deviation is ≤0.2 degrees (this is only an example and not a limitation), it indicates that the trajectory accuracy of the positioning system meets the requirements for high-precision operation.

[0033] The trajectory accuracy evaluation method based on map positioning provided in this invention breaks through the limitations of alignment operations in traditional trajectory evaluation. By calculating and optimizing the transformation matrix between the map coordinate system and the global true coordinate system, it can directly evaluate the quality of the map itself and its actual impact on positioning accuracy. It solves the problem that existing technologies cannot distinguish between map errors and positioning algorithm errors, and provides a scientific and objective technical tool for evaluating the construction quality of high-precision maps and optimizing and debugging positioning algorithms. It significantly improves the system reliability and operational accuracy of intelligent devices that rely on precise positioning, such as lawnmowers and autonomous vehicles.

[0034] This embodiment provides a trajectory accuracy evaluation method based on map positioning, which can be used in the aforementioned self-moving operation equipment. Figure 2 This is a schematic diagram of a second process for trajectory accuracy evaluation based on map positioning according to an embodiment of the present invention, as shown below. Figure 2 As shown, the process includes the following steps: Step S201: Obtain the image sequence of the mobile work device during operation and its time-synchronized true trajectory data. The true trajectory data is obtained based on a preset positioning system. For details, please refer to [link to relevant documentation]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0035] Step S202: Relocate each frame of the image sequence to the pre-created map and calculate the relocated pose of each frame in the map coordinate system.

[0036] In some optional implementations, step S202 above includes: Step S2021: Extract image feature points for each frame of image using at least two different feature extraction algorithms.

[0037] Taking a lawnmower robot operation scenario as an example, for a certain frame of grass environment image collected during robot operation, the SIFT feature extraction algorithm and the Superpoint feature extraction algorithm are used simultaneously to extract key feature points in the image. The SIFT algorithm focuses on local texture features of the image (such as grass blade edges and ground markers) to generate feature descriptors with scale invariance; the Superpoint algorithm extracts sparse key points in the image through a deep learning model, taking into account the stability and recognizability of feature points, and finally obtains two independent sets of image feature points.

[0038] Step S2022: At least two different feature matching algorithms are used to match the extracted image feature points with the 3D map points in the map.

[0039] In one optional embodiment, for the two sets of feature points mentioned above, the FLANN matching algorithm and the Lightglue matching algorithm are respectively used to match them with 3D map points in a pre-created 3D map of a grassland scene: Combination 1 (SIFT+FLANN): Utilizing the fast nearest neighbor search capability of the FLANN algorithm, it efficiently matches SIFT feature points with the ground. Figure 3 Dimensional points allow for quick preliminary matching results; Combination 2 (SIFT+Lightglue): By using the end-to-end feature matching logic of the Lightglue algorithm, the matching robustness of SIFT feature points is optimized, and false matching caused by changes in ambient lighting is reduced. Combination 3 (Superpoint + FLANN): Leveraging the efficiency of FLANN, quickly complete the mapping between Superpoint feature points and the ground. Figure 3 Dimensional relationships; Combination 4 (Superpoint + Lightglue): Leveraging Lightglue's adaptability to deep learning features, it improves the matching accuracy of Superpoint feature points, especially suitable for areas with sparse grass texture.

[0040] Step S2023: Based on different combinations of feature point extraction algorithms and feature matching algorithms, multiple candidate relocalization poses of the same frame image are calculated respectively, and the optimal pose is selected from the multiple candidate relocalization poses as the relocalization pose of the frame image in the map coordinate system.

[0041] Specifically, based on the above four algorithm combinations, the PnP (Perspective-n-Point) algorithm is used to calculate four candidate relocalization poses of the image frame in the map coordinate system; then, the reprojection error of each candidate pose (i.e., the difference between the image feature points after pose transformation and the map coordinate system) is calculated. Figure 3 (The deviation of the dimensional point projection position) selects the candidate pose with the smallest reprojection error and the most stable attitude angle as the final repositioning pose of the image frame.

