Vehicle positioning method and system based on data and physical model dual drive

Abstract

The invention provides a vehicle positioning method and system based on data and physical model dual drive. The method belongs to the technical field of vehicle positioning, and comprises the following steps: determining an initial speed, a position and an attitude of a vehicle based on speed and position information measured by a GNSS and measurement data of an IMU; recursively updating the speed, the position and the attitude of the vehicle through inertial navigation solution based on the angular increment and the speed increment measured by the IMU; constructing an error state extended Kalman filter, and updating the error state extended Kalman filter by using multi-source observation information to obtain an error state estimated value of the inertial navigation calculation result; and performing feedback correction on the result of the inertial navigation solution by using the error state estimation value, and outputting a final vehicle positioning result. According to the method, traditional physical model driving and data model driving can be taken into consideration, and the vehicle positioning precision is improved on the basis of not needing additional system hardware.

Description

Technical Field

[0001] This application relates to the field of vehicle positioning technology, and more specifically, to a vehicle positioning method and system based on a dual-drive approach of data and physical models. Background Technology

[0002] Microelectromechanical systems (MEMS) inertial measurement units (IMUs) have been widely used in vehicle positioning due to their small size, low power consumption, and low cost. IMUs can provide high-frequency acceleration and angular velocity measurements without external base stations, making them suitable for short-term, high-frequency motion tracking. However, the output of low-cost IMUs is susceptible to noise, temperature drift, random walk, and installation errors. Under long-term navigation or severe dynamic conditions, errors often accumulate, reducing the accuracy and reliability of velocity, displacement, and attitude estimations. Although GNSS (Global Navigation Satellite System) can be used to correct IMU errors, in vehicle applications, GNSS often experiences loss of lock or obstruction, leading to significant degradation or even divergence in positioning performance. Therefore, to achieve high-precision vehicle positioning, it is necessary to design a vehicle positioning method that can effectively constrain and compensate for IMU error accumulation.

[0003] Currently, vehicle localization techniques can be broadly categorized into two types: traditional methods based on physical models and filtering, and data-driven learning methods. For example, patent CN109343095B proposes a vehicle navigation combined positioning device and its combined positioning method, using GNSS / INS / ODO for multi-source combined positioning. However, this method requires high-precision encoders to be installed on the vehicle tires, increasing equipment cost and system complexity. Patent CN116664621A proposes a SLAM system based on multi-camera systems and deep neural networks, using deep learning algorithms combined with the SLAM system for vehicle localization. However, this method is affected by visual degradation scenarios, resulting in limited positioning accuracy in some situations.

[0004] Therefore, in order to achieve high-precision and robust vehicle positioning in common urban environments without increasing system complexity, a vehicle positioning method that takes into account both physical-driven and data-driven models is needed to improve vehicle positioning accuracy. Summary of the Invention

[0005] The purpose of this application is to provide a vehicle positioning method and system based on dual data and physical model driving, which can take into account both traditional physical model driving and data model driving, and improve vehicle positioning accuracy without the need for additional system hardware.

[0006] This application is implemented as follows: In a first aspect, this application provides a vehicle positioning method based on a dual-drive approach of data and physical models, comprising the following steps: S1: Initial Alignment: Based on the velocity and position information measured by GNSS and the measurement data of IMU, determine the initial velocity, position and attitude of the vehicle; S2: Inertial navigation calculation: Based on the angular and velocity increments measured by the IMU, the vehicle's speed, position, and attitude are recursively updated through inertial navigation calculation; S3: Filtering Update: Construct an error state extended Kalman filter and update the error state extended Kalman filter using multi-source observation information to obtain the error state estimate of the inertial navigation solution; wherein, the state variable of the filter is the error state of the inertial navigation solution, and the multi-source observation information includes GNSS observation information, odometer or zero-speed observation information, and odometer equivalent speed observation information generated by the data-driven model; S4: Feedback Output: The error state estimate is used to correct the inertial navigation solution and output the final vehicle positioning result.

[0007] Based on the first aspect, the steps for determining the vehicle's initial attitude, based on GNSS-measured velocity and position information and IMU measurement data, include: The initial roll angle, pitch angle, and yaw angle of the vehicle are calculated using the following expressions: ; In the formula, The initial roll angle, pitch angle, and yaw angle are... This is the output of the triaxial accelerometer. These represent the eastward and northward velocities of the GNSS.

