Unmanned aerial vehicle high-precision positioning method and system based on RTK and inertial navigation

Abstract

The present application relates to the technical field of unmanned aerial vehicle positioning, and particularly relates to a high-precision unmanned aerial vehicle positioning method and system based on RTK and inertial navigation. Time series data of unmanned aerial vehicle attitude parameters (original and smoothed angular velocity), attitude quaternions at each attitude moment, and position coordinates at each position moment are obtained. The time series data of the attitude parameters are analyzed for sudden change characteristics to obtain a characteristic time series sequence, and the body maneuvering mutation is accurately captured. The similarity of the two attitude parameter characteristic time series data is analyzed, a difference matrix is constructed and graph matching is performed to obtain a matching index pair set at the attitude moment. The time series difference and the original angular velocity characteristic value are comprehensively considered in the set to determine a more actual anti-vibration filter lag time. Based on the current position moment, RTK data transmission delay, and anti-vibration filter lag time backtracking, a query position moment is determined. The attitude quaternion at the query moment is introduced into the position coordinate correction, the influence of the attitude on the positioning is fully considered, and the position coordinate is accurately adjusted.

Description

Technical Field

[0001] This invention relates to the field of UAV positioning technology, specifically to a high-precision UAV positioning method and system based on RTK and inertial navigation. Background Technology

[0002] In the application of unmanned aerial vehicles (UAVs), high-precision positioning is a crucial prerequisite for ensuring their accurate execution of various tasks. Currently, common UAV positioning technologies mainly include the Global Positioning System (GPS) and its differential enhancement technologies, such as Real-time Dynamic Differential Positioning (RTK). RTK technology significantly improves positioning accuracy by performing carrier phase differential processing between a base station and a rover, combined with an inertial measurement unit (IMU). Under ideal conditions, it can achieve centimeter-level positioning accuracy and is widely used in scenarios with high positioning accuracy requirements, such as surveying, agricultural plant protection, and logistics distribution.

[0003] To suppress the interference of high-frequency mechanical vibrations generated by the high-speed rotation of the motor on the flight control system, digital vibration damping filters are typically integrated into the data link of the inertial measurement unit. However, while filtering out vibration noise, this filter introduces phase lag in the signal. Existing technologies usually assume that the filter's delay time is a fixed value for time compensation. However, in actual operating conditions, to adapt to vibration frequencies at different speeds, the cutoff frequency of the vibration damping filter is often dynamically adjusted with the motor speed, causing changes in the phase lag time. Therefore, using a fixed delay time for time compensation to locate the UAV will significantly reduce positioning accuracy. Summary of the Invention

[0004] To address the technical problem that in actual operating conditions, the cutoff frequency of the vibration-damping filter is often dynamically adjusted with the motor speed to adapt to vibration frequencies at different speeds, resulting in changes in phase lag time; therefore, using a fixed delay time for time compensation to locate the UAV will significantly reduce positioning accuracy, the present invention aims to provide a high-precision UAV positioning method and system based on RTK and inertial navigation. The specific technical solution adopted is as follows: A high-precision positioning method for unmanned aerial vehicles (UAVs) based on RTK and inertial navigation includes: The attitude parameters of the UAV are obtained, including the original angular velocity time series data, the smoothed angular velocity time series data after anti-vibration filtering, and the attitude quaternions at each attitude time; the position coordinates of the UAV at each position time are obtained. The numerical mutation characteristics of the raw angular velocity time series data and the smoothed angular velocity time series data of the UAV are analyzed separately to obtain the corresponding raw angular velocity feature time series sequence and smoothed angular velocity feature time series sequence. Based on the similarity between the raw angular velocity feature time series sequence and the smoothed angular velocity feature time series sequence, a difference matrix is ​​constructed and graph matching is performed to obtain the attitude time matching index pair set. In the set of attitude moment matching index pairs, the vibration-resistant filter lag time is determined based on the temporal difference characteristics between attitude moments and the original angular velocity characteristic value corresponding to the moment; the query position moment corresponding to the current position moment is determined by backtracking in time based on the current position moment, RTK link transmission delay characteristics and the vibration-resistant filter lag time. Based on the attitude quaternion of the UAV at the query location, the position coordinates of the UAV at the current location are corrected to determine the corrected observation position of the UAV at the current location.

[0005] Further, obtaining the corresponding original angular velocity feature time series sequence and the smoothed angular velocity feature time series sequence includes: For each attitude parameter, the neighborhood window corresponding to each attitude time step is determined in time series. Within the neighborhood window at each attitude time, calculate the mean of all attitude parameters, which serves as the mean feature value at each attitude time. In terms of time sequence, the mean eigenvalues ​​at each attitude time are subjected to second-order difference operation, and the resulting values ​​are used as the values ​​to be measured at each attitude time. For each attitude parameter, if the absolute value of the measured value at a certain attitude moment is less than or equal to the preset feature significance threshold, then in the feature time series corresponding to that attitude parameter, the feature value at that attitude moment is recorded as 0; otherwise, the feature value at that attitude moment is recorded as the measured value, thereby obtaining the original angular velocity feature time series and the smoothed angular velocity feature time series.

