USV-UAV cooperative positioning method based on dynamic weight adjustment

Abstract

The invention discloses a USV-UAV cooperative positioning method based on dynamic weight adjustment, and relates to the technical field of intelligent equipment positioning, an unmanned ship GNSS signal intensity sensor collects signal quality parameters such as a signal-to-noise ratio and the number of satellites in real time, the parameters serve as dynamic weight adjustment basis, complex environment parameter identification does not need to be relied on, and the positioning accuracy is improved. The problem of weight adjustment lag is solved; secondly, constructing a GNSS, inertial navigation, vision and laser ranging four-source data fusion model, performing fusion calculation on unmanned ship inertial navigation displacement data, unmanned aerial vehicle laser ranging height data and traditional GNSS and vision data by adopting a Kalman filtering algorithm, and when a GNSS signal fails, supplementing displacement information through inertial navigation data and correcting height deviation through laser ranging data so as to obtain a GNSS signal; positioning interruption or drifting is avoided, and the bottleneck of single data source of the existing algorithm is broken through.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent device positioning, and in particular to a USV-UAV cooperative positioning method based on dynamic weight adjustment. BACKGROUND

[0002] In the water operation scene, the unmanned ship often needs to perform tasks in open waters or around islands, and in waters with multiple obstacles. The positioning accuracy directly affects the operation effect. At present, the unmanned ship mainly relies on Global Navigation Satellite System (GNSS) positioning. However, in the areas blocked by high-rise buildings, bridges and islands, the GNSS signal is easily disturbed, and the positioning error can reach more than 10 meters. At the same time, when a single unmanned ship is positioned, it cannot cover a large area, and it is difficult to obtain three-dimensional spatial position information. In the case of sudden situations such as target tracking and accident rescue, the response efficiency is low. To make up for the positioning defects of a single unmanned vehicle, attempts have been made in the prior art to position the unmanned ship and the unmanned aerial vehicle cooperatively. However, the commonly used methods still have obvious shortcomings and have not formed a deep cooperative mechanism. The current mainstream cooperative methods are mainly concentrated in one-way data transmission, that is, the unmanned aerial vehicle collects environmental data such as water surface images and target position information through the camera and radar devices carried by the unmanned aerial vehicle, and transmits the data to the unmanned ship to assist the unmanned ship in optimizing the navigation path, but does not directly participate in the positioning calculation. In the GNSS signal blocked area, the unmanned ship is still given a high weight to its own positioning data, which leads to limited improvement in cooperative positioning accuracy. In the prior art, for the cooperative positioning method, for example, in the patent with the patent number CN114928881B and the name "Cooperative positioning system and positioning method based on ultra-wideband and visual intelligent device", a cooperative positioning system and positioning method based on ultra-wideband device and visual intelligent device are disclosed. The cooperative positioning system can integrate sensors on a single platform to make up for the shortcomings of the global navigation positioning system in terms of availability, reliability and vulnerability. At the same time, data sharing between multiple platforms provides further performance improvement for positioning. In the patent with the patent number CN103197279B and the name "Positioning method of mobile target cooperative positioning system", a mobile target cooperative positioning system and positioning method are disclosed. The positioning system includes a three-axis laser gyroscope, a three-axis optical fiber acceleration sensor, a data processor, an inertial navigation parameter communication module, an ultra-wideband wireless transceiver, a wireless parameter communication module and a cooperative operation processing unit. The data processor collects the attitude angle and acceleration of the mobile target. The ultra-wideband wireless receiver obtains the signal time difference and angle parameter of the target and the transmitter. The multi-sensor parameters are transmitted to the cooperative operation processing unit through the corresponding serial port module. The target position and attitude are calculated and output. The invention integrates the advantages of strapdown inertial navigation and ultra-wideband wireless sensor positioning methods, which can be used for positioning the position and attitude of mobile targets in indoor or mine environments, and can be used for positioning robots, vehicles or mine personnel. In view of the defects of the existing cooperative positioning algorithm, some researches try to optimize the data fusion logic to realize cooperative positioning, but there are still key technical bottlenecks. Some improved algorithms introduce a scene recognition module to try to adjust the data weight according to the environmental characteristics, but the environmental parameters such as the height of the shielding object and the strength of the electromagnetic interference are difficult to obtain accurately in real time, resulting in a lag in weight adjustment. Another scheme tries to use Kalman filter algorithm to fuse multi-source data, but it is limited to fusing unmanned ship GNSS and unmanned aerial vehicle vision data, and does not include inertial navigation, laser ranging and other key data. When the GNSS core fails, the algorithm is prone to positioning interruption or drift, which cannot guarantee the continuity of positioning. These technical bottlenecks make it difficult for the existing cooperative positioning scheme to achieve stable and high-precision positioning effect in complex water scene. SUMMARY

