RU2849357C2 - Method for integrating ins and gnss data to solve object navigation problems - Google Patents

Abstract

FIELD: navigation.

SUBSTANCE: the invention relates to the field of navigation, specifically to methods for integrating navigation information obtained from global navigation satellite systems (GNSS) and inertial navigation systems (INS). The method for integrating INS and GNSS data includes: obtaining information from three accelerometers and three angular velocity sensors (AVS) installed on the object in axes that are assumed to be orthogonal and form two coinciding triangles relative to the object; averaging the obtained data, performing an initial setting of the roll and pitch angles, as well as the course based on the data on the track angle obtained from the GNSS; forming a state vector by integrating the data obtained from the accelerometers and AGS. At the same time, to compensate for errors in the accelerometers and DUS, a covariance matrix is formed; the Kalman optimal filtering procedure is performed.

EFFECT: increased accuracy of navigation data determination.

1 dwg

Description

The invention relates to the field of implementing navigation systems, namely to methods for integrating and correcting navigation information obtained from satellite navigation systems and inertial sensors as part of a moving object.

A known prior art method for optimally estimating errors in an inertial navigation system and correcting it using a fixed ground reference with known geographic coordinates (RU Patent for Invention 2713582, published 05.02.2020) includes angular tracking of a fixed ground reference correction (GRC) and discrete measurement of the slant range to it in a gentle operating mode for the laser rangefinder included in the surveillance and targeting system with a frequency of emitting pulses of 0.5-1.0 Hz and is based on the combined processing of the measured current sighting angles of the GRC and the slant range to it, the current angles of the true and gyroscopic courses, roll and pitch of the object and the geographic coordinates of its location and the current baro-inertial altitude calculated by the inertial navigation system (INS). In this case, in the mode of continuous angular tracking of the object, one- to two-second time intervals between adjacent measurements of the range to the object are filled with ten-hertz calculated values, which are formed in accordance with a modified elevation procedure for determining the slant range, invariant to the relief of the underlying surface, which assumes the use of the current baroinertial altitude of the object, the cosine of the angle between the geographic vertical and the direction to the object, and the reference value of the altitude of the object above sea level formed from the measurements of the OPS, while the components of the absolute linear velocity of the object are estimated in accordance with the kinematic model of its movement relative to a stationary ground object in projections onto the axes of the inertial coordinate system. In this case, two parallel optimal assessment procedures are implemented - the main and auxiliary ones, the first of which ensures the assessment of the extended vector of INS state parameters and the subsequent correction of its navigation and piloting parameters, and the second one - the formation of adequate positional and speed signals of satellite navigation systems (SNS), used in the main optimal assessment procedure as signals of an ideal measuring instrument.

The main disadvantages of this method include:

- the impossibility of improving the accuracy of determining the location of an object in the absence of ground reference points, since in order to correct the “drift” errors of sensors in the INS from the navigation system, it is necessary to have a geographic reference to the environment;

- the need to maintain a database of fixed ground landmarks indicating their geographic coordinates;

- the need to determine the distance to the object using third-party measuring devices (laser rangefinder).

Also known is a method for determining the instrumental errors of the inertial navigation system's measuring instruments at the initial alignment stage (RU patent for invention 2300081, published 05/27/2007) including measuring the output signals of gyroscopes, accelerometers and sensors of the attitude and heading angles relative to the object, generating signals proportional to the positional and integral components of the horizontal components of the apparent acceleration and gyroscopic heading for constructing a calibration circuit, characterized in that the inertial attitude and heading system with gyroscopic angular velocity measuring instruments and accelerometers rigidly fixed to it is forcibly rotated relative to the three structural axes of the object without using gyroscopic stabilization, then the values of the ideal output signals of the gyroscopes and accelerometers are calculated and compared with the output signals of the gyroscopes and accelerometers taken from the measuring instruments during the initial preparation of the inertial navigation system, and the following is determined for the gyroscopes: drift gyroscopes, scale factor errors, scale factor asymmetry errors, and gyroscopic sensitivity axis misalignments; for accelerometers: accelerometer errors, scale factor errors, scale factor asymmetry errors, errors in the deviation of the accelerometer's center of mass from the coordinate origin of the coordinate system associated with the sensor block, and gyroscopic sensitivity axis misalignments.