[0042] This invention employs a multi-algorithm combination design to ensure that repositioning pose calculation does not rely on a single feature extraction or matching logic. This effectively avoids the bias of a single algorithm for specific scenarios (such as overfitting of traditional algorithms in specific texture environments), ensuring that the repositioning results objectively reflect the device's true motion state in the map coordinate system. The optimal repositioning pose combines accuracy and stability, providing high-quality basic data for subsequent calculation of the transformation matrix between the map coordinate system and the ground truth coordinate system. This avoids evaluation bias caused by inaccurate repositioning poses and ensures the reliability of trajectory accuracy evaluation results.

[0043] Step S203: Based on the repositioned pose and the ground truth pose in the ground truth trajectory data at the same timestamp, calculate the initial transformation matrix representing the transformation relationship between the map coordinate system and the ground truth coordinate system. For details, please refer to [link to relevant documentation]. Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0044] Step S204: The initial transformation matrix is ​​subjected to outlier filtering and optimization processes in sequence to obtain the final transformation matrix between the optimized map coordinate system and the true coordinate system.

[0045] In an optional embodiment, step S204 includes the following steps: Step S2041: Perform rotation anomaly filtering and translation anomaly filtering on the initial transformation matrix in sequence, wherein: rotation anomaly filtering includes: converting the rotation part of each pose pair into Euler angles and filtering out poses in which any component of the Euler angles exceeds a preset angle threshold; translation anomaly filtering includes: calculating the translation difference vector of the poses that have passed the rotation anomaly filtering, and using a clustering algorithm to process the translation difference vector to filter out outlier poses.

[0046] Specifically, in the embodiments of the present invention, during the rotational anomaly filtering, the rotational part of each pose pair is converted into Euler angles (roll, pitch, yaw), and poses in which any component of the Euler angles exceeds a preset angle threshold are filtered out. For example, during filtering, the angle thresholds for roll, pitch, and yaw are 3 degrees, and any value that does not meet the threshold will be removed.

[0047] In one optional embodiment, the clustering algorithm used for translation anomaly filtering is a hierarchical clustering algorithm. Its purpose is to identify "mainstream consistent solutions" from the spatial distribution of the translation difference vector and eliminate outliers, without relying on the strong assumption that "the translation should be zero". The specific clustering process includes: a1, treat each translation difference vector as an independent cluster, with the initial number of clusters equal to the number of samples of the translation difference vector; a2, calculates the distance between any two clusters from bottom to top, merges the two closest clusters, and repeats this operation until a preset condition is met. The preset condition can be a specified number of clusters or a distance threshold. a3. The clustering tree structure is truncated to obtain multiple clusters. A predetermined number of clusters based on the amount of data are selected as valid clusters. For example, two translation clusters are ultimately retained, and the cluster with the largest number of samples is considered to correspond to a consistent solution for the translation estimate, while the remaining clusters are regarded as outliers caused by relocation anomalies. This assumption is based on the fact that, in the same scenario, most correct relocation results should have high consistency in the translation space, while outlier relocation results usually present as a small number of discretely distributed translation deviations.

[0048] The embodiments of the present invention employ a translation outlier removal method based on hierarchical clustering, which can effectively reduce translation errors caused by relocation mismatch and local degradation, providing a reliable data foundation for subsequent high-precision coordinate system alignment.

[0049] Step S2042: Take the mean of all translation difference vectors after filtering out outlier poses, and use it as the prior transformation matrix for the translation relationship between the repositioned trajectory and the true trajectory.

[0050] Specifically, this invention calculates a priori transformation matrix as a priori estimate of the translation relationship between the repositioned trajectory and the true trajectory. This priori result is used to initialize subsequent optimization processes, thereby improving the stability and accuracy of the final transformation matrix estimate while suppressing the influence of translation outliers.

[0051] Step S2043: Using a nonlinear optimization method, a least-squares objective function for the matching error between the repositioned pose and the true pose is constructed. Based on the prior transformation matrix, the final transformation matrix is ​​obtained by iteratively solving for the optimal parameters.

