A scene adaptive combined positioning method and system

Abstract

This invention relates to the field of radio navigation technology, and more particularly to a scene-adaptive integrated positioning method and system. The method includes: when the unmanned vehicle's GNSS receiver is working normally, using inertial navigation and GNSS integrated navigation to output high-precision position results; calculating high-precision ranging values ​​between every two unmanned vehicles, and using these values ​​along with the data link ranging values ​​between every two unmanned vehicles as a training dataset, obtaining a data link ranging error correction model through machine learning; when the unmanned vehicle's GNSS is not working normally, calculating high-precision ranging values ​​between every two unmanned vehicles other than the unmanned vehicle itself, obtaining the data link ranging error of the unmanned vehicle based on the data link ranging error correction model, correcting the unmanned vehicle's data link ranging value, and obtaining a high-precision data link ranging value; and obtaining a high-precision position based on the high-precision data link ranging value using the inertial navigation and data link integrated navigation method. This invention improves the reliability and accuracy of navigation information.

Description

Technical Field

[0001] This invention relates to the field of radio navigation technology, and in particular to a scene-adaptive combined positioning method and system. Background Technology

[0002] Positioning and navigation technology is a core supporting technology in fields such as intelligent transportation, unmanned vehicle inspection, and indoor robots. GNSS has become a conventional positioning method due to its high positioning accuracy and low cost. However, in complex environments such as urban canyons, underground tunnels, and large indoor stadiums, satellite navigation signals are easily affected by blockage, transmission, and multipath effects, resulting in weak signals, loss of lock, or even complete failure. This leads to a significant decrease in GNSS positioning accuracy, or even renders the system inoperable. Inertial navigation systems (INS) have the advantages of complete autonomy, no dependence on external signals, high sampling frequency, and fast dynamic response, enabling continuous calculation of vehicle attitude, velocity, and position. However, the positioning error of INS accumulates exponentially over time, and its use alone cannot meet the accuracy requirements for long-term positioning.

[0003] Existing technologies often employ positioning methods that combine inertial navigation systems (INS) with single or multiple source sensors such as satellite navigation and barometric altimeters, using filtering and fusion to improve positioning accuracy. However, these methods cannot effectively suppress the accumulation of errors in the INS when external auxiliary positioning signals completely fail, leading to a rapid decline in positioning accuracy. Furthermore, existing combined positioning methods are mostly single-terminal multi-sensor fusion, lacking interaction and collaboration between multiple nodes. In complex environments with partial obstruction, they struggle to achieve comprehensive perception and positioning, exhibiting insufficient anti-interference capabilities and robustness. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a scene adaptive combined positioning method and system. When GNSS is available for the unmanned vehicle, the data link ranging value is corrected by using GNSS information; when GNSS is unavailable for the unmanned vehicle, the data link ranging value is corrected by using a previously established data link ranging error correction model, thereby realizing inertial navigation error correction and improving the reliability and accuracy of navigation information.

[0005] This invention is achieved through the following technical solution: A scene-adaptive integrated localization method includes the following steps: S1: When the GNSS receiver of the unmanned vehicle is working normally, the high-precision position result is output by using a combination of inertial navigation and GNSS. S2: Collect GNSS pseudorange information of all unmanned vehicles, calculate the GNSS high-precision ranging value between every two unmanned vehicles, and simultaneously collect the data link ranging value between every two unmanned vehicles. Use the GNSS high-precision ranging value between every two unmanned vehicles and the data link ranging value between every two unmanned vehicles as training datasets. Use a long short-term memory network model to obtain a data link ranging error correction model through machine learning. S3: When the GNSS of this unmanned vehicle is not working properly, calculate the high-precision GNSS ranging value between every two unmanned vehicles other than this one, and at the same time collect the data link ranging value between every two unmanned vehicles. Based on the data link ranging error correction model, obtain the data link ranging error of this unmanned vehicle, correct the data link ranging value of the unmanned vehicle, and obtain the high-precision data link ranging value. S4: Based on high-precision data link ranging values, a combined inertial navigation and data link navigation method is used to obtain high-precision position.

[0006] In the optimized version, in step S1, when the GNSS receiver of the unmanned vehicle is working normally, a loosely combined model of inertial navigation and GNSS is used to output high-precision position results.

