Unmanned aerial vehicle intelligent RTK inertial navigation system and method based on visual processing

Abstract

The invention discloses an unmanned aerial vehicle intelligent RTK inertial navigation system and method based on visual processing, and relates to the technical field of unmanned aerial vehicle positioning. The unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing comprises the following steps: acquiring satellite RTK signal data, attitude time sequence data and motion environment image time sequence data of a set unmanned aerial vehicle; based on satellite RTK signal data, analyzing an initial three-dimensional position coordinate and a signal confidence feature value; based on set attitude time sequence data of the unmanned aerial vehicle, analyzing an attitude comprehensive stability characteristic value; based on a pre-trained visual environment perception model, in combination with motion environment image time sequence data, environment positioning interference characteristic values are analyzed, initial three-dimensional position coordinates are corrected through the characteristic values, positioning three-dimensional position coordinates of a set unmanned aerial vehicle are obtained, and therefore self-adaptive cooperative positioning of multi-source data is achieved. Therefore, the stability of the positioning result is maintained, and the positioning precision of the unmanned aerial vehicle in a complex environment is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle positioning, in particular to an intelligent RTK inertial navigation system and method for unmanned aerial vehicles based on visual processing. BACKGROUND

[0002] With the wide application of unmanned aerial vehicles in surveying and mapping, inspection, emergency rescue and other fields, the flight positioning accuracy and attitude stability of unmanned aerial vehicles have become key factors affecting the quality of task execution. In the prior art, unmanned aerial vehicles usually obtain high-precision position coordinates through a global satellite positioning system and realize short-time continuous navigation in combination with an inertial navigation system. However, in complex environments such as densely built-up areas, forest areas and canyons, satellite signals are easily affected by shielding, reflection and multipath effect interference, resulting in a significant decrease in RTK positioning accuracy. In addition, the inertial navigation system has a cumulative drift problem during long-time operation, which will further amplify the position error if there is no external correction.

[0003] At the same time, with the development of computer vision and artificial intelligence technology, auxiliary positioning methods based on visual information have gradually attracted attention. However, existing methods are mostly limited to independent visual odometry or image matching methods, and lack deep fusion with satellite positioning signals and inertial navigation data.

[0004] Among them, the limitations of the prior art at least include the following problems: the prior art is difficult to realize adaptive correction of unmanned aerial vehicle positioning in complex environments, so that the unmanned aerial vehicle is difficult to dynamically correct in time by comprehensively considering the changes of various data sources in complex environments, which is easy to cause positioning deviation and attitude drift. For example, when the RTK signal strength fluctuates greatly, the prior art fails to combine the inertial navigation attitude change trend and the visual image stability to perform confidence correction, resulting in that the positioning result is greatly affected by signal interference. In addition, the unmanned aerial vehicle has a positioning response delay when the attitude changes dramatically, and it is difficult to extract the evolution law of environmental interference from the image time sequence, especially in scenes with more light reflection or dynamic obstacles, which will exacerbate the positioning error, and thus it is difficult to realize high-precision three-dimensional positioning. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides an intelligent RTK inertial navigation system and method for unmanned aerial vehicles based on visual processing, which solves the problem that the prior art is difficult to fuse multi-source data for dynamic correction, resulting in insufficient positioning accuracy of unmanned aerial vehicles in complex environments.

[0006] In order to achieve the above object, the present application is realized by the following technical scheme: the unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing comprises the following steps: acquiring satellite RTK signal data, attitude time sequence data and motion environment image time sequence data of the set unmanned aerial vehicle; based on the satellite RTK signal data of the set unmanned aerial vehicle, analyzing the initial three-dimensional position coordinates and signal confidence characteristic values of the set unmanned aerial vehicle; based on the attitude time sequence data of the set unmanned aerial vehicle, analyzing the attitude comprehensive stability characteristic values of the set unmanned aerial vehicle; based on the pre-trained visual environment perception model and combined with the motion environment image time sequence data, analyzing the environment positioning interference characteristic values of the set unmanned aerial vehicle; based on the signal confidence characteristic values, the attitude comprehensive stability characteristic values and the environment positioning interference characteristic values of the set unmanned aerial vehicle, correcting the initial three-dimensional position coordinates to obtain the positioning three-dimensional position coordinates of the set unmanned aerial vehicle.

[0007] Further, the satellite RTK signal data comprises RTK signal data of a plurality of satellites, and each RTK signal data is specifically signal intensity value and pseudo-random code of each signal time point; the specific steps of analyzing the initial three-dimensional position coordinates of the set unmanned aerial vehicle are as follows: based on the RTK signal data of each satellite of the set unmanned aerial vehicle, extracting the pseudo-range observation value and carrier phase observation value of the corresponding signal time point; based on the pseudo-range observation value and carrier phase observation value of each signal time point of each satellite of the set unmanned aerial vehicle, analyzing the initial three-dimensional position coordinates of the set unmanned aerial vehicle.

[0008] Further, the specific steps of analyzing the signal confidence characteristic values of the set unmanned aerial vehicle are as follows: reading the signal intensity value of each signal time point of each satellite of the set unmanned aerial vehicle, analyzing the signal positioning interference characteristic set of the corresponding satellite, including energy deflection characteristic value, energy entropy change characteristic value and main peak energy shift characteristic value; based on the signal positioning interference characteristic set of each satellite of the set unmanned aerial vehicle, analyzing the signal confidence characteristic values of the set unmanned aerial vehicle.

[0009] Further, the attitude time sequence data comprises attitude angle set, three-axis angular velocity value and three-axis linear acceleration value of each attitude time point; the specific steps of analyzing the attitude comprehensive stability characteristic values of the set unmanned aerial vehicle are as follows: based on the attitude time sequence data of the set unmanned aerial vehicle, analyzing the attitude stability response characteristic set of the set unmanned aerial vehicle, including attitude disturbance entropy characteristic value, attitude coordination fluctuation characteristic value and attitude response synchronization characteristic value; fusing the attitude stability response characteristic set of the set unmanned aerial vehicle to obtain the attitude comprehensive stability characteristic values of the set unmanned aerial vehicle.

[0010] Further, the specific steps of analyzing the attitude stability response feature set of the set unmanned aerial vehicle are as follows: time sequence processing is performed on the attitude angle set of each attitude time point of the set unmanned aerial vehicle to obtain attitude disturbance entropy feature values and attitude coordination fluctuation feature values of the set unmanned aerial vehicle; and correlation processing is performed on the three-axis angular velocity values and three-axis linear acceleration values of each attitude time point of the set unmanned aerial vehicle to obtain attitude response synchronization feature values of the set unmanned aerial vehicle.

[0011] Further, the motion environment image time sequence data includes a plurality of frames of motion environment image data, and each frame of motion environment image data specifically includes a pixel value, a two-dimensional coordinate and corresponding depth information value of each pixel point in the motion environment image, and the visual environment perception model includes an input layer, a feature decomposition layer, an LSTM layer and an output layer.