[0003] The purpose of the present application is to provide a USV-UAV cooperative positioning method based on dynamic weight adjustment to solve the problems in the background art.

[0004] In order to achieve the above purpose, the present application provides the following technical scheme: a USV-UAV cooperative positioning method based on dynamic weight adjustment, the positioning method comprising the following steps: Step 1: The cloud control center sends a start command to the unmanned surface vessel and drone, collects initial GNSS positioning data and uploads it to the cloud, and simultaneously collects signal quality parameters and the number of visible satellites through the GNSS signal strength sensor, and calculates the time per unit of time. Displacement data within the vessel, and synchronously collect the heading angle of the unmanned vessel. The UAV subsystem collects its own GNSS positioning data; Step 2: The cloud-based positioning algorithm server calculates the weight of the GNSS data uploaded by the unmanned surface vessel (USV) based on the signal quality parameter Q. Weighting of UAV-assisted positioning data ; Step 3: Cloud-based GNSS data uploaded by the drone Calculate the auxiliary positioning data of the unmanned vessel. ; Step 4: The cloud-based Kalman filter algorithm is used to process the GNSS data of the unmanned vessel. Auxiliary positioning data and inertial navigation displacement data The data is then fused to obtain the final localization result. ; Step 5: Repeat steps 1-4 every 30 seconds in the cloud, outputting the final positioning result of the unmanned vessel in real time. This information is then fed back to unmanned ships and drones for navigation control and position correction.

[0005] Preferably, in step 2, the cloud-based positioning algorithm server calculates the weight of the unmanned surface vessel's GNSS data based on the signal quality parameter Q uploaded by the unmanned surface vessel. Weighting of UAV-assisted positioning data The weight redistribution formula is: when , , ,when , , ,when , , .

[0006] Preferably, in step 3: the cloud uses the drone GNSS data uploaded by the drone. Calculate the auxiliary positioning data of the unmanned vessel. The expression for calculating latitude and longitude is: , The height calculation expression is: ,in, The relative position difference between the unmanned ship and the drone in the eastward direction. Latitude of the drone For the Earth's radius, These are the corrected longitude and latitude of the unmanned vessel. The relative position difference between the unmanned ship and the drone in the north direction. This is a correction value for atmospheric refraction error, based on the drone's altitude. Retrieve from the preset correction table.

[0007] Preferably, in step 4: during the prediction phase, based on the final positioning result of the previous moment... With inertial navigation displacement The predicted location value at the current moment is expressed as: ,in, Let be the heading angle of the unmanned vessel at the current moment, and let the predicted covariance matrix be . ,in, Let be the state transition matrix.

[0008] Preferably, in step 4, during the update phase, the Kalman filter gain is calculated and expressed as: Combined with observations .

[0009] Preferably, step 5 also includes the UAV flight control unit collecting the remaining battery power in real time. ,when A return-to-home request is sent to the cloud. After receiving the request, the cloud generates a return-to-home route and schedules a backup drone to take off. Steps 1-4 are repeated to continue collaborative positioning. Once the backup drone establishes a data connection with the unmanned vessel, the original drone performs the return-to-home landing process on the deck of the unmanned vessel to recharge.