[0007] Furthermore, in step S1, when the unmanned vehicle's GNSS receiver is working normally, the method for outputting high-precision position results using inertial navigation and GNSS combined navigation is as follows: S111: Calculate the three-dimensional attitude error, three-dimensional velocity error, three-dimensional position error, three-dimensional gyroscope drift error, and three-dimensional zero-bias error of the inertial navigation system according to equation (1): (1); in: This represents the state variables of the inertial navigation and GNSS combined model. Indicates the heading angle error. Indicates the pitch angle error. Indicates the roll angle error. This indicates the eastward velocity error of the inertial navigation system. This indicates the northbound velocity error of the inertial navigation system. This indicates the inertial navigation system's upward velocity error. This indicates the positional error in the latitude direction of the inertial navigation system. This indicates the positional error in the longitude direction of the inertial navigation system. This indicates the position error in the altitude direction of the inertial navigation system. Inertial navigation Axial gyroscope drift error, Inertial navigation Axial gyroscope drift error, This represents the gyroscope drift error in the z-axis direction of the inertial navigation system. Inertial navigation Add zero offset error to the axial direction. Inertial navigation Add zero offset error to the axial direction. This indicates the zero-bias error added to the z-axis direction of the inertial navigation system; S112: Establish the observation model as Equation (2), and obtain the position and velocity errors of the inertial navigation and GNSS combined model based on the observation model. Then, after error compensation, output the high-precision position results: (2); in: This indicates that the inertial navigation system outputs the northeast-sky velocity vector. This indicates the three-dimensional position output by the inertial navigation system. This indicates that the satellite's output velocity vector is in the northeast direction. This indicates the three-dimensional position output by the satellite navigation system. This indicates the velocity error between inertial navigation and satellite navigation. This indicates the positional error between the inertial navigation system and the satellite navigation system.

[0008] In the optimized step S2, the acquisition frequency of GNSS pseudorange information for all unmanned vehicles is 1 Hz.

[0009] In the optimized step S2, the data link ranging value acquisition frequency between every two unmanned vehicles is 1 Hz.

[0010] In the optimized step S2, the GNSS high-precision ranging value between every two unmanned vehicles is calculated using pseudorange double difference.

[0011] Furthermore, in step S2, the GNSS high-precision ranging value between every two unmanned vehicles is calculated according to equation (3): (3); in: Indicates pseudo-distance double difference, This indicates the first driverless car and the... The unit vector between GNSS stars, This indicates the first driverless car and the... The unit vector between GNSS stars, This represents the transformation matrix from the carrier coordinate system to the ECEF coordinate system. This represents the GNSS high-precision ranging vector between every two unmanned vehicles. Indicates the first antenna, the second antenna, and the third antenna. Measurement noise generated between GNSS satellites Indicates the first antenna, the second antenna, and the third antenna. Measurement noise generated between GNSS satellites Indicates the first GNSS stars in Position coordinates along the axis This indicates the carrier obtained by multi-navigation sensor fusion under ECEF. Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis This indicates the carrier obtained by multi-navigation sensor fusion under ECEF. Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis This indicates the carrier obtained by multi-navigation sensor fusion under ECEF. Position coordinates along the axis Indicates the carrier and the first The distance between GNSS stars Indicates the first GNSS stars in Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis Indicates the carrier and the first The distance between GNSS stars.

[0012] Furthermore, in step S4, based on the high-precision data link ranging value, the inertial navigation and data link combined navigation method is used to obtain the high-precision position according to equation (4): (4); in: This represents the prior state prediction value of the inertial navigation and data link combination. The error equation transfer matrix represents the combination of inertial navigation and data link. This represents the optimal state estimate of the inertial navigation and data link combination at time k-1. This represents the noise allocation matrix for the inertial navigation and data link combination. This represents a noise sequence representing the combination of inertial navigation and data link. This represents the pseudorange difference between two unmanned vehicles in the inertial navigation and data link integrated navigation algorithm. This represents the high-precision GNSS ranging value between two unmanned vehicles. This represents the high-precision data link ranging value between two unmanned vehicles. The matrix representing the pseudorange difference between two unmanned vehicles in an inertial navigation and data link integrated navigation algorithm. Represents the observation matrix. Indicates the serial number of the driverless vehicle. This indicates the number of driverless cars.