[0012] Further, the specific steps of analyzing the environment positioning interference feature value of the set unmanned aerial vehicle are as follows: the motion environment image time sequence data of the set unmanned aerial vehicle is input into the pre-trained visual environment perception model to analyze the environment interference mapping feature set of the set unmanned aerial vehicle, including propagation obstruction evolution feature values, scattering distribution evolution feature values and environment interference coordination feature values; and based on the environment interference mapping feature set of the set unmanned aerial vehicle, the environment positioning interference feature value of the set unmanned aerial vehicle is analyzed.

[0013] Further, the specific steps of analyzing the environment interference mapping feature set of the set unmanned aerial vehicle are as follows: in the input layer of the visual environment perception model, each frame of motion environment image data of the set unmanned aerial vehicle is received and preprocessed; in the feature decomposition layer of the visual environment perception model, feature extraction processing is performed on the preprocessed each frame of motion environment image data of the set unmanned aerial vehicle to obtain an environment interference time sequence feature vector in the corresponding frame of motion environment image thereof; in the LSTM layer of the visual environment perception model, time sequence analysis is performed on the environment interference time sequence feature vector in each frame of motion environment image of the set unmanned aerial vehicle to obtain an environment evolution feature vector of the set unmanned aerial vehicle; and in the output layer of the visual environment perception model, based on the environment evolution feature vector of the set unmanned aerial vehicle, the environment interference mapping feature set of the set unmanned aerial vehicle is output.

[0014] Further, the specific steps of obtaining the positioning three-dimensional position coordinates of the set unmanned aerial vehicle are as follows: based on the signal confidence feature value, the attitude comprehensive stability feature value and the environment positioning interference feature value of the set unmanned aerial vehicle, the positioning signal feature value of the set unmanned aerial vehicle is analyzed; and based on the positioning signal feature value of the set unmanned aerial vehicle and in combination with the initial three-dimensional position coordinates, the positioning three-dimensional position coordinates of the set unmanned aerial vehicle are analyzed.

[0015] The unmanned aerial vehicle intelligent RTK inertial navigation system based on visual processing comprises a data acquisition module, an initial positioning module, an attitude analysis module, an environment positioning interference analysis module and a comprehensive positioning analysis module.

[0016] The application has the following beneficial effects: The unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing extracts the initial three-dimensional position coordinates and the signal confidence characteristic value by deeply analyzing the satellite RTK signal data, and on this basis, deeply fuses the characteristic values extracted in combination with the attitude time series data and the motion environment image time series data and the signal confidence characteristic value to generate the positioning signal characteristic value, thereby accurately correcting the initial three-dimensional position coordinates, realizing adaptive collaborative positioning of multi-source data, maintaining the stability of the positioning result when the signal fluctuates or the environment changes, significantly reducing the positioning drift, and effectively improving the positioning accuracy of the unmanned aerial vehicle in a complex environment.

[0017] (2) The unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing analyzes the attitude time series data, introduces the attitude stability response feature set to generate the attitude comprehensive stability characteristic value, thereby being capable of identifying the smooth coordination of the attitude change in real time, thereby being capable of sensing the attitude change trend in time when the unmanned aerial vehicle rapidly turns, climbs or is disturbed by the attitude fluctuation, inhibiting the transmission effect of the inertial navigation cumulative error, effectively improving the attitude control accuracy of the unmanned aerial vehicle in a high-speed maneuvering or complex airflow environment, and making the attitude comprehensive stability characteristic value capable of assisting in correcting the initial three-dimensional position coordinates, thereby making the unmanned aerial vehicle still maintain accurate three-dimensional positioning ability under the disturbed condition.

[0018] (3) The unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing, through constructing a visual environment perception model, performs multi-layer feature decomposition and time sequence evolution analysis on motion environment image time sequence data, realizes environment interference recognition and mapping, the model extracts environment interference time sequence feature vectors of each image in the feature decomposition layer, realizes time sequence correlation modeling of image features in the LSTM layer, and then outputs environment interference mapping feature sets, so as to reflect the influence of light, reflection and dynamic obstacle changes in the visual scene on the positioning process, thereby generating environment positioning interference feature values, and based on this, dynamically correcting the initial three-dimensional position coordinates, so that the unmanned aerial vehicle has environment adaptive ability under visual perception, and the positioning anti-interference performance of the unmanned aerial vehicle in a complex visual environment is significantly improved.

[0019] (4) The unmanned aerial vehicle intelligent RTK inertial navigation system based on visual processing realizes hierarchical analysis and interlayer cooperation of multi-source data through modular structure design, the data acquisition module ensures that the RTK signal, attitude data and motion environment image data participate in positioning analysis in a unified cycle, the initial positioning module, the attitude analysis module and the environment positioning interference analysis module independently extract features such as signal confidence, attitude stability and environment interference, and the comprehensive positioning analysis module realizes accurate correction of the initial three-dimensional position coordinates based on the fusion analysis of the above features, thereby significantly improving the operation efficiency of the unmanned aerial vehicle positioning, and further improving the task execution reliability of the unmanned aerial vehicle in complex working conditions.

[0020] Of course, implementing any product of the present application does not necessarily need to achieve all the advantages described above at the same time. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 The flow chart of the unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing of the present application.

[0022] Figure 2 The data schematic diagram of the satellite sequence signal positioning interference feature set of the unmanned aerial vehicle in the unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing of the present application.

[0023] Figure 3 The specific step flow chart of analyzing and setting the environment interference mapping feature set of the unmanned aerial vehicle in the unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing of the present application.

[0024] Figure 4 The block diagram of the unmanned aerial vehicle intelligent RTK inertial navigation system based on visual processing of the present application. DETAILED DESCRIPTION

[0025] Please refer to Figure 1The embodiment of the application provides a technical scheme: a visual processing-based unmanned aerial vehicle intelligent RTK inertial navigation method, comprising the following steps: acquiring satellite RTK signal data, attitude time sequence data and motion environment image time sequence data of a set unmanned aerial vehicle within a set period (the period is the least common multiple of the sampling frequencies of the RTK signal, the attitude time sequence data and the motion environment image time sequence data); analyzing initial three-dimensional position coordinates and signal confidence characteristic values of the set unmanned aerial vehicle based on the satellite RTK signal data of the set unmanned aerial vehicle; analyzing attitude comprehensive stability characteristic values of the set unmanned aerial vehicle based on the attitude time sequence data of the set unmanned aerial vehicle; analyzing environment positioning interference characteristic values of the set unmanned aerial vehicle based on a pre-trained visual environment perception model and in combination with the motion environment image time sequence data; and correcting the initial three-dimensional position coordinates based on the signal confidence characteristic values, the attitude comprehensive stability characteristic values and the environment positioning interference characteristic values of the set unmanned aerial vehicle to obtain positioning three-dimensional position coordinates of the set unmanned aerial vehicle.