[0010] Preferably, the normalization formula for the signal quality parameters is: ,in, These are the normalized values ​​of the signal quality parameters. , These are the weighting coefficients. , For maximum signal-to-noise ratio, For real-time signal-to-noise ratio, The maximum number of visible satellites, This represents the number of satellites visible in real time.

[0011] Preferably, step 1 further includes collecting the angular velocity of the unmanned surface vessel. acceleration And calculate the unit time. Displacement data within The calculation formula is: , ,in and The x-axis and y-axis accelerations were collected, and the heading angle of the unmanned vessel was collected simultaneously. .

[0012] Preferably, the UAV subsystem collects its own GNSS positioning data, and the visual and lidar positioning is represented as follows: , ,in, The baseline distance of the binocular cameras. For camera focal length, , The parallaxes along the x and y axes are measured, and the vertical distance between the UAV and the unmanned vessel is collected.

[0013] The technical effects and advantages provided by the present invention in the above technical solution are as follows: (1) Integrating GNSS, inertial navigation, visual recognition and laser ranging multi-source data to break through the scenario limitations of single GNSS positioning and improve the positioning accuracy in obscured areas; (2) Adjust the fusion weight in real time according to the GNSS signal strength of the unmanned vessel to balance positioning accuracy and stability, and avoid positioning interruption caused by the failure of a single data source.

[0014] (3) The UAV uses the unmanned vessel as a ground reference to correct its own positioning error, and the unmanned vessel uses the UAV as a high-altitude perspective to supplement three-dimensional positioning information, forming a two-way collaboration and improving the overall positioning reliability of the system. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0016] Figure 1 This is a flowchart of the positioning method of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example: This example provides a USV-UAV cooperative localization method based on dynamic weight adjustment. Please refer to [link / reference]. Figure 1 As shown, the positioning method includes the following steps: Step 1: The cloud control center sends a start command to the unmanned surface vessel and drone, collects initial GNSS positioning data and uploads it to the cloud, and simultaneously collects signal quality parameters and the number of visible satellites through the GNSS signal strength sensor, and calculates the time per unit of time. Displacement data within the vessel, and synchronously collect the heading angle of the unmanned vessel. The UAV subsystem collects its own GNSS positioning data; Step 2: The cloud-based positioning algorithm server calculates the weight of the GNSS data uploaded by the unmanned surface vessel (USV) based on the signal quality parameter Q. Weighting of UAV-assisted positioning data ; Step 3: Cloud-based GNSS data uploaded by the drone Calculate the auxiliary positioning data of the unmanned vessel. ; Step 4: The cloud-based Kalman filter algorithm is used to process the GNSS data of the unmanned vessel. Auxiliary positioning data and inertial navigation displacement data The data is then fused to obtain the final localization result. ; Step 5: Repeat steps 1-4 every 30 seconds in the cloud, outputting the final positioning result of the unmanned vessel in real time. This information is then fed back to unmanned ships and drones for navigation control and position correction.

[0019] Taking marine environmental monitoring as an example, the implementation process is as follows.

[0020] Equipment Deployment: Select one unmanned surface vessel (USV) equipped with dual-mode GNSS, IMU, and millimeter-wave radar, and one unmanned aerial vehicle (UAV) equipped with a binocular camera, laser rangefinder, and 4G communication module. Deploy a USV-UAV cooperative localization algorithm based on dynamic weight adjustment at the cloud control center. The process noise covariance matrix is ​​as follows: The GNSS observation noise matrix is The auxiliary positioning observation noise matrix is .