[0013] A scene-adaptive integrated positioning system, characterized in that: it is used to execute a scene-adaptive integrated positioning method as described in any one of the above, which includes a GNSS receiver, an inertial navigation system, a data link device, a data acquisition module, a data link ranging error correction model construction module, and a data processing module; The data acquisition module is used to collect GNSS pseudorange information of all unmanned vehicles and the distance measurement value of the data link between every two unmanned vehicles. The data link ranging error correction model construction module is used to take the GNSS high-precision ranging value between every two unmanned vehicles and the data link ranging value between every two unmanned vehicles as training datasets when the unmanned vehicle GNSS receiver is working normally, and use the long short-term memory network model to obtain the data link ranging error correction model through machine learning. The data processing module is used to calculate the high-precision GNSS ranging value between every two unmanned vehicles using the GNSS pseudorange information of the unmanned vehicle, and to obtain the data link ranging error of the unmanned vehicle based on the data link ranging error correction model. The unmanned vehicle data link ranging value is then corrected using the data link ranging error of the unmanned vehicle to obtain the high-precision data link ranging value.

[0014] Beneficial effects of the invention: The present invention provides a scene-adaptive combined positioning method and system. When GNSS is available for the unmanned vehicle, the data link ranging value is corrected using GNSS information. When GNSS is unavailable for the unmanned vehicle, the data link ranging value is corrected using a previously established data link ranging error correction model. The corrected data link observation value is used to correct the inertial navigation error, thereby improving the reliability and accuracy of navigation information. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall process of the present invention. Detailed Implementation

[0016] A scene-adaptive combined localization method, the overall flowchart of which is shown in Figure 1, specifically includes the following steps: S1: When the GNSS receiver of the unmanned vehicle is working normally, the high-precision position result is output by using a combination of inertial navigation and GNSS. In an autonomous vehicle swarm, each vehicle is typically equipped with inertial navigation, a GNSS receiver, and a data link. Considering the limited power of autonomous vehicles, when the GNSS receiver is functioning normally, inertial navigation errors are corrected through a combination of inertial navigation and GNSS. Simultaneously, high-precision distance measurements between autonomous vehicles are achieved by transmitting GNSS pseudorange information between different vehicles via the data link. This information serves as the observation value for the data link measurements, correcting for data link ranging errors.

[0017] The optimized approach allows for the use of a loosely coupled inertial navigation and GNSS model to output high-precision position results when the unmanned vehicle's GNSS receiver is operating normally.

[0018] Furthermore, when the GNSS receiver of the unmanned vehicle is working normally, the method for outputting high-precision position results using inertial navigation and GNSS combined navigation is as follows: S111: Calculate the three-dimensional attitude error, three-dimensional velocity error, three-dimensional position error, three-dimensional gyroscope drift error, and three-dimensional zero-bias error of the inertial navigation system according to equation (1): (1); in: This represents the state variables of the inertial navigation and GNSS combined model. Indicates the heading angle error. Indicates the pitch angle error. Indicates the roll angle error. This indicates the eastward velocity error of the inertial navigation system. This indicates the northbound velocity error of the inertial navigation system. This indicates the inertial navigation system's upward velocity error. This indicates the positional error in the latitude direction of the inertial navigation system. This indicates the positional error in the longitude direction of the inertial navigation system. This indicates the position error in the altitude direction of the inertial navigation system. Inertial navigation Axial gyroscope drift error, Inertial navigation Axial gyroscope drift error, This represents the gyroscope drift error in the z-axis direction of the inertial navigation system. Inertial navigation Add zero offset error to the axial direction. Inertial navigation Add zero offset error to the axial direction. This indicates the zero-bias error added to the z-axis direction of the inertial navigation system; S112: Establish the observation model as Equation (2), and obtain the position and velocity errors of the inertial navigation and GNSS combined model based on the observation model. Then, after error compensation, output the high-precision position results: (2); in: This indicates that the inertial navigation system outputs the northeast-sky velocity vector. This indicates the three-dimensional position output by the inertial navigation system. This indicates that the satellite's output velocity vector is in the northeast direction. This indicates the three-dimensional position output by the satellite navigation system. This indicates the velocity error between inertial navigation and satellite navigation. This indicates the positional error between the inertial navigation system and the satellite navigation system.

[0019] Here, the satellite navigation outputs the northeastern celestial velocity vector. and satellite navigation output three-dimensional position The observations are for the combined inertial navigation and GNSS model.

