Protection level compensation method, device, positioning terminal and readable storage medium

Abstract

The present invention provides a compensation method, device, positioning terminal and readable storage medium for protection level, which relate to the field of positioning. The positioning terminal first obtains the current positioning result at the current epoch, and then uses the RAIM algorithm to calculate the initial protection level data of the current positioning result; then uses a pre-trained decision tree model to process the current solution feature set of the current positioning result to obtain deviation data; finally, uses the deviation data to compensate the initial protection level data to determine the target protection level data of the current positioning result. Since the deviation data can represent the deviation between the preliminary positioning error and the actual positioning error of the current positioning result, in the case of distortion of the initial protection level data, the deviation data can be used to compensate the initial protection level data to obtain the final target protection level data, thereby correctly evaluating the positioning accuracy of the terminal.

Description

Technical Field

[0001] The present invention relates to the field of positioning, and in particular to a protection level compensation method, device, positioning terminal and readable storage medium. Background Art

[0002] RTK (Real Time Kinematics) positioning is a real-time dynamic differential positioning technology. This technology offers advantages such as high precision, real-time performance, and dynamic operation. It has wide applications in land surveying, measurement, geographic information, and autonomous driving. Providing higher accuracy than standard GNSS positioning, it enables precise control and navigation. The basic principle of RTK positioning is to utilize the distance differences between two or more Global Navigation Satellite System (GNSS) receivers to eliminate atmospheric delay and other errors, thereby achieving high-precision real-time dynamic positioning. RTK positioning requires at least two receivers: a base station and a rover. The base station receives satellite signals and measures their arrival time and phase. This data is then transmitted in real time to a rover via a wireless link. The rover measures the relative distance to the base station and, combined with the satellite signals received by the base station, calculates the position to achieve high-precision real-time dynamic positioning.

[0003] In the automotive sector, navigation is typically achieved through combined positioning using a Global Navigation Satellite System-Inertial Navigation System (GNSS / INS). This system must also comply with automotive functional safety (FuSa) requirements. For example, ISO 26262 classifies systems or their components into Automotive Safety Integrity Levels (ASILs) from A to D, based on the degree of safety risk. This ASIL focuses on limiting risks associated with potential hardware and software failures within a device. Another standard, ISO / PAS 21448, addresses safety of the intended function (SOTIF), focusing on limiting objective risks even when the device is performing its intended function perfectly. In the case of GNSS, these risks can be caused by positioning errors, which can arise from satellite failures, satellite signal transmission failures, and atmospheric interference, such as tropospheric and ionospheric storms, ionospheric scintillation, and, most commonly, multipath distortion. At the same time, since RTK technology uses a virtual reference station (VRS), although the VRS system is generally reliable, there are also some factors that may cause VRS system failures, such as base station failures, network problems, data transmission delays, multipath effects, GNSS signal interruptions, etc. For this reason, the VRS system usually adopts corresponding fault tolerance and recovery mechanisms to deal with system failures caused by these factors, such as automatic switching to a backup base station, data caching and recovery, etc. In addition, the reliability and performance of the VRS system are also related to the specific implementation and deployment. Therefore, when using the VRS system, it is necessary to identify faults, because any measurement failure of any satellite may have a huge impact on the final GNSS positioning results.

[0004] To address this issue, the concept of Protect Level (PL) has been introduced for positioning terminals. PL represents a dynamic upper limit for positioning error. This technology was first applied in the aviation field. However, in the autonomous driving field, the accuracy requirements for positioning terminals are much higher than in aviation, with accuracies reaching tens of centimeters or even centimeters, and an alarm threshold of 3 meters. RTK positioning typically uses Kalman filtering for state estimation, requiring data preprocessing before and after filtering. Traditional quality control and detection algorithms, such as ambiguity cycle slip detection and receiver autonomous integrity monitoring, can largely ensure the probability of positioning anomalies.

[0005] In the positioning terminal, real-time integrity monitoring results are quantified by outputting alarm states and calculating the Protection Level. The Protection Level represents the accuracy of the RTK positioning algorithm. However, if erroneous observations are missed or if the observation noise does not conform to a normal distribution or empirical evidence, the Protection Level output from the real-time integrity monitoring results will be distorted, leading to an incorrect assessment of the terminal's positioning accuracy and potentially posing a safety risk to the vehicle. Summary of the Invention

[0006] The object of the present invention is to provide a protection level compensation method, device, positioning terminal and readable storage medium to improve the problems existing in the prior art.

[0007] The embodiments of the present invention can be implemented as follows:

[0008] In a first aspect, the present invention provides a protection level compensation method, applied to a positioning terminal, the method comprising:

[0009] Get the current positioning result of the current epoch;

[0010] Calculating initial protection level data of the current positioning result using a RAIM algorithm; the initial protection level data represents a preliminary positioning error of the current positioning result;

[0011] Processing a current solution feature set of the current positioning result using a pre-trained decision tree model to obtain deviation data; the deviation data represents a deviation between the preliminary positioning error and a true positioning error of the current positioning result;

[0012] The initial protection level data is compensated using the deviation data to determine the target protection level data of the current positioning result.