[0006] Furthermore, the method for obtaining the set of attitude moment matching index pairs includes: In the original angular velocity feature time series, the attitude moments corresponding to all maxima are obtained and used as the first attitude moments. In the smooth angular velocity feature time series, the attitude moments corresponding to all maxima are obtained and used as the second attitude moments. In all first attitude moments, after arranging the original angular velocity feature values ​​in descending order, the attitude moments corresponding to the first preset first number of original angular velocity feature values ​​are taken as the first attitude moments to be matched. In all second attitude moments, after arranging the smooth angular velocity feature values ​​in descending order, the attitude moments corresponding to the first preset first number of smooth angular velocity feature values ​​are taken as the second attitude moments to be matched. The index of the first pose to be matched is used as the column of the difference matrix, and the index of the second pose to be matched is used as the row of the difference matrix; For any position in the difference matrix, if the timing of the first matching attitude time corresponding to that position is greater than the timing of the second matching attitude time, then the difference at that position is recorded as a preset value. If the timing of the first matching attitude moment corresponding to this position is less than or equal to the timing of the second matching attitude moment, then analyze the timing difference between the first matching attitude moment and the second matching attitude moment to determine the time dimension difference value corresponding to this position; analyze the difference between the original angular velocity feature value corresponding to the first matching attitude moment and the smooth angular velocity feature value corresponding to the second matching attitude moment to determine the amplitude dimension difference value corresponding to this position. The normalized sum of the time dimension difference value and the amplitude dimension difference value corresponding to this position is taken as the difference degree at that position. The Hungarian algorithm is used to solve the difference matrix, thereby outputting all attitude time-matching index pairs, forming a set of attitude time-matching index pairs.

[0007] Furthermore, the method for obtaining the time dimension difference value includes: The time difference between the second matching posture time and the first matching posture time is used as the deviation factor, and the absolute value of the difference between the deviation factor and the preset delay factor is used as the time dimension difference value corresponding to that position.

[0008] Furthermore, the method for obtaining the amplitude dimension difference value includes: The original angular velocity feature value corresponding to the first matching attitude moment is multiplied by a preset passband coefficient to obtain an adjustment feature value. The absolute value of the difference between the adjustment feature value and the smooth angular velocity feature value corresponding to the second matching attitude moment is taken as the amplitude dimension difference value corresponding to that position.

[0009] Furthermore, the method for obtaining the vibration-damping filter hysteresis time includes: In the set of attitude moment matching index pairs, in each attitude moment matching index pair, the product of the deviation factor and the original angular velocity feature value corresponding to the first matching attitude moment is used as the feature weighted deviation of each attitude moment matching index pair. The sum of the feature weighted deviations of all attitude moment matching index pairs is used as the numerator, and the sum of the original angular velocity feature values ​​corresponding to the first matching attitude moment in all attitude moment matching index pairs is used as the denominator. The resulting ratio is used as the vibration-damping filter hysteresis time.

[0010] Furthermore, the method for obtaining the query location time includes: The difference between the current location time and the preset transmission delay time of the RTK link is used as the physical sampling location time corresponding to the drone's position coordinates at the current location time; The sum of the physical sampling time and the vibration-resistant filtering lag time is used as the query time corresponding to the current position time.

[0011] Furthermore, the method for obtaining the corrected observation position includes: Spherical linear interpolation is performed using the attitude quaternions at the second number of positions closest to the query position time to obtain the attitude quaternions at the query position time. These quaternions are then used to rotate the preset antenna installation vector of the RTK from the body coordinate system to the navigation coordinate system to obtain the antenna offset. The difference between the UAV's position coordinates at the current location and the antenna offset is used as the corrected observation position of the UAV at the current location.

[0012] Furthermore, the method for setting the neighborhood window includes: The neighborhood window is 3 time units in size, including each pose time and the two pose times that are closest to each pose time in time sequence.

[0013] A high-precision positioning system for unmanned aerial vehicles (UAVs) based on RTK and inertial navigation includes a processor and a memory. The memory stores at least one instruction, at least one program, code set, or instruction set. When the processor loads and executes the at least one instruction, at least one program, code set, or instruction set, it implements the steps of a high-precision positioning method for UAVs based on RTK and inertial navigation.