[0021] The cloud control center sends start commands to the unmanned surface vessel (USV) and the drone. The USV enters the operating area, and the drone takes off from the USV's deck. The two establish a data connection via wireless communication modules and collect initial GNSS positioning data, denoted as [data missing]. , The data is uploaded to the cloud. Simultaneously, signal quality parameters and the number of visible satellites are collected using a GNSS signal strength sensor. The normalization formula for the signal quality parameters is as follows: ,in , These are the weighting coefficients. ; The maximum signal-to-noise ratio is set to 45 dB. This represents the maximum number of visible satellites. The angular velocity of the unmanned surface vessel is collected via its inertial navigation module. acceleration And calculate the unit time. Displacement data within The calculation formula is as follows: .

[0022] in and The x-axis and y-axis accelerations were collected by the inertial navigation module, and the heading angle of the unmanned vessel was collected simultaneously. .

[0023] The UAV subsystem collects its own GNSS positioning data through the GNSS positioning module. Visual and lidar positioning can be represented as... , ,in, The baseline distance of the binocular cameras. For camera focal length, , The parallaxes are measured along the x-axis and y-axis, respectively. The vertical distance between the UAV and the unmanned surface vessel is collected using a laser ranging module. .

[0024] The cloud-based positioning algorithm server calculates the weight of the GNSS data uploaded by the unmanned surface vessel (USV) based on the signal quality parameter Q. Weighting of UAV-assisted positioning data Weighting formula when , , ; when , , ; when , , ; Cloud-based drone uploads Calculate the auxiliary positioning data of the unmanned vessel. Latitude and longitude calculation: ; ; Height calculation: ;in, This is a correction value for atmospheric refraction error, based on the drone's altitude. Retrieve from the preset correction table.

[0025] The cloud-based Kalman filter algorithm is used to process GNSS data from unmanned vessels. Auxiliary positioning data and inertial navigation displacement data The data is then fused to obtain the final localization result. .

[0026] Prediction phase: Based on the final positioning result of the previous moment With inertial navigation displacement Predict the current location value: ,in, Let be the heading angle of the unmanned vessel at the current moment. The predicted covariance matrix is: ,in, Let be the state transition matrix.

[0027] During the update phase, calculate the Kalman filter gain: Combined with observations: .

[0028] Operational Procedure: The unmanned surface vessel (USV) departs from the port and heads to the target monitoring area; the unmanned aerial vehicle (UAV) takes off from the USV's deck, flies to an altitude of 80 meters above the USV, and initiates visual recognition and laser ranging; the USV's GNSS signal strength sensor collects signal-to-noise ratio, number of visible satellites, calculates signal quality parameters, and corresponding weights; the UAV collects its own GNSS data, uses visual positioning to obtain the ship's x and y coordinates, and performs laser ranging. The latitude, longitude, and altitude of the auxiliary positioning data are calculated, and the cloud uses Kalman filtering to fuse the USV's GNSS data and auxiliary positioning data to obtain the final USV positioning. When the UAV's remaining battery power drops to 20%, the cloud schedules the UAV to return to the USV for charging, while a backup UAV is activated to continue collaborative positioning; after the operation is completed, the cloud generates a complete high-precision positioning trajectory for the USV.

[0029] In this application: The process noise covariance matrix is ​​set based on the following: In the method of this invention, the process noise mainly comes from the unexpected motion disturbances (such as water flow impact, heading deviation or speed fluctuation caused by wind and waves) that the unmanned vessel experiences during operation, as well as the cumulative error (such as zero bias drift) generated by the inertial measurement unit (IMU) due to long-term operation.

[0030] The values ​​of the matrix need to be determined based on the actual motion state of the unmanned vessel and the hardware characteristics of the IMU: Typical scenario example: When operating in calm waters (such as inland lakes), the unmanned surface vessel (USV) is less affected by external interference, and the noise of the IMU's accelerometer and gyroscope is low. In this case, the parameters in the matrix that describe the uncertainty of position and velocity can be set to small values ​​(for example, the parameter reflecting the uncertainty of position corresponds to an actual error range of about 0.1 to 0.5 meters, and the uncertainty of velocity is about 0.01 to 0.05 meters / second). However, in complex waters (such as nearshore tidal areas or offshore waters during typhoon weather), the USV is easily affected by strong currents or waves, and its motion state changes drastically. It is necessary to increase the relevant parameters in the matrix (for example, the uncertainty of position is expanded to 0.5 to 2 meters, and the uncertainty of velocity is expanded to 0.05 to 0.2 meters / second) to reflect stronger dynamic disturbances.