[0052] Specifically, this invention employs a nonlinear optimization method (such as one based on Ceres Solver) to minimize the matching error between the repositioned pose and the ground truth pose at the same timestamp. A least-squares objective function is constructed, and the prior transformation matrix obtained after outlier filtering is used as the initial iteration value. Through multiple iterations, the optimal parameters are solved, ultimately yielding the final transformation matrix that accurately represents the mapping relationship between the map coordinate system and the ground truth coordinate system. The matching error encompasses positional deviation (the coordinate difference between the repositioned pose and the ground truth pose) and attitude deviation (the difference in rotation angle). The objective function quantifies and minimizes this error, achieving precise correction of the transformation matrix parameters.

[0053] Nonlinear optimization can effectively fit the complex mapping relationship between the repositioned pose and the ground truth pose. Compared with the initial transformation matrix, it significantly reduces the impact of data noise and algorithm errors, enabling the final transformation matrix to more accurately align the two coordinate systems and provide a high-precision benchmark for subsequent trajectory comparison. Using the prior transformation matrix as the initial value for iteration avoids getting trapped in local optima during the optimization process, shortens the iteration convergence time, and ensures the consistency of optimization results. Even in complex operating scenarios, it can output a stable and reliable transformation matrix. Based on the iterative solution of the least squares objective function, the final transformation matrix can fit the real coordinate system mapping relationship to the greatest extent, making the subsequent comparison results between the ground truth trajectory and the map positioning trajectory more objective and truly reflecting the actual accuracy of the positioning system.

[0054] Step S205: The true trajectory data is transformed to the map coordinate system using the final transformation matrix and compared with the map-based positioning trajectory data to evaluate the trajectory accuracy of the positioning system. For details, please refer to [link to relevant documentation]. Figure 1 Step S105 of the illustrated embodiment will not be described again here.

[0055] This embodiment provides a trajectory accuracy evaluation method based on map positioning, which can be used in the aforementioned self-moving operation equipment. Figure 3 This is a schematic diagram of a second process for trajectory accuracy evaluation based on map positioning according to an embodiment of the present invention, as shown below. Figure 3 As shown, the process includes the following steps: Step S301: Obtain the image sequence of the mobile work device during operation and its time-synchronized true trajectory data. The true trajectory data is obtained based on a preset positioning system. For details, please refer to [link to relevant documentation]. Figure 2 Step S201 of the illustrated embodiment will not be described again here.

[0056] Step S302 involves relocalizing each frame of the image sequence with the pre-created map, calculating the relocalized pose of each frame in the map coordinate system. For details, please refer to [link to relevant documentation]. Figure 2 Steps S2021-S2023 of the illustrated embodiment will not be repeated here.

[0057] Step S303: Based on the repositioned pose and the true pose in the ground truth trajectory data at the same timestamp, calculate the initial transformation matrix representing the transformation relationship between the map coordinate system and the ground truth coordinate system; for details, please refer to [link to relevant documentation]. Figure 2 Step S203 of the illustrated embodiment will not be described again here.

[0058] Step S304 involves sequentially filtering out outliers and optimizing the initial transformation matrix to obtain the final transformation matrix between the optimized map coordinate system and the ground truth coordinate system; for details, please refer to [link to relevant documentation]. Figure 2 Steps S2041-S2043 of the illustrated embodiment will not be described again here.

[0059] Step S305: Use at least one preset quality inspection method to check the correctness of the final transformation matrix. If the quality inspection result of any preset quality inspection method does not meet the corresponding preset requirements, the final transformation matrix is ​​determined to be invalid, and a new transformation matrix is ​​recalculated.

[0060] In one optional embodiment, to verify the validity of the final transformation matrix, at least three verification methods can be used: The relative verification method includes: performing multiple mappings in the same scene, calculating the transformation matrix between different map trajectories, verifying the consistency of the transformation matrix obtained from multiple calculations in translation and rotation, and determining whether the final transformation matrix is ​​valid according to preset consistency requirements.

