Carrier integrated navigation method, device, equipment and medium

Abstract

The embodiment of the invention discloses a carrier integrated navigation method, device and equipment and a medium, and relates to the technical field of intelligent driving navigation, the method comprises the following steps: acquiring wheel speed data of a current epoch output by a wheel speed measurement unit, and acquiring inertia measurement data of the current epoch output by an inertia measurement unit; inputting a non-integrity constraint prediction model to obtain a non-integrity constraint component of the current epoch; and constructing a three-dimensional motion constraint observation value based on the non-integrity constraint component and the wheel speed data, performing measurement updating on a kinematics observation equation of the Kalman filter, obtaining a state estimation result obtained by updating, and outputting a positioning result in the current epoch. According to the method, the lateral and vertical velocity components are dynamically predicted through the non-integrity constraint prediction model, the three-dimensional motion constraint observation value is constructed, and the state estimation of the Kalman filter is optimized, so that the problem of carrier motion constraint performance degradation under a complex motion condition can be effectively relieved, and the integrated navigation precision can be further improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving navigation technology, and in particular to a vehicle combined navigation method, device, equipment and medium. Background Art

[0002] Intelligent vehicle applications in environmental perception, obstacle avoidance, and path planning all require accurate position and posture. However, existing multi-sensor fusion methods struggle to achieve continuous, high-precision state estimation in complex observation scenarios and under variable motion conditions. While AI has brought new capabilities to perception and planning, its potential for improving position and posture estimation has yet to be fully explored.

[0003] Currently, traditional multi-sensor fusion navigation methods primarily rely on the absolute positioning capabilities of the Global Navigation Satellite System (GNSS) and the relative positioning capabilities of systems such as the Inertial Navigation System (INS) to achieve state estimation in all scenarios. For vehicles, their Controller Area Network Bus (CAN Bus) typically integrates a speed information interface for real-time monitoring of the vehicle's motion state and can be used as an odometer (Odometer) to couple with the GNSS and INS. Furthermore, based on the laws of rigid body kinematics, non-holonomic constraints (NHCs) are applied to the vehicle's drive wheels to form three-dimensional motion constraints in the vehicle's coordinate system (body frame, b-frame). These NHCs, a traditional kinematic constraint, are simple in principle and effective, and therefore widely used in vehicle navigation.

[0004] However, the combined navigation methods described in these related technologies also have significant drawbacks. The calculated NHC constraints are based on a strong assumption of zero lateral and vertical velocity. In reality, due to complex road conditions and the vehicle's complex motion states, such as turning and vibration, the zero velocity assumption in the non-moving directions cannot always be met. Imposing incorrect motion constraints on the vehicle can degrade the combined navigation system's constraint performance and even lead to erroneous state estimation results.

[0005] Therefore, there is currently a lack of a method that can predict and improve NHC constraints based on the vehicle's actual lateral and vertical velocities to alleviate the problem of vehicle motion constraint performance degradation in complex motion situations and improve the vehicle's combined navigation accuracy in complex motion situations. Summary of the Invention

[0006] The embodiments of the present application provide a vehicle integrated navigation method, apparatus, device, and medium to address the deficiencies of the above-mentioned related technologies. The technical solutions are as follows: In a first aspect, an embodiment of the present application provides a vehicle combined navigation method, comprising: Obtaining wheel speed data of a current epoch output by a wheel speed measurement unit of the vehicle, and obtaining inertial measurement data of the current epoch output by an inertial measurement unit of the vehicle; Inputting the inertial measurement data and the wheel speed data into a nonholonomic constraint prediction model, and calculating the nonholonomic constraint component of the current epoch through the nonholonomic constraint prediction model; Constructing a three-dimensional motion constraint observation value based on the non-holonomic constraint component and the wheel speed data; performing measurement updates on the kinematic observation equations of a Kalman filter based on the three-dimensional motion constraint observation values, obtaining a state estimation result obtained by the Kalman filter update, and outputting a positioning result of the vehicle at a current epoch based on the state estimation result; The non-holonomic constraint prediction model is trained based on a given sample set.

[0007] In an optional solution of the first aspect, inputting the inertial measurement data and the wheel speed data into a nonholonomic constraint prediction model, and calculating the nonholonomic constraint component of the current epoch using the nonholonomic constraint prediction model includes: The inertial measurement data of the current epoch output by the inertial measurement unit of the vehicle includes an angular velocity output by a three-axis gyroscope and a specific force output by a three-axis accelerometer; Inputting the angular velocity output by the three-axis gyroscope, the specific force output by the three-axis accelerometer, and the wheel speed data of the current epoch into the nonholonomic constraint prediction model, so that the nonholonomic constraint prediction model predicts nonholonomic constraint components of the current epoch based on the input data, the nonholonomic constraint components including a lateral velocity component and a vertical velocity component; The three-dimensional motion constraint observation value is constructed based on the nonholonomic constraint component and the wheel speed data, and the formula is applied: ; in, is the three-dimensional motion constraint observation value, is the wheel speed data, is the lateral velocity component, is the vertical velocity component, t represents the corresponding epoch, and the superscript b indicates that the coordinate system of the corresponding parameter is the carrier coordinate system.

