A method for correcting the flight trajectory of a target drone based on multi-source fusion navigation

Abstract

This invention discloses a target drone flight trajectory correction method based on multi-source fusion navigation, belonging to the field of trajectory correction technology. This invention can fully integrate multi-source navigation information such as AI visual navigation, inertial measurement, and barometric altimeter. Through joint analysis of short-term and trend envelopes, it effectively distinguishes between short-term disturbances and long-term deviations, significantly improving the sensitivity and reliability of trajectory anomaly detection. By setting a direction consistency screening and hierarchical weight suppression mechanism, it can suppress false alarms caused by barometric pressure fluctuations or instantaneous noise in low-altitude flight, thereby reducing the false alarm rate. Furthermore, by utilizing envelope difference accumulation evidence and threshold linkage triggering mechanisms, it achieves early warning of trajectory deviations, improving the timeliness of anomaly response in flight missions. Moreover, it can dynamically generate stable and accurate trajectory correction values ​​in a variable flight environment, ensuring the robustness and convergence of the correction process.

Description

Technical Field

[0001] This invention relates to the field of trajectory correction technology, and in particular to a method for correcting the flight trajectory of a target drone based on multi-source fusion navigation. Background Technology

[0002] With the widespread application of unmanned aerial vehicles (UAVs) in military training, range testing, and scientific research verification, the demand for precise control and correction of flight trajectories is increasing. During flight missions, target drones are often affected by various factors such as satellite navigation signal obstruction, multipath interference, cumulative errors in inertial navigation systems, and fluctuations in barometric altimeter measurements, leading to deviations in heading, normal, and altitude, which in turn affect the accuracy of mission completion. In recent years, multi-source navigation fusion technology has become an important research direction for flight trajectory correction because it can comprehensively utilize multiple information sources such as visual navigation, inertial measurement, and barometric altimeter measurements to achieve highly robust and accurate navigation calculations. However, in dynamic and complex environments, effectively identifying and suppressing trajectory deviations caused by environmental disturbances, sensor noise, or abnormal maneuvers, while ensuring real-time performance and achieving precise correction, remains a technological bottleneck in this field.

[0003] CN113095504B discloses a target trajectory prediction system and method. By establishing a target maneuver feature model set and combining forward prediction and backward feedback correction mechanisms, it achieves target trajectory prediction and correction. This method has certain advantages in trajectory modeling and maneuver intent recognition, and is particularly suitable for targets with obvious maneuver patterns. However, trajectory correction relies on calculating the matching degree between the maneuver feature model and the actual trajectory, as well as posterior parameter correction based on a globally important map model, making it difficult to respond promptly to instantaneous disturbances or atypical maneuver deviations occurring during flight. Furthermore, the correction process of this method relies more on adjusting global prediction parameters and lacks a real-time, localized deviation identification and correction mechanism based on multi-source navigation data, resulting in insufficient sensitivity and accuracy in detecting trajectory deviations within a short time window.

[0004] CN111857177B discloses a method, apparatus, device, and medium for generating remotely controlled target commands. This method generates modified equivalent control commands by combining pilot control commands with target dynamics, kinematic characteristics, and the motion characteristics of the simulated target, thus achieving target flight control. This method emphasizes human-machine collaboration and matching simulation characteristics, making it suitable for remote control missions. However, the trajectory correction process relies on manual control and pre-designed control laws, lacking the ability to dynamically analyze multi-source navigation data during flight. It cannot effectively identify and suppress deviations caused by navigation signal disturbances, sensor errors, or sudden environmental changes. Furthermore, this method is more focused on control command-level correction, failing to address refined, multi-stage processing of deviations in flight trajectory across different directions (heading, normal, altitude), and lacking an automated correction mechanism for real-time envelope analysis and trend monitoring. Summary of the Invention

[0005] In view of the problems existing in the correction of target drone flight trajectory in the prior art, the present invention is proposed.

[0006] Therefore, the problem to be solved by this invention is how to achieve high-precision, real-time correction of flight trajectories.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] In a first aspect, the present invention provides a target drone flight trajectory correction method based on multi-source fusion navigation, comprising: S1: determining heading direction deviation, normal direction deviation, and altitude deviation based on multi-source navigation data streams, and generating short-time envelopes and trend envelopes respectively within a sliding time window; S2: calculating envelope differences based on short-time envelopes and trend envelopes, and generating one-way cumulative evidence; when the one-way cumulative evidence exceeds a set cumulative threshold within a continuous time period and the short-time envelope threshold and trend envelope threshold are triggered, generating an early warning flag; and defining the continuous time period as the current time period; S3: when the early warning flag is in a valid state, selecting the time anchor point closest to the current time period from the task segment anchor point set, performing segment integration on the heading parameters and envelope differences from the time anchor point to the current time period, and generating a candidate correction quantity sequence; S4: using the optimal correction quantity as a constraint condition in the multi-source navigation data fusion solution process, and outputting the corrected trajectory data and state estimation update value.