The disadvantage of this method is the inability to use the mathematical models used to determine the errors of the INS sensors during the operation of the technical equipment, as a result of which the accuracy of determining the navigation position decreases due to the accumulation of measurement errors of the said sensors.

In addition, a disadvantage of this method is the need to install additional specialized equipment (rotary platform elements, motors, etc.), which leads to a decrease in operational reliability, difficulty in maintenance, and an increase in the cost of the product.

A method for estimating errors in inertial information and its correction based on SNS measurements is known (Russian Federation patent for invention No. 2617565, published on April 25, 2017). The method consists in using the traditional procedure of optimal filtering and Kalman identification, for which the measurement signals of the optimal filter-identifier are formed by comparing the same geographic coordinates of the location and the horizontal components of the absolute linear velocity in projections on the axes of the reference trihedron of the gyroplatform (GP) of the INS, formed from the measurements of the satellite navigation system (SNS), and its structure is synthesized in accordance with the traditional INS error model, while the nature of the flight is methodically organized in such a way that after 270 seconds of straight and horizontal flight, during which precise "horizontalization" of the gyroplatform is implemented and the well-observed parameters of the horizontal channels of the INS are estimated, a maneuver such as a "snake", coordinated or combat turn is carried out, after which the active phase of the optimal filtering and identification procedure is suspended and the filter-identifier is transferred to the long-term mode - until the next correction session, forecast, for the implementation of which the measurement signals are reset to zero, and the values of the estimates at the time of completion of the active phase of the evaluation procedure are used as initial conditions in the forecast procedure, while the forecast itself is carried out in accordance with discrete equations for calculating a priori estimates of ANN errors, and the correction of the ANN output parameters - the geographic coordinates of the location and the components of the absolute linear velocity - is implemented in an open ANN circuit, for which the current predicted values of the estimates of the ANN state parameters are used. In this case, the INS error model is expanded by including a mathematical description of the coordinates of its location relative to the antenna unit (AU) of the SNS and presenting them in the form of a system of three interconnected first-order differential equations in projections onto the axes of the reference trihedron of the INS GP, which simultaneously describe the kinematic components of the relative speed of the INS movement that are additively included in the speed measurement signals, and when generating the measurement signals and the observation matrix, kinematic relationships are used that relate the errors Δφ, Δλ, Δχ of the dead reckoning of the geographic coordinates of the location and the angle of azimuthal orientation of the reference trihedron of the INS GP with the errors of maintaining the vertical αx, αy and the angle αz of the azimuthal drift of the INS GP with an accuracy of up to second-order smallness values relative to such parameters as Δφ, Δλ, αх, αy, αz, ensuring the determination of the current values of the elements of the message and observation matrices.

The well-known patent discusses a gyroplatform (GP), which has several disadvantages compared to strapdown inertial navigation systems (SINS): high cost, difficulty in maintenance, and the inability to apply the proposed maneuvers to ground vehicles. Furthermore, the system's redundancy due to the use of two INSs is a drawback.

Also known is an integrated inertial-satellite navigation system (RU Patent for Invention No. 2277696, published on 10.06.2006). The system comprises a radio receiver connected via an amplifier to an antenna, the outputs of which are connected to a navigation satellite positioning computer, and the inputs to a satellite orbit data almanac initial setup unit, the outputs of which are connected to the inputs of a radio-visible satellite selection unit, the outputs of which are connected to the first group of inputs of a working satellite constellation selection unit, the outputs of which are connected to the inputs of a user positioning unit, as well as an absolute angular velocity projection meter and an apparent acceleration vector projection meter, their outputs respectively connected via an angular velocity corrector and an apparent acceleration corrector to the first group of inputs of the navigation parameter computer, the outputs of which are connected to the first group of outputs of the system. The system also includes an initial data computer, connected by three groups of inputs respectively to the outputs of the absolute angular velocity projection meter and the apparent acceleration vector projection meter, the outputs of the data integration unit, and the outputs of the user location calculation unit. Some of the outputs of the initial data computer are connected to the inputs of the navigation parameter computer, and all of the outputs are connected to the first group of inputs of the data integration unit, the second group of inputs of which is connected to the outputs of the angular velocity corrector and the apparent acceleration corrector, and the third group of inputs is connected to the outputs of the user location calculation unit. One group of outputs of the data integration unit is connected to the second group of inputs of the working satellite constellation selection unit, another group of outputs is directly connected to the second group of system outputs, the third group of outputs is connected to the inputs of the apparent acceleration corrector, and the fourth group of outputs is connected to the inputs of the angular velocity corrector and the second group of inputs of the initial data computer.