[0013] In an optional embodiment, the deviation data includes a predicted horizontal deviation and a predicted vertical deviation; and the decision tree model is trained in the following manner:

[0014] Acquire a training data set, wherein the training data set includes a plurality of training samples, and the training samples include a historical solution feature set and historical deviation data of historical positioning results;

[0015] Based on a decision tree generation algorithm, the decision tree model is generated using the training data set. The decision tree model includes a first decision tree and a second decision tree. The first decision tree and the second decision tree are used to predict the predicted horizontal deviation and the predicted vertical deviation, respectively.

[0016] In an optional embodiment, the step of obtaining a training data set includes:

[0017] Obtaining an original solution feature set, initial protection level data, and true positioning error data corresponding to a number of the historical positioning results; wherein the original solution feature set includes original data of a plurality of solution feature factors; the initial protection level data of the historical positioning results includes an initial horizontal protection level and an initial vertical protection level; and the true positioning error data includes a horizontal positioning error value and a vertical positioning error value;

[0018] For any of the historical positioning results, normalize the original data of all solution feature factors in the original solution feature set of the historical positioning result to obtain a historical solution feature set corresponding to the historical positioning result; the historical solution feature set includes normalized feature values ​​of each of the plurality of solution feature factors;

[0019] Calculating the difference between the horizontal positioning error value corresponding to the historical positioning result and the initial horizontal protection level to obtain the original horizontal deviation of the historical positioning result;

[0020] Calculating the difference between the vertical positioning error value corresponding to the historical positioning result and the initial vertical protection level to obtain the original vertical deviation of the historical positioning result; the historical deviation data includes the original horizontal deviation and the original vertical deviation;

[0021] Each of the historical positioning results is traversed to obtain the training data set.

[0022] In an optional embodiment, the step of generating the decision tree model using the training data set based on a decision tree generation algorithm includes:

[0023] Based on the training data set, a first training set and a second training set are determined; the first training set includes a historical solution feature set and an original horizontal deviation of a plurality of the historical positioning results; the second training set includes a historical solution feature set and an original vertical deviation of a plurality of the historical positioning results;

[0024] Based on the decision tree generation algorithm, the first decision tree and the second decision tree are generated using the first training set and the second training set respectively.

[0025] In an optional embodiment, the current solution feature set includes normalized feature values ​​of each of the plurality of solution feature factors; the decision tree model includes a first decision tree and a second decision tree;

[0026] The first decision tree and the second decision tree are both binary trees, and both include a root node, multiple branch nodes, and multiple leaf nodes; the branch node includes a characteristic condition related to the solution characteristic factor, the leaf node of the first decision tree includes a predicted horizontal deviation, and the leaf node of the second decision tree includes a predicted vertical deviation;

[0027] The step of using a pre-trained decision tree model to process the current solution feature set of the current positioning result to obtain deviation data includes:

[0028] Inputting all the normalized feature values ​​of the current solution feature set into the first decision tree and the second decision tree;

[0029] Based on all the normalized eigenvalues, performing matching searches on the first decision tree and the second decision tree respectively to find a first target leaf node and a second target leaf node;

[0030] Obtain the predicted horizontal deviation of the first target leaf node and the predicted vertical deviation of the second target leaf node to obtain deviation data corresponding to the current positioning result.

[0031] In an optional embodiment, the initial protection level data includes an initial horizontal protection level and an initial vertical protection level; the deviation data includes a predicted horizontal deviation and a predicted vertical deviation; and the step of compensating the initial protection level data using the deviation data to determine the target protection level data of the current positioning result includes:

[0032] Adding the initial horizontal protection level to the horizontal deviation to obtain a target horizontal protection level;

[0033] The initial vertical protection level is added to the vertical deviation to obtain a target vertical protection level; wherein the target protection level data includes the target horizontal protection level and the target vertical protection level.

[0034] In a second aspect, the present invention provides a protection level compensation device, which is applied to a positioning terminal, wherein:

[0035] Positioning module, used to obtain the current positioning result of the current epoch;

[0036] a calculation module, configured to calculate initial protection level data of the current positioning result using a RAIM algorithm; wherein the initial protection level data represents a preliminary positioning error of the current positioning result;

[0037] Compensation modules for:

[0038] Processing a current solution feature set of the current positioning result using a pre-trained decision tree model to obtain deviation data; the deviation data represents a deviation between the preliminary positioning error and a true positioning error of the current positioning result;

[0039] The initial protection level data is compensated using the deviation data to determine the target protection level data of the current positioning result.