[0009] As a preferred embodiment of the target drone flight trajectory correction method based on multi-source fusion navigation described in this invention, wherein: S1 includes: multi-source navigation data stream including AI visual navigation data, inertial measurement data, and barometric altitude data; calculating heading deviation, normal deviation, and altitude deviation within a sliding time window to initially form a heading deviation sequence, a normal deviation sequence, and an initial altitude deviation sequence; and dynamically correcting the altitude deviation using a hierarchical weight suppression method based on the initial altitude deviation sequence, assigning dynamic weight factors to the altitude deviation according to the altitude change rate to suppress instantaneous peaks caused by barometric pressure fluctuations during low-altitude flight.

[0010] As a preferred embodiment of the target drone flight trajectory correction method based on multi-source fusion navigation described in this invention, step S1 further includes: after weight adjustment, within the sliding time window, performing direction consistency screening on the heading direction deviation sequence, normal direction deviation sequence, and altitude deviation sequence respectively, eliminating disturbance points with opposite directions to adjacent time slices and amplitudes lower than the adjacent median, to obtain corresponding short-time candidate sequences; on the short-time candidate sequences, extracting upper and lower bounds according to preset upper and lower quantiles, and setting dual constraints of the lower limit of envelope width and the upper limit of the rate of change of adjacent time slices to prevent the envelope from being too narrow or instantaneous jumps, outputting the heading short-time envelope, normal short-time envelope, and altitude short-time envelope; within the same sliding time window... Within the time window, the short-time envelope centerline sequence of each deviation direction is taken as the trend sequence, and median sliding smoothing is used to obtain the heading trend sequence, normal trend sequence, and altitude trend sequence. The short-time envelope centerline is the arithmetic mean of the upper and lower bounds of the heading short-time envelope, normal short-time envelope, and altitude short-time envelope, forming three centerline curves that change with time. The local slope and local variance of each trend sequence within the current sliding time window are calculated. The product of the absolute value of the local slope and the slope coefficient and the product of the local variance and the variance coefficient are added to the base offset to obtain the upper offset. The lower offset is calculated using the same method but with the direction negative. The heading trend envelope, normal trend envelope, and altitude trend envelope are generated.

[0011] As a preferred embodiment of the target drone flight trajectory correction method based on multi-source fusion navigation described in this invention, the generation of unidirectional cumulative evidence includes: calculating the centerline difference sequence between the short-term envelope of the heading and the heading trend envelope, the centerline difference sequence between the short-term envelope of the normal and the normal trend envelope, and the centerline difference sequence between the short-term envelope of the altitude and the altitude trend envelope, and using each difference sequence as input for unidirectional cumulative evidence calculation; for the envelope difference sequence in each direction, only the positive difference consistent with the historical deviation trend direction is retained, and the unidirectional cumulative evidence sequence is constructed using an exponential decay accumulation method within the sliding time window.

[0012] As a preferred embodiment of the target drone flight trajectory correction method based on multi-source fusion navigation described in this invention, the setting of the short-time envelope threshold includes: extracting the difference between the maximum and minimum values ​​from the short-time envelope midline sequence of the current sliding time window, and multiplying it by a fluctuation sensitivity coefficient, the resulting value is the short-time envelope threshold; the fluctuation sensitivity coefficient is preset according to the flight mission type; the setting of the trend envelope threshold includes: performing sign consistency detection on the trend envelope slope within the current sliding time window; within the sign consistency segment, superimposing a variance correction term on the mean of the trend envelope slope, and multiplying it by the directional offset limit set by the mission, as the trend envelope threshold; the calculation process of the variance correction term includes: normalizing the value of the local variance within the current sliding time window according to the maximum and minimum values ​​within the corresponding window, and multiplying it by the stability weight calculated from the trend envelope directional consistency.

[0013] As a preferred embodiment of the target drone flight trajectory correction method based on multi-source fusion navigation described in this invention, the generation of the candidate correction quantity sequence includes: using the time anchor point as the starting boundary of the segment integration; extracting the heading parameter sequence and the corresponding centerline difference sequence within the time range from the time anchor point to the current time period; performing an integration operation on the heading parameter sequence based on time decay weight within the time range, and simultaneously performing an integration operation on the centerline difference sequence after trend direction consistency screening, and combining the integration results of the two according to stability weight to form the candidate correction quantity sequence; the heading parameter sequence is the extracted sequence of the heading direction deviation sequence within the time range.

[0014] As a preferred embodiment of the target drone flight trajectory correction method based on multi-source fusion navigation described in this invention, the selection of the optimal correction amount includes: dividing the candidate correction amount sequence into several continuous sub-segments according to a sliding time window, with each sub-segment covering candidate correction amounts of a set of continuous or partially overlapping time slices; for each sub-segment, statistically analyzing the proportion of positive differences between the heading parameter sequence and the corresponding centerline difference sequence after trend direction consistency screening, using this as the trajectory correction consistency index of the sub-segment; multiplying the trajectory correction consistency index of each sub-segment by a stability weight to obtain a weighted consistency value; comparing the weighted consistency values ​​of all sub-segments, selecting the candidate correction amount corresponding to the sub-segment with the highest weighted consistency value as the optimal correction amount.