[0026] The specific steps for obtaining the positioning three-dimensional position coordinates of the set unmanned aerial vehicle are as follows: analyzing positioning confidence characteristic values of the set unmanned aerial vehicle based on the signal confidence characteristic values, the attitude comprehensive stability characteristic values and the environment positioning interference characteristic values of the set unmanned aerial vehicle, and the specific formula is as follows: ; wherein, the positioning confidence characteristic values of the set unmanned aerial vehicle, the signal confidence characteristic values of the set unmanned aerial vehicle, a signal confidence adjustment coefficient stored in a database, the attitude comprehensive stability characteristic values of the set unmanned aerial vehicle, an attitude stability adjustment coefficient stored in the database, the environment positioning interference characteristic values of the set unmanned aerial vehicle, an environment positioning interference adjustment coefficient stored in the database, , an interaction adjustment coefficient stored in the database.

[0027] It should be explained that the signal confidence adjustment coefficient stored in the database, the attitude stability adjustment coefficient stored in the database and the environment positioning interference adjustment coefficient The acquisition steps are as follows: acquiring signal confidence characteristic values, attitude comprehensive stability characteristic values and environment positioning interference characteristic values of a plurality of historical periods, extracting signal confidence characteristic mean values, attitude comprehensive stability characteristic mean values and environment positioning interference characteristic mean values, performing summation processing to obtain a confidence sum value, and performing ratio processing on the signal confidence characteristic mean values, the attitude comprehensive stability characteristic mean values and the environment positioning interference characteristic mean values and the confidence sum value respectively, and taking the corresponding results as the signal confidence adjustment coefficient , the attitude stability adjustment coefficient , environmental positioning interference adjustment coefficient .

[0028] interaction adjustment coefficient stored in the database The acquisition steps are as follows: the signal confidence feature value, the attitude comprehensive stability feature value, and the environmental positioning interference feature value in each historical period are normalized, the Pearson correlation coefficient between the signal confidence feature value and the attitude compensation response feature value is extracted to obtain the synergy strength, the environmental positioning interference feature value and the negative correlation coefficient (taking the absolute value) of the synergy strength are extracted, and ratio processing is performed, that is, the synergy strength / (1+ negative correlation coefficient), and the result is taken as the interaction adjustment coefficient .

[0029] Based on the positioning confidence feature value of the set unmanned aerial vehicle, and combined with the initial three-dimensional position coordinates, the positioning three-dimensional position coordinates of the set unmanned aerial vehicle are analyzed, which specifically are: the positioning confidence feature value of the set unmanned aerial vehicle is normalized and mapped to 0-1, and compared with the preset positioning confidence feature threshold value; if the positioning confidence feature value of the set unmanned aerial vehicle is higher than the preset positioning confidence feature threshold value, the initial three-dimensional position coordinates of the set unmanned aerial vehicle are marked as the positioning three-dimensional position coordinates; If the positioning confidence feature value of the set unmanned aerial vehicle is lower than or equal to the preset positioning confidence feature threshold value, the roll angle value, the pitch angle value, and the yaw angle value of each attitude time point of the set unmanned aerial vehicle are read, the difference values of the attitude angles of adjacent attitude time points are processed in turn, and the sliding average values thereof are taken to obtain the roll angle change trend value, the pitch angle change trend value, and the yaw angle change trend value, and the three-dimensional position coordinates of the set unmanned aerial vehicle at each signal time point are read, the difference values of the three-dimensional position coordinates of adjacent signal time points are processed in turn, and the sliding average processing is performed to obtain the change amounts of the initial three-dimensional position coordinates on the X, Y, and Z axes, and the signs of the roll angle change trend value, the pitch angle change trend value, and the yaw angle change trend value are analyzed whether they are the same as the signs of the change amounts on the X, Y, and Z axes; If the signs of the roll angle change trend value and the change amount on the X axis are the same, it indicates that the unmanned aerial vehicle is consistent with the change trend of the initial three-dimensional position coordinates in the lateral attitude change direction, and it is determined that the RTK signal response in this direction is synchronous, and reverse correction needs to be performed to offset the excessive drift bias, that is, the coordinate component in the lateral X axis direction is reduced by a certain proportion based on the initial three-dimensional position coordinates, and the X axis coordinate value in the initial three-dimensional position coordinates is (- (1-positioning confidence feature value))×(the absolute value of the difference between the X axis coordinate values in the three-dimensional position coordinates of the last signal time point and the previous signal time point in this period). If the roll angle trend value and the change amount on the X-axis are opposite in sign, it indicates that the lateral attitude change direction of the UAV is opposite to the change direction of the initial three-dimensional position coordinates, and it is determined that the RTK signal in this direction has a lag or drift phenomenon, and same direction compensation correction needs to be performed to offset the delay error, that is, the X-axis coordinate value in the initial three-dimensional position coordinates + (1-position signal characteristic value) x (the absolute value of the difference between the X-axis coordinate value in the three-dimensional position coordinates at the last signal time point and the previous signal time point in this period); Similarly, if the pitch angle trend value and the change amount on the Y-axis are the same in sign, reverse correction is performed, otherwise same direction compensation correction is performed; if the yaw angle trend value and the change amount on the Z-axis are the same in sign, reverse correction is performed, otherwise same direction compensation correction is performed, so as to obtain the positioning three-dimensional position coordinates of the set UAV.

[0030] Specifically, the satellite RTK signal data includes RTK signal data of several satellites (obtained by the RTK receiving device equipped on the UAV), and each RTK signal data is specifically the signal strength value and the pseudo-random code at each signal time point. The specific steps of analyzing the initial three-dimensional position coordinates of the set UAV are as follows: Based on the RTK signal data of each satellite of the set UAV, the pseudo-range observation value and the carrier phase observation value at the corresponding signal time point are extracted, which are specifically: the RTK signal data of each satellite of the set UAV is subjected to frequency conversion processing, and the time offset correlation analysis is performed on the pseudo-random code at each signal time point after frequency conversion processing and the reference pseudo-random code sequence of the corresponding satellite stored in the data (including the reference pseudo-random code of several signal time points, which is the standard code sequence generated and stored locally by the RTK receiving device equipped on the set UAV according to the pseudo-random code identification number of the satellite stored in the database), that is, the time offset amount of the reference pseudo-random code sequence is adjusted in fixed steps on the time axis, and the correlation value of the pseudo-random code and the reference pseudo-random code is analyzed at each time offset amount (the correlation value can be extracted by using the cross-correlation matching algorithm), when the correlation value reaches the maximum, the corresponding time offset amount is taken as the propagation delay of the satellite signal, and the light speed value stored in the database is multiplied with the propagation delay to extract the pseudo-range observation value at the corresponding signal time point; at the same time, the carrier phase error estimation and correction are performed on each satellite signal data after frequency conversion processing by using the phase-locked loop (PLL), so as to track the change of the satellite carrier phase in the continuous time sequence in real time, so as to extract the carrier phase observation value corresponding to each signal time point; Based on the pseudorange observation value, carrier phase observation value of each signal time point of each satellite of the set unmanned aerial vehicle, the initial three-dimensional position coordinates of the set unmanned aerial vehicle are analyzed, which is specifically: based on the pseudorange observation value and carrier phase observation value of each signal time point of each satellite of the set unmanned aerial vehicle, combined with the three-dimensional spatial coordinates of each satellite stored in the database, using GNSS positioning algorithm (such as constructing the observation equation of pseudorange and carrier phase, and using least square method, extended Kalman filter or double difference RTK solving method, etc.), the three-dimensional position coordinates (including the corresponding coordinate values of X, Y, Z axes) of the set unmanned aerial vehicle at each signal time point are extracted, and weighted average processing is performed to extract the initial three-dimensional position coordinates of the set unmanned aerial vehicle.