[0061] The relative verification method provided in this invention uses one set of map trajectories as a reference to calculate the transformation matrix between the other sets of map trajectories and the reference trajectory. Through quantitative analysis, the transformation matrix between different map trajectories is calculated, taking into account the mean error in the translation dimension and the angular error in the rotation dimension. Combined with a preset consistency threshold, it is determined whether the final transformation matrix can achieve stable alignment between the map coordinate system and the true coordinate system. If the error meets the threshold requirement, it is considered valid; otherwise, it is invalid. In one example, the threshold corresponding to the mean translation error can be taken as an empirical value of 0.05 cm; the threshold for the rotation angle error can be taken as an empirical value of 0.1 to 0.3 degrees.

[0062] The point cloud verification method includes: using the final transformation matrix to transform the map point clouds obtained from multiple mappings to the same coordinate system, checking the degree of overlap between point clouds, and judging whether the final transformation matrix is ​​valid according to the preset overlap requirements.

[0063] Specifically, by quantitatively analyzing the degree of spatial overlap between different point clouds and combining it with a preset overlap threshold, it is determined whether the final transformation matrix can achieve precise alignment between the map coordinate system and the ground truth coordinate system. In an example scenario, taking a lawnmower robot operation as an example, the specific implementation process is as follows: b1. For the same lawn in the courtyard, control the lawn mowing robot to complete independent mapping in 3 separate operations to obtain 3 different lawn scene map point clouds (denoted as point cloud A, point cloud B, and point cloud C). Each point cloud contains environmental three-dimensional spatial information such as lawn boundary, tree position, and flower bed outline. b2, Select the map corresponding to point cloud A as the reference map, and apply the final transformation matrix calculated by the method of this invention to point cloud B and point cloud C respectively, transforming them from their original coordinate systems to the northeast-sky coordinate system of the reference map; b3, adopt the point cloud registration error quantification index (such as calculating the mean Euclidean distance between the corresponding 3D points in the converted point clouds B and C and the reference point cloud A), and set the preset overlap requirement as "mean Euclidean distance ≤ 0.1cm"; b4. If the average Euclidean distances between point clouds B and C and point cloud A after transformation are 0.08cm and 0.06cm respectively, both meeting the preset requirements, then the final transformation matrix is ​​deemed valid; if the overlap of a certain group of point clouds does not meet the standard (e.g., the average distance is 0.2cm), then the transformation matrix is ​​deemed invalid and needs to be recalculated.

[0064] Point clouds contain 3D data of the entire work area. The verification process covers global environmental features, rather than local pose data, and can comprehensively check alignment deviations of transformation matrices in different areas, ensuring the rigor of the evaluation. It is not affected by environmental factors such as sparse grass texture and changes in lighting. Even in complex work scenarios, it can still achieve effective verification through spatial distribution comparison of 3D point clouds, enhancing the robustness of the method. The judgment standard is quantified by a preset overlap threshold to avoid subjective judgment errors. It ensures that only transformation matrices that can achieve high-precision point cloud overlap are used for subsequent trajectory evaluation, providing a solid guarantee for the accuracy of the final positioning accuracy evaluation results.

[0065] The pose verification method includes: analyzing the covariance matrix of the pose residuals during the optimization process, evaluating the credibility distribution of the true pose in the map coordinate system, and determining whether the final transformation matrix is ​​valid according to the preset credibility distribution requirements.

[0066] Specifically, in the nonlinear optimization process based on Ceres Solver, pose verification analyzes the covariance matrix of the matching error (residual) between the repositioned pose and the true pose, focusing on the diagonal elements of the matrix to quantitatively evaluate the confidence distribution of the true pose in the map coordinate system. If the confidence distribution meets the preset requirements (such as uniform error distribution and no obvious discrete peaks), the final transformation matrix is ​​determined to be valid; otherwise, it is determined to be invalid.

[0067] By delving into the error distribution characteristics during the optimization process through the covariance matrix, rather than relying solely on surface matching results, the reliability of the true pose in the map coordinate system can be accurately identified, avoiding evaluation distortion caused by local pose deviations. Supplementing the verification dimension with the error distribution dimension complements relative verification and point cloud verification, comprehensively identifying potential problems with the transformation matrix, and ensuring that only transformation matrices with uniform error distribution and acceptable reliability are used for subsequent trajectory evaluation. If the reliability distribution does not meet the requirements, the pose quality can be re-evaluated or parameter settings optimized, providing a clear direction for recalculating the transformation matrix and improving the overall closure and rigor of the evaluation method.