[0015] As a preferred embodiment of the target drone flight trajectory correction method based on multi-source fusion navigation described in this invention, the output of the corrected trajectory data and state estimation update values ​​includes: taking the optimal correction amount as a constraint input, combining it with the multi-source navigation data stream of the current time period, performing constraint updates in the Kalman-type fusion solution process, generating a corrected trajectory data sequence, and extracting the estimated update values ​​of position, velocity, and attitude, which are then used together with the corrected trajectory data sequence as the output results.

[0016] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the target drone flight trajectory correction method based on multi-source fusion navigation as described in the first aspect of the present invention are implemented.

[0017] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the target drone flight trajectory correction method based on multi-source fusion navigation as described in the first aspect of the present invention are implemented.

[0018] The beneficial effects of this invention are as follows: This invention can fully integrate multi-source navigation information such as AI visual navigation, inertial measurement, and barometric altimeter. Through joint analysis of short-term and trend envelopes, it effectively distinguishes between short-term disturbances and long-term deviations, significantly improving the sensitivity and reliability of trajectory anomaly detection. By setting up directional consistency screening and hierarchical weight suppression mechanisms, it can suppress false alarms caused by barometric pressure fluctuations or instantaneous noise during low-altitude flight. Furthermore, by utilizing envelope difference accumulation evidence and threshold linkage triggering mechanisms, it can achieve early warning of trajectory deviations, improving the timeliness of anomaly response during flight missions. It can also dynamically generate stable and accurate trajectory correction values ​​in variable flight environments, ensuring the robustness and convergence of the correction process. This invention significantly enhances the navigation reliability and mission execution safety of target drones in complex mission environments. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0020] Figure 1 This is a flowchart of a target drone flight trajectory correction method based on multi-source fusion navigation.

[0021] Figure 2 A flowchart for generating one-way cumulative evidence. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Figure 1 This is a flowchart illustrating a target drone flight trajectory correction method based on multi-source fusion navigation according to an embodiment of the present invention. Figure 1 As shown, the target drone flight trajectory correction method based on multi-source fusion navigation includes:

[0026] S1: Determine the heading direction deviation, normal direction deviation, and altitude deviation based on multi-source navigation data streams, and generate short-time envelopes and trend envelopes respectively within the sliding time window.

[0027] The multi-source navigation data stream includes AI visual navigation data, inertial measurement data, and barometric altitude data.

[0028] S1.1: Calculate the heading deviation, normal deviation, and altitude deviation within the sliding time window to initially form the heading deviation sequence, normal deviation sequence, and altitude deviation original sequence.

[0029] Specifically, within the sliding time window, for each time slice, the preset heading obtained from the heading calculation model based on AI visual navigation data is time-by-time differencing with the actual heading output by the inertial measurement unit to obtain the heading deviation value; the actual position of the flight trajectory in the normal direction is time-by-time differencing with the preset trajectory position to obtain the normal deviation value; and the barometric altitude data is converted into altitude values ​​and time-by-time differencing with the preset flight altitude to obtain the altitude deviation value. The three-direction deviation values ​​for each time slice are stored in chronological order, initially forming a heading deviation sequence, a normal deviation sequence, and a raw altitude deviation sequence, which serve as inputs for subsequent differential processing and weight correction. The length and time step of the sliding window can be preset according to the flight mission type to ensure that the data within the window covers minor trajectory fluctuations while smoothing out occasional disturbances.

[0030] S1.2: Based on the original altitude deviation sequence, a hierarchical weighted suppression method is used for dynamic correction. Dynamic weight factors are assigned to the altitude deviation according to the altitude change rate to suppress the instantaneous peak caused by air pressure fluctuations during low-altitude flight.

[0031] In practice, within the sliding time window, for each time slice's altitude deviation value, three factors are first calculated: the altitude change rate, the proportion of the current altitude deviation amplitude to the entire window's altitude deviation range, and the consistency of the deviation change trend between adjacent time slices. The highest weight is assigned when the altitude change rate is small, the deviation amplitude is in the upper range of the window's range, and the trend direction is continuous and consistent; the lowest weight is assigned when the altitude change rate is large, or the deviation amplitude is in the lower range of the window's range, and the trend direction is opposite to adjacent segments; for other intermediate cases, the weight is calculated using linear interpolation based on the altitude change rate and amplitude proportion. This weight is multiplied by the corresponding time slice's altitude deviation value to obtain the weighted altitude deviation. This corrected sequence replaces the original altitude deviation sequence and, together with the heading deviation sequence and normal deviation sequence, serves as the final input for generating the short-time envelope and trend envelope.