[0008] Based on the first aspect, the steps for recursively updating the vehicle's speed, position, and attitude through inertial navigation calculations based on the angular and velocity increments measured by the IMU include: The expressions for calculating the angular increment and velocity increment are set as follows: ; In the formula, Let be the angular increment vector at time k. Let be the angular velocity vector at time k. The time interval between two epochs. Let k be the velocity increment vector. Let be the specific force vector at time k; The expression for speed update is: ; In the formula, Let k be the velocity vector at time k. Let k be the ratio integral term at time k. The velocity increment caused by Coriolis acceleration, centripetal acceleration, and gravitational acceleration; The expression for position update is: ; In the formula, Let k be the quaternion corresponding to the latitude and longitude at time k. and This is a quaternion rotation vector between the e-frame and n-frame within adjacent time intervals. The height at time k, Let be the velocity component in the height direction at time k. The time interval between two epochs. This is the quaternion multiplication operator; The expression for attitude update is: ; In the formula, Let be the quaternion corresponding to the pose at time k. and This is a quaternion rotation vector between b-frame and n-frame within adjacent time intervals. This is the quaternion multiplication operator.

[0009] Based on the first aspect, the state equation and measurement equation of the error-state extended Kalman filter are respectively expressed as: Equations of state: ; In the formula, Let be the system state vector at time k. Let k be the state transition matrix at time k. Let k be the system noise driving matrix at time k-1. The system noise matrix at time k-1; Measurement equation: ; In the formula, Let k be the observation vector at time k. Let be the observation matrix at time k. Let be the system state vector at time k. Let be the observation noise matrix at time k.

[0010] Based on the first aspect, the observation matrix of the measurement equation is: ; In the formula, Let be the observation matrix at time k. Let be the GNSS position observation matrix at time k. Let be the GNSS velocity observation matrix at time k. The IMU zero-rate update observation matrix at time k. The velocity observation matrix is ​​the output of the data-driven model at time k. It is a third-order identity matrix. It is a third-order zero matrix. This is the rotation matrix between the b-frame and the n-frame. This is the lever vector from the center of the GNSS antenna to the center of the IMU.

[0011] Based on the first aspect, the feedback correction includes error feedback for position, velocity, and attitude, and is implemented respectively through the following expressions: The expression for position error feedback is: ; In the formula, The quaternion corresponding to the latitude and longitude after error feedback. The quaternion corresponding to the latitude and longitude before error feedback. The horizontal position error rotation vector, The elevation after error feedback. The elevation before error feedback. For elevation position error, This is the quaternion multiplication operator; The expression for speed error feedback is: ; In the formula, The velocity vector after error feedback. This is the rotation matrix between the c-frame and the n-frame. The velocity vector before error feedback. This is the velocity error vector; The expression for attitude error feedback is: ; In the formula, This is the quaternion corresponding to the attitude after error feedback. The attitude error quaternion, The quaternion corresponding to the attitude before error feedback. This is the quaternion multiplication operator.

[0012] Based on the first aspect, feedback correction also includes error feedback for the IMU zero bias; The expression for IMU zero bias error feedback is: ; In the formula, This is the zero-bias vector of the gyroscope after error feedback. This is the zero-bias vector of the gyroscope before error feedback. This is the zero bias vector of the gyroscope at the current moment. This is the zero bias vector of the accelerometer after error feedback. The zero bias vector of the accelerometer before error feedback. This is the zero bias vector of the accelerometer at the current moment.

[0013] Secondly, this application also provides a vehicle positioning system based on a dual-drive approach of data and physical models, comprising: Initial alignment module: It is configured to determine the vehicle's initial speed, position, and attitude based on GNSS-measured speed and position information and IMU measurement data; Inertial navigation calculation module: It is configured to recursively update the vehicle's speed, position, and attitude based on the angular and velocity increments measured by the IMU through inertial navigation calculation; The filter update module is configured to construct an error state extended Kalman filter and update the error state extended Kalman filter using multi-source observation information to obtain the error state estimate of the inertial navigation solution. The state variable of the filter is the error state of the inertial navigation solution, and the multi-source observation information includes GNSS observation information, odometer or zero-speed observation information, and odometer equivalent speed observation information generated by the data-driven model. Feedback output module: It is configured to use the error state estimate to correct the inertial navigation solution and output the final vehicle positioning result.