[0031] Setting logic: Specific values ​​need to be determined through actual testing—for example, let the unmanned ship move at a constant speed in a straight line under specific conditions, compare the deviation between the Kalman filter predicted position and the actual position, and gradually adjust the matrix parameters until the filtering result is stable and the error is minimized.

[0032] The observation noise matrix is ​​set based on the following: Observation noise mainly originates from three types of data: GNSS positioning data (affected by satellite signal blockage, ionospheric interference, etc.), data from the laser rangefinder and visual sensor onboard the UAV (affected by lighting conditions, target reflectivity, etc.), and auxiliary positioning data (such as calculation errors when the UAV uses its own GNSS to infer the position of the unmanned vessel). The matrix values ​​need to be calibrated according to the actual accuracy of each observation source. GNSS observation noise: In open water (≥8 visible satellites), the horizontal positioning accuracy of GNSS is usually 1-3 meters, and the vertical accuracy is 2-5 meters. Therefore, the noise parameters corresponding to the GNSS position observation in the matrix can be set to approximately 1-4 square meters in the horizontal direction (corresponding to a standard deviation of 1-2 meters) and approximately 4-25 square meters in the vertical direction (corresponding to a standard deviation of 2-5 meters). If the GNSS signal is blocked (e.g., ≤4 visible satellites), the positioning error will increase significantly (horizontally it may reach 5-10 meters). In this case, the corresponding parameters in the matrix need to be increased (e.g., 10-100 square meters in the horizontal direction).

[0033] Noise in laser ranging and visual observation: When the laser rangefinder on the UAV measures the vertical distance of the unmanned vessel, the typical accuracy is ±0.05 to 0.2 meters (affected by the surface material and reflectivity of the target). Therefore, the noise parameter for the height observation in the matrix can be set to 0.0025 to 0.04 square meters (standard deviation 0.05 to 0.2 meters). The accuracy of visual positioning (such as the estimation of the unmanned vessel's planar position by a binocular camera) is usually ±0.1 to 0.5 meters (affected by illumination and image clarity). The noise parameter for the planar position observation in the matrix can be set to 0.01 to 0.25 square meters (standard deviation 0.1 to 0.5 meters).

[0034] Auxiliary positioning observation noise: When the UAV calculates the latitude and longitude of the unmanned vessel using its own GNSS data and the relative position difference (such as the east-west / north-south distance between the UAV and the unmanned vessel), the error mainly comes from the measurement deviation of the relative position difference (such as the calibration error of ±0.5 to 2 meters in the actual east-west distance between the UAV and the unmanned vessel). Therefore, the noise parameters corresponding to the auxiliary positioning latitude and longitude in the matrix need to be set according to the calibration accuracy of the relative position difference (for example, about 0.25 to 4 square meters in the horizontal direction).

[0035] Atmospheric refraction error is a correction factor that needs to be considered when UAVs calculate the latitude and longitude of unmanned vessels using GNSS altitude data (because the path of electromagnetic waves bends when propagating in the atmosphere, causing a deviation between the measured altitude and the actual geometric altitude). In the method of this invention, this correction value is obtained by querying a "preset correction table," and the specific logic is as follows: The correction table is generated based on the internationally recognized "Hopfield model" or "Saastamoinen model" (both are classic calculation models for atmospheric refraction error, and the refraction correction amount can be inferred from the UAV altitude). In practical applications, the correction table can be generated in advance using simulation software (such as the MATLAB Atmospheric Model Toolbox) or measured data (such as collecting the difference between GNSS altitude and the actual altitude at a fixed altitude point with known geographic coordinates).