[0031] The specific steps of analyzing the signal confidence characteristic value of the set unmanned aerial vehicle are as follows: reading the signal strength value of each signal time point of each satellite of the set unmanned aerial vehicle, analyzing the signal positioning interference characteristic set of the corresponding satellite, including energy deflection characteristic value, energy entropy change characteristic value, main peak energy shift characteristic value, which is specifically: for the signal strength value of each signal time point of each satellite, taking every 3 signal time points as a sliding window and one signal time point as a sliding step, the deviation (i.e. difference taking average) between the signal strength value of the middle signal time point and the average signal strength of the adjacent front and rear signal time points is extracted in each sliding window, and the sum is processed to obtain the deviation sum value. The deviation of each sliding window is processed by ratio with the deviation sum value, and the result is weighted to extract the energy deflection characteristic value of each satellite, which is used to represent the disturbance of signal strength. When it is larger, it means that the signal has a mutation or asymmetric fluctuation in the current period, and the propagation path has disturbance or reflection phenomenon. The signal strength value of each satellite of the set unmanned aerial vehicle is divided into several short time subintervals, and in each short time subinterval, the signal strength value is divided into several strength intervals according to the value range of signal strength, the probability value of the occurrence of signal strength value in each strength interval is counted, and the signal strength information entropy of the corresponding short time subinterval is extracted, and the information entropy difference (taking absolute value) between adjacent subintervals is extracted in turn, and the average processing is performed to extract the energy entropy change characteristic value of each satellite, which is used to represent the change strength of energy distribution disorder degree in the current period, that is, to reflect the dynamic fluctuation of signal strength distribution in short time scale. When it is larger, it means that the disorder degree of signal energy distribution changes sharply in the period, which represents that the signal is significantly affected by external disturbance, multipath reflection or shielding; Based on the signal strength value of each satellite of the unmanned aerial vehicle at each signal time point, the maximum signal strength value of each satellite in the set period and its corresponding signal time point are extracted, the signal time point is recorded as the main peak time point, and the time difference (taking the absolute value) between the main peak time point and the center signal time point of the set period (i.e. the signal time point corresponding to the median of the first signal time point and the last signal time point of the set period) is calculated, and the time difference is processed by ratio with the total length of the set period (i.e. the total number of signal time points), to extract the main peak energy offset characteristic value of each satellite, which is used to represent the time domain offset degree of the main energy distribution of the satellite signal in the current period, i.e. to reflect the consistency of the signal propagation path and phase, when it is smaller, it means that the main energy peak of the signal is close to the center of the period, the signal propagation path is stable, and the phase consistency is high. Based on the signal positioning interference feature set of each satellite of the set unmanned aerial vehicle, the signal confidence feature value of the set unmanned aerial vehicle is analyzed, which is specifically: based on the energy deflection characteristic value, energy entropy change characteristic value and main peak energy offset characteristic value of each satellite of the set unmanned aerial vehicle, the signal stability characteristic value of the corresponding satellite is analyzed, and the mean value is taken to obtain the signal confidence feature value of the set unmanned aerial vehicle. The specific formula for calculating the signal stability characteristic value of a certain satellite of the set unmanned aerial vehicle is as follows: ; wherein, is the signal stability characteristic value of a certain satellite of the set unmanned aerial vehicle, is the energy deflection characteristic value of a certain satellite of the set unmanned aerial vehicle, is the energy deflection adjustment coefficient stored in the database, is the energy entropy change characteristic value of a certain satellite of the set unmanned aerial vehicle, is the energy entropy change adjustment coefficient stored in the database, is the main peak energy offset characteristic value of a certain satellite of the set unmanned aerial vehicle, is the main peak energy offset adjustment coefficient stored in the database, is the cooperative adjustment coefficient stored in the database, .

[0032] It should be noted that the energy deflection adjustment coefficient , the energy entropy change adjustment coefficient , the main peak energy offset adjustment coefficient stored in the database are obtained as follows: The energy deflection characteristic value, energy entropy change characteristic value and main peak energy offset characteristic value of each satellite are read, the energy deflection characteristic mean value, energy entropy change characteristic mean value and main peak energy offset characteristic mean value are extracted respectively, and the sum is processed to obtain the signal sum value, and the energy deflection characteristic mean value, energy entropy change characteristic mean value and main peak energy offset characteristic mean value are respectively processed by ratio with the signal sum value to obtain the energy deflection adjustment coefficient , energy entropy change adjustment coefficient , main peak energy offset adjustment coefficient .

[0033] Synergistic adjustment coefficient stored in the database The acquisition steps are as follows: reading the energy deflection characteristic value, the energy entropy change characteristic value, and the main peak energy offset characteristic value of each satellite, using the Pearson correlation coefficient for any two characteristic values, analyzing the corresponding correlation value (taking the absolute value), and taking the mean value as the synergistic adjustment coefficient .

[0034] The specific implementation example of calculating the signal stability characteristic value of a satellite of the set unmanned aerial vehicle is as follows, and the existing data includes: the energy deflection characteristic value, the energy entropy change characteristic value, and the main peak energy offset characteristic value of the four satellites (randomly selected) of the set unmanned aerial vehicle, as shown in Table 1 and Figure 2 . Table 1 Signal positioning interference characteristic set data example of satellite sequence of the set unmanned aerial vehicle Energy deflection characteristic value Energy entropy change characteristic value Main peak energy shift characteristic value Satellite 1 0.344 0.421 0.213 Satellite 2 0.189 0.327 0.087 Satellite 3 0.246 0.358 0.124 Satellite 4 0.302 0.394 0.149 Energy deflection adjustment coefficient stored in the database about: 0.342; Energy entropy change adjustment coefficient about: 0.372; Main peak energy offset adjustment coefficient about: 0.286; Synergistic adjustment coefficient stored in the database about: 0.784; Substitute the data in Table 1 and the above coefficients into the specific formula for calculating the signal stability characteristic value of a satellite of the set unmanned aerial vehicle to obtain: The signal stability characteristic value of the first satellite of the set unmanned aerial vehicle = exp (-0.784 x (0.342 x 0.344 + 0.372 x 0.421 + 0.286 x 0.213)) ≈ 0.769; The signal stability characteristic value of the second satellite of the set unmanned aerial vehicle = exp (-0.784 x (0.342 x 0.189 + 0.372 x 0.327 + 0.286 x 0.087)) ≈ 0.847; The signal stability characteristic value of the third satellite of the set unmanned aerial vehicle = exp (-0.784 x (0.342 x 0.246 + 0.372 x 0.358 + 0.286 x 0.124)) ≈ 0.820; Set the signal stability eigenvalue of the fourth satellite of the unmanned aerial vehicle = exp (-0.784 x (0.342 x 0.302 + 0.372 x 0.394 + 0.286 x 0.149)) ≈ 0.795.