[0020] S2: Collect GNSS pseudorange information of all unmanned vehicles, calculate the GNSS high-precision ranging value between every two unmanned vehicles, and collect the data link ranging value between every two unmanned vehicles. Use the GNSS high-precision ranging value between every two unmanned vehicles and the data link ranging value between every two unmanned vehicles as training datasets. Use the Long Short-Term Memory Network (LSTM) model to obtain the data link ranging error correction model through machine learning. Specifically, the acquisition frequency of GNSS pseudorange information of all unmanned vehicles can preferably be 1 Hz, and the acquisition frequency of data link ranging values ​​between two unmanned vehicles can preferably be 1 Hz.

[0021] The optimized GNSS high-precision ranging value between every two unmanned vehicles can be calculated using pseudorange double difference.

[0022] The pseudorange measured by a GNSS receiver is mainly affected by receiver clock error, satellite clock error, the ionosphere, and the troposphere. Therefore, the first... The pseudorange between a GNSS satellite and the receiver can be calculated using the following formula: ; in: Indicates the first The pseudorange between a GNSS satellite and the receiver Indicates the first The actual distance between a GNSS satellite and the receiver. Indicates receiver clock bias. Indicates the first GNSS star clock bias, Indicates the first Ionospheric delay corresponding to each GNSS satellite Indicates the first The tropospheric delay corresponding to each GNSS satellite. Indicates the first Multipath delay of a GNSS satellite Represents the speed of light. Indicates the first Measurement noise of a single GNSS satellite.

[0023] The two antennas receive the first GNSS satellite signal pseudorange single difference for: ; in: Indicates that the two antennas receive the first The pseudorange of the GNSS satellite signal is single-difference. Indicates the first antenna number The actual distance between a GNSS satellite and the receiver. Indicates the second antenna The actual distance between a GNSS satellite and the receiver. This indicates the clock bias of the first antenna receiver. This indicates the clock bias of the second antenna receiver. Indicates the first antenna number Multipath delay of a GNSS satellite Indicates the second antenna Multipath delay of a GNSS satellite This represents the measurement noise generated between the first and second antennas.

[0024] Assuming that in the vehicle coordinate system, the GNSS high-precision ranging vector between every two unmanned vehicles is... The transformation matrix from the carrier coordinate system to the ECEF coordinate system is: The first driverless car and the first Unit vector between GNSS stars Since the two unmanned vehicles are close to each other, they can be considered to have the same unit vector for the GNSS satellite. Therefore, the two antennas receive the first... The pseudorange single difference of the GNSS satellite signal can be rewritten as Assuming a common-view GNSS satellite is selected. As a double-difference object, when both antennas receive real sky signals, pseudorange double-difference... It can be represented as ; Therefore, the GNSS high-precision ranging value between every two unmanned vehicles can be calculated according to equation (3): (3); in: Indicates pseudo-distance double difference, This indicates the first driverless car and the... The unit vector between GNSS stars, This indicates the first driverless car and the... The unit vector between GNSS stars, This represents the transformation matrix from the carrier coordinate system to the ECEF coordinate system. This represents the GNSS high-precision ranging vector between every two unmanned vehicles. Indicates the first antenna, the second antenna, and the third antenna. Measurement noise generated between GNSS satellites Indicates the first antenna, the second antenna, and the third antenna. Measurement noise generated between GNSS satellites Indicates the first GNSS stars in Position coordinates along the axis This indicates the carrier obtained by multi-navigation sensor fusion under ECEF. Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis This indicates the carrier obtained by multi-navigation sensor fusion under ECEF. Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis This indicates the carrier obtained by multi-navigation sensor fusion under ECEF. Position coordinates along the axis Indicates the carrier and the first The distance between GNSS stars Indicates the first GNSS stars in Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis Indicates the carrier and the first The distance between GNSS stars.

[0025] When the GNSS receiver is working normally, pseudorange double-difference can accurately calculate the distance between any two unmanned vehicles (UAVs). Simultaneously, the distance between two UAVs can also be measured via data link, although the error in this measurement is larger than that obtained using GNSS pseudorange double-difference. In this case, the distance measured by GNSS pseudorange double-difference can be used as the observation, which, for ease of understanding, can be considered the true value. The distance between any two UAVs measured via GNSS pseudorange double-difference is taken as the true value, and the distance measured by the data link, which has a larger error, is simultaneously input into a Long Short-Term Memory (LSTM) network model. Through machine learning, a data link ranging error correction model is obtained.