[0036] The correction table uses the drone's altitude (in meters) as the horizontal axis and the corresponding refraction error correction value (in meters) as the vertical axis, listing specific values ​​at fixed altitude intervals (e.g., every 10 meters or 50 meters). For example, when the drone's altitude is 100 meters, the correction value might be +0.02 meters; at 500 meters, the correction value might be +0.1 meters; and at 1000 meters, the correction value might be +0.2 meters (the specific values ​​need to be calculated based on the actual model or calibrated by actual measurement).

[0037] When calculating unmanned surface vessel (USV) assisted positioning data, the cloud-based algorithm reads the current altitude of the USV, directly obtains the corresponding refraction error correction value by looking up a table (without needing to calculate the model in real time), and substitutes it into the latitude and longitude calculation formula to compensate for the impact of altitude error on planar position.

[0038] The relative position difference between the unmanned surface vessel (USV) and the unmanned aerial vehicle (UAV) (including east-west and north-south differences) is a key parameter for the UAV to calculate the latitude and longitude of the USV using its own GNSS data (used to convert the absolute positioning data of the UAV into relative reference frame data for the USV). In the method of this invention, this relative position difference is obtained through one of the following two typical methods: Pre-calibration method (recommended): Before the first collaborative operation between the unmanned surface vessel (USV) and the unmanned aerial vehicle (UAV), the ground control center guides both to complete relative position measurements while stationary at a known location. For example, the USV is moored at a fixed dock, and the UAV hovers at a specific position directly above or to the side of the USV (e.g., 10 meters vertically above and 5 meters east-west offset). Using a high-precision GNSS receiver (e.g., centimeter-level RTK equipment), the absolute latitude and longitude of the USV and the UAV are measured respectively. The east-west and north-south coordinate differences between them (in meters) are calculated, and these differences are stored as fixed parameters in a cloud-based algorithm. Subsequent operations directly use this pre-calibration value, eliminating the need for repeated calculations.

[0039] Real-time perception method (backup): If no pre-calibration conditions are available, specific marker points on the unmanned surface vessel (USV) (such as high-contrast color blocks or reflective markings on the hull) can be identified in real time using the visual sensors (e.g., binocular cameras) or LiDAR on the UAV. The 3D coordinates of the marker points are then calculated by combining the UAV's own pose (obtained by fusing IMU and GNSS data), and the relative position difference between the USV and the UAV can be deduced. For example, the UAV can capture images of the USV marker points using its binocular camera, calculate the horizontal (east-west / north-south) and vertical (height) distances of the marker points relative to the UAV using the parallax principle, and then deduce the relative coordinate difference between the USV and the USV based on the UAV's current position. This method relies on the accuracy of the visual / LiDAR sensors and environmental conditions (e.g., sufficient lighting and clearly visible marker points), and is suitable for temporary operational scenarios without pre-calibration conditions.

[0040] Comparison of positioning errors under GNSS signal obstruction conditions Test scenario: In a near-shore island obstruction area (simulating a weak / failed GNSS signal environment), the unmanned vessel's navigation area contains obstacles such as tall buildings and mountains, causing GNSS signal obstruction at certain times (measured signal-to-noise ratio SNR < 30dB, visible satellite count < 4). The test is divided into two groups: Control group: The traditional single GNSS positioning method was used (without UAV collaboration and without dynamic weight adjustment), which directly relied on raw GNSS data to output positioning results; Experimental group: The method of this invention (dynamic weight adjustment + four-source data fusion + Kalman filtering) is used to calculate the results by fusing UAV-assisted positioning with UAV inertial navigation, vision and laser ranging data.