[0008] In an optional solution of the first aspect, before measuring and updating the kinematic observation equation of the Kalman filter based on the three-dimensional motion constraint observation value, the method further includes: Calculating the velocity of the center of the inertial measurement unit in the navigation coordinate system based on the inertial measurement data, and converting the velocity of the navigation coordinate system to the velocity of the wheel speed measurement unit to obtain the converted velocity of the vehicle coordinate system; The measuring and updating of the kinematic observation equation of the Kalman filter based on the three-dimensional motion constraint observation value includes: Subtracting the three-dimensional motion constraint observation value from the reduced velocity to calculate the prior residual of the kinematic observation equation; The prior residual is used as the velocity information observation of the Kalman filter, and the state estimation result of the current epoch is obtained by updating the Kalman filter.

[0009] In an optional solution of the first aspect, the training process of the non-holonomic constraint prediction model includes the steps of: driving the vehicle for a typical convergence time of the GNSS solution strategy while meeting the preset GNSS observation conditions, and then driving the vehicle for a preset time while meeting the preset GNSS observation conditions; During the driving process of the preset duration, obtaining inertial measurement data of each epoch output by the inertial measurement unit of the vehicle, wheel speed data of the corresponding epoch output by the wheel speed measurement unit, and GNSS positioning, speed, and attitude data of the corresponding epoch output by the GNSS signal receiving unit; Constructing a sample set for training the non-holonomic constraint prediction model based on the inertial measurement data of each epoch, the wheel speed data of the corresponding epoch, and the GNSS positioning speed and attitude data of the corresponding epoch; The non-holonomic constraint prediction model is trained online based on the sample set, and the weight parameters of the converged non-holonomic constraint prediction model are determined to obtain a trained non-holonomic constraint prediction model.

[0010] In an optional solution of the first aspect, the step of causing the vehicle to travel under preset GNSS observation conditions until a typical convergence time of a GNSS solution strategy is reached further includes: Calibrate and obtain lever arm information from the inertial measurement unit to the driving wheel; When the vehicle is in a stationary state and meets preset GNSS observation conditions, obtaining initial GNSS positioning, speed and attitude measurement data output by the GNSS signal receiving unit and initial inertial measurement data output by the inertial measurement unit; The navigation state of the vehicle is initialized based on the lever arm information, the initial GNSS observation value and the initial inertial measurement data, and an initial state estimation result of a Kalman filter is obtained, including initial position information, initial velocity information and initial attitude information of the vehicle.

[0011] In an optional solution of the first aspect, the sample set for training the non-holonomic constraint prediction model is constructed based on the inertial measurement data of each epoch, the wheel speed data of the corresponding epoch, and the GNSS positioning speed and attitude data of the corresponding epoch, including: Calculating the navigation coordinate system velocity of the vehicle at the corresponding epoch based on the GNSS positioning, speed, and attitude measurement data, converting the navigation coordinate system velocity to a vehicle coordinate system to obtain a vehicle coordinate system velocity, and calculating a lateral velocity component and a vertical velocity component of the vehicle coordinate system velocity; The wheel speed data, angular velocity, and specific force of the same epoch are used as sample inputs, and the lateral velocity component and vertical velocity component of the same epoch are used as sample labels of the samples to construct the sample set.

[0012] In an optional solution of the first aspect, the preset GNSS observation conditions include: The number of available GNSS satellites observed by the vehicle's GNSS signal receiving unit is greater than a preset number threshold, the dilution of precision factor of the GNSS positioning result is less than a preset dilution factor threshold, and the sum of the main diagonal elements of the position covariance matrix of the GNSS positioning result is less than a preset threshold; After obtaining the trained non-holonomic constraint prediction model, the method further includes: During the driving of the vehicle, under the condition that the preset GNSS observation conditions are met, wheel speed data, inertial measurement data, and GNSS positioning speed and attitude data are obtained for each epoch; New training samples are constructed based on wheel speed data, inertial measurement data, and GNSS positioning speed and attitude data of the same epoch; Supplementing the new training samples to the sample set, and performing online training on the non-holonomic constraint prediction model using the supplemented sample set to update the weight parameters of the non-holonomic constraint prediction model to obtain an updated non-holonomic constraint prediction model; The step of calculating the nonholonomic constraint component of the current epoch by using the nonholonomic constraint prediction model is performed based on the updated nonholonomic constraint prediction model.

[0013] In a second aspect, an embodiment of the present application further provides a vehicle combined navigation device, comprising: a data acquisition unit, configured to acquire wheel speed data of a current epoch outputted by a wheel speed measurement unit of the vehicle, and acquire inertial measurement data of a current epoch outputted by an inertial measurement unit of the vehicle; a constraint calculation unit, configured to input the inertial measurement data and the wheel speed data into a nonholonomic constraint prediction model, and calculate the nonholonomic constraint component of the current epoch using the nonholonomic constraint prediction model; The constraint calculation unit is further configured to construct a three-dimensional motion constraint observation value based on the non-holonomic constraint component and the wheel speed data; a positioning result calculation unit, configured to measure and update the kinematic observation equation of a Kalman filter based on the three-dimensional motion constraint observation value, obtain a state estimation result obtained by the Kalman filter update, and output a positioning result of the vehicle at a current epoch based on the state estimation result; The non-holonomic constraint prediction model is trained based on a given sample set.