[0040] In an optional embodiment, the deviation data includes a predicted horizontal deviation and a predicted vertical deviation; and the apparatus further includes a training module for:

[0041] Acquire a training data set, wherein the training data set includes a plurality of training samples, and the training samples include a historical solution feature set and historical deviation data of historical positioning results;

[0042] Based on a decision tree generation algorithm, the decision tree model is generated using the training data set. The decision tree model includes a first decision tree and a second decision tree. The first decision tree and the second decision tree are used to predict the predicted horizontal deviation and the predicted vertical deviation, respectively.

[0043] In a third aspect, the present invention provides a positioning terminal, comprising: a memory and a processor, wherein the memory stores a software program, and when the positioning terminal is running, the processor executes the software program to implement the protection level compensation method as described in the first aspect above.

[0044] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the protection level compensation method described in the first aspect.

[0045] Compared to the prior art, the embodiments of the present invention provide a protection level compensation method, apparatus, positioning terminal, and readable storage medium. The positioning terminal first obtains the current positioning result at the current epoch, then uses the RAIM algorithm to calculate the initial protection level data for the current positioning result. The terminal then uses a pre-trained decision tree model to process the current solution feature set of the current positioning result to obtain deviation data. Finally, the deviation data is used to compensate the initial protection level data to determine the target protection level data for the current positioning result. Because the deviation data can represent the deviation between the preliminary positioning error and the actual positioning error of the current positioning result, if the initial protection level data is distorted, the deviation data can be used to compensate the initial protection level data to obtain the final target protection level data, thereby accurately evaluating the terminal's positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0047] Figure 1 A flowchart of a protection level compensation method provided by an embodiment of the present invention.

[0048] Figure 2 A schematic diagram of the tree structure of a subtree where a branch node K is located in a first decision tree provided in an embodiment of the present invention.

[0049] Figure 3 A schematic diagram of the overall flow of a protection level compensation method provided by an embodiment of the present invention.

[0050] Figure 4 A schematic structural diagram of a protection level compensation device provided by an embodiment of the present invention.

[0051] Figure 5 A schematic structural diagram of a positioning terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0054] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0055] In addition, the terms "first", "second", etc., if used, are merely used to distinguish and describe, and should not be understood as indicating or implying relative importance.

[0056] It should be noted that, in the absence of conflict, the features in the embodiments of the present invention may be combined with each other.

[0057] RTK positioning engines typically use Kalman filtering for state estimation, requiring data preprocessing before and after filtering. Traditional quality control and detection algorithms, such as ambiguity cycle slip detection and receiver autonomous integrity monitoring (RAIM), can largely guarantee the probability of positioning anomalies. Furthermore, based on the mathematical relationship between innovation verification, residual / innovation, and state estimation error during the filtering process, observation noise is set based on empirical values ​​for unknown observation failures, and the corresponding protection level is solved, which can be called the original protection level.

[0058] The original protection level can represent the accuracy index of the RTK positioning algorithm. However, when incorrect observation measurements are missed, the observation noise does not conform to the normal distribution or experience, etc., the protection level output by the real-time monitoring results of the integrity will be distorted, and the positioning accuracy of the terminal will be incorrectly evaluated, resulting in vehicle safety risks.

[0059] In the prior art, the patent "CN115062097A - A Positioning Integrity Monitoring Method, Terminal and Computer Storage Medium" uses coordinate verification under different positioning solution states and a subset classification principle set based on experience to further verify the positioning solution state.

[0060] Although the use of external verification of different positioning solution states and the positioning solution verification rate of subsets can improve the efficiency and accuracy of positioning integrity monitoring and quantify integrity indicators, when GNSS is in a complex environment, the probability of observation anomalies caused by obstruction, multipath, NLOS, and cycle slips increases significantly. In urban environments, it may also be affected by deception interference. In this case, the number of observations received by the positioning terminal will be significantly reduced, and the error in the observation amount will be large, so there will be only one or two positioning solution states. In this case, the positioning solution results of external verification and subset classification will have a large deviation, and the estimated protection level will fail or be wrong.

[0061] Based on the discovery of the above technical problems, the inventors have proposed the following technical solutions after creative work to solve or improve the above problems. It should be noted that the defects existing in the solutions in the above prior art are the results obtained by the inventors after practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed in the embodiments of this application below for the above problems should all be the contributions made by the inventors to this application in the process of invention and creation, and should not be understood as technical contents known to those skilled in the art.

[0062] The protection level compensation method provided in the embodiments of the present invention can be applied to a positioning terminal, which may be, but is not limited to, an unmanned vehicle, a drone, a robot, etc. The protection level compensation method provided by the present invention is described in detail below through examples and in conjunction with the accompanying drawings.