[0041] Test results: During the GNSS signal obstruction period (lasting approximately 120 seconds), the control group exhibited significant fluctuations in positioning error, with an average positioning error of 4.2 meters (range 3.5–6.8 meters) and a vertical error of 5.1 meters (range 4.0–7.2 meters). Furthermore, due to signal loss, the control group experienced multiple positioning drifts (the trajectory deviated from the actual path by more than 10 meters). In the same time period, the experimental group increased the weight of UAV-assisted positioning data to the dominant position through dynamic weight adjustment (GNSS weight was reduced to 0), and combined Kalman filtering to fuse and correct the inertial navigation displacement and laser ranging altitude data. The horizontal positioning error was reduced to 1.1 meters (range 0.8 to 1.5 meters), the vertical error was reduced to 0.9 meters (range 0.7 to 1.2 meters), and the trajectory continuity and the degree of consistency with the actual path were significantly improved (maximum deviation not exceeding 2 meters).

[0042] Conclusion: Under conditions of severely limited GNSS signals, the method of this invention reduces the positioning error from 4-6 meters to 1-2 meters by dynamic weight allocation and multi-source data fusion, effectively avoiding positioning interruption and drift problems.

[0043] Comparative Experiment: Compared with a single GNSS fusion method and a fusion method without dynamic weights Experimental Design: Single GNSS positioning method: using only the unmanned vessel's own GNSS data (no UAV collaboration, no dynamic weight adjustment); The fusion method did not use dynamic weights: GNSS, inertial navigation, vision, and laser ranging data were fused, but the weights were fixed (e.g., GNSS weight was fixed at 70%, and UAV-assisted weight was fixed at 30%), and were not dynamically adjusted according to SNR. The method of this invention is: dynamic weight adjustment + four-source fusion + Kalman filtering (same as Example 1).

[0044] Test conditions: Simulated multiple scenarios (open water SNR>40dB, island obstruction area SNR<30dB, transition area 30dB≤SNR<40dB), and the average positioning error and number of positioning interruptions for each method were statistically analyzed (an error greater than 5 meters for 3 consecutive seconds was considered an interruption). The experimental results are shown in Table 1: Table 1: Experimental Results

[0045] In open water, the method of this invention has similar errors to the traditional method (due to good GNSS signal, the dynamic weights are biased towards GNSS primary positioning), but its stability is slightly better. In obstructed areas and transition zones, the method of this invention uses dynamic weights to elevate UAV-assisted positioning (high reliability) to the dominant data source, and integrates inertial navigation / laser ranging correction, significantly reducing errors (from 5.8 meters to 1.1 meters in obstructed areas) and the number of interruptions (from 8 times to 1 time). Its overall performance is better than that of single GNSS and fixed-weight fusion methods. The fixed-weight fusion method does not adjust the data source priority according to SNR, and still relies on some GNSS data (weight 30%) in obstructed areas, resulting in higher errors and more interruptions than the method of this invention.

[0046] Through the above quantitative examples and comparative experiments, it can be seen that the method of the present invention, in complex scenarios with weak / failed GNSS signals, accurately allocates the priority of data sources through a dynamic weight adjustment mechanism (prioritizing the use of highly reliable UAV-assisted positioning), and combines Kalman filtering to fuse multi-source data (inertial navigation displacement, laser ranging altitude, etc.), achieving a leap in positioning error from "meter level" to "sub-meter level" (typical value 1-2 meters), while avoiding positioning interruption and drift problems. It can provide clear implementation reference and expected results for those skilled in the art.