[0014] Thirdly, this application also provides an electronic device, comprising: Memory, used to store one or more programs; processor; The above method is implemented when one or more programs are executed by the processor.

[0015] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0016] Compared with the prior art, this application has at least the following advantages or beneficial effects: This application provides a vehicle positioning method and system based on a dual-drive approach of data and physical models. By utilizing velocity and position information measured by GNSS and measurement data from an IMU, the initial velocity, position, and attitude of the vehicle are determined. This provides an accurate and reliable initial reference, fundamentally eliminating the rapid divergence problem caused by excessive initial attitude errors. Traditional methods may rely solely on IMU static base alignment (calculating horizontal attitude but failing to determine heading) or require specific vehicle maneuvers (such as figure-eight movement) for initialization. This step cleverly utilizes the velocity vector directions (eastward and northward velocities) provided by GNSS, combined with the gravity direction sensed by the IMU accelerometer, to quickly and directly calculate the vehicle's three complete attitude angles (roll, pitch, and yaw) without requiring additional vehicle maneuvers.

[0017] By using angular and velocity increments measured by IMU and recursively updating the vehicle's speed, position, and attitude through inertial navigation calculations, the system provides high-frequency, continuous, and externally independent calculations of vehicle motion states, ensuring the smoothness and real-time nature of the positioning results.

[0018] By constructing an error state extended Kalman filter and updating it using multi-source observation information, the error state estimate of the inertial navigation solution is obtained. Multi-source information fusion is performed to effectively estimate and suppress the accumulation of inertial navigation errors. By introducing a data-driven model, the endurance and robustness in the event of GNSS failure are greatly enhanced.

[0019] By using error state estimates to correct the inertial navigation solution, the final vehicle positioning result is output. This achieves closed-loop error control, ensuring that the error does not grow unbounded, thus outputting a long-term stable and reliable final positioning result. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of an embodiment of a vehicle positioning method based on dual-drive of data and physical model according to this application; Figure 2 This is a horizontal positioning result diagram of a vehicle positioning method based on dual data and physical model driving according to an embodiment of the present application. Figure 3This is an elevation positioning result diagram of a vehicle positioning method based on dual data and physical model driving according to an embodiment of the present application. Figure 4 This is a schematic diagram of an embodiment of a vehicle positioning system based on dual data and physical model driving according to this application; Figure 5 This is a schematic diagram of the structure of an electronic device according to this application.

[0022] icon: 1. Initial alignment module; 2. Inertial navigation calculation module; 3. Filtering update module; 4. Feedback output module; 5. Processor; 6. Memory; 7. Communication interface. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0024] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the various embodiments and features described below can be combined with each other.

[0025] Example Through long-term research and practice, the inventors have discovered that existing vehicle positioning technologies mainly fall into two categories: traditional methods based on physical models and filtering, and data-driven learning methods. For example, published invention patent CN109343095B proposes a vehicle navigation combined positioning device and its combined positioning method, using GNSS / INS / ODO for multi-source combined positioning. However, this method requires high-precision encoders to be installed on the vehicle tires, increasing the cost and system complexity. Public invention patent CN116664621A proposes a SLAM system based on multi-camera systems and deep neural networks, using deep learning algorithms combined with the SLAM system for vehicle positioning. However, this method is affected by visual degradation scenarios, resulting in limited positioning accuracy in some situations.

[0026] In view of this, the embodiments of this application provide a vehicle positioning method and system based on dual data and physical model driving, which can take into account both traditional physical model driving and data model driving, and improve vehicle positioning accuracy without the need for additional system hardware.

[0027] The physical model in this embodiment refers to a mathematical model based on classical physics laws and kinematic principles. For example, inertial navigation calculations use acceleration and angular velocity measured by an integrating IMU to deduce position, velocity, and attitude. Another example is the state equation of an error-state extended Kalman filter, which describes how inertial navigation errors (such as position error, velocity error, attitude error, and sensor bias error) evolve and propagate over time.

[0028] The data model in this embodiment refers to a model that does not rely on explicit physical laws, but rather learns or summarizes the relationship between input and output from historical data. For example, a data-driven model can be a trained neural network or regression model that directly estimates the vehicle's equivalent speed by analyzing the complex, nonlinear relationship between vehicle CAN bus data (such as wheel speed pulses, steering wheel angle, accelerator and brake signals, etc.) and the actual vehicle speed. It does not concern itself with specific physical details such as wheel slippage or tire radius changes, but instead directly establishes a mapping relationship from vehicle dynamic signals to speed through a data "black box" or "grey box."