[0063] Please refer to Figure 1 , Figure 1 A flowchart of a protection level compensation method provided by an embodiment of the present invention is provided. The method includes the following steps S201 to S204:

[0064] S201: Obtain the current positioning result of the current epoch.

[0065] In this embodiment, the positioning terminal can be equipped with a receiver as a mobile receiver. Based on the observation data received by the mobile receiver and the data transmitted from the base station at a fixed position, the positioning terminal can perform RTK positioning based on the positioning engine to obtain the current positioning result of the current epoch.

[0066] S202: Calculate initial protection level data of the current positioning result using the RAIM algorithm.

[0067] Receiver Autonomous Integrity Monitoring (RAIM) monitors the integrity of positioning results based on redundant receiver observations. Its purpose is to detect satellite failures during navigation and ensure navigation accuracy. Integrity quantifies the confidence level in the position information provided by the positioning system. Its direct output is called the Protection Level (PL), which is generated along with the positioning results and provides an upper limit on the positioning error at the current moment.

[0068] In this embodiment, RAIM may be used to calculate initial protection level data of the current positioning result, where the initial protection level data represents a preliminary positioning error of the current positioning result.

[0069] S203: Using a pre-trained decision tree model to process the current solution feature set of the current positioning result, to obtain deviation data.

[0070] As you can understand, the RTK positioning process consists of data preprocessing, Kalman float filter update, wide-term ambiguity fixation, and ambiguity fixation. The corresponding current positioning results can include coordinate results for floating-point solutions, wide-term fixed solutions, and narrow-lane fixed solutions. The current solution feature set can characterize the characteristics of different solution results.

[0071] In this embodiment, the deviation data represents the deviation between the preliminary positioning error and the actual positioning error of the current positioning result.

[0072] S204: Compensate the initial protection level data using the deviation data to determine the target protection level data of the current positioning result.

[0073] In this embodiment, after compensation, the target protection level data may be close to the actual positioning error of the current positioning result, and the target protection level data may be used to evaluate the positioning accuracy of the positioning terminal.

[0074] When the positioning terminal performs real-time integrity monitoring, it is expected that the output initial protection level data can be close to the actual positioning error of the positioning terminal. However, in cases where erroneous observation quantities are missed, the observation noise does not conform to the normal distribution or experience, the initial protection level data is easily distorted and has a large deviation from the actual positioning error. By using a pre-trained decision tree model, the deviation can be predicted based on the current solution feature set of the current positioning result, thereby compensating the initial protection level data and correctly evaluating the positioning accuracy of the positioning terminal.

[0075] In the protection level compensation method provided by an embodiment of the present invention, the positioning terminal first obtains the current positioning result at the current epoch, then uses the RAIM algorithm to calculate the initial protection level data of the current positioning result; then uses a pre-trained decision tree model to process the current solution feature set of the current positioning result to obtain deviation data; finally, the deviation data is used to compensate the initial protection level data to determine the target protection level data of the current positioning result. Because the deviation data can represent the deviation between the preliminary positioning error and the actual positioning error of the current positioning result, if the initial protection level data is distorted, the deviation data can be used to compensate the initial protection level data to obtain the final target protection level data, thereby accurately evaluating the positioning accuracy of the terminal.

[0076] In an optional implementation, the initial protection level data includes a vertical protection level (VPL) and a horizontal protection level (HPL). Correspondingly, the deviation data also includes a predicted horizontal deviation and a predicted vertical deviation. The predicted horizontal deviation can represent the deviation between the VPL in the initial protection level data and the actual horizontal positioning error; the predicted vertical deviation can represent the deviation between the HPL in the initial protection level data and the actual vertical positioning error.

[0077] The initial protection level data is calculated using a RAIM algorithm and is self-consistent. The least squares method is used in the calculation process. The specific calculation logic is based on existing technology and will not be elaborated on here.

[0078] The following describes the training process of the decision tree model, which includes the following steps S101-S102.

[0079] S101. Obtain a training dataset.

[0080] In this embodiment, the training data set may include several training samples, and the training samples may include a historical solution feature set and historical deviation data of a historical positioning result.

[0081] Optionally, the sub-steps of S101 may include S1011 to S1015:

[0082] S1011. Obtain an original solution feature set, initial protection level data, and actual positioning error data corresponding to several historical positioning results.

[0083] In this embodiment, the initial protection level data of the historical positioning result may include the initial horizontal protection level and the initial vertical protection level, the actual positioning error data may include the horizontal positioning error value and the vertical positioning error value, and the original solution feature set may include the original data of multiple solution feature factors.

[0084] It can be understood that the solution characteristic factor is a characteristic of the solution results such as the floating point solution, the wide term fixed solution, and the narrow lane fixed solution. The solution characteristic factors may include but are not limited to the 25 types listed in the following table:

[0085]

[0086] It should be noted that the solution characteristic factors listed in the above table are only examples and are not limiting here.