[0047] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0048] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for USV-UAV cooperative positioning based on dynamic weight adjustment, characterized in that: The positioning method comprises the following steps: Step 1: the cloud control center sends start instructions to the unmanned ship and the unmanned aerial vehicle, collects initial GNSS positioning data and uploads it to the cloud, collects signal quality parameters and the number of visible satellites through the GNSS signal strength sensor, and calculates the displacement data in the unit time synchronously collects the heading angle of the unmanned ship , and the unmanned aerial vehicle subsystem collects its own GNSS positioning data; Step 2: The cloud positioning algorithm server calculates the weight of the unmanned ship GNSS data according to the signal quality parameter Q uploaded by the unmanned ship With unmanned aerial vehicle auxiliary positioning data weight ; Step 3: Cloud computes the auxiliary positioning data of the unmanned ship based on the GNSS data uploaded by the unmanned aerial vehicle ;​ Step 4: The cloud uses Kalman filtering algorithm to fuse GNSS data, auxiliary positioning data and inertial navigation displacement data of the unmanned ship to obtain the final positioning result ;​​​ Step 5: The cloud repeats steps 1-4 every 30 seconds, and outputs the final positioning result of the unmanned ship in real time And feedback to the unmanned ship and the unmanned aerial vehicle for navigation control and position correction.

2. The USV-UAV cooperative positioning method based on dynamic weight adjustment according to claim 1, characterized in that: In step 2, the cloud positioning algorithm server calculates the unmanned ship GNSS data weight according to the signal quality parameter Q uploaded by the unmanned ship With unmanned aerial vehicle auxiliary positioning data weight , the weight redistribution formula is: when , , When , , When , , .

3. The USV-UAV cooperative positioning method based on dynamic weight adjustment according to claim 2, characterized in that: In step 3: The cloud uses the drone GNSS data uploaded by the drone. Calculate the auxiliary positioning data of the unmanned vessel. The expression for calculating latitude and longitude is: , The height calculation expression is: ,in, The relative position difference between the unmanned ship and the drone in the eastward direction. Latitude of the drone For the Earth's radius, These are the corrected longitude and latitude of the unmanned vessel. The relative position difference between the unmanned ship and the drone in the north direction. This is a correction value for atmospheric refraction error, based on the drone's altitude. Retrieve from the preset correction table.

4. The USV-UAV cooperative positioning method based on dynamic weight adjustment according to claim 3, characterized in that: In step 4: in the prediction stage, based on the final positioning result at the last time , and the inertial displacement The positioning value at the current time is predicted, expressed as: , wherein is the heading angle of the unmanned ship at the current time, and the prediction covariance matrix is , wherein is the state transition matrix.

5. The USV-UAV cooperative positioning method based on dynamic weight adjustment according to claim 3, characterized in that: In step 4, in the update phase, the Kalman filter gain is computed, denoted as: , combining the observation .

6. The USV-UAV cooperative positioning method based on dynamic weight adjustment according to claim 1, characterized in that: Step 5 also includes the UAV flight control unit collecting the remaining power in real time When Send the return request to the cloud, the cloud receives the request, generates the return route and dispatches the standby UAV to take off, repeats steps 1-4 to continue cooperative positioning, and when the standby UAV establishes data connection with the unmanned ship, the original UAV performs the return landing process on the unmanned ship deck.

7. The USV-UAV cooperative positioning method based on dynamic weight adjustment according to any one of claims 1-6, characterized in that: The signal quality parameter normalization processing formula is: Wherein, is a signal quality parameter normalized value, , is a weight coefficient, , is a maximum signal-to-noise ratio, is a real-time signal-to-noise ratio, is a maximum number of visible satellites, is a real-time number of visible satellites.

8. The USV-UAV cooperative positioning method based on dynamic weight adjustment according to claim 7, characterized in that: Step 1 also includes collecting the angular velocity of the unmanned surface vessel. acceleration And calculate the unit time. Displacement data within The calculation formula is: , ,in and The x-axis and y-axis accelerations were collected, and the heading angle of the unmanned vessel was collected simultaneously. .

9. The USV-UAV cooperative positioning method based on dynamic weight adjustment according to claim 7, characterized in that: The drone subsystem collects its own GNSS positioning data, vision and laser radar positioning is expressed as: , , , wherein, is the baseline distance of the binocular camera, , are the x-axis and y-axis parallax respectively, and the vertical distance between the drone and the unmanned ship is collected.

Images