[0014] The present invention has the following beneficial effects: The time-series data of the UAV's attitude parameters (including raw angular velocity and smoothed angular velocity after vibration suppression filtering), attitude quaternions at various attitude moments, and the UAV's position coordinates at various location moments were acquired, providing comprehensive raw information for subsequent positioning processing. To eliminate the influence of gyroscope zero-point drift and the DC component during steady-state flight on the reference, the numerical abrupt change characteristics in the attitude parameter time-series data were analyzed, resulting in corresponding characteristic time-series sequences, which help accurately capture abrupt changes in aircraft maneuvers. To adapt to waveform detail distortion caused by vibration suppression filtering, the similarity of the characteristic time-series data of the two attitude parameters was analyzed to construct a difference matrix and perform graph matching. The optimal time correspondence was found from a global perspective, and a set of attitude moment matching index pairs was obtained based on a causal relationship model of discrete events. Furthermore, in the set of attitude moment matching index pairs, the time-series differences between the attitude moment matching index pairs and the characteristic values ​​of the raw angular velocity were comprehensively considered. This allows for a more comprehensive consideration of factors affecting the vibration suppression filtering lag time, making the determined vibration suppression filtering lag time more consistent with reality and providing accurate time parameters for subsequent positioning correction. Then, based on three key factors—the current position time, the unavoidable delay in RTK data transmission, and the vibration-resistant filtering lag time—backtracking is performed in time sequence to determine the query position time corresponding to the current position time. Finally, since attitude quaternions can accurately describe the UAV's attitude information, incorporating the UAV's attitude quaternions at the query position time into the position coordinate correction process can fully consider the impact of UAV attitude on positioning, enabling more precise adjustment of the UAV's position coordinates, eliminating positioning errors caused by attitude changes and various time delay factors, and obtaining a more accurate corrected observation position. Attached Figure Description

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

[0016] Figure 1 This is a flowchart illustrating a high-precision positioning method for unmanned aerial vehicles (UAVs) based on RTK and inertial navigation, provided in one embodiment of the present invention. Figure 2 This is a system block diagram of a high-precision positioning system for unmanned aerial vehicles (UAVs) based on RTK and inertial navigation, provided as an embodiment of the present invention. Figure 3 This is a schematic diagram of the system structure of a high-precision positioning system for unmanned aerial vehicles based on RTK and inertial navigation, provided as an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a high-precision positioning method and system for unmanned aerial vehicles based on RTK and inertial navigation proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of a high-precision positioning method and system for unmanned aerial vehicles (UAVs) based on RTK and inertial navigation provided by this invention.

[0020] Please see Figure 1 The diagram illustrates a flowchart of a high-precision UAV positioning method based on RTK and inertial navigation, according to an embodiment of the present invention. The method includes the following steps: Step S1: Obtain the attitude parameters of the UAV, including the original angular velocity time series data, the smoothed angular velocity time series data after anti-vibration filtering, and the attitude quaternions at each attitude time; obtain the position coordinates of the UAV at each position time.

[0021] The IMU is the core device for acquiring the attitude parameters (raw angular velocity and smooth angular velocity) of the UAV, and typically includes a gyroscope and an accelerometer. The gyroscope is used to measure the rotational angular velocity of the UAV around three axes (roll axis, pitch axis, and yaw axis). Since the inertial measurement unit (IMU) and the RTK positioning module usually have independent internal clocks and different sampling frequencies (e.g., IMU is 1000Hz, RTK is 10Hz), in this embodiment of the invention, for ease of distinction, the time of data acquired by the inertial measurement unit is taken as the attitude time, and the time of UAV position information acquired by the RTK positioning module is taken as the position time.

[0022] A high-precision hardware counter of a microcontroller maintains a locally monotonically increasing microsecond-level time axis with a fixed sampling period (1 millisecond in this embodiment). In this embodiment, the rotational angular velocities along the three axes are moduloed to obtain the sum of the angular velocities, which serve as the raw angular velocity time series data. The attitude quaternions at each attitude moment are obtained using a strapdown inertial navigation algorithm. A high-precision, low-noise gyroscope, such as a fiber optic gyroscope or a MEMS (Micro-Electro-Mechanical Systems) gyroscope, is selected to ensure the accuracy and reliability of the acquired angular velocity data. A digital vibration damping filter is also integrated into the IMU to suppress the interference of high-frequency mechanical vibrations generated by the high-speed rotation of the motor on the flight control system. After processing by this filter, the raw angular velocity data is smoothed. The cutoff frequency of the vibration damping filter can be dynamically adjusted according to the motor speed to adapt to vibration frequencies under different operating conditions.

[0023] At the same time, it is also necessary to capture the pulse signal output by the RTK module, establish a linear mapping relationship between RTK time and local time axis, and record the position time of receiving the pulse signal and the position coordinates of the UAV at the position time (the position coordinates are determined based on the navigation coordinate system, that is, the GPS positioning system).

[0024] In this embodiment of the invention, the position time of the UAV's position coordinates is a discrete time point, which can be stored when the RTK system receives it; the length of the timing data of the attitude parameters is set to 500 milliseconds, which can be traced back to the history from the position coordinates at the current position time. This length is sufficient to cover the sum of the RTK link transmission delay and the vibration filtering lag.

[0025] Step S2: Analyze the numerical mutation characteristics of the raw angular velocity time series data and the smoothed angular velocity time series data of the UAV respectively, so as to obtain the corresponding raw angular velocity feature time series sequence and smoothed angular velocity feature time series sequence; construct the difference matrix based on the similarity between the raw angular velocity feature time series sequence and the smoothed angular velocity feature time series sequence and perform graph matching to obtain the attitude time matching index pair set.