[0087] S1012: For any historical positioning result, perform normalization conversion on the original data of all solution feature factors in the original solution feature set of the historical positioning result to obtain the historical solution feature set corresponding to the historical positioning result.

[0088] In this embodiment, the historical solution feature set may include normalized feature values ​​of each of a plurality of solution feature factors.

[0089] It is understandable that the dimensions of the raw data from different feature factor solutions may differ. Therefore, normalization (scaling, feature encoding, feature selection, or dimensionality reduction) can be performed to unify the dimensions. This can be achieved through various transformation methods or algorithms, such as standardization, normalization, logarithmic transformation, and principal component analysis (PCA).

[0090] S1013. Calculate the difference between the horizontal positioning error value corresponding to the historical positioning result and the initial horizontal protection level to obtain the original horizontal deviation of the historical positioning result.

[0091] S1014. Calculate the difference between the vertical positioning error value corresponding to the historical positioning result and the initial vertical protection level to obtain the original vertical deviation of the historical positioning result.

[0092] In this embodiment, the historical deviation data includes the original horizontal deviation and the original vertical deviation.

[0093] S1015. Traverse each historical positioning result to obtain a training data set.

[0094] Therefore, for each historical positioning result, by executing the above steps S1012 to S1014, the original solution feature set and historical deviation data of each historical positioning result can be obtained, that is, the training data set is obtained.

[0095] S102: Based on a decision tree generation algorithm, a decision tree model is generated using a training data set.

[0096] In this embodiment, the decision tree model may include a first decision tree and a second decision tree. The first decision tree and the second decision tree may be used to predict the predicted horizontal deviation and the predicted vertical deviation, respectively.

[0097] Optionally, the decision tree generation algorithm may be a CART algorithm, and the sub-steps of S102 may include S1021 and S1022:

[0098] S1021. Determine a first training set and a second training set based on the training data set.

[0099] In this embodiment, the first training set may include a historical solution feature set and an original horizontal deviation of several historical positioning results; the second training set may include a historical solution feature set and an original vertical deviation of several historical positioning results.

[0100] S1022: Based on a decision tree generation algorithm, generate a first decision tree and a second decision tree using the first training set and the second training set respectively.

[0101] In this embodiment, the first and second decision trees are part of a decision tree regression model (regression tree for short). The goal of a decision tree regression model is to predict a continuous numerical output. Therefore, both the first and second decision trees are binary trees, each consisting of a root node, multiple branch nodes, and multiple leaf nodes.

[0102] In the decision tree regression model, a tree structure is constructed based on the characteristics of the input data, in which each branch node represents a characteristic condition related to the characteristic factor to be solved, and each leaf node represents an output value, that is: the leaf node of the first decision tree includes the predicted horizontal deviation, and the leaf node of the second decision tree includes the predicted vertical deviation.

[0103] Taking the first decision tree as an example, during the training process, appropriate features and thresholds can be automatically selected to divide the first training set into different subsets so that the data in each subset has similar output values. The entire first training set is input for each training session. For the historical solution feature set of one of the historical positioning results, the decision tree structure is traversed according to the values ​​of each solution feature factor, and finally a leaf node is reached. The predicted value in the leaf node is returned as the output of the model. All outputs are compared with the original horizontal deviation of each historical positioning result in the first training set input, and the root mean square error is calculated to evaluate the model accuracy. If the expected model accuracy is not met, the feature values ​​extracted by the algorithm are adjusted to find the feature value related to the actual deviation, and the next training is repeated until the model accuracy meets the standard.

[0104] The prediction process of a decision tree regression model is to traverse the tree from the root node downward, compare the input data based on the feature conditions, and finally reach the leaf node and return the output value of the leaf node as the prediction result. The decision tree regression model has the following advantages: it is easy to understand and interpret, can generate a visual tree structure, can handle nonlinear relationships and interaction effects, and is robust to outliers and missing values.

[0105] Optionally, the decision tree generation algorithm can be: bagging, boosting, or stacking. Taking the bagging algorithm as an example, its principle is to sample the training set to obtain n sub-training sets, then use each of the n sub-training sets to train n decision trees. During prediction, the data is input into each of the n decision trees to obtain n prediction results. Finally, these n prediction results are combined to determine the final output.

[0106] In an optional implementation, for the current positioning result of the positioning terminal at the current epoch, the corresponding solution feature set may include the normalized feature values ​​of multiple solution feature factors during the solution process. After positioning at the current epoch, the original data of each solution feature factor may be obtained and normalized to obtain the current solution feature set corresponding to the current positioning result. Correspondingly, the sub-steps of the above step S203 may include S2031 to S2033:

[0107] S2031, inputting all normalized eigenvalues ​​of the current solved feature set into the first decision tree and the second decision tree;

[0108] S2032: Based on all normalized eigenvalues, perform matching searches on the first decision tree and the second decision tree respectively to find a first target leaf node and a second target leaf node;

[0109] S2033. Obtain the predicted horizontal deviation of the first target leaf node and the predicted vertical deviation of the second target leaf node to obtain deviation data corresponding to the current positioning result.