[0014] In a third aspect, an embodiment of the present application further provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method provided in the first aspect of the embodiment of the present application or any one of the implementations of the first aspect is implemented.

[0015] In a fourth aspect, the present application also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements the method provided by the first aspect of the embodiment of the present application or any one of the implementation methods of the first aspect.

[0016] The beneficial effects of the technical solutions provided by some embodiments of the present application include at least: The present application provides a vehicle combined navigation method, device, electronic device and storage medium, which dynamically predicts the lateral and vertical velocity components through a non-holonomic constraint prediction model, constructs three-dimensional motion constraint observation values ​​and optimizes the state estimation of the Kalman filter, which can effectively alleviate the problem of carrier motion constraint performance degradation in complex motion conditions, avoids the defect that the NHC constraints calculated in related technologies are based on strong lateral and vertical zero-speed assumptions and cannot reflect the real motion constraints of the vehicle in complex motion conditions, and thus can improve the accuracy of combined navigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in this application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 This is a flow chart of a vehicle combined navigation method provided in an embodiment of the present application; Figure 2 This is a schematic structural diagram of a vehicle combined navigation device provided in an embodiment of the present application; Figure 3 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0020] The terms "including" and "having," and any variations thereof, in the specification and claims of this application and the accompanying drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or modules is not limited to the listed steps or modules, but may optionally include steps or modules not listed, or may optionally include other steps or modules inherent to the process, method, product, or apparatus.

[0021] It should be noted that the terms "first" and "second" used in this application are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It is understood that the terms "first" and "second" may interchangeably represent a specific order or precedence, where permitted. It should be understood that the objects distinguished by "first" and "second" may interchangeably represent a specific order or precedence, where appropriate, such that the embodiments of the present application described herein can be implemented in an order other than that described or illustrated herein.

[0022] It should be noted that vehicles can include various types of vehicles, such as cars, buses, and trucks. These vehicles are typically equipped with an INS system and a GNSS signal receiving unit. GNSS can provide accurate position and velocity information, but the signal may be subject to interference or unavailable in certain environments (such as tunnels, indoors, or in urban areas with tall buildings). The INS system utilizes the accelerometer and gyroscope in the Inertial Measurement Unit (IMU) to calculate the vehicle's position and orientation. Although its errors accumulate over time, it can provide continuous navigation information when the GNSS signal is lost. Combining the two systems leverages their respective strengths to provide more stable and reliable navigation services, enabling integrated navigation.

[0023] However, the NHC constraints calculated in the combined navigation by the related technology are based on strong lateral and vertical zero-speed assumptions, which do not take into account that in actual situations, due to complex road conditions, the vehicle may experience complex motion states such as turning and vibration while driving, and it is impossible to guarantee that the zero-speed assumption in the non-motion direction is always met. That is, the related technology does not take into account the actual lateral and vertical speeds of the vehicle in complex motion conditions, which easily leads to the accumulation of errors and thus inaccurate navigation positioning results.

[0024] Based on this, the inventors discovered a fundamental contradiction between the traditional zero-speed assumption and the actual state of motion. By analyzing the correlation between multi-source sensor data, they found that the angular velocity and specific force information output by the inertial measurement unit can reflect the changing patterns of the vehicle's motion pattern. Considering the machine learning model's ability to fit nonlinear relationships, they proposed using historical GNSS positioning, speed, and attitude data to construct a supervisory signal (as a sample label) and train a non-complete constraint component prediction model. This model perceives the vehicle's motion state in real time and dynamically adjusts the constraint equation parameters, thereby establishing a three-dimensional observation model that is more consistent with actual kinematics, providing technical support for precise integrated navigation of vehicles in complex environments and under complex motion conditions.

[0025] The present application is described in detail below with reference to specific embodiments.

[0026] Next, combine Figure 1 , introduces a vehicle combined navigation method, device, equipment and medium provided by the embodiment of this application. For details, please refer to Figure 1 , Figure 1 FIG. 1 shows a flow chart of a vehicle combined navigation method provided by an embodiment of the present application. Figure 1 As shown, the method includes the following steps: S101, obtaining wheel speed data of a current epoch output by a wheel speed measurement unit of a vehicle, and obtaining inertial measurement data of a current epoch output by an inertial measurement unit of the vehicle; S102, inputting the inertial measurement data and the wheel speed data into a nonholonomic constraint prediction model, and calculating the nonholonomic constraint component of the current epoch using the nonholonomic constraint prediction model; S103, constructing a three-dimensional motion constraint observation value based on the nonholonomic constraint component and the wheel speed data; S104: measuring and updating the kinematic observation equation of the Kalman filter based on the three-dimensional motion constraint observation value, obtaining a state estimation result obtained by the Kalman filter update, and outputting the positioning result of the vehicle at the current epoch based on the state estimation result.