[0068] In an optional embodiment, the relative verification method, point cloud verification method, and pose verification method described above can be executed independently, with a single verification result serving as the basis for judging the validity of the final transformation matrix; alternatively, at least two of the verification methods can be combined and executed according to actual application requirements, and a comprehensive judgment can be made based on multiple verification results, thereby improving the reliability of the final transformation matrix validity determination.

[0069] The above quality inspection methods can be combined in any way to inspect the final transformation matrix. If any quality inspection result fails to meet the corresponding preset requirements, the final transformation matrix is ​​deemed invalid, and a new transformation matrix is ​​recalculated. Preferably, the three quality inspection methods are executed simultaneously and verified collaboratively: rotational consistency of relative verification is used as the core judgment criterion; point cloud verification verifies spatial alignment through point cloud overlap; and pose verification relies on the covariance matrix to analyze pose reliability distribution. These three methods form a comprehensive verification system from different dimensions. During the verification process, if the result of any method fails to meet the corresponding preset requirements, the final transformation matrix is ​​deemed invalid and discarded, and the transformation matrix calculation process must be restarted. Only when the verification results of all three methods meet the preset standards can the transformation matrix be confirmed as valid and used for subsequent trajectory accuracy evaluation.

[0070] Step S306: The true trajectory data is transformed to the map coordinate system using the final transformation matrix and compared with the map-based positioning trajectory data to evaluate the trajectory accuracy of the positioning system. For details, please refer to [link to relevant documentation]. Figure 2 Step S205 of the illustrated embodiment will not be described again here.

[0071] This embodiment also provides a trajectory accuracy evaluation device based on map positioning, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0072] This embodiment provides a trajectory accuracy evaluation device based on map positioning, such as... Figure 4 As shown, it includes: The data acquisition module 401 is used to acquire the image sequence of the mobile operation equipment during operation and the true trajectory data synchronized with its time, wherein the true trajectory data is acquired based on a preset positioning system; The relocation calculation module 402 is used to relocate each frame of the image sequence with a pre-created map and calculate the relocation pose of each frame in the map coordinate system. The initial matrix calculation module 403 is used to calculate an initial transformation matrix representing the transformation relationship between the map coordinate system and the true value coordinate system based on the repositioning pose and the true value pose in the true value trajectory data at the same timestamp. The final transformation matrix acquisition module 404 is used to sequentially perform outlier filtering and optimization processing on the initial transformation matrix to obtain the final transformation matrix between the optimized map coordinate system and the true coordinate system. The accuracy evaluation module 405 is used to transform the true trajectory data to the map coordinate system using the final transformation matrix, and compare it with the map-based positioning trajectory data to evaluate the trajectory accuracy of the positioning system. In some alternative implementations, the relocation calculation module 402 includes: The image feature point extraction unit is used to extract image feature points for each frame of image using at least two different feature extraction algorithms. A matching unit is used to match extracted image feature points with 3D map points in a map using at least two different feature matching algorithms; The repositioning pose acquisition unit is used to calculate multiple candidate repositioning poses for the same frame image based on different combinations of feature point extraction algorithms and feature matching algorithms, and select the optimal pose from the multiple candidate repositioning poses as the repositioning pose of the frame image in the map coordinate system.

[0073] In some alternative implementations, the initial matrix calculation module 403 includes: An anomaly handling unit is used to sequentially perform rotation anomaly filtering and translation anomaly filtering on the initial transformation matrix, wherein: Rotation anomaly filtering includes: converting the rotation portion of each pose pair into Euler angles and filtering out poses in which any component of the Euler angles exceeds a preset angle threshold; Translation anomaly filtering includes: calculating the translation difference vector of the poses that have passed the rotation anomaly filtering and using a clustering algorithm to process the translation difference vector to filter out outlier poses.

[0074] In some optional implementations, the initial matrix calculation module 403 further includes: a priori transformation matrix acquisition unit, used to take the average of all translation difference vectors after filtering out outlier poses, as the priori transformation matrix of the translation relationship between the repositioned trajectory and the true trajectory.