[0029] Please refer to Figure 1 This vehicle positioning method, driven by both data and physical models, includes the following steps: S1: Initial Alignment: Based on the velocity and position information measured by GNSS and the measurement data of IMU, determine the initial velocity, position and attitude of the vehicle; Specifically, this step provides an accurate and reliable initial reference, fundamentally eliminating the rapid divergence problem caused by excessive initial attitude errors. Traditional methods may only rely on IMU for static base alignment (calculating horizontal attitude but not determining heading), or require the vehicle to perform specific maneuvers (such as figure-eight movement) for initialization. This step cleverly utilizes the velocity vector directions (eastward and northward velocities) provided by GNSS, combined with the gravity direction sensed by the IMU accelerometer, to quickly and directly calculate the vehicle's complete three attitude angles (roll, pitch, and yaw) without requiring additional vehicle maneuvers.

[0030] As one implementation method, the initial attitude of the vehicle can be determined in the following ways: The calculation expressions for the vehicle's initial roll angle, pitch angle, and yaw angle are set as follows: ; In the formula, The initial roll angle, pitch angle, and yaw angle are... This is the output of the triaxial accelerometer. These represent the eastward and northward velocities of the GNSS.

[0031] S2: Inertial navigation calculation: Based on the angular and velocity increments measured by the IMU, the vehicle's speed, position, and attitude are recursively updated through inertial navigation calculation; Specifically, IMU data sampling rates are typically as high as hundreds of Hz. Through inertial navigation calculations (mechanical orchestration), it can provide speed, position, and attitude change information at rates up to hundreds of Hz between two GNSS updates (usually 1 Hz or 10 Hz). During the extremely short period when GNSS completely fails, it is the only source of position information. Inertial navigation calculations provide high-frequency, continuous, and externally independent vehicle motion state estimations, ensuring the smoothness and real-time nature of the positioning results.

[0032] As one implementation method, this step can be achieved in the following way: The expressions for calculating the angular increment and velocity increment are set as follows: ; In the formula, Let be the angular increment vector at time k. Let be the angular velocity vector at time k. The time interval between two epochs. Let k be the velocity increment vector. Let be the specific force vector at time k; The expression for speed update is: ; In the formula, Let k be the velocity vector at time k. Let k be the ratio integral term at time k. The velocity increment caused by Coriolis acceleration, centripetal acceleration, and gravitational acceleration; The expression for position update is: ; In the formula, Let k be the quaternion corresponding to the latitude and longitude at time k. and This is a quaternion rotation vector between the e-frame and n-frame within adjacent time intervals. The height at time k, Let be the velocity component in the height direction at time k. The time interval between two epochs. This is the quaternion multiplication operator; The expression for attitude update is: ; In the formula, Let be the quaternion corresponding to the pose at time k. and This is a quaternion rotation vector between b-frame and n-frame within adjacent time intervals. This is the quaternion multiplication operator.

[0033] S3: Filtering Update: Construct an error state extended Kalman filter and update the error state extended Kalman filter using multi-source observation information to obtain the error state estimate of the inertial navigation solution; wherein, the state variable of the filter is the error state of the inertial navigation solution, and the multi-source observation information includes GNSS observation information, odometer or zero-speed observation information, and odometer equivalent speed observation information generated by the data-driven model; Specifically, considering that the results of inertial navigation calculations inevitably diverge over time, and errors (especially position errors) can grow indefinitely, this step effectively estimates and suppresses the accumulation of inertial navigation errors through multi-source information fusion. By introducing a data-driven model, the system's endurance and robustness in the event of GNSS failure are greatly enhanced. By constructing an error state extended Kalman filter, an optimal estimation framework is provided, which does not directly correct the inertial navigation results but intelligently estimates the magnitude of the errors, providing a basis for subsequent accurate correction. By introducing GNSS observation information, an absolute, drift-free position and velocity reference is provided when the signal is good, serving as an "anchor point" for error correction. By introducing odometer or zero-speed observation information, accurate velocity / zero-speed constraints can be provided, effectively suppressing velocity errors and altitude channel divergence. By introducing odometer equivalent velocity observation information generated by the data-driven model, when the GNSS signal is lost (entering a tunnel, underground parking garage) and the vehicle is not stationary (ZUPT cannot be used), the traditional system will rely entirely on the error-divergent INS, and the positioning results will rapidly deviate from the true trajectory. This step analyzes dynamic data such as the vehicle's CAN bus to generate a reliable speed observation in real time, which serves as input to the physical model (EKF). This observation provides a crucial "speed anchor" for the filter in GNSS blind zones, significantly extending the "endurance" of high-precision positioning and solving the pain point of traditional integrated navigation in dynamic GNSS blind zones.

[0034] As one implementation method, the state equation and measurement equation of the error-state extended Kalman filter are respectively expressed as: Equations of state: ; In the formula, Let be the system state vector at time k. Let k be the state transition matrix at time k. Let k be the system noise driving matrix at time k-1. The system noise matrix at time k-1; Measurement equation: ; In the formula, Let k be the observation vector at time k. Let be the observation matrix at time k. Let be the system state vector at time k. Let be the observation noise matrix at time k.

[0035] Furthermore, the observation matrix of the measurement equation is: ; In the formula, Let be the observation matrix at time k. Let be the GNSS position observation matrix at time k. Let be the GNSS velocity observation matrix at time k. The IMU zero-rate update observation matrix at time k. The velocity observation matrix is ​​the output of the data-driven model at time k. It is a third-order identity matrix. It is a third-order zero matrix. This is the rotation matrix between the b-frame and the n-frame. This is the lever vector from the center of the GNSS antenna to the center of the IMU.

[0036] S4: Feedback Output: The error state estimate is used to correct the inertial navigation solution and output the final vehicle positioning result.

[0037] Specifically, this step implements closed-loop error control, ensuring that the error does not grow unbounded, thus outputting a long-term stable and reliable final positioning result. The error state estimated in step S3 is used to compensate / correct the inertial navigation solution result in step S2. This forms a negative feedback closed loop of "inertial solution - error estimation - feedback correction - more accurate inertial solution," fundamentally solving the divergence problem of pure inertial navigation. Furthermore, the final output positioning result is the best estimate after optimal fusion of multi-source information, and its accuracy and reliability are far higher than the results of any single sensor or open-loop solution. Moreover, this feedback loop can also estimate and compensate for the sensor zero bias of the IMU in real time, further improving the inherent accuracy of the system.

[0038] As one implementation method, feedback correction includes error feedback for position, velocity, and attitude, and is implemented through the following expressions respectively: The expression for position error feedback is: ; In the formula, The quaternion corresponding to the latitude and longitude after error feedback. The quaternion corresponding to the latitude and longitude before error feedback. The horizontal position error rotation vector, The elevation after error feedback. The elevation before error feedback. For elevation position error, This is the quaternion multiplication operator; The expression for speed error feedback is: ; In the formula, The velocity vector after error feedback. This is the rotation matrix between the c-frame and the n-frame. The velocity vector before error feedback. This is the velocity error vector; The expression for attitude error feedback is: ; In the formula, This is the quaternion corresponding to the attitude after error feedback. The attitude error quaternion, The quaternion corresponding to the attitude before error feedback. This is the quaternion multiplication operator.

[0039] Based on the first aspect, feedback correction also includes error feedback for the IMU zero bias; The expression for IMU zero bias error feedback is: ; In the formula, This is the zero-bias vector of the gyroscope after error feedback. This is the zero-bias vector of the gyroscope before error feedback. This is the zero bias vector of the gyroscope at the current moment. This is the zero bias vector of the accelerometer after error feedback. The zero bias vector of the accelerometer before error feedback. This is the zero bias vector of the accelerometer at the current moment.

[0040] To verify the correctness of this invention, actual tests were conducted. The performance parameters of the IMU are listed in Table 1, and the performance parameters of the GNSS module are listed in Table 2.

[0041] Table 1: Performance parameters of the IMU in the actual experiment

[0042] Table 2: Performance parameters of the GNSS module in the actual experiment

[0043] Three sets of field tests were conducted using vehicle positioning data collected from the modules corresponding to Tables 1 and 2. The experiments included a satellite-denied environment, and the vehicle position was calculated using the method proposed in this application. The three-dimensional positioning error table of the method proposed in this application is shown in Table 3.

[0044] Table 3: Three-dimensional positioning error table of the algorithm in this application

[0045] As shown in Table 3, the average three-dimensional positioning error of the vehicle using the method proposed in this application is 0.2m in a test environment including satellite denial. This indicates that the method proposed in this application can achieve high-precision vehicle positioning in a satellite-denied environment without the need for additional sensors. Therefore, the effectiveness and correctness of the method provided by this invention are verified.

[0046] Furthermore, to verify the vehicle positioning accuracy using the method described in this application, the applicant conducted a vehicle positioning simulation using the method described in this application. For the simulation diagram of the horizontal positioning, please refer to... Figure 2 Please refer to the simulation diagram for elevation positioning. Figure 2 .Depend on Figure 2 and Figure 3 As can be seen, the horizontal and vertical positioning results obtained by the method of this application almost coincide with the true values, which once again verifies that the vehicle positioning using the method of this application has high accuracy.

[0047] Please refer to Figure 4 The application also provides a vehicle positioning system based on a dual-drive approach of data and physical models, including: Initial alignment module 1: It is configured to determine the initial speed, position and attitude of the vehicle based on the speed and position information measured by GNSS and the measurement data of IMU; Inertial navigation calculation module 2: It is configured to recursively update the vehicle's speed, position, and attitude based on the angular and velocity increments measured by the IMU through inertial navigation calculation; Filtering update module 3: It is configured to construct an error state extended Kalman filter and update the error state extended Kalman filter using multi-source observation information to obtain the error state estimate of the inertial navigation solution; wherein, the state variable of the filter is the error state of the inertial navigation solution, and the multi-source observation information includes GNSS observation information, odometer or zero-speed observation information, and odometer equivalent speed observation information generated by the data-driven model; Feedback output module 4: It is configured to use the error state estimate to correct the inertial navigation solution and output the final vehicle positioning result.

[0048] For a detailed implementation of the vehicle positioning system based on both data and physical models, please refer to the detailed implementation of a vehicle positioning method based on both data and physical models; further details will not be provided here.

[0049] Please refer to Figure 5 This application also provides an electronic device, including: Memory 6 is used to store one or more programs; Processor 5; Processor 5 and memory 6 are connected via communication interface 7; When one or more programs are executed by processor 5, all or some of the above methods are implemented.

[0050] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor 5, implements all or part of the methods described above.

[0051] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A vehicle positioning method based on dual-drive of data and physical model, characterized in that, Includes the following steps: S1: Initial Alignment: Based on the velocity and position information measured by GNSS and the measurement data of IMU, determine the initial velocity, position and attitude of the vehicle; S2: Inertial navigation calculation: Based on the angular and velocity increments measured by the IMU, the vehicle's speed, position, and attitude are recursively updated through inertial navigation calculation; S3: Filtering Update: Construct an error state extended Kalman filter and update the error state extended Kalman filter using multi-source observation information to obtain the error state estimate of the inertial navigation solution; wherein, the state variable of the filter is the error state of the inertial navigation solution, and the multi-source observation information includes GNSS observation information, odometer or zero-speed observation information, and odometer equivalent speed observation information generated by the data-driven model; S4: Feedback Output: The error state estimate is used to correct the inertial navigation solution result, and the final vehicle positioning result is output.

2. The vehicle positioning method based on dual-drive of data and physical model according to claim 1, characterized in that, The steps for determining the vehicle's initial attitude based on GNSS-measured velocity and position information and IMU measurement data include: The initial roll angle, pitch angle, and yaw angle of the vehicle are calculated using the following expressions: ； In the formula, The initial roll angle, pitch angle, and yaw angle are... This is the output of the triaxial accelerometer. These represent the eastward and northward velocities of the GNSS.