[0110] In this embodiment, the current solution feature set is input into the first decision tree and the second decision tree respectively, and the first target leaf node and the second target leaf node can be found by traversing the tree from the root node downward. The deviation data corresponding to the current positioning result can include: the predicted horizontal deviation of the first target leaf node and the predicted vertical deviation of the second target leaf node.

[0111] Assuming there are 10 solution characteristic factors, the corresponding current solution characteristic set is:

[0112]

[0113] Taking the first decision tree as an example, the partial tree structure of the first decision tree can be as follows Figure 2 As shown, Figure 2 The tree structure of the subtree where the branch node K is located in the first decision tree, and the characteristic conditions of the branch nodes K, K1, and K2 are respectively: 、 、 The predicted level deviations (values) of leaf nodes K3, K4, K5, and K6 are 6.282, 14.346, 0.817, and 9.959, respectively.

[0114] When explanation is needed, Figure 2 The examples shown are only examples. The number and size of the characteristic factors, the characteristic conditions of each branch node in the decision tree, and the predicted values ​​in the leaf nodes shall be based on the actual application and are not limited here.

[0115] In an optional implementation, the initial protection level data of the current positioning result may include an initial horizontal protection level and an initial vertical protection level. Correspondingly, the sub-steps of the above step S204 may include S2041-S2042:

[0116] S2041. Add the initial horizontal protection level and the horizontal deviation to obtain a target horizontal protection level;

[0117] S2042: Add the initial vertical protection level and the vertical deviation to obtain a target vertical protection level.

[0118] In this embodiment, the target protection level data includes a target horizontal protection level and a target vertical protection level.

[0119] The overall calculation of the target protection level data in the compensation method of the protection level introduced above is as follows: Figure 3As shown, a decision tree model is used to predict the deviation between the initial protection level data and the actual positioning error of the current positioning result. This deviation data is then used to compensate for the initial protection level data to obtain the final target protection level data for the current positioning result. This method can estimate the protection level in real time with extremely low computing power, optimize the initial protection level, and improve the accuracy of protection level estimation, thus meeting the stringent integrity requirements of autonomous driving.

[0120] Optionally, the target protection level data may be obtained in another manner, which involves the following steps S1 to S3:

[0121] S1. Collect a set of historical solution features (including normalized eigenvalues ​​of multiple solution feature factors) corresponding to several historical positioning results and real positioning error data (including horizontal positioning error values ​​and vertical positioning error values) to construct a training data set;

[0122] S2. Use the training data set to train a first decision tree and a second decision tree, where the first decision tree and the second decision tree are used to predict a predicted horizontal positioning error value and a predicted vertical positioning error value, respectively.

[0123] S3. Input the current solution feature set of the current positioning result into the first decision tree and the second decision tree, and traverse the tree from the root node downward to find the first target leaf node and the second target leaf node. The target protection level data corresponding to the current positioning result may include: the predicted horizontal positioning error value of the first target leaf node and the predicted vertical positioning error value of the second target leaf node.

[0124] It can be understood that compared with the existing method of calculating the protection level based on the RAIM algorithm, this method can accurately estimate the target protection level data in real time. Even in cases where erroneous observations are missed, the observation noise does not conform to the normal distribution or experience, the target protection level data predicted by this method will not be distorted, thereby meeting the strict integrity requirements of autonomous driving.

[0125] Optionally, there is another implementation method to determine the target protection level data. The following is a brief description of the implementation:

[0126] 1. Calculate the initial protection level data of the current epoch;

[0127] 2. Based on the above steps S3-S3, the predicted protection level data predicted by the decision tree model (including the predicted horizontal positioning error value and the predicted vertical positioning error value) are obtained;

[0128] 3. Compare the initial protection level data with the predicted protection level data, and take the maximum value of the two as the final target protection level data.

[0129] It should be noted that the execution order of the steps in the above method embodiment is not limited to that shown in the drawings, and the execution order of the steps shall be based on actual application conditions.

[0130] In order to execute the corresponding steps in the above method embodiment and various possible implementations, an implementation of a protection level compensation device is provided below.

[0131] See Figure 4 , Figure 4 The protection level compensation device 200 is applied to a positioning terminal and includes a positioning module 210 , a calculation module 220 , and a compensation module 230 .