[0075] In some optional implementations, the initial matrix calculation module 403 further includes: an optimization processing unit, used to construct a least-squares objective function for the matching error between the repositioned pose and the true pose using a nonlinear optimization method, and to obtain the final transformation matrix by iteratively solving for the optimal parameters based on the prior transformation matrix.

[0076] In some alternative implementations, such as Figure 5 As shown, it also includes a quality inspection module 406, which is used to inspect the correctness of the final transformation matrix using at least one preset quality inspection method. If the quality inspection result of any preset quality inspection method does not meet the corresponding preset requirements, the final transformation matrix is ​​determined to be invalid, and a new transformation matrix is ​​recalculated.

[0077] In some optional implementations, the quality inspection module 406 includes: a relative verification unit, used to perform multiple mappings in the same scene, calculate the transformation matrix between different map trajectories, verify the consistency of the transformation matrix obtained from multiple calculations in translation and rotation, and determine whether the final transformation matrix is ​​valid according to preset consistency requirements.

[0078] In some optional implementations, the quality inspection module 406 includes: a point cloud verification unit, used to convert the map point clouds obtained from multiple mappings to the same coordinate system using the final transformation matrix, verify the degree of overlap between point clouds, and determine whether the final transformation matrix is ​​valid according to a preset degree of overlap requirement.

[0079] In some optional implementations, the quality inspection module 406 includes: a pose verification unit, used to analyze the covariance matrix of the pose residuals during the optimization process, evaluate the credibility distribution of the true pose in the map coordinate system, and determine whether the final transformation matrix is ​​valid according to the preset credibility distribution requirements.

[0080] The trajectory accuracy evaluation device based on map positioning provided in this embodiment of the invention can execute the trajectory accuracy evaluation method based on map positioning provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method. Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0081] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0082] The following is a detailed reference. Figure 6 This diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 601, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 602 or a program loaded from memory 608 into random access memory (RAM) 603. RAM 603 also stores various programs and data required for the operation of the electronic device. The processor 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0083] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 4 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0084] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a memory 608, or installed from a ROM 602. When the computer program is executed by the processor 601, it performs the functions defined in the map-based positioning trajectory accuracy evaluation method of the embodiments of the present invention.

[0085] Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0086] This invention also provides a self-moving work device, illustrated by a lawnmower, such as... Figure 7 As shown, it includes a mobile component 701, an operation component 702, a vision component 703, a positioning component 704, a radar component 705, and a controller 706, wherein: The moving component 701 provides the equipment with the power and ability to move, and is the actuator that enables the equipment to move autonomously. In one example, it includes a drive motor, a reducer, wheels / tracks, a steering mechanism, and a power source (such as a battery), which can perform actions such as forward movement, backward movement, steering, and speed adjustment, allowing the equipment to be freed from manual traction, efficiently adapt to complex work paths, and improve movement efficiency and position adjustment accuracy.

[0087] The operation component 702 is used to execute specific operation actions according to the operation instructions. It varies depending on the type of operation, such as the roller brush, suction port, and water tank of a sweeping robot; the cutting blade of a lawnmower, etc. It is the core functional component that directly completes the target operation task. It executes specific operation actions according to the operation instructions, such as sweeping garbage and mowing lawns. Its action parameters (such as operation intensity, frequency, and stroke) can be adjusted by the controller to adapt to different operation objects and working conditions.

[0088] The vision component 703 is used to acquire images and depth information of the working environment in real time. It can be a monocular / binocular camera, depth camera, image acquisition card, image processing module, etc. It can realize visual SLAM (simultaneous localization and mapping), obstacle recognition (such as distinguishing pedestrians, equipment, and steps), work object detection (such as identifying garbage and weeds), scene feature extraction and localization assistance, and provide visual data support for path planning, obstacle avoidance decision-making and improvement of work accuracy. In some scenarios, it can also be used for real-time monitoring and feedback of work results.