[0027] Specifically, in S101, the inertial measurement data output by the inertial measurement unit includes specific force and angular velocity. The inertial measurement unit generally measures specific force using an onboard accelerometer, which represents acceleration relative to an inertial reference frame. The inertial measurement unit generally measures angular velocity using a gyroscope. The wheel speed data output by the wheel speed measurement unit represents the speed of the vehicle's drive wheels.

[0028] It should be noted that the measurement data output by the inertial measurement unit is relative to the carrier coordinate system b (body frame), and the specific force corresponding to epoch t is , the angular velocity corresponding to epoch t is .

[0029] Specifically, the wheel speed measurement unit may include an odometer (OD) installed on the vehicle's driving wheel or non-steering wheel, through which the wheel speed in the vehicle's coordinate system b can be obtained. The wheel speed can also be obtained through the wheel speed acquisition interface on the CAN bus.

[0030] In some embodiments, in S102, the inertial measurement data and the wheel speed data are input into a nonholonomic constraint prediction model, and the nonholonomic constraint components of the current epoch are calculated using the nonholonomic constraint prediction model, specifically including: The inertial measurement data of the current epoch t output by the vehicle's inertial measurement unit includes the angular velocity output by the three-axis gyroscope , the relative force of the three-axis accelerometer output .

[0031] Specifically, the angular velocity output by the three-axis gyroscope can be , the specific force output by the three-axis accelerometer and the wheel speed data of the current epoch Construct the vector Then, it is used as the input of the non-holonomic constraint prediction model.

[0032] Furthermore, the non-integrity constraint prediction model is based on the input data The non-holonomic constraint components of the current epoch are predicted, including the lateral velocity component. and vertical velocity component .

[0033] Specifically, in S103, based on the lateral velocity component and vertical velocity component and wheel speed data Construct the three-dimensional motion constraint observation value, and the specific application formula is: ; in, is the three-dimensional motion constraint observation value, is the wheel speed data, is the lateral velocity component, is the vertical velocity component, t represents the corresponding epoch, and the superscript b indicates that the coordinate system of the corresponding parameter is the carrier coordinate system.

[0034] In some embodiments, before performing the step of measuring and updating the kinematic observation equation of the Kalman filter based on the three-dimensional motion constraint observation value in S104, the following steps may be performed: The velocity of the center of the inertial measurement unit in the navigation coordinate system n is calculated based on the inertial measurement data. and the center speed The calculated speed of the carrier coordinate system is obtained by converting it to the wheel speed measurement unit .

[0035] Specifically, the earth-centered earth-fixed frame E can be selected as the reference coordinate system N for navigation. It can be understood that when the E frame is selected as the navigation coordinate system, the velocity of the center of the inertial measurement unit in the navigation coordinate system N is expressed as .

[0036] For example, taking the e-frame as the navigation coordinate system n-frame as an example, the steps for calculating the velocity of the center of the inertial measurement unit in the navigation coordinate system are as follows: Based on the inertial measurement data, mechanical arrangement is performed to output the position information of the inertial measurement unit, specifically according to the specific force and angular velocity For mechanical arrangement, the error state differential equation of mechanical arrangement can be expressed as: ; The posture information of the inertial measurement unit includes: position ,speed ,attitude .

[0037] Furthermore, the velocity of the center of the inertial measurement unit can be calculated based on the relationship between the velocity of the center of the circle and the velocity of the point on the circle in circular motion. Calculate the speed to the wheel speed measurement unit to get the calculated speed of the carrier coordinate system , apply the formula: ; in, is the position information of the inertial measurement unit in the Earth-centered Earth-fixed coordinate system, For location information The differential of Differential representing position information The error, is the velocity information of the inertial measurement unit in the Earth-centered Earth-fixed coordinate system, Display speed information The error, For speed information The differential of Differential representing velocity information The error, is the attitude error of the inertial measurement unit, is the attitude error The differential of is the Earth's rotation angular velocity, is the rotation matrix from the carrier coordinate system to the Earth-centered Earth-fixed coordinate system, is the acceleration due to gravity, is the error in gravitational acceleration, is the velocity information in the carrier coordinate system, is the specific force in the carrier coordinate system, is the error in the specific force, is the angular velocity in the carrier coordinate system, is the error of the angular velocity in the carrier coordinate system, is the rotation matrix from the carrier coordinate system to the vehicle coordinate system, and the superscript T represents the transposed matrix. It is the lever arm information from the inertial measurement unit to the drive wheel.

[0038] Specifically, the lever arm information from the inertial measurement unit to the driving wheel can be measured before the vehicle moves. The arm information can also be obtained by looking up it in the vehicle's product manual or other instruction documents. This application does not limit the method of obtaining the arm information and the installation angle error.