3. The vehicle positioning method based on dual-drive of data and physical model according to claim 1, characterized in that, Based on the angular and velocity increments measured by the IMU, the steps for recursively updating the vehicle's speed, position, and attitude through inertial navigation calculations include: The expressions for calculating the angular increment and velocity increment are set as follows: ； In the formula, Let be the angular increment vector at time k. Let be the angular velocity vector at time k. The time interval between two epochs. Let k be the velocity increment vector. Let be the specific force vector at time k; The expression for speed update is: ； In the formula, Let k be the velocity vector at time k. Let k be the ratio integral term at time k. The velocity increment caused by Coriolis acceleration, centripetal acceleration, and gravitational acceleration; The expression for position update is: ； In the formula, Let k be the quaternion corresponding to the latitude and longitude at time k. and This is a quaternion rotation vector between the e-frame and n-frame within adjacent time intervals. The height at time k, Let be the velocity component in the height direction at time k. The time interval between two epochs. This is the quaternion multiplication operator; The expression for attitude update is: ； In the formula, Let be the quaternion corresponding to the pose at time k. and This is a quaternion rotation vector between b-frame and n-frame within adjacent time intervals. This is the quaternion multiplication operator.

4. The vehicle positioning method based on dual-drive of data and physical model according to claim 1, characterized in that, The state equation and measurement equation of the error-state extended Kalman filter are respectively expressed as: Equations of state: ； In the formula, Let be the system state vector at time k. Let k be the state transition matrix at time k. Let k be the system noise driving matrix at time k-1. Let k be the system noise matrix at time k-1; Measurement equation: ； In the formula, Let k be the observation vector at time k. Let be the observation matrix at time k. Let be the system state vector at time k. Let be the observation noise matrix at time k.

5. The vehicle positioning method based on dual-drive of data and physical model according to claim 4, characterized in that, The observation matrix of the measurement equation is: ； In the formula, Let be the observation matrix at time k. Let be the GNSS position observation matrix at time k. Let be the GNSS velocity observation matrix at time k. The IMU zero-rate update observation matrix at time k. The velocity observation matrix is ​​the output of the data-driven model at time k. It is a third-order identity matrix. It is a third-order zero matrix. This is the rotation matrix between the b-frame and the n-frame. This is the lever vector from the center of the GNSS antenna to the center of the IMU.

6. The vehicle positioning method based on dual-drive of data and physical model according to claim 1, characterized in that, The feedback correction includes error feedback for position, velocity, and attitude, and is implemented through the following expressions respectively: The expression for position error feedback is: ； In the formula, The quaternion corresponding to the latitude and longitude after error feedback. The quaternion corresponding to the latitude and longitude before error feedback. The horizontal position error rotation vector, The elevation after error feedback. The elevation before error feedback. For elevation position error, This is the quaternion multiplication operator; The expression for speed error feedback is: ； In the formula, The velocity vector after error feedback. This is the rotation matrix between the c-frame and the n-frame. The velocity vector before error feedback. This is the velocity error vector; The expression for attitude error feedback is: ； In the formula, This is the quaternion corresponding to the attitude after error feedback. The attitude error quaternion, The quaternion corresponding to the attitude before error feedback. This is the quaternion multiplication operator.

7. The vehicle positioning method based on dual-drive of data and physical model according to claim 6, characterized in that, The feedback correction also includes error feedback for the IMU zero bias; The expression for IMU zero bias error feedback is: ； In the formula, This is the zero-bias vector of the gyroscope after error feedback. This is the zero-bias vector of the gyroscope before error feedback. This is the zero bias vector of the gyroscope at the current moment. This is the zero bias vector of the accelerometer after error feedback. The zero bias vector of the accelerometer before error feedback. This is the zero bias vector of the accelerometer at the current moment.

8. A vehicle positioning system based on a dual-drive approach of data and physical models, characterized in that, include: Initial alignment module: It is configured to determine the vehicle's initial speed, position, and attitude based on GNSS-measured speed and position information and IMU measurement data; Inertial navigation calculation module: It is configured to recursively update the vehicle's speed, position, and attitude based on the angular and velocity increments measured by the IMU through inertial navigation calculation; The filter update module is configured to construct an error state extended Kalman filter and update the error state extended Kalman filter using multi-source observation information to obtain the error state estimate of the inertial navigation solution result. The state variable of the filter is the error state of the inertial navigation solution result, and the multi-source observation information includes GNSS observation information, odometer or zero-speed observation information, and odometer equivalent velocity observation information generated by the data-driven model. Feedback output module: It is configured to use the error state estimate to correct the result of the inertial navigation solution and output the final vehicle positioning result.

9. An electronic device, characterized in that, include: Memory, used to store one or more programs; processor; When the one or more programs are executed by the processor, the method as described in any one of claims 1-7 is implemented.

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 method as described in any one of claims 1-7.

Images