[0132] Positioning module 210, used to obtain the current positioning result of the current epoch;

[0133] The calculation module 220 is configured to calculate initial protection level data of the current positioning result using a RAIM algorithm; the initial protection level data represents a preliminary positioning error of the current positioning result;

[0134] The compensation module 230 is used to: use a pre-trained decision tree model to process the current solution feature set of the current positioning result to obtain deviation data; the deviation data represents the deviation between the preliminary positioning error and the actual positioning error of the current positioning result; use the deviation data to compensate the initial protection level data to determine the target protection level data of the current positioning result.

[0135] Optionally, the deviation data may include a predicted horizontal deviation and a predicted vertical deviation; the protection level compensation device 200 may further include a training module 240, which is used to: obtain a training data set, the training data set including a number of training samples, the training samples including a historical solution feature set and historical deviation data of historical positioning results; based on a decision tree generation algorithm, the decision tree model is generated using the training data set, the decision tree model including a first decision tree and a second decision tree; the first decision tree and the second decision tree are used to predict the predicted horizontal deviation and the predicted vertical deviation, respectively.

[0136] Those skilled in the art will readily appreciate that compensation module 230 can be used to implement steps S203, S204, and their sub-steps, and training module 240 can be used to implement steps S101-S102, and their sub-steps. For ease and brevity, the specific operating procedures of protection level compensation device 200 described above can be referenced to the corresponding procedures in the aforementioned method embodiments and will not be further elaborated upon here.

[0137] See Figure 5 , Figure 5 The present invention provides a schematic structural diagram of a positioning terminal 300. The positioning terminal 300 includes a processor 310, a memory 320, and a bus 330. The processor 310 is connected to the memory 320 via the bus 330.

[0138] The memory 320 may be used to store software programs, such as the software program corresponding to the protection level compensation device 200 provided in the embodiment of the present invention. The processor 310 executes the software program stored in the memory 320 to perform various functional applications and data processing to implement the protection level compensation method provided in the embodiment of the present invention.

[0139] Among them, the memory 320 can be, but is not limited to, random access memory (RAM), read only memory (ROM), flash memory (Flash), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0140] The processor 310 can be an integrated circuit chip with signal processing capabilities. The processor 310 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0141] I understand. Figure 5 The structure shown is for illustration only. The positioning terminal 300 may also include Figure 5 More or fewer components than shown, or with Figure 5 Different configurations shown. Figure 5Each component shown in the figure can be implemented by hardware, software or a combination thereof.

[0142] Embodiments of the present invention further provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the protection level compensation method disclosed in the above embodiments. The computer-readable storage medium may be, but is not limited to, a USB flash drive, a mobile hard drive, ROM, RAM, PROM, EPROM, EEPROM, a FLASH disk, or an optical disk, among other media capable of storing program code.

[0143] In summary, the embodiments of the present invention provide a protection level compensation method, device, positioning terminal, and readable storage medium. The positioning terminal first obtains the current positioning result at the current epoch, then uses the RAIM algorithm to calculate the initial protection level data of the current positioning result; then uses a pre-trained decision tree model to process the current solution feature set of the current positioning result to obtain deviation data; finally, the deviation data is used to compensate the initial protection level data to determine the target protection level data of the current positioning result. Since the deviation data can represent the deviation between the preliminary positioning error and the actual positioning error of the current positioning result, if the initial protection level data is distorted, the deviation data can be used to compensate the initial protection level data to obtain the final target protection level data, thereby correctly evaluating the positioning accuracy of the terminal.

[0144] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A protection level compensation method, characterized in that: Applied to a positioning terminal, the method includes: Get the current positioning result of the current epoch; Calculating initial protection level data of the current positioning result using a RAIM algorithm; the initial protection level data represents a preliminary positioning error of the current positioning result; Processing a current solution feature set of the current positioning result using a pre-trained decision tree model to obtain deviation data; the deviation data represents a deviation between the preliminary positioning error and a true positioning error of the current positioning result; Compensating the initial protection level data using the deviation data to determine target protection level data of the current positioning result; The deviation data includes the predicted horizontal deviation and the predicted vertical deviation; the decision tree model is trained in the following way: Obtaining an original solution feature set, initial protection level data, and true positioning error data corresponding to a number of historical positioning results; wherein the original solution feature set includes original data of a plurality of solution feature factors; the initial protection level data of the historical positioning results includes an initial horizontal protection level and an initial vertical protection level; and the true positioning error data includes a horizontal positioning error value and a vertical positioning error value; For any of the historical positioning results, normalize the original data of all solution feature factors in the original solution feature set of the historical positioning result to obtain a historical solution feature set corresponding to the historical positioning result; the historical solution feature set includes normalized feature values ​​of each of the plurality of solution feature factors; Calculating the difference between the horizontal positioning error value corresponding to the historical positioning result and the initial horizontal protection level to obtain the original horizontal deviation of the historical positioning result; Calculating the difference between the vertical positioning error value corresponding to the historical positioning result and the initial vertical protection level to obtain the original vertical deviation of the historical positioning result; the historical deviation data includes the original horizontal deviation and the original vertical deviation; Traversing each of the historical positioning results to obtain a training data set, wherein the training data set includes a plurality of training samples, and the training samples include a historical solution feature set and historical deviation data of the historical positioning results; Based on a decision tree generation algorithm, the decision tree model is generated using the training data set. The decision tree model includes a first decision tree and a second decision tree. The first decision tree and the second decision tree are used to predict the predicted horizontal deviation and the predicted vertical deviation, respectively.