[0039] Specifically, in S104, the kinematic observation equation of the Kalman filter is measured and updated based on the three-dimensional motion constraint observation value, including: The 3D motion constraint observation and reduction speed The prior residual of the kinematic observation equation is calculated by performing subtraction.

[0040] Specifically, based on the 3D motion constraint observation For ODO / NHC measurement update, the observation equation can be expressed as: ; in, is the prior residual of the above motion constraint observation equation, for n Tie to bThe rotation matrix of the system, for n The speed error under the system, for n The attitude error of the system, is the lever arm vector calibrated offline, is the bias error of the inertial measurement unit gyroscope, is the motion constraint observation noise, Represents the antisymmetric matrix operation of a vector.

[0041] Specifically, the arm distance from the IMU center to the GNSS antenna phase center (Antenna Reference Point, ARP) can be measured. b Projection under the system , and the lever arm from the IMU center to the drive wheel is b Projection under the system , and then complete the offline calibration.

[0042] After calculating the a priori residual of the current epoch, the a priori residual can be used as the velocity information observation of the Kalman filter, and the state estimation result of the current epoch is obtained by updating the Kalman filter, specifically including: The observation equation is: ; in, , Specifically, it is the observation value vector in the Kalman filter at time k , This includes the position error in the n-frame , speed error , attitude error and gyro bias , accelerometer bias The state quantity included.

[0043] is the coefficient matrix corresponding to each state quantity, which can be obtained according to the ODO / NHC observation equation. Those skilled in the art know the specific form of the ODO / NHC observation equation, and this application does not limit it.

[0044] is the observation noise, and the corresponding covariance matrix is The time update result before the measurement update is recorded as , the corresponding covariance is , then the measurement update process can be expressed as: ; ; ; Calculate the above intermediate variables Then, update the filter variables: ; ; in, The state estimation results obtained by ODO / NHC measurement update, The two together constitute the navigation output of the Kalman filter, and then the positioning result of the vehicle in the current epoch can be output based on the state estimation result.

[0045] In some embodiments, the training process of the non-holonomic constraint prediction model in S102 includes the following steps: First, initialize the static navigation of the vehicle, including: Obtain the calibrated lever arm information from the inertial measurement unit to the drive wheel ; Please refer to the above embodiment for the specific process of calibration, which will not be repeated here.

[0046] When the vehicle is in a stationary state and meets preset GNSS observation conditions, obtaining initial GNSS positioning, speed and attitude measurement data output by the GNSS signal receiving unit and initial inertial measurement data output by the inertial measurement unit; Initialize the navigation state of the vehicle based on the lever arm information, initial GNSS observations, and initial inertial measurement data, and obtain the initial state estimation result of the Kalman filter, including the initial position information of the vehicle. , initial velocity information and initial posture information .

[0047] It should be noted that the Kalman filter is a recursive algorithm that relies on the state estimate at the previous moment to infer the current state. Therefore, before the specific calculation, it is possible to provide an initial state estimate in a static state. This ensures the normal operation of the recursive process, accelerates the convergence of the filter, improves overall performance, and prevents the filter from diverging due to excessive initial errors.

[0048] The specific training process includes the following steps: S1021, driving the vehicle for a typical convergence time of the GNSS solution strategy while meeting the preset GNSS observation conditions, and then driving the vehicle for a preset time while meeting the preset GNSS observation conditions.

[0049] Specifically, the preset GNSS observation conditions may include the following conditions: The number of available GNSS satellites observed by the vehicle's GNSS signal receiving unit is greater than a preset number threshold, the dilution of precision factor of the GNSS positioning result is less than a preset dilution factor threshold, and the sum of the main diagonal elements of the position covariance matrix of the GNSS positioning result is less than a preset threshold; Among them, the number of available GNSS satellites observed by the GNSS signal receiving unit is specifically the satellite number NSAT (Number of SATellites). Generally, in order to achieve three-dimensional positioning (longitude, latitude, altitude), at least 4 satellites are required. The embodiment of the present application does not limit the value of the preset number threshold, and can be set considering different accuracy requirements and observation conditions and other factors.

[0050] It should be noted that the typical convergence time of the GNSS solution strategy is Refers to the time required from the receiver starting to receive satellite signals to the time the positioning solution reaches the required level of accuracy.

[0051] Specifically, the preset time length depends on the size of the sample set used for training, and the preset time length can be denoted as s.

[0052] S1022, during the driving process of the preset time length, obtain the inertial measurement data of each epoch output by the inertial measurement unit of the vehicle, the wheel speed data of the corresponding epoch output by the wheel speed measurement unit, and the GNSS positioning, speed and attitude data of the corresponding epoch output by the GNSS signal receiving unit.

[0053] Specifically, the preset driving time corresponds to the time interval ,in, is the end time of the typical convergence time of the GNSS solution strategy. , the velocity and attitude data of each epoch t can be calculated based on GNSS positioning b Vehicle speed , get the posterior corresponding to epoch t b The angular velocity output by the IMU three-axis gyroscope corresponding to the vehicle speed , the relative force of the three-axis accelerometer output , wheel speed output by the wheel speed measurement unit .