[0035] In the embodiment, the initial three-dimensional coordinates are calculated by deeply analyzing the RTK signal, and the signal confidence analysis is introduced to the RTK signal, the fluctuation law and energy distribution change of the signal strength in the time sequence and the main peak time domain offset are comprehensively considered, and then the subtle changes of the signal stability can be sensitively captured, and on this basis, the energy deflection adjustment coefficient and the energy entropy change adjustment coefficient and the main peak energy offset adjustment coefficient and the cooperative adjustment coefficient are set, so that the characteristic results of each satellite have self-adaptive balance ability, the influence of abnormal signal on the overall positioning accuracy is avoided, the positioning accuracy of the unmanned aerial vehicle in complex scenes is enhanced, and the unmanned aerial vehicle can still maintain high confidence navigation stability when flying dynamically and changing attitude quickly.

[0036] Specifically, the attitude time sequence data includes an attitude angle set at each attitude time point, a three-axis angular velocity value, and a three-axis linear acceleration value. The specific steps of analyzing the attitude comprehensive stability eigenvalue of the set unmanned aerial vehicle are as follows: based on the attitude time sequence data of the set unmanned aerial vehicle, analyzing the attitude stability response eigenvalue set of the set unmanned aerial vehicle, including an attitude disturbance entropy eigenvalue, an attitude coordination fluctuation eigenvalue, and an attitude response synchronization eigenvalue; performing fusion processing on the attitude stability response eigenvalue set of the set unmanned aerial vehicle to obtain the attitude comprehensive stability eigenvalue of the set unmanned aerial vehicle, that is, performing weighted processing on the attitude disturbance entropy eigenvalue, the attitude coordination fluctuation eigenvalue, and the attitude response synchronization eigenvalue of the set unmanned aerial vehicle, and performing reciprocal processing on the result, that is, taking the reciprocal of the weighted processing result + 1 to obtain the attitude comprehensive stability eigenvalue of the set unmanned aerial vehicle, which is used to represent the attitude dynamic stability degree of the set unmanned aerial vehicle in the set period. The greater the value is, the more stable the attitude of the unmanned aerial vehicle is and the smaller the compensation demand is.

[0037] The attitude angle set includes a roll angle value, a pitch angle value, and a yaw angle value, and can be obtained by an IMU installed on the unmanned aerial vehicle body.

[0038] The three-axis angular velocity value is the angular velocity of the set unmanned aerial vehicle in X, Y, and Z three axes, which is obtained by a gyroscope in real time.

[0039] The three-axis linear acceleration value is the linear acceleration of the set unmanned aerial vehicle in X, Y, and Z three-axis directions, which is obtained by an accelerometer in real time.

[0040] The specific steps of analyzing the attitude stability response eigenvalue set of the set unmanned aerial vehicle are as follows: performing time sequence processing on the attitude angle set of each attitude time point of the set unmanned aerial vehicle to obtain the attitude disturbance entropy eigenvalue and the attitude coordination fluctuation eigenvalue, which are specifically: The roll angle value, the pitch angle value and the yaw angle value of each attitude time point are respectively subjected to differential processing (taking absolute value), to obtain a plurality of groups of roll angle change values, pitch angle change values and yaw angle change values of adjacent attitude time points, and the sample entropy processing is performed, the window length (for example, 3 groups of adjacent attitude time points) is used to perform sliding segmentation processing on the roll angle change value, the pitch angle change value and the yaw angle change value of each group of adjacent attitude time points, to obtain a plurality of data segments, for any one data segment, the maximum absolute difference of the corresponding data points is calculated by comparing with the remaining data segments in the same sequence, and compared with the preset threshold value, if the maximum absolute difference is less than the preset threshold value, it is determined as a similar data segment pair, the number of similar data segment pairs is counted, the similar probability is obtained, the window length is expanded by 1, the above process is repeated to obtain the expanded similar probability, and the sample entropy value is analyzed, to obtain the roll angle change entropy value, the pitch angle change entropy value and the yaw angle change entropy value, and the weighted average processing is performed, to obtain the attitude disturbance entropy characteristic value, which is used to represent the complexity of the attitude change of the set unmanned aerial vehicle in the set period, the larger the value is, the more disordered the attitude change is, and the worse the attitude keeping stability is; Based on the roll angle value, the pitch angle value and the yaw angle value of each attitude time point, the variances of each attitude angle and the covariance values between any two attitude angles are extracted, to form a covariance matrix, and the trace value (the three elements of the main diagonal line of the covariance matrix, that is, the algebraic sum of the variance values of each attitude angle, which can be obtained by sequentially accumulating the diagonal line elements of the matrix) and the determinant value of the covariance matrix are extracted, and the weighted processing is performed, to extract the attitude coordination fluctuation characteristic value, which is used to represent the fluctuation degree of the overall coordinated stability of the attitude of the set unmanned aerial vehicle in the set period; The three-axis angular velocity value and the three-axis linear acceleration value of each attitude time point of the setting unmanned aerial vehicle are associated to obtain an attitude response synchronization characteristic value of the setting unmanned aerial vehicle, which is specifically: the angular velocity and the linear acceleration of each attitude time point are cross-correlated, that is, the angular velocity and the linear acceleration of the same axis are arranged in time sequence point by point, then the time position relationship between the two groups of data is changed in turn, the angular velocity data is moved forward or backward by a certain time interval, at each moving position, the angular velocity data and the linear acceleration data are multiplied point by point at the same time point and summed to compare the similarity degree of the change trend of the two groups of data under different time offsets; the similarity degree results under all time offsets are compared to find the group of data with the highest similarity degree, and the corresponding time interval is the similar time difference of the axis direction, so as to obtain the similar time difference of each same axis, and the corresponding synchronization value of the same axis is extracted, that is, 1 / (1+similar time difference), and weighted average processing is performed to extract the attitude response synchronization characteristic value, which is used to represent the synchronization coordination and time sequence matching degree between the attitude change and the power response of the setting unmanned aerial vehicle within the setting period. The greater the attitude response synchronization characteristic value is, the more synchronized the angular velocity and the linear acceleration change, and the higher the attitude and power coordination is.