[0032] S1.3: After weight adjustment, within the sliding time window, the heading direction deviation sequence, normal direction deviation sequence, and altitude deviation sequence are screened for directional consistency. Disturbance points that are opposite in direction to the adjacent time slices and whose amplitude is lower than the adjacent median are removed to obtain the corresponding short-term candidate sequences.

[0033] The screening results generate short-term candidate sequences for the heading, normal, and altitude. This process eliminates instantaneous disturbances while preserving continuous trend changes, ensuring that only effective deviation fluctuations are considered when generating the short-term envelope.

[0034] S1.4: On the short-time candidate sequence, extract the upper and lower bounds according to the preset upper and lower quantiles, and set the double constraint of the lower limit of the envelope width and the upper limit of the rate of change of adjacent time slices to prevent the envelope from being too narrow or instantaneous jumps, and output the heading short-time envelope, normal short-time envelope and altitude short-time envelope.

[0035] Specifically, the two constraints include: Constraint 1: Lower limit constraint on envelope width, logically requiring that the difference between the upper and lower limits must not be less than the deviation range within the current sliding window multiplied by the width adjustment coefficient. If this is not met, the boundary is extended to the lower limit. Constraint 2: Change rate constraint between adjacent time slices, logically requiring that the increase or decrease of the current time slice envelope boundary must not exceed the increase or decrease of the previous time slice envelope boundary multiplied by the change rate coefficient. If this is exceeded, it is corrected proportionally. The two constraints are combined to generate the final short-term envelope boundary, with the heading, normal, and altitude directions processed independently.

[0036] It can be seen that by constraining the lower limit of the width to ensure that the short-term envelope covers the fluctuations of the real trajectory, the correction error caused by excessive narrowness can be prevented; by constraining the rate of change of adjacent time slices to avoid instantaneous jumps, a continuous and smooth envelope can be formed.

[0037] S1.5: Within the same sliding time window, the short-time envelope midline sequence of each deviation direction is taken as the trend sequence, and median sliding smoothing is used to obtain the heading trend sequence, normal trend sequence, and altitude trend sequence. The trend sequence reflects the medium- and long-term trend of deviation and provides a basis for calculating local offsets.

[0038] S1.6: The short-time envelope centerline is the arithmetic mean of the upper and lower bounds of the heading short-time envelope, the normal short-time envelope, and the altitude short-time envelope, forming three centerline curves that change with time.

[0039] Calculate the local slope and local variance for each time slice within the current sliding time window for each trend sequence; add the product of the absolute value of the local slope and the slope coefficient, and the product of the local variance and the variance coefficient to the base offset to obtain the upper offset; the lower offset is calculated using the same method but with the direction negative; the resulting offset will be dynamically adjusted according to the trend change amplitude (variance) and trend change speed (slope), with the envelope tightly attached during slow changes and automatically widened during rapid changes to avoid over-cutting the actual trajectory changes; generate the heading trend envelope, normal trend envelope, and altitude trend envelope.

[0040] For example, the local slope is calculated as follows:

[0041] Trend sequence Each time slice within the sliding time window Use fixed front and back windows Linear regression is performed on n time slices, where n is a constant and the slope is defined as the coefficient of the fitted line:

[0042]

[0043] in, Indicates the first The deviation direction in time slice The local slope; For trend sequence in time slice Deviation values ​​(heading, normal, or altitude); The mean value over the window period; This represents the trend sequence mean within the window.

[0044] The local variance is calculated as follows:

[0045] Trend base sequence within the same sliding time window Local variance represents the fluctuation range of the deviation within the window, and the calculation formula is:

[0046]

[0047] in, For the first The deviation direction in time slice The local variance; this variance is used to reflect the fluctuation range of the trend within the sliding window and is used to dynamically adjust the envelope offset.

[0048] Based on local slope and local variance, the upper offset is... and Defined as:

[0049]

[0050]

[0051] in, For the first The basic offset in each deviation direction is preset by the task type; The slope coefficient is used to adjust the effect of the rate of change of the trend on the envelope; The variance coefficient is used to adjust for the effect of trend fluctuations on the envelope. The dynamic extension quantity representing the contribution of the rate of trend change. This represents the dynamic expansion amount contributing to the trend fluctuation amplitude.

[0052] S2: Calculate the envelope difference based on the short-term envelope and the trend envelope, and generate one-way cumulative evidence. When the one-way cumulative evidence exceeds the set cumulative threshold within a continuous time period and the short-term envelope threshold and the trend envelope threshold are triggered, generate an early warning indicator; and define the continuous time period as the current time period.

[0053] The generation of one-way cumulative evidence includes the following steps:

[0054] First, the midline difference sequences between the short-term envelope of the heading and the heading trend envelope, the midline difference sequences between the short-term envelope of the normal and the normal trend envelope, and the midline difference sequences between the short-term envelope of the altitude and the altitude trend envelope are calculated separately, and each difference sequence is used as input for the calculation of one-way cumulative evidence.