This patent considers two gyroscope units, which is redundant. Furthermore, this system is aimed at overcoming fault tolerance rather than navigation accuracy.

The goal of the navigation task is to determine the geographic coordinates (latitude, longitude, altitude) of the controlled object, the velocity vector, and the attitude angles (heading, roll, pitch) of the object. This task is accomplished by integrating data from the INS (inertial navigation system) and GNSS (global navigation satellite systems). The GNSS generates only the coordinates and velocity of the object (the use of multi-antenna satellite receivers is not considered), while the INS generates acceleration and angular velocity.

In practice, each sensor has errors that must be assessed during integration. Furthermore, due to loss of satellite signals, existing solutions quickly degrade in accuracy, which is unacceptable for agricultural robots.

The technical result achieved by implementing this invention consists in increasing the accuracy of determining navigation data of a moving object, due to the possibility of integrating INS and GNSS data with compensation for the measurement errors of the sensors within the INS, the possibility of autonomous movement when the satellite signal is lost, the absence of the need to use multi-antenna satellite receivers, and an increase in the stability of the navigation solution due to the thermostatting of inertial sensors.

The specified technical result is achieved by a method of integrating INS and GNSS data to solve the problem of object navigation, including obtaining output information from three accelerometers and three angular rate sensors (ARS) installed on the object in axes assumed to be orthogonal and forming two coinciding trihedrons relative to the object, the obtained data are averaged and an initial pitch and roll angle setting is performed, then an initial heading setting is performed based on the data received from the GNSS on the course angle, then a state vector is formed by integrating the data received from the accelerometers and ARS using Poisson's kinematic equations, after which, to compensate for errors in the composition of the accelerometers and ARS, a covariance matrix is formed based on the data received from the GNSS in accordance with the following measurement model:

Next, optimal Kalman filtering procedures are carried out, with the state vector updated in accordance with the following formulas:

The technical solutions used in this invention allow for the integration of INS and GNSS data, taking into account errors arising in the INS sensors, to solve object navigation problems.

The inertial sensors used consist of three accelerometers, positioned in orthogonal axes, and three RUSs, also positioned in orthogonal axes. These two trihedrons are assumed to coincide. The INS sensors are rigidly mounted on the control object. The coordinates of the INS's reduced center and installation angles are assumed to be known in the object's structural axes. The GNSS antenna is mounted on the roof of the object, with known coordinates in the object's structural axes.

Temperature calibration of the sensors is not performed. The inertial sensors are heated to a constant temperature of 80°C, which is maintained with a high accuracy of 0.01°C by a thermostatting system to prevent temperature errors. Thermostatting is achieved by surrounding the inertial sensor with resistors, and it is controlled by an integrated temperature sensor using a PID controller. The heating temperature is selected so that the sensors cannot exceed this temperature due to external factors. The thermostatting unit is not shown separately in the figure.

The accelerometer readings are assumed to contain errors in the form of apparent acceleration, an unknown zero offset that is constant from run to run, and random noise, assumed to be white, of known intensity.

The VRS readings are subject to the same error model as the accelerometer. Note that we will use the historically justified term for the VRS reading's zero offset—drift.

The method is based on a linear discrete Kalman filter (DKF) in real-time feedback loops. The DKF is based on a state vector x and a covariance matrix P. The state vector includes coordinate and velocity errors, angular errors, and current estimates of the VSS drifts and accelerometer biases. The covariance matrix is a matrix consisting of pairwise error covariances of all components of the state vector.

The functional comprehension (FC) consists of three stages: initialization, correction, and prediction. At each step of the algorithm, we have a current estimate of the state vector and a current estimate of the covariance matrix.

Data from the INS and GNSS are sent to the control unit - a computer that processes and integrates the data in accordance with the algorithm described below.

The technical essence of the invention is explained by the drawings. Fig. 1 shows a block diagram of the algorithm for integrating INS and GNSS data.