[0041] In the embodiment, the attitude time sequence data is analyzed and processed to reveal the attitude evolution trend of the unmanned aerial vehicle in the flight process, such as introducing the attitude disturbance entropy, the coordination fluctuation and the response synchronization, and comprehensively evaluating the flight stability. When the unmanned aerial vehicle is in rapid turning, climbing or is disturbed by airflow, the degree of disorder and structural deviation of the attitude change can be identified in time, and the inertial navigation cumulative error can be corrected immediately. At the same time, the combination analysis of the covariance matrix and the sample entropy can judge the overall coordination trend of the attitude and find the abnormal fluctuation of the single axis to avoid local instability causing overall deviation. Through the cross-correlation analysis of the angular velocity and the linear acceleration, the time lag relationship between the power response and the attitude change can be identified to provide a lead judgment for the attitude stability, so that the unmanned aerial vehicle can still maintain stable flight in the environment of high-speed maneuvering or wind field disturbance, and provide a more reliable data basis for positioning correction.

[0042] Specifically, as shown in Figure 3 The motion environment image time sequence data includes a plurality of frames of motion environment image data, and each frame of motion environment image data specifically includes the pixel value, the two-dimensional coordinates and the corresponding depth information value of each pixel point in the motion environment image. The visual environment perception model includes an input layer, a feature decomposition layer, an LSTM layer and an output layer.

[0043] The specific steps of analyzing the environment positioning interference feature value of the setting unmanned aerial vehicle are as follows: inputting the motion environment image time series data of the setting unmanned aerial vehicle into the pre-trained visual environment perception model, analyzing the environment interference mapping feature set of the setting unmanned aerial vehicle, including the propagation obstacle evolution feature value, the scattering distribution evolution feature value, and the environment interference coordination feature value; based on the environment interference mapping feature set of the setting unmanned aerial vehicle, analyzing the environment positioning interference feature value of the setting unmanned aerial vehicle, which is specifically: weighting the propagation obstacle evolution feature value, the scattering distribution evolution feature value, and the environment interference coordination feature value of the setting unmanned aerial vehicle to obtain the environment positioning interference feature value of the setting unmanned aerial vehicle.

[0044] It should be explained that in the weighting process involved in the present embodiment example, the weight coefficient corresponding to each parameter can be obtained by the proportional self-normalization method, and the weighting process of the environment positioning interference feature value of the setting unmanned aerial vehicle is taken as an example: The propagation obstacle evolution feature value, the scattering distribution evolution feature value, and the environment interference coordination feature value of a plurality of historical periods are obtained, the propagation obstacle evolution feature mean value, the scattering distribution evolution feature mean value, and the environment interference coordination feature mean value are extracted, and the sum is processed to obtain the interference sum value. The propagation obstacle evolution feature mean value, the scattering distribution evolution feature mean value, and the environment interference coordination feature mean value are respectively processed by ratio with the interference sum value to obtain the weight coefficients corresponding to the propagation obstacle evolution feature value, the scattering distribution evolution feature value, and the environment interference coordination feature value.

[0045] The specific steps of analyzing the environment interference mapping feature set of the setting unmanned aerial vehicle are as follows: in the input layer of the visual environment perception model, receiving each frame of motion environment image data of the setting unmanned aerial vehicle and performing preprocessing, such as normalizing each input frame of motion environment image to uniformly map the pixel value to the [0, 1] interval, resampling the size of each frame of motion environment image to unify the input resolution, removing random noise points in the image by using Gaussian filtering or median filtering, etc. In the feature decomposition layer of the visual environment perception model, the preprocessed each frame of motion environment image data of the setting unmanned aerial vehicle is processed for feature extraction to obtain the environment interference time series feature vector in the corresponding frame of motion environment image, which is specifically: The pixel value of each pixel point in each frame of moving environment image is subjected to HSI conversion processing to extract the brightness component of the corresponding pixel point, and subjected to convolution operation to extract the horizontal brightness gradient and vertical brightness gradient of the corresponding pixel point, and the square root of the square sum of the two is taken as the brightness change rate of the pixel. At the same time, taking each pixel point as the center, a number of pixel points (such as 3x3 or 5x5 window) around it are taken as the neighborhood window, and the brightness gradient direction vector is extracted, that is, the horizontal brightness gradient value and the vertical brightness gradient value of each pixel point in the neighborhood window are read, and the brightness gradient direction angle of the pixel point is calculated by taking the arctangent value of the two, and the average value is taken to obtain the local texture direction similarity value of the corresponding pixel point, and the brightness change rate is weighted and processed, and the average value of the result is taken as the light reflection feature (used to represent the surface reflection intensity of the region where the pixel is located and the potential interference degree to the satellite signal propagation path); The depth information value of each pixel point is subjected to difference processing (taking absolute value) with the depth information value distribution of each pixel point in the neighborhood window, and the average depth difference of the pixel point is obtained by taking the average value, and the average depth difference of all pixel points higher than the preset shielding threshold value is extracted (higher than the preset shielding threshold value, it is determined that the pixel is at the edge of the obstacle or the sharply fluctuating region), and the average value is taken as the space shielding feature (used to represent the blocking strength of the obstacle to the signal propagation direction), and the light reflection feature is weighted and processed to extract the environment propagation hindering feature; The brightness component of each pixel point in each frame of moving environment image is subjected to discrete Fourier transform (DFT) to extract the high frequency energy and total energy in each frame of moving environment image, and subjected to ratio processing to obtain the high frequency energy proportion (representing the proportion of high frequency details in the image, such as rough surface), and the local brightness variance (i.e. the brightness component of the brightness component of each pixel point is subjected to variance processing) is extracted by taking a fixed size window (such as 5x5), and the average value is taken, and the high frequency energy proportion is weighted and processed to extract the environment scattering distribution feature (used to represent the non-directional scattering intensity of the environment surface to the satellite signal); The environment propagation hindering feature and the environment scattering distribution feature in each frame of moving environment image are spliced into an environment interference time sequence feature vector in time sequence; In the LSTM layer of the visual environment perception model, the time sequence characteristics vector of the environmental interference in each frame of motion environment image of the set unmanned aerial vehicle is analyzed, and the environmental evolution characteristics vector of the set unmanned aerial vehicle is obtained, which is specifically: LSTM updates the hidden state information of each time step through its internal gating unit structure (including input gate, forget gate and output gate) in the processing process; this process can learn the time dependence of the characteristics between consecutive frames, capture the long-term evolution law of environmental interference, and automatically filter short-term noise and unstable fluctuations; at each time step, the LSTM layer calculates the new hidden state representation according to the hidden state of the last time and the current input feature vector, so as to retain the key environmental change characteristics in the whole time sequence, and finally the LSTM layer outputs the environmental dynamic interference characteristics vector which can comprehensively reflect the influence of the external environment on the satellite signal propagation, such as: For the environmental propagation obstruction characteristics in each frame of motion environment image, sliding average processing is performed, that is, the sliding window length is set to several frames (for example, 3 frames or 5 frames), the sliding step is one frame, the environmental propagation obstruction characteristic mean is extracted in each sliding window, and the variance is processed to extract the propagation obstruction variance. The mean value of the environmental propagation obstruction characteristics of adjacent sliding windows is processed by difference, and the mean value is taken, and the propagation obstruction variance is weighted to extract the propagation obstruction evolution characteristics, which are used to represent the time evolution trend of the reflection, shielding and other interference factors in the external environment on the stability of the satellite signal propagation path. When it is larger, it means that the propagation obstruction strength shows an upward trend over time, and the external interference gradually increases. For the environmental scattering distribution characteristics in each frame of motion environment image, sliding average processing is performed to extract the scattering distribution evolution characteristics, which are used to represent the time sequence change trend of the surface roughness and texture structure in the external environment on the signal multipath scattering intensity. When it is larger, it means that the roughness of the environment surface increases and the signal scattering effect intensifies. For the environmental propagation obstruction characteristics and the environmental scattering distribution characteristics in each frame of motion environment image, correlation analysis (Pearson correlation coefficient method can be used) is performed to extract the environmental interference coordination characteristics, which are used to represent the interference degree of the synchronous change degree of the reflection and shielding effect and the scattering effect in the external environment on the signal propagation; and the propagation obstruction evolution characteristics, the scattering distribution evolution characteristics and the environmental interference coordination characteristics are spliced into the environmental evolution characteristics vector. In the output layer of the visual environment perception model, based on the environmental evolution characteristics vector of the set unmanned aerial vehicle, the environmental interference mapping feature set of the set unmanned aerial vehicle is output, which is specifically: the propagation obstruction evolution characteristics, the scattering distribution evolution characteristics and the environmental interference coordination characteristics in the environmental evolution characteristics vector are respectively activated by the Sigmoid function to obtain the propagation obstruction evolution characteristic value, the scattering distribution evolution characteristic value and the environmental interference coordination characteristic value between 0 and 1.

[0046] The pre-training step of the visual environment perception model is as follows: An annotated data set is obtained, which is composed of motion environment image time series data collected by a UAV in different complex environments and its corresponding environment interference true value label. The data collection scenarios include densely built-up areas, forest areas, canyons, open areas, and strong light reflection areas. The data set is jointly labeled by flight tests and ground light shielding observation results to ensure that the training data cover various typical interference patterns. Each sample in the annotated data set includes a plurality of consecutive frames of motion environment images. Each frame of image includes pixel values, two-dimensional coordinates, and depth information values of each pixel point, and is attached with its corresponding true interference degree label (including propagation obstruction level, scattering distribution level, and cooperative interference intensity).

[0047] Secondly, the annotated data set is preprocessed, including pixel value normalization, size resampling, brightness standardization, and noise removal. The data set is divided into a training set, a validation set, and a test set in chronological order, for example, in a ratio of 80%, 10%, and 10%, to ensure that the chronological continuity is not disturbed.

[0048] In the model training phase, each preprocessed image data is input into the input layer of the visual environment perception model. The environmental propagation obstruction features and scattering distribution features are extracted through the feature decomposition layer, and the time sequence feature learning is performed in the LSTM layer. The LSTM optimizes the gate unit weight through the back propagation algorithm (BPTT) to capture the time dependence and evolution trend of the environmental interference features. In the training process, the Adam optimizer is used to minimize the loss function (such as mean square error MSE or cross entropy), and the learning rate and LSTM hidden layer unit number are continuously adjusted to improve the convergence speed and generalization ability of the model.

[0049] The prediction error of the model in different interference scenarios is monitored in real time through the validation set, and the training process is early stopped and the hyperparameters are fine-tuned. Finally, the performance of the model in unseen data is verified using the test set to evaluate its stability and robustness in multiple scenarios and multiple lighting conditions. After training, the optimal model parameters are saved for environment perception and positioning interference feature extraction in actual UAV flight, realizing intelligent adaptation and robust recognition of complex visual environments.

[0050] In the embodiment, the motion environment image time series data is analyzed by introducing a visual environment perception model, so that the unmanned aerial vehicle has adaptive recognition and perception ability to external interference in a complex visual environment. For example, the model can extract actual spatial interference features from details such as light reflection and shielding structure, and can identify the influence trend of satellite signals such as reflection, shielding and scattering when the unmanned aerial vehicle passes through a forest, a valley or a city building group, so as to correct the positioning. The introduction of the LSTM layer enables the model to understand the evolution process of environmental interference, and to construct an environmental interference mapping feature set, so as to accurately evaluate the interference strength when the signal distortion or multipath effect is obvious, thereby significantly improving the environmental adaptability of the unmanned aerial vehicle in a complex scene such as strong light reflection and building group, making the positioning more stable, and ensuring that the flight process is not affected by the sudden burst of visual interference.

[0051] Please refer to Figure 4 The embodiment of the present application provides a technical solution: an unmanned aerial vehicle intelligent RTK inertial navigation system based on visual processing, comprising: a data acquisition module for acquiring satellite RTK signal data, attitude time series data and motion environment image time series data of a set unmanned aerial vehicle; an initial positioning module for analyzing the initial three-dimensional position coordinates and signal confidence characteristic values of the set unmanned aerial vehicle based on the satellite RTK signal data of the set unmanned aerial vehicle; an attitude analysis module for analyzing the attitude comprehensive stability characteristic values of the set unmanned aerial vehicle based on the attitude time series data of the set unmanned aerial vehicle; an environmental positioning interference analysis module for analyzing the environmental positioning interference characteristic values of the set unmanned aerial vehicle based on a pre-trained visual environment perception model and in combination with the motion environment image time series data; and a comprehensive positioning analysis module for correcting and processing the initial three-dimensional position coordinates based on the signal confidence characteristic values, the attitude comprehensive stability characteristic values and the environmental positioning interference characteristic values of the set unmanned aerial vehicle to obtain the positioning three-dimensional position coordinates of the set unmanned aerial vehicle.

[0052] Although the preferred embodiments of the present application have been described, those skilled in the art can make further changes and modifications to the embodiments once they know the basic inventive concept. Therefore, the appended claims are intended to be interpreted as including all changes and modifications falling within the scope of the present application.

[0053] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application also intends to include these modifications and variations.

Claims

1. An unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing, characterized in that, The method comprises the following steps: Obtain satellite RTK signal data, attitude time series data and motion environment image time series data of the set unmanned aerial vehicle; Based on the satellite RTK signal data of the set unmanned aerial vehicle, analyze the initial three-dimensional position coordinates and signal confidence characteristic values of the set unmanned aerial vehicle; Based on the attitude time series data of the set unmanned aerial vehicle, analyze the attitude comprehensive stability characteristic values of the set unmanned aerial vehicle; Based on the pre-trained visual environment perception model and combined with the motion environment image time series data, analyze the environment positioning interference characteristic values of the set unmanned aerial vehicle; Based on the signal confidence characteristic values, attitude comprehensive stability characteristic values and environment positioning interference characteristic values of the set unmanned aerial vehicle, correct the initial three-dimensional position coordinates to obtain the positioning three-dimensional position coordinates of the set unmanned aerial vehicle. 2.The visual processing based intelligent RTK and inertial navigation method for UAVs according to claim 1, wherein, The satellite RTK signal data comprises RTK signal data of several satellites, and each RTK signal data is specifically signal intensity value and pseudo-random code at each signal time point. The specific steps for analyzing the initial three-dimensional position coordinates of the set unmanned aerial vehicle are as follows: Based on the RTK signal data of each satellite of the set unmanned aerial vehicle, extract the pseudo-range observation value and carrier phase observation value at the corresponding signal time point; Based on the pseudo-range observation value and carrier phase observation value at each signal time point of each satellite of the set unmanned aerial vehicle, analyze the initial three-dimensional position coordinates of the set unmanned aerial vehicle. 3.The visual processing based UAV intelligent RTK INS method of claim 2, wherein, The specific steps for analyzing the signal confidence characteristic values of the set unmanned aerial vehicle are as follows: Read the signal intensity value at each signal time point of each satellite of the set unmanned aerial vehicle, analyze the signal positioning interference characteristic set of the corresponding satellite, including energy deflection characteristic value, energy entropy change characteristic value and main peak energy shift characteristic value; Based on the signal positioning interference characteristic set of each satellite of the set unmanned aerial vehicle, analyze the signal confidence characteristic values of the set unmanned aerial vehicle. 4.The visual processing based UAV intelligent RTK INS method of claim 1, wherein, The attitude time series data comprises attitude angle set, three-axis angular velocity value and three-axis linear acceleration value at each attitude time point. The specific steps for analyzing the attitude comprehensive stability characteristic values of the set unmanned aerial vehicle are as follows: Based on the attitude time series data of the set unmanned aerial vehicle, analyze the attitude stability response characteristic set, including attitude disturbance entropy characteristic value, attitude coordination fluctuation characteristic value and attitude response synchronization characteristic value; Fuse the attitude stability response characteristic set of the set unmanned aerial vehicle to obtain the attitude comprehensive stability characteristic values of the set unmanned aerial vehicle. 5.The visual processing based intelligent RTK and inertial navigation method for UAVs according to claim 4, wherein, The specific steps for analyzing the attitude stability response characteristic set of the set unmanned aerial vehicle are as follows: Perform time series processing on the attitude angle set at each attitude time point of the set unmanned aerial vehicle to obtain the attitude disturbance entropy characteristic value and attitude coordination fluctuation characteristic value; And perform correlation processing on the three-axis angular velocity value and three-axis linear acceleration value at each attitude time point of the set unmanned aerial vehicle to obtain the attitude response synchronization characteristic value. 6.The visual processing based intelligent RTK and inertial navigation method for UAVs according to claim 1, wherein, The motion environment image time series data comprises several frames of motion environment image data, and each frame of motion environment image data specifically comprises pixel value, two-dimensional coordinates and corresponding depth information value of each pixel point in the motion environment image. The visual environment perception model comprises an input layer, a feature decomposition layer, an LSTM layer and an output layer. 7.The visual processing based UAV intelligent RTK INS method of claim 6, wherein, The specific steps for analyzing the environment positioning interference characteristic values of the set unmanned aerial vehicle are as follows: The motion environment image time series data of the set unmanned aerial vehicle is input into the pre-trained visual environment perception model, and the environment interference mapping feature set of the set unmanned aerial vehicle is analyzed, including the propagation obstacle evolution characteristic value, the scattering distribution evolution characteristic value and the environment interference cooperative characteristic value. Based on the environment interference mapping feature set of the set unmanned aerial vehicle, the environment positioning interference characteristic value of the set unmanned aerial vehicle is analyzed. 8.The visual processing based UAV intelligent RTK INS method of claim 7, wherein, The specific steps of analyzing the environment interference mapping feature set of the set unmanned aerial vehicle are as follows: In the input layer of the visual environment perception model, each frame of motion environment image data of the set unmanned aerial vehicle is received and preprocessed; In the feature decomposition layer of the visual environment perception model, the feature extraction processing is performed on the preprocessed each frame of motion environment image data of the set unmanned aerial vehicle, and the environment interference time series feature vector in the corresponding frame of motion environment image is obtained; In the LSTM layer of the visual environment perception model, the environment interference time series feature vector in each frame of motion environment image of the set unmanned aerial vehicle is analyzed, and the environment evolution feature vector of the set unmanned aerial vehicle is obtained; In the output layer of the visual environment perception model, based on the environment evolution feature vector of the set unmanned aerial vehicle, the environment interference mapping feature set of the set unmanned aerial vehicle is output. 9.The visual processing based UAV intelligent RTK INS method of claim 1, wherein, The specific steps of obtaining the positioning three-dimensional position coordinates of the set unmanned aerial vehicle are as follows: Based on the signal confidence feature value, the attitude comprehensive stability feature value and the environment positioning interference feature value of the set unmanned aerial vehicle, the positioning signal feature value of the set unmanned aerial vehicle is analyzed; Based on the positioning signal feature value of the set unmanned aerial vehicle, and combined with the initial three-dimensional position coordinates, the positioning three-dimensional position coordinates of the set unmanned aerial vehicle are analyzed.

10. An unmanned aerial vehicle intelligent RTK inertial navigation system based on visual processing, applying the unmanned aerial vehicle intelligent RTK inertial navigation method based on visual processing of any one of claims 1-9, characterized in that, Comprise: The data acquisition module is used for acquiring satellite RTK signal data, attitude time series data and motion environment image time series data of the set unmanned aerial vehicle; The initial positioning module is used for analyzing the initial three-dimensional position coordinates and the signal confidence feature value of the set unmanned aerial vehicle based on the satellite RTK signal data of the set unmanned aerial vehicle; The attitude analysis module is used for analyzing the attitude comprehensive stability feature value of the set unmanned aerial vehicle based on the attitude time series data of the set unmanned aerial vehicle; The environment positioning interference analysis module is used for analyzing the environment positioning interference feature value of the set unmanned aerial vehicle based on the pre-trained visual environment perception model and combined with the motion environment image time series data; The comprehensive positioning analysis module is used for correcting the initial three-dimensional position coordinates based on the signal confidence feature value, the attitude comprehensive stability feature value and the environment positioning interference feature value of the set unmanned aerial vehicle, and obtaining the positioning three-dimensional position coordinates of the set unmanned aerial vehicle.

Images