[0026] Because the intervention of filters can cause distortions in signal waveforms, such as peak clipping or broadening, conventional time-domain alignment methods based on waveform geometric similarity are difficult to accurately identify the actual lag time. Therefore, in this embodiment of the invention, to adapt to the waveform detail distortions caused by anti-vibration filtering, the dependence on the geometry of continuous waveforms is abandoned, and a matching model based on discrete-time causal relationships is constructed instead. In this step, the continuous angular velocity time series data is transformed into discrete features that can reflect abrupt changes in the aircraft's maneuvering. At the same time, since the abrupt changes in numerical characteristics can reflect significant changes in the UAV's angular velocity at different times, and these changes are often related to events such as changes in the UAV's flight state, the numerical abrupt changes in the UAV's original angular velocity time series data and smoothed angular velocity time series data can be analyzed first to obtain the corresponding original angular velocity feature time series sequence and smoothed angular velocity feature time series sequence, which are used to capture key information in the data.

[0027] Preferably, in one embodiment of the present invention, obtaining the corresponding original angular velocity feature time series sequence and the smoothed angular velocity feature time series sequence includes: In order to eliminate the zero-bias drift and linear trend term of the gyroscope and thus extract the purest maneuver mutation features, second-order difference operation can be used in this embodiment of the invention. However, since second-order difference operation is more sensitive to noise, the data can be smoothed and preprocessed before performing the difference calculation.

[0028] For each attitude parameter's time-series data, a neighborhood window is determined temporally for each attitude moment. Here, the size of the neighborhood window is set to 3 time units, including each attitude moment and the two attitude moments most closely related to it. Then, within the neighborhood window of each attitude moment, the mean of all attitude parameters is calculated as the mean feature value for each attitude moment. The mean feature value can eliminate local fluctuations in the data within the neighborhood window, resulting in a relatively stable feature value.

[0029] Next, to eliminate the influence of gyroscope zero-point drift, in terms of timing, a second-order difference operation is performed on the mean characteristic value at each attitude moment, and the obtained value is used as the test value at each attitude moment. At this time, the test value can convert the slowly changing signal components into 0 values, and only generate significant pulse values ​​at the moment when the body is subjected to a sudden change in external torque.

[0030] Finally, to further identify the minute fluctuations caused by sensor noise and prevent them from being misidentified as maneuvering events, a salience threshold filtering mechanism can be introduced: under each attitude parameter, if the absolute value of the measured value at a certain attitude moment is less than or equal to a preset feature salience threshold, then in the feature time series corresponding to that attitude parameter, the feature value at that attitude moment is recorded as 0; otherwise, the feature value at that attitude moment is recorded as the measured value. This yields the original angular velocity feature time series and the smoothed angular velocity feature time series. At this point, the original angular velocity feature series and the smoothed angular velocity feature series only retain significant maneuvering events, transforming the originally dense time series into a sparse pulse sequence. This allows for the extraction of the purest and most critical maneuvering features from the complex data, laying a more accurate data foundation for subsequent graph matching.

[0031] It should be noted that the preset feature saliency threshold in this embodiment of the present invention can be set to 5 times the standard deviation of the sensor's static noise floor. The specific value can also be adjusted according to the implementation scenario, and is not limited here.

[0032] Graph matching algorithms are an effective sequence matching method. They treat elements in a sequence as nodes in a graph and similarity relationships between elements as edges. Sequence matching is achieved by finding the optimal matching path within the graph. Graph matching algorithms consider both local similarity and global structural information within the sequence, exhibiting high matching accuracy and robustness. Therefore, the similarity between the original angular velocity characteristic time series and the smoothed angular velocity characteristic time series can be analyzed to construct a difference matrix, enabling graph matching. This yields an accurate set of attitude time-matching index pairs, which is helpful in subsequently calculating the nonlinear dynamic hysteresis characteristics caused by vibration damping.

[0033] Preferably, in one embodiment of the present invention, the method for obtaining the set of attitude moment matching index pairs includes: In a set of data, the maximum value usually represents the most significant feature. Therefore, in the original angular velocity feature time series, the attitude time corresponding to all the maximum values ​​is obtained and used as the first attitude time. In the smooth angular velocity feature time series, the attitude time corresponding to all the maximum values ​​is obtained and used as the second attitude time. The first attitude time and the second attitude time can enable the subsequent analysis process to focus on the key moments, reduce the amount of data to be processed, and improve efficiency.