[0026] The Long Short-Term Memory (LSTM) network model consists of a six-layer architecture, as follows: 1. Input layer: Used to receive low-precision ranging raw data acquired in real time from the data link; 2. Normalization layer: Used to scale ranging values ​​with different ranges to a uniform range, smooth out differences in numerical magnitude, speed up training, and ensure computational stability; 3. LSTM layer: Relying on the gated memory mechanism, it mines the patterns of temporal changes, error drift, and distance measurement deviation, and extracts deep error features; 4. Dropout layer: used to randomly block some neurons during training to prevent the model from memorizing sample noise and improve the compensation generalization ability under unfamiliar data; 5. Fully connected layer: Used to fuse multi-dimensional features and convert temporal features into a unified error estimate; 6. Output layer: Used to output the final prediction error value, which is used to correct low-precision ranging data.

[0027] S3: When the GNSS of this unmanned vehicle is not working properly, calculate the high-precision GNSS ranging value between every two unmanned vehicles other than this one, and at the same time collect the data link ranging value between every two unmanned vehicles. Based on the data link ranging error correction model, obtain the data link ranging error of this unmanned vehicle, correct the data link ranging value of the unmanned vehicle, and obtain the high-precision data link ranging value. When a GNSS receiver for a particular unmanned vehicle (UAV) is unavailable, this UAV lacks high-precision ranging values ​​with the other n-1 UAVs. However, high-precision ranging values ​​exist between the other n-1 UAVs. Based on the data link ranging error correction model trained when the GNSS receiver is working normally, and using the current data link ranging value and the high-precision ranging values ​​of the n-1 UAVs, the data link ranging error of the UAV with the unavailable GNSS receiver is determined. The data link ranging value is then corrected to obtain a high-precision data link ranging value. .

[0028] S4: Based on high-precision data link ranging values, a combined inertial navigation and data link navigation method is used to obtain high-precision position.

[0029] Specifically, based on high-precision data link ranging values, the inertial navigation and data link combined navigation method can obtain high-precision position according to equation (4): (4); in: This represents the prior state prediction value of the inertial navigation and data link combination. The error equation transfer matrix represents the combination of inertial navigation and data link. This represents the optimal state estimate of the inertial navigation and data link combination at time k-1. This represents the noise allocation matrix for the inertial navigation and data link combination. This represents a noise sequence representing the combination of inertial navigation and data link. This represents the pseudorange difference between two unmanned vehicles in the inertial navigation and data link integrated navigation algorithm. This represents the high-precision GNSS ranging value between two unmanned vehicles. This represents the high-precision data link ranging value between two unmanned vehicles. The matrix representing the pseudorange difference between two unmanned vehicles in an inertial navigation and data link integrated navigation algorithm. Represents the observation matrix. Indicates the serial number of the driverless vehicle. This indicates the number of driverless cars.

[0030] Here The state variables of the inertial navigation and data link integrated navigation model mainly include 15-dimensional inertial navigation state error and one-dimensional data link clock error. The inertial navigation state variables are the same as those of the inertial and GNSS integrated model, and the data link error is the range error equivalent to the clock error. This represents the state variables of the inertial navigation and GNSS combined model. This represents the state variables of the data chain.

[0031] ,in, This represents the transfer matrix of the inertial navigation error equation. This represents the transition matrix of the data link error equation. , Represents the inertial navigation noise allocation matrix. Represents the data link noise distribution matrix. , Represents the inertial navigation noise sequence. This represents a data link noise sequence.

[0032] In the inertial navigation and data link integrated navigation algorithm, the pseudorange difference between two unmanned vehicles is used. As the observation vector, a high-precision position can be obtained according to equation (4).

[0033] This invention addresses the problem of GNSS signal obstruction and unavailability in scenarios such as canyons and dense forests for unmanned vehicle swarms, proposing an adaptive combined positioning method. When GNSS is available, machine learning is used to generate a data link ranging error model; when GNSS is unavailable, the trained data link ranging error model is used to correct the data link ranging values, obtaining high-precision data link ranging values. Then, inertial navigation and data link combined navigation are performed, thereby improving the positioning accuracy of the system when GNSS is unavailable.

[0034] A scene-adaptive integrated positioning system, characterized in that: it is used to execute a scene-adaptive integrated positioning method as described in any one of the above, which includes a GNSS receiver, an inertial navigation system, a data link device, a data acquisition module, a data link ranging error correction model construction module, and a data processing module; The data acquisition module is used to collect GNSS pseudorange information of all unmanned vehicles and the distance measurement value of the data link between every two unmanned vehicles. The data link ranging error correction model construction module is used to obtain the data link ranging error correction model through machine learning when the GNSS receiver of the unmanned vehicle is working normally, using the GNSS high-precision ranging value between every two unmanned vehicles and the data link ranging value between every two unmanned vehicles as training datasets. The data processing module is used to calculate the high-precision GNSS ranging value between every two unmanned vehicles using the GNSS pseudorange information of the unmanned vehicle, and to obtain the data link ranging error of the unmanned vehicle based on the data link ranging error correction model. The unmanned vehicle data link ranging value is then corrected using the data link ranging error of the unmanned vehicle to obtain the high-precision data link ranging value.

[0035] In summary, the scene-adaptive combined positioning method and system provided by this invention corrects the data link ranging value using GNSS information when the GNSS of the unmanned vehicle is available; when the GNSS of the unmanned vehicle is unavailable, adaptive adjustment is performed, and the data link ranging value is corrected using the previously established data link ranging error correction model. The corrected high-precision data link ranging value is used to correct the inertial navigation error, thereby improving the reliability and accuracy of navigation information.

[0036] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A scene-adaptive combined localization method, characterized in that: Includes the following steps: S1: When the GNSS receiver of the unmanned vehicle is working normally, the high-precision position result is output by using a combination of inertial navigation and GNSS. S2: Collect GNSS pseudorange information of all unmanned vehicles, calculate the GNSS high-precision ranging value between every two unmanned vehicles, and simultaneously collect the data link ranging value between every two unmanned vehicles. Use the GNSS high-precision ranging value between every two unmanned vehicles and the data link ranging value between every two unmanned vehicles as training datasets. Use a long short-term memory network model to obtain a data link ranging error correction model through machine learning. S3: When the GNSS of this unmanned vehicle is not working properly, calculate the high-precision GNSS ranging value between every two unmanned vehicles other than this one, and at the same time collect the data link ranging value between every two unmanned vehicles. Based on the data link ranging error correction model, obtain the data link ranging error of this unmanned vehicle, correct the data link ranging value of the unmanned vehicle, and obtain the high-precision data link ranging value. S4: Based on high-precision data link ranging values, a combined inertial navigation and data link navigation method is used to obtain high-precision position.

2. The scene adaptive combined localization method according to claim 1, characterized in that: In step S1, when the GNSS receiver of the unmanned vehicle is working normally, a loosely combined model of inertial navigation and GNSS is used to output high-precision position results.

3. The scene adaptive combined localization method according to claim 1, characterized in that: In step S1, when the GNSS receiver of the unmanned vehicle is working normally, the method for outputting high-precision position results using inertial navigation and GNSS combined navigation is as follows: S111: Calculate the three-dimensional attitude error, three-dimensional velocity error, three-dimensional position error, three-dimensional gyroscope drift error, and three-dimensional zero-bias error of the inertial navigation system according to equation (1): （1）； in: This represents the state variables of the inertial navigation and GNSS combined model. Indicates the heading angle error. Indicates the pitch angle error. Indicates the roll angle error. This indicates the eastward velocity error of the inertial navigation system. This indicates the northbound velocity error of the inertial navigation system. This indicates the inertial navigation system's upward velocity error. This indicates the positional error in the latitude direction of the inertial navigation system. This indicates the positional error in the longitude direction of the inertial navigation system. This indicates the position error in the altitude direction of the inertial navigation system. Inertial navigation Axial gyroscope drift error, Inertial navigation Axial gyroscope drift error, This represents the gyroscope drift error in the z-axis direction of the inertial navigation system. Inertial navigation Add zero offset error to the axial direction. Inertial navigation Add zero offset error to the axial direction. This indicates the zero-bias error added to the z-axis direction of the inertial navigation system; S112: Establish the observation model as Equation (2), and obtain the position and velocity errors of the inertial navigation and GNSS combined model based on the observation model. Then, after error compensation, output the high-precision position results: （2）； in: This indicates that the inertial navigation system outputs the northeast-sky velocity vector. This indicates the three-dimensional position output by the inertial navigation system. This indicates that the satellite's output velocity vector is in the northeast direction. This indicates the three-dimensional position output by the satellite navigation system. This indicates the velocity error between inertial navigation and satellite navigation. This indicates the positional error between the inertial navigation system and the satellite navigation system.