[0055] The calculation of each median difference sequence is performed according to the time slice correspondence, that is, the short-term envelope median and the trend envelope median are subtracted on the same time slice to ensure data logical alignment, and the output results are used as input for subsequent one-way cumulative evidence construction.

[0056] To ensure the stability of the difference values, for time slices where the trend envelope direction changes abruptly, the moving median of adjacent time slices is used to replace outliers, preventing the difference sequence from being amplified by instantaneous noise. The units of the difference sequence are kept consistent with the original deviation sequence, so that subsequent cumulative calculations do not require unit conversion, and thus it can be directly applied to the construction of unidirectional cumulative evidence.

[0057] Secondly, for the envelope difference sequence in each direction, only the positive difference that is consistent with the historical deviation trend direction is retained, and the accumulated evidence is focused on the abnormal changes in the trend that deviate from the positive direction, eliminating misjudgments of normal fluctuations; and a one-way accumulated evidence sequence is constructed by using an exponential decay accumulation method within the sliding time window.

[0058] Specifically, within the sliding time window, the filtered differences are accumulated using an exponentially decaying accumulation method to generate a one-way cumulative evidence sequence. The exponential decay coefficient ensures that the differences in recent time slices contribute more to the cumulative evidence, while the influence of the differences in more distant time slices gradually weakens. The specific calculation formula is as follows:

[0059]

[0060] in, Indicates time slice Upper The cumulative evidence value in each direction, This represents the corresponding filter difference. The exponential decay coefficient, ranging from 0 to 1, is set according to the mission type. For example, it is 0.9 for precision guidance missions (emphasizing recent data), 0.7 for cruise surveillance missions (balancing historical data), and 0.5 for maneuver training missions (responding quickly to changes). This method ensures that the accumulated evidence reflects both current biases and recent trend changes, achieving dynamic weight fusion.

[0061] In this embodiment of the invention, the setting of the short-time envelope threshold includes: extracting the difference between the maximum and minimum values ​​from the short-time envelope midline sequence of the current sliding time window, and multiplying it by the fluctuation sensitivity coefficient, the resulting value is the short-time envelope threshold; the fluctuation sensitivity coefficient is preset according to the flight mission type.

[0062] In this embodiment of the invention, setting the trend envelope threshold includes: performing sign consistency detection on the trend envelope slope within the current sliding time window; within the sign consistency segment, superimposing a variance correction term on the mean of the trend envelope slope, and multiplying it by the direction offset limit set by the task, as the trend envelope threshold.

[0063] In this embodiment of the invention, the calculation process of the variance correction term includes: normalizing the local variance within the current sliding time window according to the maximum and minimum values ​​within the corresponding window, and multiplying it by the stability weight calculated from the consistency of the trend envelope direction, so as to reflect the dynamic adjustment of the threshold by the trend fluctuation amplitude. This ensures that the threshold adapts to the trend stability, avoids false alarms when the stability is low, and enhances sensitivity when the trend deviates significantly.

[0064] Furthermore, the continuous time period is defined as the current time period and is used to determine the unidirectional cumulative evidence. For example, assuming multi-source navigation data is continuously collected during a flight mission, short-time envelopes and trend envelopes are continuously calculated according to a sliding time window, and differential and cumulative evidence analysis is performed. The timeline includes... Wait, and when and The envelope difference begins to exceed the threshold during the middle period. The continuous time period refers to the time interval from the moment the envelope difference first exceeds the threshold until the end of the current detection cycle or the triggering of a preset condition (e.g., ...). → The current time period is an identified time period that needs to be directly referenced by subsequent steps in this flight data processing flow. Simply stating a continuous time period is merely a statistical result and hasn't been explicitly assigned as the "current" target interval to be processed. However, this interval will be used for task segment backtracking and correction quantity generation, so a unique and callable identifier is necessary. This assigns the identity of "current time period" to the continuous time period, essentially a variable binding. Without this definition, multiple continuous time periods may exist during the flight, and subsequent correction steps might reference the wrong time period, leading to miscalculations in correction quantities. For example… Exceeding the threshold → defined as the current time period. If the threshold is exceeded again, a new current time period is added (overwriting the old one), while S3 only processes the most recent current time period to ensure that the correction amount is consistent with the real-time state.

[0065] S3: When the warning indicator is in a valid state, select the time anchor point closest to the current time period from the task segment anchor point set, perform segment integration on the heading parameters and envelope difference between the time anchor point and the current time period, and generate a candidate correction sequence.

[0066] S3.1: Using the time anchor point as the starting boundary for segment integration, and within the time range from the time anchor point to the current time period, extract the heading parameter sequence and the corresponding centerline difference sequence. The heading parameter sequence is the extracted sequence of the heading direction deviation sequence within the time range.

[0067] The mission segment anchor point set consists of key time points pre-defined during the flight mission planning phase. Each time anchor point corresponds to a location with typical constraints in the flight path, such as a segment turn, the start point of a climb, and the end point of a descent. Time anchor points serve as a reference throughout the navigation correction process, and their selection is based on ensuring the complete coverage and data consistency of the correction calculation interval.