[0089] The positioning component 704 is used to determine the position, attitude, and trajectory of the equipment in the working environment in real time. It may include, for example, an inertial measurement unit (IMU), a GPS / BeiDou module, a QR code reader, a UWB positioning module, and an odometer, often employing a multi-sensor fusion scheme. Its core task is to determine the precise position, attitude (such as angle and acceleration), and trajectory of the equipment in the working environment in real time, providing basic data for navigation and path planning. By matching with map data, it corrects movement errors, achieving centimeter-level or even millimeter-level positioning, ensuring the equipment moves along a preset route and avoiding positional deviations that could affect operational results.

[0090] The radar component 705 is used for environmental perception and positioning assistance; for example, lidar, and may also include millimeter-wave radar. Lidar scans the surrounding environment by emitting laser beams to obtain three-dimensional data such as the distance, angle, and contour of obstacles, which is used to build high-precision environmental maps, real-time obstacle avoidance (setting deceleration / stop protection zones), and can assist SLAM in achieving positioning and path planning; millimeter-wave radar is suitable for obstacle detection in adverse weather conditions (such as rain, snow, and fog), making up for the limitations of vision components.

[0091] The controller executes the trajectory accuracy evaluation method based on map positioning described above or in any of its corresponding embodiments. It can employ embedded processors (such as ARM, DSP), PLCs, etc., to receive perception data from components such as vision, positioning, and radar, fuse and analyze the data to generate decision commands, and coordinate the movement path and speed of mobile components and the operational actions of work components. Simultaneously, it handles equipment status monitoring, fault diagnosis, and task scheduling. It can also connect to a host computer or cloud system via a communication interface to receive work tasks and provide feedback on execution status, achieving fully autonomous operation control. Furthermore, the controller can perform real-time trajectory accuracy evaluation by calculating and verifying the transformation matrix between the map coordinate system and the true coordinate system, accurately quantifying the positioning system accuracy and map construction quality. Based on the evaluation results, the equipment can promptly adjust its positioning strategy and optimize map data, continuously improving the accuracy of the work path and the completeness of the work area coverage, effectively avoiding problems such as missed or repeated work caused by positioning deviations, and significantly improving operational efficiency and effectiveness.

[0092] The integrated design does not affect the original functions of the equipment. Its compact structure and high compatibility allow it to adapt to various self-propelled mobile equipment such as lawnmowers and sweepers, as well as different operating scenarios including lawns and indoor spaces. Fully automated assessment can be completed without manual intervention, reducing operational complexity. At the same time, the assessment results are objective and reliable, providing accurate data support for equipment maintenance and algorithm iteration. This helps to provide users with a more stable and efficient intelligent operating experience, enhancing the product's market competitiveness.

[0093] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the map-based trajectory accuracy evaluation method shown in the above embodiments is implemented.

[0094] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0095] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A trajectory accuracy evaluation method based on map positioning, characterized in that, include: Image sequences and time-synchronized true trajectory data of the mobile work equipment during operation are acquired, the true trajectory data being acquired based on a preset positioning system; Each frame of the image sequence is relocated to a pre-created map, and the relocation pose of each frame in the map coordinate system is calculated. Based on the repositioning pose and the true pose in the true trajectory data at the same timestamp, calculate the initial transformation matrix representing the transformation relationship between the map coordinate system and the true coordinate system; The initial transformation matrix is ​​sequentially subjected to outlier filtering and optimization processes to obtain the final transformation matrix between the optimized map coordinate system and the ground value coordinate system. The true trajectory data is transformed to a map coordinate system using the final transformation matrix and compared with the positioning trajectory data based on the map to evaluate the trajectory accuracy of the positioning system.

2. The method according to claim 1, characterized in that, The step of relocalizing each frame of the image sequence with a pre-created map and calculating the relocalized pose of each frame in the map coordinate system includes: For each frame of the image, at least two different feature extraction algorithms are used to extract image feature points; At least two different feature matching algorithms are used to match the extracted image feature points with the 3D map points in the map; Based on different combinations of feature point extraction algorithms and feature matching algorithms, multiple candidate relocation poses of the same frame image are calculated respectively. The optimal pose is selected from the multiple candidate relocation poses as the relocation pose of the frame image in the map coordinate system.