4. The scene adaptive combined localization method according to claim 1, characterized in that: In step S2, the frequency of collecting GNSS pseudorange information for all unmanned vehicles is 1 Hz.

5. The scene adaptive combined localization method according to claim 1, characterized in that: In step S2, the data link ranging value between every two unmanned vehicles is collected at a frequency of 1 Hz.

6. The scene adaptive combined localization method according to claim 1, characterized in that: In step S2, the GNSS high-precision ranging value between every two unmanned vehicles is calculated using pseudorange double difference.

7. The scene adaptive combined localization method according to claim 1, characterized in that: In step S2, the GNSS high-precision ranging value between every two unmanned vehicles is calculated according to equation (3): （3）； in: Indicates pseudo-distance double difference, This indicates the first driverless car and the... The unit vector between GNSS stars, This indicates the first driverless car and the... The unit vector between GNSS stars, This represents the transformation matrix from the carrier coordinate system to the ECEF coordinate system. This represents the GNSS high-precision ranging vector between every two unmanned vehicles. Indicates the first antenna, the second antenna, and the third antenna. Measurement noise generated between GNSS satellites Indicates the first antenna, the second antenna, and the third antenna. Measurement noise generated between GNSS satellites Indicates the first GNSS stars in Position coordinates along the axis This indicates the carrier obtained by multi-navigation sensor fusion under ECEF. Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis This indicates the carrier obtained by multi-navigation sensor fusion under ECEF. Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis This indicates the carrier obtained by multi-navigation sensor fusion under ECEF. Position coordinates along the axis Indicates the carrier and the first The distance between GNSS stars Indicates the first GNSS stars in Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis Indicates the first GNSS stars in Position coordinates along the axis Indicates the carrier and the first The distance between GNSS stars.

8. The scene adaptive combined localization method according to claim 1, characterized in that: In step S4, based on the high-precision data link ranging value, the high-precision position is obtained using the inertial navigation and data link combined navigation method according to equation (4): (4)； in: This represents the prior state prediction value of the inertial navigation and data link combination. The error equation transfer matrix represents the combination of inertial navigation and data link. This represents the optimal state estimate of the inertial navigation and data link combination at time k-1. This represents the noise allocation matrix for the inertial navigation and data link combination. This represents a noise sequence representing the combination of inertial navigation and data link. This represents the pseudorange difference between two unmanned vehicles in the inertial navigation and data link integrated navigation algorithm. This represents the high-precision GNSS ranging value between two unmanned vehicles. This represents the high-precision data link ranging value between two unmanned vehicles. The matrix representing the pseudorange difference between two unmanned vehicles in an inertial navigation and data link integrated navigation algorithm. Represents the observation matrix. Indicates the serial number of the driverless vehicle. This indicates the number of driverless cars.

9. A scene-adaptive integrated positioning system, characterized in that: The method for performing a scene adaptive combined positioning method as described in any one of claims 1 to 8 includes a GNSS receiver, an inertial navigation system, a data link device, a data acquisition module, a data link ranging error correction model construction module, and a data processing module. The data acquisition module is used to collect GNSS pseudorange information of all unmanned vehicles and the distance measurement value of the data link between every two unmanned vehicles. The data link ranging error correction model construction module is used to take the GNSS high-precision ranging value between every two unmanned vehicles and the data link ranging value between every two unmanned vehicles as training datasets when the unmanned vehicle GNSS receiver is working normally, and use the long short-term memory network model to obtain the data link ranging error correction model through machine learning. The data processing module is used to calculate the high-precision GNSS ranging value between every two unmanned vehicles using the GNSS pseudorange information of the unmanned vehicle, and to obtain the data link ranging error of the unmanned vehicle based on the data link ranging error correction model. The unmanned vehicle data link ranging value is then corrected using the data link ranging error of the unmanned vehicle to obtain the high-precision data link ranging value.

Images