[0068] In this invention, the selection rule for time anchor points is defined as follows: find the time point closest to the current warning time period in the task segment anchor point set, and use this time point as the starting boundary for integration calculation. This ensures that the correction calculation interval covers the entire process from the previous critical point to the current anomaly point, thus avoiding the omission of historical deviation information that affects correction judgment. If the interval between the time anchor point and the current time period is too short, it indicates that the flight changes too rapidly in that segment. In this case, the invention automatically extends the integration interval to the previous anchor point to ensure that the integration interval has sufficient length and avoids the correction amount being unrepresentative due to an excessively small segment.

[0069] It should be noted that the truncation operation must be fully aligned with the heading parameter sequence, that is, the difference must be obtained under the same time index, so as to achieve strict consistency between the two types of sequences in the time dimension.

[0070] S3.2: Within the time range, perform integral calculation on the heading parameter sequence based on time decay weight, and simultaneously perform integral calculation on the centerline difference sequence after trend direction consistency screening. Then, combine the integral results of the two according to stability weight to form a candidate correction quantity sequence.

[0071] In this context, trend direction consistency refers to whether the direction of change of the current median difference is consistent with the overall deviation trend. If the direction of the median difference in a certain time slice is opposite to the trend direction, that data point will be set to zero to avoid misleading effects during integration. The filtered median difference sequence is then integrated based on time decay weights. This method ensures that the integration result consists only of the accumulated deviations consistent with the trend direction, thereby enhancing the reliability and directional indicativeness of the integration result in the generation of corrections.

[0072] S4: The multi-source navigation data fusion solution process uses the optimal correction amount as a constraint condition, and outputs the corrected trajectory data and state estimation update value.

[0073] The selection of the optimal correction amount includes the following steps:

[0074] S4.1: Divide the candidate correction sequence into several continuous sub-segments according to the sliding time window. Each sub-segment covers the candidate corrections of a set of continuous or partially overlapping time slices.

[0075] The basic principle of the partitioning operation is that each sub-segment covers the candidate correction values ​​for consecutive time slices, and there is partial overlap between adjacent sub-segments to ensure that the changing trends of candidate correction values ​​are smoothly reflected in adjacent segments. This invention, through segmentation, decomposes large interval problems into multiple sub-problems, effectively avoiding misjudgments caused by excessively long overall intervals. Through these operations, each sub-segment can be used independently as an evaluation unit for subsequent trajectory correction consistency calculations, thereby improving the precision and stability of optimal correction value selection.

[0076] Once the optimal correction is successfully applied, the current time period identifier, one-way cumulative evidence counter, and warning identifier status are reset.

[0077] S4.2: For each sub-segment, the proportion of positive differences between the heading parameter sequence and the corresponding centerline difference sequence after trend direction consistency screening is used as the track correction consistency index for the sub-segment.

[0078] The specific operation is as follows: Within the time range of this sub-segment, firstly, the heading parameter sequence and the corresponding centerline difference sequence are extracted. Then, using a trend direction consistency screening method, only positive differences whose direction is consistent with the overall trend are retained. Subsequently, the proportion of positive differences in this sub-segment to the total number of differences in this segment is calculated as the track correction consistency index. This index ranges from 0 to 1; a larger value indicates a more concentrated correction direction in this segment, and a higher correction reliability. This avoids the problem of conventional methods that generally only judge the magnitude of the correction while ignoring directional consistency, which can easily lead to incorrect correction judgments under local fluctuations.

[0079] S4.3: Multiply the trajectory correction consistency index and stability weight of each sub-segment to obtain the weighted consistency value.

[0080] The weighted consistency value reflects two dimensions simultaneously: the concentration of the correction direction and the stability level of the flight trajectory. Using only the consistency index may lead to falsely high consistency in low-stability scenarios; using only the stability weight fails to reflect the reliability of the correction direction. This invention combines both to ensure that the evaluation value comprehensively describes the effectiveness of the correction amount.

[0081] S4.4: Compare the weighted consistency values ​​of all sub-segments, and select the candidate correction amount corresponding to the sub-segment with the highest weighted consistency value as the optimal correction amount.

[0082] After calculating the weighted consistency value for all sub-segments, a horizontal comparison is needed. The segment with the highest weighted consistency value is selected as the segment corresponding to the optimal correction amount, and the final optimal correction amount is chosen from the candidate correction amounts for that segment. The specific steps include: first, sorting the weighted consistency values ​​of all sub-segments and selecting the segment with the largest value as the target segment; then, calculating the weighted average value in the candidate correction amount sequence for that segment according to the attenuation weighting method of the time slice, and using this average value as the final optimal correction amount. This method ensures that the selected correction amount reflects the directional consistency characteristics within the local segment while avoiding interference from individual anomalous time slices through weighted averaging.