3. The method according to claim 1, characterized in that, The outlier filtering process for the initial transformation matrix includes: The initial transformation matrix is ​​subjected to rotation anomaly filtering and translation anomaly filtering in sequence, wherein: The rotation anomaly filtering includes: converting the rotation portion of each pose pair into Euler angles, and filtering out poses in which any component of the Euler angles exceeds a preset angle threshold. The translation anomaly filtering includes: calculating the translation difference vector of the pose that has been filtered out by rotation anomaly, and using a clustering algorithm to process the translation difference vector to filter out outlier poses.

4. The method according to claim 3, characterized in that, Also includes: The mean of all translation difference vectors after filtering out outliers is taken as the prior transformation matrix for the translation relationship between the relocalized trajectory and the true trajectory.

5. The method according to claim 4, characterized in that, The initial transformation matrix is ​​optimized, including: A nonlinear optimization method is used to construct a least-squares objective function for the matching error between the repositioned pose and the true pose. Based on the prior transformation matrix, the final transformation matrix is ​​obtained by iteratively solving for the optimal parameters.

6. The method according to any one of claims 1-5, characterized in that, After obtaining the final transformation matrix between the optimized map coordinate system and the true coordinate system, the method further includes: using at least one preset quality inspection method to check the correctness of the final transformation matrix; if the quality inspection result of any preset quality inspection method does not meet the corresponding preset requirements, the final transformation matrix is ​​determined to be invalid, and a new transformation matrix is ​​recalculated.

7. The method according to claim 6, characterized in that, The preset quality inspection method includes a relative verification method, which includes: performing multiple mappings in the same scene, calculating the transformation matrix between different map trajectories, verifying the consistency of the transformation matrix obtained from multiple calculations in translation and rotation, and determining whether the final transformation matrix is ​​valid according to preset consistency requirements.

8. The method according to claim 6, characterized in that, The preset quality inspection method includes: Point cloud verification includes: using the final transformation matrix to transform the map point clouds obtained from multiple mappings to the same coordinate system, checking the degree of overlap between point clouds, and determining whether the final transformation matrix is ​​valid according to the preset degree of overlap requirements.

9. The method according to claim 6, characterized in that, The preset quality inspection method includes: Pose verification includes: analyzing the covariance matrix of the pose residuals during the optimization process, evaluating the credibility distribution of the true pose in the map coordinate system, and determining whether the final transformation matrix is ​​valid according to the preset credibility distribution requirements.

10. A trajectory accuracy evaluation device based on map positioning, characterized in that, The device includes: The data acquisition module is used to acquire image sequences from the mobile work equipment during operation and real-value trajectory data synchronized with its time, wherein the real-value trajectory data is acquired based on a preset positioning system; The relocation calculation module is used to relocate each frame of the image sequence with a pre-created map and calculate the relocation pose of each frame in the map coordinate system. The initial matrix calculation module is used to calculate an initial transformation matrix representing the transformation relationship between the map coordinate system and the true coordinate system based on the repositioning pose and the true pose in the true trajectory data at the same timestamp. The final transformation matrix acquisition module is used to sequentially perform outlier filtering and optimization processing on the initial transformation matrix to obtain the final transformation matrix between the optimized map coordinate system and the true coordinate system. The accuracy evaluation module is used to transform the true trajectory data to a map coordinate system using the final transformation matrix, and compare it with the positioning trajectory data based on the map to evaluate the trajectory accuracy of the positioning system.

11. A self-moving operating device, characterized in that, include: The components include a mobile component, an operational component, a vision component, a positioning component, a radar component, and a controller, among which: The mobile component is used to provide the device with mobility and motion capability; The task component is used to execute specific task actions according to task instructions; The vision component is used to acquire images and depth information of the working environment in real time; The positioning component is used to determine the position, attitude and movement trajectory of the device in the working environment in real time; The radar component is used for environmental perception and positioning assistance; The controller performs the trajectory accuracy evaluation method based on map positioning as described in any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the trajectory accuracy evaluation method based on map positioning as described in any one of claims 1 to 9.

Images