[0034] Then, in all first attitude moments, the raw angular velocity feature values ​​are sorted in descending order, and the attitude moments corresponding to the first preset first number of raw angular velocity feature values ​​are selected as the first attitude moments to be matched. Similarly, in all second attitude moments, the smoothed angular velocity feature values ​​are sorted in descending order, and the attitude moments corresponding to the first preset first number of smoothed angular velocity feature values ​​are selected as the second attitude moments to be matched. Since not all maxima are equally important for UAV attitude matching, sorting in descending order and selecting attitude moments with larger feature values ​​as the attitude moments to be matched allows for prioritizing the moments with the most drastic angular velocity changes and the most significant impact on attitude. It should be noted that the preset first number can be set to 20, and the specific value can be adjusted according to the implementation scenario. It is not limited here, but it is necessary to ensure that the total number of first attitude moments and the total number of second attitude moments are both greater than or equal to the preset first number.

[0035] To quantify the probability that the raw angular velocity at any attitude moment and the smooth angular velocity at any attitude moment belong to the same maneuver, a difference matrix is ​​constructed: the index of the first attitude moment to be matched is used as the column of the difference matrix, and the index of the second attitude moment to be matched is used as the row of the difference matrix.

[0036] At this point, since physically, the response time of the filtered signal is necessarily later than the original excitation time, for any position in the difference matrix, if the timing of the first matching attitude time corresponding to that position is greater than the timing of the corresponding second matching attitude time, it indicates that the smoothing time occurred earlier than the original time, violating physical laws. Therefore, the matching is directly determined to be invalid. So, the difference at that position is recorded as a preset value. This preset value should be set to infinity. In actual engineering, it can be taken as a maximum value, such as 10. 9 .

[0037] Conversely, if the timing of the first matching attitude moment corresponding to that position is less than or equal to the timing of the corresponding second matching attitude moment, meaning the timing sequence conforms to physical causality, then the timing difference between the first and second matching attitude moments is analyzed to determine the time dimension difference value corresponding to that position. The timing difference between the second and first matching attitude moments is used as a deviation factor. The deviation factor characterizes the instantaneous lag time when the first and second matching attitude moments truly represent the same maneuvering event. A larger deviation factor indicates a more significant lag. However, although the lag time is dynamic, it often does not undergo large abrupt changes. Therefore, the absolute value of the difference between the deviation factor and the preset delay factor is calculated as the time dimension difference value corresponding to that position. The preset delay factor can be obtained based on the system's historical state or prior knowledge. When the time dimension difference value is larger, it indicates that the timing relationship between the first and second matching attitude moments does not conform to the system's historical delay pattern, and the matching probability is lower. It should be noted that the preset delay factor can be taken as the theoretical design constant of the vibration-damping filter, such as 15ms. The specific value can be adjusted according to the implementation scenario and is not limited here.

[0038] Next, we continue to analyze the difference between the original angular velocity characteristic value corresponding to the first matching attitude moment and the smoothed angular velocity characteristic value corresponding to the second matching attitude moment, and determine the amplitude dimension difference value corresponding to this position: any vibration-damping filter will attenuate the signal amplitude, so the original angular velocity characteristic value corresponding to the first matching attitude moment is multiplied by the preset passband coefficient as the adjustment characteristic value. That is to say, if the theoretical amplitude of the original data after filtering should be near the adjustment characteristic value, the absolute value of the difference between the adjustment characteristic value and the smoothed angular velocity characteristic value corresponding to the second matching attitude moment is taken as the amplitude dimension difference value corresponding to this position. The smaller the amplitude dimension difference value, the more the amplitudes at the first matching attitude moment and the second matching attitude moment are matched in energy, and the greater the probability of representing the same event.

[0039] It should be noted that the preset passband factor is set to 0.95, and the specific value can be adjusted according to the implementation scenario, without limitation here.

[0040] The sum of the time dimension difference value and the amplitude dimension difference value corresponding to this position is normalized and used as the difference degree at that position. This difference degree serves as a comprehensive indicator, representing the degree of matching between the two matching pose moments corresponding to this position. Based on the aforementioned logic, the larger the difference degree, the lower the degree of matching. Normalization is a technique well-known to those skilled in the art. The normalization function can be linear normalization or standard normalization, such as maximum-minimum value normalization, where the maximum and minimum values ​​in the difference degree matrix are all not equal to the preset value of 10. 9The selection is made from the data, and the specific normalization method is not limited here.

[0041] Finally, the Hungarian algorithm is used to solve the difference matrix, thereby outputting all attitude time-matching index pairs, forming a set of attitude time-matching index pairs. Thus, each attitude time-matching pair in the set of attitude time-matching index pairs represents the corresponding original event and smoothed event.

[0042] It should be noted that the Hungarian algorithm is a well-known technique, and the specific process will not be elaborated here.

[0043] Step S3: In the set of attitude time matching index pairs, determine the vibration damping lag time based on the temporal difference characteristics between attitude times and the original angular velocity characteristic value corresponding to the time; perform time backtracking based on the current position time, RTK link transmission delay characteristics and vibration damping lag time to determine the query position time corresponding to the current position time.