[0083] This invention uses a comprehensive mechanism of partitioning, weighting, and filtering to ensure the stability and anti-interference of the selection process for the optimal correction amount, making it suitable for complex and ever-changing flight environments.

[0084] Furthermore, outputting the corrected trajectory data and updated state estimates involves the following steps:

[0085] S4.5: Using the optimal correction amount as the constraint input, combined with the multi-source navigation data stream of the current time period, the constraint is updated in the Kalman-type fusion solution process to generate the corrected trajectory data sequence, and the estimated update values ​​of position, velocity and attitude are extracted and used as the output results together with the corrected trajectory data sequence.

[0086] A better approach is to use a pseudo-observation equation to inject the optimal correction as a constraint, and combine it with multi-source navigation data streams collected within the current time period to complete the constraint update operation within a Kalman-type fusion solution framework. This method allows for dynamic correction of the navigation solution's state variables, thereby simultaneously outputting the corrected trajectory data sequence in the solution results and extracting updated estimates of core states such as position, velocity, and attitude.

[0087] This method can reduce the cumulative error caused by a single data source to a certain extent and improve the overall accuracy of trajectory calculation. After completing the constraint fusion calculation, the corrected trajectory data sequence and state estimate update values ​​can be output. The trajectory data includes a corrected three-dimensional position point sequence, used to characterize the actual motion trajectory of the aircraft; the state estimate update values ​​include the optimal estimates of position, velocity, and attitude within the current time period. These outputs not only provide real-time correction information for the navigation system but also provide accurate state support for the upper-level flight control module.

[0088] While outputting the corrected results, it is also necessary to record key auxiliary data for subsequent self-calibration operations on thresholds and window lengths within the time window. Specific recorded content includes: the selection results of time anchor points, the calculation results of cumulative evidence values, and the values ​​of short-term envelope thresholds and trend envelope thresholds. By saving this historical information, the threshold size and window length can be dynamically adjusted in subsequent sliding time window updates, allowing for adaptation to changes in the flight environment.

[0089] This embodiment also provides a computer device applicable to a target drone flight trajectory correction method based on multi-source fusion navigation, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the target drone flight trajectory correction method based on multi-source fusion navigation as proposed in the above embodiment.

[0090] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0091] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the target drone flight trajectory correction method based on multi-source fusion navigation as proposed in the above embodiments.

[0092] In summary, this invention fully integrates multi-source navigation information such as AI visual navigation, inertial measurement, and barometric altimeter. Through joint analysis of short-term and trend envelopes, it effectively distinguishes between short-term disturbances and long-term deviations, significantly improving the sensitivity and reliability of trajectory anomaly detection. By setting up directional consistency screening and hierarchical weight suppression mechanisms, it can suppress false alarms caused by barometric pressure fluctuations or instantaneous noise during low-altitude flight, thereby reducing the false alarm rate. Furthermore, by utilizing envelope difference accumulation evidence and threshold-linked triggering mechanisms, it achieves early warning of trajectory deviations, improving the timeliness of anomaly response during flight missions. Moreover, it can dynamically generate stable and accurate trajectory correction values ​​in variable flight environments, ensuring the robustness and convergence of the correction process. This invention significantly enhances the navigation reliability and mission execution safety of target drones in complex mission environments.

[0093] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for correcting the flight trajectory of a target drone based on multi-source fusion navigation, characterized in that: include: S1: Determine the heading direction deviation, normal direction deviation, and altitude deviation based on multi-source navigation data streams, and generate short-time envelopes and trend envelopes respectively within the sliding time window; S2: Calculate the envelope difference based on the short-term envelope and the trend envelope, and generate one-way cumulative evidence. When the one-way cumulative evidence exceeds the set cumulative threshold within a continuous time period and the short-term envelope threshold and the trend envelope threshold are triggered, generate an early warning indicator; and define the continuous time period as the current time period. S3: When the warning sign is in a valid state, select the time anchor point closest to the current time period from the task segment anchor point set, perform segment integration on the heading parameters and envelope difference from the time anchor point to the current time period, and generate a candidate correction quantity sequence. S4: The multi-source navigation data fusion solution process uses the optimal correction amount as a constraint condition, and outputs the corrected trajectory data and state estimation update value.

2. The target drone flight trajectory correction method based on multi-source fusion navigation as described in claim 1, characterized in that: S1 includes: Multi-source navigation data streams include AI visual navigation data, inertial measurement data, and barometric altitude data; Within the sliding time window, the heading deviation, normal deviation, and altitude deviation are calculated to initially form the heading deviation sequence, normal deviation sequence, and altitude deviation original sequence; Based on the original altitude deviation sequence, a hierarchical weighted suppression method is used for dynamic correction. Dynamic weight factors are assigned to the altitude deviation according to the altitude change rate to suppress the instantaneous peak caused by air pressure fluctuations during low-altitude flight.