2. The method according to claim 1, characterized in that The step of generating the decision tree model using the training data set based on the decision tree generation algorithm includes: Based on the training data set, a first training set and a second training set are determined; the first training set includes a historical solution feature set and an original horizontal deviation of a plurality of the historical positioning results; the second training set includes a historical solution feature set and an original vertical deviation of a plurality of the historical positioning results; Based on the decision tree generation algorithm, the first decision tree and the second decision tree are generated using the first training set and the second training set respectively.

3. The method according to claim 1, characterized in that The current solution feature set includes normalized feature values ​​of each of the plurality of solution feature factors; the decision tree model includes a first decision tree and a second decision tree; The first decision tree and the second decision tree are both binary trees, and both include a root node, multiple branch nodes, and multiple leaf nodes; the branch node includes a characteristic condition related to the solution characteristic factor, the leaf node of the first decision tree includes a predicted horizontal deviation, and the leaf node of the second decision tree includes a predicted vertical deviation; The step of using a pre-trained decision tree model to process the current solution feature set of the current positioning result to obtain deviation data includes: Inputting all the normalized feature values ​​of the current solution feature set into the first decision tree and the second decision tree; Based on all the normalized eigenvalues, performing matching searches on the first decision tree and the second decision tree respectively to find a first target leaf node and a second target leaf node; Obtain the predicted horizontal deviation of the first target leaf node and the predicted vertical deviation of the second target leaf node to obtain deviation data corresponding to the current positioning result.

4. The method according to claim 1, wherein The initial protection level data includes an initial horizontal protection level and an initial vertical protection level; and the step of compensating the initial protection level data using the deviation data to determine the target protection level data of the current positioning result includes: Adding the initial horizontal protection level to the horizontal deviation to obtain a target horizontal protection level; The initial vertical protection level is added to the vertical deviation to obtain a target vertical protection level; wherein the target protection level data includes the target horizontal protection level and the target vertical protection level.

5. A protection level compensation device, characterized in that, Applied to a positioning terminal, the device: Positioning module, used to obtain the current positioning result of the current epoch; A calculation module, configured to calculate initial protection level data of the current positioning result using a RAIM algorithm; The initial protection level data represents a preliminary positioning error of the current positioning result; Compensation modules for: Processing a current solution feature set of the current positioning result using a pre-trained decision tree model to obtain deviation data; the deviation data represents a deviation between the preliminary positioning error and a true positioning error of the current positioning result; Compensating the initial protection level data using the deviation data to determine target protection level data of the current positioning result; The deviation data includes a predicted horizontal deviation and a predicted vertical deviation; and the device further includes a training module for: Obtaining an original solution feature set, initial protection level data, and true positioning error data corresponding to a number of historical positioning results; wherein the original solution feature set includes original data of a plurality of solution feature factors; the initial protection level data of the historical positioning results includes an initial horizontal protection level and an initial vertical protection level; and the true positioning error data includes a horizontal positioning error value and a vertical positioning error value; For any of the historical positioning results, normalize the original data of all solution feature factors in the original solution feature set of the historical positioning result to obtain a historical solution feature set corresponding to the historical positioning result; the historical solution feature set includes normalized feature values ​​of each of the plurality of solution feature factors; Calculating the difference between the horizontal positioning error value corresponding to the historical positioning result and the initial horizontal protection level to obtain the original horizontal deviation of the historical positioning result; Calculating the difference between the vertical positioning error value corresponding to the historical positioning result and the initial vertical protection level to obtain the original vertical deviation of the historical positioning result; the historical deviation data includes the original horizontal deviation and the original vertical deviation; Traversing each of the historical positioning results to obtain a training data set, wherein the training data set includes a plurality of training samples, and the training samples include a historical solution feature set and historical deviation data of the historical positioning results; Based on a decision tree generation algorithm, the decision tree model is generated using the training data set. The decision tree model includes a first decision tree and a second decision tree. The first decision tree and the second decision tree are used to predict the predicted horizontal deviation and the predicted vertical deviation, respectively.

6. A positioning terminal, characterized in that: include: A memory and a processor, wherein the memory stores a software program, and when the positioning terminal is running, the processor executes the software program to implement the protection level compensation method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the protection level compensation method according to any one of claims 1 to 4 is implemented.

Images