[0044] Based on the aforementioned steps, a set of attitude time matching index pairs can be obtained. At this time, each attitude time matching index pair will have a temporal difference feature, which is an important basis for quantifying the anti-vibration filter lag time. Meanwhile, since small-amplitude maneuvers are easily affected by noise, the temporal difference feature is not accurate enough. However, the signal-to-noise ratio of large-amplitude maneuvers is more reliable than the corresponding time difference. Therefore, in this step, the temporal difference feature between attitude times in the set of attitude time matching index pairs and the original angular velocity feature value corresponding to the time can be analyzed to determine the anti-vibration filter lag time.

[0045] Preferably, in one embodiment of the present invention, the method for obtaining the vibration-damping filter hysteresis time includes: In the set of attitude moment matching index pairs, the deviation factor in each attitude moment matching index pair reflects the instantaneous lag time between the two matching attitude moments in this attitude moment matching index pair, while the original angular velocity feature value at the first matching attitude moment reflects the maneuver intensity of this attitude moment matching index pair. The larger the value, the more significant the information contained and the stronger the noise resistance, so the weight should be higher. Therefore, the product of the deviation factor and the original angular velocity feature value corresponding to the first matching attitude moment is used as the feature weighted deviation of each attitude moment matching index pair. The larger the feature weighted deviation, the greater the influence of this attitude moment matching index pair on the final result.

[0046] Finally, the sum of the feature weighted deviations of all attitude moment matching index pairs is used as the numerator, and the sum of the original angular velocity feature values ​​corresponding to the first matching attitude moment in all attitude moment matching index pairs is used as the denominator. Essentially, this is a weighted average. The resulting ratio is used as the anti-vibration filter lag time. The larger the anti-vibration filter lag time, the greater the delay.

[0047] When the drone's position coordinates at the current time are received from the RTK module, it is necessary to obtain the actual attitude of the drone corresponding to those position coordinates. Since the RTK module has an inherent link transmission delay from satellite signal sampling and calculation to transmission to the microcontroller via serial port / bus, when determining the query position time corresponding to the current position time, it is necessary to consider not only the vibration filtering lag time, but also the RTK link transmission delay characteristics.

[0048] Preferably, in one embodiment of the present invention, the method for obtaining the query location time includes: First, eliminate the inherent link transmission delay: take the difference between the current position time and the preset transmission delay of the RTK link as the physical sampling position time corresponding to the position coordinates of the UAV at the current position time. At this time, the physical sampling position time eliminates the systematic time deviation caused by the transmission link.

[0049] Then, in order to obtain the actual motion attitude of the aircraft at the physical sampling position, it is also necessary to cancel the filter lag: Since there is a time lag between the smoothed attitude data and the actual physical motion, that is, the actual action occurs at time t, but will appear at time (t + filter lag) in the smoothed attitude data, the sum of the physical sampling position time and the anti-vibration filter lag time is used as the query position time corresponding to the current position time. The query position time at this time represents that, under the action of the anti-vibration filter, the attitude action corresponding to the position coordinates of the UAV at the current position time received by RTK is delayed until the query position time.

[0050] It should be noted that the preset transmission delay duration in this embodiment of the invention is set to 60ms. The specific value can be adjusted according to the implementation scenario and is not limited here.

[0051] Step S4: Based on the attitude quaternion of the UAV at the query location, correct the position coordinates of the UAV at the current location to determine the corrected observation position of the UAV at the current location.

[0052] RTK antennas are typically mounted on the top or tail of the fuselage, at a physical distance (the distance of the stick arm) from the flight control core. Therefore, after the analysis and calculations described above, the query position time corresponding to the current position time can be obtained. The position coordinates of the UAV at the current position time can be corrected using the attitude quaternion of the UAV at the query position time, thus obtaining the corrected observation position of the UAV at the current position time, thereby achieving accurate positioning.

[0053] Preferably, in one embodiment of the present invention, the method for obtaining the corrected observation position includes: Since the acquisition frequencies of position time and attitude time are inconsistent, the attitude quaternion may not exist at the aforementioned position time. Therefore, spherical linear interpolation is performed using the attitude quaternions at the second preset number of position times (set to 4 in this embodiment) that are closest to the query position time (a well-known technique, which will not be elaborated here) to obtain the attitude quaternion at the query position time. This quaternion is then used to rotate the preset antenna installation vector of RTK (that is, the position coordinates of the RTK antenna relative to the fuselage) from the body coordinate system to the navigation coordinate system to obtain the antenna offset (this process can be implemented by the quaternion rotation formula, which is a well-known technique, and will not be elaborated here).

[0054] Finally, the difference between the UAV's position coordinates and the antenna offset at the current location is used as the corrected observation position of the UAV at the current location. This corrected observation position not only finds the aircraft attitude at the accurate time point, but also solves the system deviation caused by the antenna installation position. Therefore, it can more accurately characterize the UAV's position characteristics at the current location.