3. The target drone flight trajectory correction method based on multi-source fusion navigation as described in claim 2, characterized in that: S1 further includes: After weight adjustment, within the sliding time window, the heading direction deviation sequence, normal direction deviation sequence, and altitude deviation sequence are screened for directional consistency. Disturbance points that are opposite in direction to the adjacent time slices and whose amplitude is lower than the adjacent median are removed to obtain the corresponding short-term candidate sequences. On the short-time candidate sequence, the upper and lower bounds are extracted according to the preset upper and lower quantiles, and the double constraints of the lower limit of the envelope width and the upper limit of the rate of change of adjacent time slices are set to prevent the envelope from being too narrow or instantaneous jumps, and the heading short-time envelope, normal short-time envelope and altitude short-time envelope are output. Within the same sliding time window, the short-time envelope midline sequence of each deviation direction is taken as the trend sequence, and median sliding smoothing is used to obtain the heading trend sequence, normal trend sequence, and altitude trend sequence. The short-time envelope centerline is the arithmetic mean of the upper and lower bounds of the heading short-time envelope, the normal short-time envelope, and the altitude short-time envelope, forming three centerline curves that change with time. Calculate the local slope and local variance of each trend sequence within the current sliding time window; add the product of the absolute value of the local slope and the slope coefficient and the product of the local variance and the variance coefficient to the base offset to obtain the upper offset; the lower offset is calculated using the same method but with the direction reversed. Generate the heading trend envelope, normal trend envelope, and altitude trend envelope.

4. The target drone flight trajectory correction method based on multi-source fusion navigation as described in claim 1, characterized in that: The generation of the one-way cumulative evidence includes: The centerline difference sequences between the short-term envelope of the heading and the heading trend envelope, the centerline difference sequences between the short-term envelope of the normal and the normal trend envelope, and the centerline difference sequences between the short-term envelope of the altitude and the altitude trend envelope are calculated separately, and each difference sequence is used as input for the calculation of one-way cumulative evidence. For each direction of the envelope difference sequence, only the positive difference that is consistent with the historical deviation trend direction is retained, and a one-way cumulative evidence sequence is constructed by exponential decay accumulation within the sliding time window.

5. The target drone flight trajectory correction method based on multi-source fusion navigation as described in claim 4, characterized in that: The setting of the short-time envelope threshold includes: extracting the difference between the maximum and minimum values ​​from the short-time envelope midline sequence of the current sliding time window, and multiplying it by the fluctuation sensitivity coefficient. The resulting value is the short-time envelope threshold. The fluctuation sensitivity coefficient is preset according to the flight mission type. The setting of the trend envelope threshold includes: performing sign consistency detection on the trend envelope slope within the current sliding time window; within the sign consistency segment, superimposing a variance correction term on the mean of the trend envelope slope, and multiplying it by the direction offset limit set by the task, as the trend envelope threshold. The calculation process of the variance correction term includes: normalizing the value of the local variance in the current sliding time window according to the maximum and minimum values ​​in the corresponding window, and multiplying it by the stability weight calculated from the consistency of the trend envelope direction.

6. The target drone flight trajectory correction method based on multi-source fusion navigation as described in claim 1, characterized in that: The generation of the candidate correction sequence includes: The time anchor point is used as the starting boundary for segment integration; Within the time range from the aforementioned time anchor point to the current time period, extract the heading parameter sequence and the corresponding centerline difference sequence; Within the time range, the heading parameter sequence is integrally calculated based on time decay weight, and the centerline difference sequence is integrally calculated after trend direction consistency screening. The integral results of the two are then weighted and synthesized according to stability weight to form a candidate correction quantity sequence. The heading parameter sequence is a truncated sequence of the heading direction deviation sequence within the specified time range.

7. The target drone flight trajectory correction method based on multi-source fusion navigation as described in claim 1, characterized in that: The selection of the optimal correction amount includes: The candidate correction sequence is divided into several continuous sub-segments according to the sliding time window. Each sub-segment covers the candidate corrections of a set of continuous or partially overlapping time slices. For each sub-segment, the proportion of positive differences between the heading parameter sequence and the corresponding centerline difference sequence after trend direction consistency screening is used as the track correction consistency index for the sub-segment. The track correction consistency index and stability weight of each sub-segment are multiplied together to obtain the weighted consistency value; By comparing the weighted consistency values ​​of all sub-segments, the candidate correction amount corresponding to the sub-segment with the highest weighted consistency value is selected as the optimal correction amount.

8. The target drone flight trajectory correction method based on multi-source fusion navigation as described in claim 7, characterized in that: The output corrected trajectory data and state estimation update values ​​include: The optimal correction value is used as a constraint input. Combined with the multi-source navigation data stream of the current time period, the constraint is updated in the Kalman-type fusion solution process to generate a corrected trajectory data sequence. The estimated update values ​​of position, velocity and attitude are extracted and used together with the corrected trajectory data sequence as the output result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the target drone flight trajectory correction method based on multi-source fusion navigation as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the target drone flight trajectory correction method based on multi-source fusion navigation as described in any one of claims 1 to 8.

Images