[0054] S1023: Constructing a sample set for training the non-holonomic constraint prediction model based on the inertial measurement data of each epoch, the wheel speed data of the corresponding epoch, and the GNSS positioning speed and attitude data of the corresponding epoch.

[0055] Specifically, the sample set construction process includes: The navigation coordinate system velocity of the vehicle at the corresponding epoch is calculated based on the GNSS positioning, speed and attitude measurement data, the navigation coordinate system velocity is converted to the carrier coordinate system to obtain the carrier coordinate system velocity, and the lateral velocity component and vertical velocity component of the carrier coordinate system velocity are calculated, specifically including: Get GNSS measurement update posterior solution speed ,attitude Calculate the direction cosine matrix , calculate the posterior b-system velocity , apply the formula: .

[0056] based on Calculate the lateral velocity component and vertical velocity component , the wheel speed data of the same epoch , angular velocity , compare As the input of the sample, the lateral velocity component of the same epoch and vertical velocity component As the sample label of the sample, the sample set is constructed .

[0057] S1024 , performing online training on the non-holonomic constraint prediction model based on the sample set, determining weight parameters of the converged non-holonomic constraint prediction model, and obtaining a trained non-holonomic constraint prediction model.

[0058] Specifically, taking the Long Short-Term Memory (LSTM) network model as an example, the training process applies the formula: ; Among them, the subscript Indicates the time corresponding to the epoch, subscript i indicates the input gate, subscript f indicates the forget gate, subscript c indicates the unit state, and subscript o indicates the output gate. represents the weight matrix, U represents the weight matrix, V represents the weight matrix, b represents the bias vector, x represents the input vector, Represents the hidden state. The functions involved include: Sigmoid activation function , hyperbolic tangent activation function . is the Hadamard product operation.

[0059] In some embodiments, after obtaining the trained non-holonomic constraint prediction model in S1024, the method further includes: During the driving of the vehicle, under the condition that the preset GNSS observation conditions are met, wheel speed data, inertial measurement data, and GNSS positioning speed and attitude data are obtained for each epoch; New training samples are constructed based on wheel speed data, inertial measurement data, and GNSS positioning speed and attitude data of the same epoch; Supplementing the new training samples to the sample set, and performing online training on the non-holonomic constraint prediction model using the supplemented sample set to update the weight parameters of the non-holonomic constraint prediction model to obtain an updated non-holonomic constraint prediction model; The step of calculating the nonholonomic constraint component of the current epoch by using the nonholonomic constraint prediction model is performed based on the updated nonholonomic constraint prediction model.

[0060] Specifically, in this way, the number of samples in the sample set can be expanded, and the non-complete constraint prediction model can be continuously updated as the vehicle is running, so that the trained model can adapt to the specific mechanical conditions of the vehicle and the driving habits of the driver, avoiding the defect that the non-complete constraint prediction model trained in batches can only provide accurate prediction results in the initial stage. The parameters of the model can be adjusted according to the specific mechanical conditions of the vehicle and the driving habits of the driver, thereby improving the calculation accuracy of the model and further improving the navigation effect of the combined navigation.

[0061] In some embodiments, the weight parameters of the non-completeness constraint prediction model obtained by training each vehicle at the corresponding time can also be saved, and a mapping relationship between each vehicle and the weight parameters of the non-completeness constraint prediction model can be established. When the driver changes to different vehicles, the non-completeness constraint prediction model can be adjusted in real time according to the mapping relationship between each vehicle and the weight parameters of the non-completeness constraint prediction model to adapt to the changes in the vehicle, improve the adaptability of the non-completeness constraint prediction model, and eliminate the impact caused by changes in the vehicle hardware conditions as much as possible.

[0062] The following are device embodiments of the present application, which can be used to implement the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0063] See next Figure 2 , is a schematic diagram of the structure of a vehicle integrated navigation device provided by an exemplary embodiment of the present application. The device can be implemented as all or part of a terminal through software, hardware, or a combination of both, and can also be integrated into a server as an independent module. A vehicle integrated navigation device in an embodiment of the present application can be applied to a terminal or the cloud. The device 20 includes a data acquisition unit 201, a constraint calculation unit 202, and a positioning result calculation unit 203, wherein: The data acquisition unit 201 is used to acquire the wheel speed data of the current epoch output by the wheel speed measurement unit of the vehicle, and acquire the inertial measurement data of the current epoch output by the inertial measurement unit of the vehicle; The constraint calculation unit 202 is configured to input the inertial measurement data and the wheel speed data into a nonholonomic constraint prediction model, and calculate the nonholonomic constraint components of the current epoch through the nonholonomic constraint prediction model; The constraint calculation unit 202 is further configured to construct a three-dimensional motion constraint observation value based on the non-holonomic constraint component and the wheel speed data; The positioning result calculation unit 203 is configured to measure and update the kinematic observation equation of the Kalman filter based on the three-dimensional motion constraint observation value, obtain a state estimation result obtained by the Kalman filter update, and output the positioning result of the vehicle at the current epoch based on the state estimation result; The non-holonomic constraint prediction model is trained based on a given sample set.

[0064] It should be noted that when the device 20 provided in the above embodiment executes a vehicle combination navigation method, the division of the above functional modules is only used as an example to illustrate. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the device provided in the above embodiment and the embodiment of a vehicle combination navigation method belong to the same concept. The implementation process is detailed in the method embodiment and will not be repeated here.

[0065] An embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the method of any of the above embodiments are implemented.

[0066] See Figure 3 , is a structural block diagram of an electronic device provided in an embodiment of the present application.

[0067] like Figure 3 As shown, the electronic device 300 includes a processor 301 and a memory 302 .

[0068] In the embodiment of the present application, the processor 301 is the control center of the computer system and can be the processor of a physical machine or the processor of a virtual machine. The processor 301 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 301 can be implemented in the form of at least one hardware of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), or PLA (Programmable Logic Array).

[0069] The processor 301 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state.

[0070] The memory 302 may include one or more computer-readable storage media, which may be non-transitory. The memory 302 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices and flash memory storage devices. In some embodiments of the present application, the non-transitory computer-readable storage medium in the memory 302 is used to store at least one instruction, which is used to be executed by the processor 301 to implement the method in the embodiment of the present application.

[0071] In some embodiments, the electronic device 300 further includes a peripheral device interface 303 and at least one peripheral device 304. The processor 301, memory 302, and peripheral device interface 303 may be connected via a bus or signal lines. Each peripheral device 304 may be connected to the peripheral device interface 303 via a bus, signal lines, or circuit boards. Specifically, the peripheral devices 304 include a display screen, a camera, and an audio circuit. The peripheral device interface 303 may be used to connect at least one I / O (Input / Output)-related peripheral device to the processor 301 and memory 302.

[0072] In some embodiments of the present application, the processor 301, the memory 302, and the peripheral device interface 303 are integrated on the same chip or circuit board; in some other embodiments of the present application, any one or two of the processor 301, the memory 302, and the peripheral device interface 303 may be implemented on separate chips or circuit boards. This embodiment of the present application is not specifically limited to this.

[0073] The electronic device structure block diagram shown in the embodiment of the present application does not constitute a limitation on the electronic device 300. The electronic device 300 may include more or fewer components than shown in the figure, or combine certain components, or adopt a different component arrangement.

[0074] The present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method of any of the aforementioned embodiments. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a microdrive and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic or optical card, a nanosystem (including a molecular memory IC), or any other type of medium or device suitable for storing instructions and / or data.

[0075] Through the description of the above embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the relevant technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A vehicle combined navigation method, characterized in that: include: Obtaining wheel speed data of a current epoch output by a wheel speed measurement unit of the vehicle, and obtaining inertial measurement data of the current epoch output by an inertial measurement unit of the vehicle; Inputting the inertial measurement data and the wheel speed data into a nonholonomic constraint prediction model, and calculating the nonholonomic constraint component of the current epoch through the nonholonomic constraint prediction model; Constructing a three-dimensional motion constraint observation value based on the nonholonomic constraint component and the wheel speed data; performing measurement updates on the kinematic observation equations of a Kalman filter based on the three-dimensional motion constraint observation values, obtaining a state estimation result obtained by the Kalman filter update, and outputting a positioning result of the vehicle at a current epoch based on the state estimation result; The non-holonomic constraint prediction model is trained based on a given sample set.

2. A vehicle combined navigation method according to claim 1, characterized in that: Inputting the inertial measurement data and the wheel speed data into a nonholonomic constraint prediction model, and calculating the nonholonomic constraint component of the current epoch by the nonholonomic constraint prediction model, includes: The inertial measurement data of the current epoch output by the inertial measurement unit of the vehicle includes an angular velocity output by a three-axis gyroscope and a specific force output by a three-axis accelerometer; Inputting the angular velocity output by the three-axis gyroscope, the specific force output by the three-axis accelerometer, and the wheel speed data of the current epoch into the nonholonomic constraint prediction model, so that the nonholonomic constraint prediction model predicts nonholonomic constraint components of the current epoch based on the input data, the nonholonomic constraint components including a lateral velocity component and a vertical velocity component; The three-dimensional motion constraint observation value is constructed based on the nonholonomic constraint component and the wheel speed data, and the formula is applied: ； in, is the three-dimensional motion constraint observation value, is the wheel speed data, is the lateral velocity component, is the vertical velocity component, t represents the corresponding epoch, and the superscript b indicates that the coordinate system of the corresponding parameter is the carrier coordinate system.

3. The vehicle combined navigation method according to claim 2, characterized in that: Before measuring and updating the kinematic observation equation of the Kalman filter based on the three-dimensional motion constraint observation value, the method further includes: Calculating the velocity of the center of the inertial measurement unit in the navigation coordinate system based on the inertial measurement data, and converting the velocity of the navigation coordinate system to the velocity of the wheel speed measurement unit to obtain the converted velocity of the vehicle coordinate system; The measuring and updating of the kinematic observation equation of the Kalman filter based on the three-dimensional motion constraint observation value includes: Subtracting the three-dimensional motion constraint observation value from the reduced velocity to calculate the prior residual of the kinematic observation equation; The prior residual is used as the velocity information observation of the Kalman filter, and the state estimation result of the current epoch is obtained by updating the Kalman filter.

4. The vehicle combined navigation method according to claim 3, characterized in that: The training process of the non-holonomic constraint prediction model includes the following steps: driving the vehicle for a typical convergence time of the GNSS solution strategy while meeting the preset GNSS observation conditions, and then driving the vehicle for a preset time while meeting the preset GNSS observation conditions; During the driving process of the preset duration, obtaining inertial measurement data of each epoch output by the inertial measurement unit of the vehicle, wheel speed data of the corresponding epoch output by the wheel speed measurement unit, and GNSS positioning, speed, and attitude data of the corresponding epoch output by the GNSS signal receiving unit; Constructing a sample set for training the non-holonomic constraint prediction model based on the inertial measurement data of each epoch, the wheel speed data of the corresponding epoch, and the GNSS positioning speed and attitude data of the corresponding epoch; The non-holonomic constraint prediction model is trained online based on the sample set, and the weight parameters of the converged non-holonomic constraint prediction model are determined to obtain a trained non-holonomic constraint prediction model.

5. The vehicle combined navigation method according to claim 4, characterized in that: The method of causing the vehicle to travel under preset GNSS observation conditions before the typical convergence time of the GNSS solution strategy is reached also includes: Calibrate and obtain lever arm information from the inertial measurement unit to the driving wheel; When the vehicle is in a stationary state and meets preset GNSS observation conditions, obtaining initial GNSS positioning, speed and attitude measurement data output by the GNSS signal receiving unit and initial inertial measurement data output by the inertial measurement unit; The navigation state of the vehicle is initialized based on the lever arm information, the initial GNSS observation value and the initial inertial measurement data, and an initial state estimation result of a Kalman filter is obtained, including initial position information, initial velocity information and initial attitude information of the vehicle.

6. The vehicle combined navigation method according to claim 4, characterized in that: The sample set for training the non-holonomic constraint prediction model is constructed based on the inertial measurement data of each epoch, the wheel speed data of the corresponding epoch, and the GNSS positioning speed and attitude data of the corresponding epoch, including: Calculating the navigation coordinate system velocity of the vehicle at the corresponding epoch based on the GNSS positioning, speed, and attitude measurement data, converting the navigation coordinate system velocity to a vehicle coordinate system to obtain a vehicle coordinate system velocity, and calculating a lateral velocity component and a vertical velocity component of the vehicle coordinate system velocity; The wheel speed data, angular velocity, and specific force of the same epoch are used as sample inputs, and the lateral velocity component and vertical velocity component of the same epoch are used as sample labels of the samples to construct the sample set.

7. A vehicle combined navigation method according to any one of claims 4 to 6, characterized in that: The preset GNSS observation conditions include: The number of available GNSS satellites observed by the vehicle's GNSS signal receiving unit is greater than a preset number threshold, the dilution of precision factor of the GNSS positioning result is less than a preset dilution factor threshold, and the sum of the main diagonal elements of the position covariance matrix of the GNSS positioning result is less than a preset threshold; After obtaining the trained non-holonomic constraint prediction model, the method further includes: During the driving of the vehicle, under the condition that the preset GNSS observation conditions are met, wheel speed data, inertial measurement data, and GNSS positioning speed and attitude data are obtained for each epoch; New training samples are constructed based on wheel speed data, inertial measurement data, and GNSS positioning speed and attitude data of the same epoch; Supplementing the new training samples to the sample set, and performing online training on the non-holonomic constraint prediction model using the supplemented sample set to update the weight parameters of the non-holonomic constraint prediction model to obtain an updated non-holonomic constraint prediction model; The step of calculating the nonholonomic constraint component of the current epoch by using the nonholonomic constraint prediction model is performed based on the updated nonholonomic constraint prediction model.

8. A vehicle combined navigation device, characterized in that: include: a data acquisition unit, configured to acquire wheel speed data of a current epoch outputted by a wheel speed measurement unit of the vehicle, and acquire inertial measurement data of a current epoch outputted by an inertial measurement unit of the vehicle; a constraint calculation unit, configured to input the inertial measurement data and the wheel speed data into a nonholonomic constraint prediction model, and calculate the nonholonomic constraint component of the current epoch using the nonholonomic constraint prediction model; The constraint calculation unit is further configured to construct a three-dimensional motion constraint observation value based on the nonholonomic constraint component and the wheel speed data; a positioning result calculation unit, configured to measure and update the kinematic observation equation of a Kalman filter based on the three-dimensional motion constraint observation value, obtain a state estimation result obtained by the Kalman filter update, and output a positioning result of the vehicle at a current epoch based on the state estimation result; The non-holonomic constraint prediction model is trained based on a given sample set.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images