[0055] In summary, acquiring the time-series data of the UAV's attitude parameters (including the original angular velocity and the smoothed angular velocity after vibration suppression filtering), the attitude quaternions at each attitude moment, and the UAV's position coordinates at each position moment provides comprehensive raw information for subsequent positioning processing. To eliminate the influence of gyroscope zero-point drift and the DC component during steady-state flight on the reference, the numerical abrupt change characteristics in the attitude parameter time-series data were analyzed, resulting in corresponding characteristic time-series sequences, which helps to accurately capture abrupt changes in aircraft maneuvers. To adapt to waveform detail distortion caused by vibration suppression filtering, the similarity of the characteristic time-series data of the two attitude parameters was analyzed to construct a difference matrix and perform graph matching. The optimal time correspondence was found from a global perspective, and a set of attitude moment matching index pairs was obtained based on a causal relationship model of discrete events. Furthermore, in the set of attitude moment matching index pairs, comprehensively considering the time-series differences between the attitude moment matching index pairs and the characteristic values ​​of the original angular velocity, the factors affecting the vibration suppression filtering lag time can be more comprehensively considered, making the determined vibration suppression filtering lag time more consistent with the actual situation and providing accurate time parameters for subsequent positioning correction. Then, based on three key factors—the current position time, the unavoidable delay in RTK data transmission, and the vibration-resistant filtering lag time—backtracking is performed in time sequence to determine the query position time corresponding to the current position time. Finally, since attitude quaternions can accurately describe the UAV's attitude information, incorporating the UAV's attitude quaternions at the query position time into the position coordinate correction process can fully consider the impact of UAV attitude on positioning, enabling more precise adjustment of the UAV's position coordinates, eliminating positioning errors caused by attitude changes and various time delay factors, and obtaining a more accurate corrected observation position.

[0056] This invention also provides a high-precision positioning system for unmanned aerial vehicles (UAVs) based on RTK and inertial navigation. Please refer to [link to relevant documentation]. Figure 2 The diagram shows a system block diagram, including a data acquisition module 201 for implementing step S1 in the above method embodiment; a time matching module 202 for implementing step S2 in the above method embodiment; a hysteresis compensation module 203 for implementing step S3 in the above method embodiment; and a positioning module 204 for implementing step S4 in the above method embodiment.

[0057] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the UAV high-precision positioning system based on RTK and inertial navigation and the UAV high-precision positioning method based on RTK and inertial navigation provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiment, and will not be repeated here.

[0058] Please see Figure 3 This diagram illustrates a system architecture of a high-precision UAV positioning system based on RTK and inertial navigation, according to an embodiment of the present invention. The system includes a processor 300, a memory 301, a bus 302, and a communication interface 303. The processor 300, communication interface 303, and memory 301 are connected via the bus 302. The memory 301 may contain high-speed random access memory, and the bus 302 may be an ISA bus, PCI bus, or EISA bus, etc. The processor 300 may be an integrated circuit chip with signal processing capabilities. The memory 301 stores at least one instruction, at least one program, code set, or instruction set. When the processor loads and executes the at least one instruction, at least one program, code set, or instruction set, it implements the steps of a high-precision UAV positioning method based on RTK and inertial navigation.

[0059] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0060] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A high-precision positioning method for unmanned aerial vehicles (UAVs) based on RTK and inertial navigation, characterized in that, The method includes: The attitude parameters of the UAV are obtained, including the original angular velocity time series data, the smoothed angular velocity time series data after anti-vibration filtering, and the attitude quaternions at each attitude time; the position coordinates of the UAV at each position time are obtained. The numerical mutation characteristics of the raw angular velocity time series data and the smoothed angular velocity time series data of the UAV are analyzed separately to obtain the corresponding raw angular velocity feature time series sequence and smoothed angular velocity feature time series sequence. Based on the similarity between the raw angular velocity feature time series sequence and the smoothed angular velocity feature time series sequence, a difference matrix is ​​constructed and graph matching is performed to obtain the attitude time matching index pair set. In the set of attitude moment matching index pairs, the vibration-resistant filter lag time is determined based on the temporal difference characteristics between attitude moments and the original angular velocity characteristic value corresponding to the moment; the query position moment corresponding to the current position moment is determined by backtracking in time based on the current position moment, RTK link transmission delay characteristics and the vibration-resistant filter lag time. Based on the attitude quaternion of the UAV at the query location, the position coordinates of the UAV at the current location are corrected to determine the corrected observation position of the UAV at the current location.

2. The high-precision positioning method for UAVs based on RTK and inertial navigation according to claim 1, characterized in that, The process of obtaining the corresponding original angular velocity feature time series sequence and smoothed angular velocity feature time series sequence includes: For each attitude parameter, the neighborhood window corresponding to each attitude time step is determined in time series. Within the neighborhood window at each attitude time, calculate the mean of all attitude parameters, which serves as the mean feature value at each attitude time. In terms of time sequence, the mean eigenvalues ​​at each attitude time are subjected to second-order difference operation, and the resulting values ​​are used as the values ​​to be measured at each attitude time. For each attitude parameter, if the absolute value of the measured value at a certain attitude moment is less than or equal to the preset feature significance threshold, then in the feature time series corresponding to that attitude parameter, the feature value at that attitude moment is recorded as 0; otherwise, the feature value at that attitude moment is recorded as the measured value, thereby obtaining the original angular velocity feature time series and the smoothed angular velocity feature time series.

3. The high-precision positioning method for UAVs based on RTK and inertial navigation according to claim 1, characterized in that, The method for obtaining the set of attitude moment matching index pairs includes: In the original angular velocity feature time series, the attitude moments corresponding to all maxima are obtained and used as the first attitude moments. In the smooth angular velocity feature time series, the attitude moments corresponding to all maxima are obtained and used as the second attitude moments. In all first attitude moments, after arranging the original angular velocity feature values ​​in descending order, the attitude moments corresponding to the first preset first number of original angular velocity feature values ​​are taken as the first attitude moments to be matched. In all second attitude moments, after arranging the smooth angular velocity feature values ​​in descending order, the attitude moments corresponding to the first preset first number of smooth angular velocity feature values ​​are taken as the second attitude moments to be matched. The index of the first pose to be matched is used as the column of the difference matrix, and the index of the second pose to be matched is used as the row of the difference matrix. For any position in the difference matrix, if the timing of the first matching attitude time corresponding to that position is greater than the timing of the second matching attitude time, then the difference at that position is recorded as a preset value. If the timing of the first matching attitude moment corresponding to this position is less than or equal to the timing of the second matching attitude moment, then analyze the timing difference between the first matching attitude moment and the second matching attitude moment to determine the time dimension difference value corresponding to this position; analyze the difference between the original angular velocity feature value corresponding to the first matching attitude moment and the smooth angular velocity feature value corresponding to the second matching attitude moment to determine the amplitude dimension difference value corresponding to this position. The normalized sum of the time dimension difference value and the amplitude dimension difference value corresponding to this position is taken as the difference degree at that position. The Hungarian algorithm is used to solve the difference matrix, thereby outputting all attitude time-matching index pairs, forming a set of attitude time-matching index pairs.

4. A high-precision positioning method for UAVs based on RTK and inertial navigation according to claim 3, characterized in that, The method for obtaining the time dimension difference value includes: The time difference between the second matching posture time and the first matching posture time is used as the deviation factor, and the absolute value of the difference between the deviation factor and the preset delay factor is used as the time dimension difference value corresponding to that position.

5. A high-precision positioning method for UAVs based on RTK and inertial navigation according to claim 3, characterized in that, The method for obtaining the amplitude dimension difference value includes: The original angular velocity feature value corresponding to the first matching attitude moment is multiplied by a preset passband coefficient to obtain an adjustment feature value. The absolute value of the difference between the adjustment feature value and the smooth angular velocity feature value corresponding to the second matching attitude moment is taken as the amplitude dimension difference value corresponding to that position.

6. A high-precision positioning method for UAVs based on RTK and inertial navigation according to claim 4, characterized in that, The method for obtaining the vibration-damping filter hysteresis time includes: In the set of attitude moment matching index pairs, in each attitude moment matching index pair, the product of the deviation factor and the original angular velocity feature value corresponding to the first matching attitude moment is used as the feature weighted deviation of each attitude moment matching index pair. The sum of the feature weighted deviations of all attitude moment matching index pairs is used as the numerator, and the sum of the original angular velocity feature values ​​corresponding to the first matching attitude moment in all attitude moment matching index pairs is used as the denominator. The resulting ratio is used as the vibration-damping filter hysteresis time.

7. A high-precision positioning method for UAVs based on RTK and inertial navigation according to claim 1, characterized in that, The method for obtaining the query location time includes: The difference between the current location time and the preset transmission delay time of the RTK link is used as the physical sampling location time corresponding to the drone's position coordinates at the current location time; The sum of the physical sampling time and the vibration-resistant filtering lag time is used as the query time corresponding to the current position time.

8. A high-precision positioning method for UAVs based on RTK and inertial navigation according to claim 1, characterized in that, The method for obtaining the corrected observation position includes: Spherical linear interpolation is performed using the attitude quaternions at the second number of positions closest to the query position time to obtain the attitude quaternions at the query position time. These quaternions are then used to rotate the preset antenna installation vector of the RTK from the body coordinate system to the navigation coordinate system to obtain the antenna offset. The difference between the UAV's position coordinates at the current location and the antenna offset is used as the corrected observation position of the UAV at the current location.

9. A high-precision positioning method for UAVs based on RTK and inertial navigation according to claim 2, characterized in that, The method for setting the neighborhood window includes: The neighborhood window is 3 time units in size, including each pose time and the two pose times that are closest to each pose time in time sequence.

10. A high-precision positioning system for unmanned aerial vehicles (UAVs) based on RTK and inertial navigation, characterized in that, It includes a processor and a memory, the memory storing at least one instruction, at least one program, code set or instruction set, and when the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor, it implements the steps of the high-precision positioning method for UAVs based on RTK and inertial navigation as described in any one of claims 1-9.

Images