The algorithm for integrating INS and GNSS data contains:

1. 1. Initial pitch and roll display;

2. 2. Initial exhibition of the course;

3. 3. Autonomous calculation algorithm;

4. 4. Checking the availability of GNSS data;

5. 5. Correction according to GNSS decision;

6. 6. Taking into account feedback;

7. 7. Covariance prediction.

In a preferred embodiment, the invention is carried out as follows.

At the first stage, data from the INS sensors is fed to the control unit, where it is averaged and the initial pitch and roll angles are set (1). Subsequently, as the object moves, data from the GNSS system is fed to the control unit, which sets the initial heading (2). The general data processing algorithm then switches to the autonomous dead reckoning algorithm (3). The input to the autonomous dead reckoning algorithm (3) is the next INS sensor readings and the navigation solution from the previous iteration, and the output is the navigation solution for the current step. One cycle of the autonomous dead reckoning is the numerical integration of the accelerometer and RUS readings, where the initial conditions are the navigation solution from the previous iteration. The integration of the accelerometer readings is performed using the Euler method. The integration of the RUS readings is understood as the numerical integration of Poisson's kinematic equations. Following this, the availability of GNSS data is checked (4). If no data is available, the obtained navigation solution is considered valid, and the algorithm proceeds to a new autonomous dead reckoning cycle. If GNSS data is available, the algorithm proceeds to the GNSS decision correction procedure (5). The GNSS decision refers to the geographic coordinates, speed, and course angle obtained by the satellite navigation algorithm. During the GNSS decision correction stage (5), the current estimates of the state vector and covariance matrix are modified in accordance with the measurement model (in our case, the measurement is the GNSS decision) to ensure the ability to correct for errors in the GNSS sensor array:

Where:

- gain matrix;

- vector of measurement at a moment in time ;

- estimate of the state vector at a given moment in time , using measurements (a priori estimate);

- estimate of the state vector at a given moment in time , using measurements (a posteriori estimate);

- estimate of the covariance matrix of the estimate error at a point in time taking into account measurements ;

- estimate of the covariance matrix of the estimate error at a point in time taking into account measurements ;

Matrix associates the components of the state vector with the measurement , in accordance with the measurement model;

- measurement noise covariance matrix.

After performing the correction based on the GNSS solution (5), the algorithm proceeds to the feedback accounting stage (6). At this stage, the components of the current state vector (recall that the state vector contains errors) are subtracted from the corresponding components of the navigation solution, and the current estimates of the VSS drifts and accelerometer biases are added to the corresponding integral estimates. The state vector is reset to zero. At the covariance prediction stage (7), optimal Kalman filtering procedures are performed, based on the INS error equations. The state vector and covariance matrix are updated according to the formulas:

,

Where:

- the covariance matrix of the noise of the dynamic system (in our case, the noise of the inertial sensors).

The proposed method involves thermostatting the INS unit, maintaining the inertial sensors at a specific temperature. This eliminates the need for temperature calibration of the inertial sensors.

The inertial sensor unit is located inside the navigation unit, together with the GNSS antenna.

The proposed method uses the properties of vehicle kinematics - correction by zero components of the velocity vector of that point of the vehicle, which is determined by the kinematic model of the vehicle.

Thus, the described method is applied to the task of object navigation, in which the integration of data obtained from inertial sensors installed on the object and the GNSS system is used to improve navigation accuracy. Improved accuracy is also achieved through the use of a covariance matrix to compensate for measurement errors in these inertial sensors.

Claims (4)

A method for integrating INS and GNSS data to solve the problem of object navigation, characterized by the fact that output information is obtained from three accelerometers and three angular rate sensors installed on the object in axes assumed to be orthogonal and forming two coinciding trihedrons relative to the object, the obtained data are averaged and an initial pitch and roll angle setting is performed, then an initial heading setting is performed based on the data received from the GNSS on the course angle, then a state vector is formed by integrating the data received from the accelerometers and the RRS, using Poisson's kinematic equations, after which, to compensate for errors in the composition of the accelerometers and RRS, based on the data received from the GNSS, a covariance matrix is formed in accordance with the following measurement model: Next, optimal Kalman filtering procedures are carried out, with the state vector updated in accordance with the following formulas: