Attitude and direction estimation device, satellite communications earth station, and attitude and direction estimation method

Abstract

To improve a posture angle estimation error.SOLUTION: A posture azimuth estimation device comprises: an arrival angle measurement unit; an inertia measuring unit; an absolute coordinate output unit; and a posture azimuth calculation unit. The arrival angle measurement unit is configured to receive a radio wave transmitted from a satellite and measure an arrival angle of the received radio wave. The inertia measuring unit at least includes an accelerometer, and an angular velocity meter. The absolute coordinate output unit is configured to acquire an absolute coordinate in a current place. The posture azimuth calculation unit is configured to: subtract in-advance estimated linear acceleration from acceleration measured by the accelerometer to calculate a gravity direction measurement value; generate a first difference vector serving as a difference in a gravity direction estimated from the in-advance estimated posture; generate a second difference vector serving as a difference between a measurement arrival angle the arrival angle measurement unit acquires and an arrival angle estimated from the absolute coordinate, orbit information on the satellite, and the in-advance estimated posture; and, based on the first difference vector and the second difference vector, update the in-advance estimated posture by a complementary filter to acquire an estimation posture.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] Embodiments of the present invention relate to an attitude and direction estimation device, a satellite communications earth station, and an attitude and direction estimation method. [Background technology]

[0002] Satellite communications allow communication even in environments where conventional communication lines, fiber optic cables, mobile phones, and Wi-Fi are unavailable, as long as there is a line of sight to the satellite. There is particularly high demand for satellites mounted on vehicles and ships for communication while on the move. When the target satellite is a geostationary satellite, the angle between geostationary satellites is small (only a few degrees), requiring high-precision beam targeting to prevent radio waves from leaking to other satellites. If the earth station is attached to a mobile device, it must continuously maintain high-precision beam targeting of the satellite while moving.

[0003] Linear polarization is often used in satellite communications using geostationary satellites. Polarization has two orthogonal axes, each assigned to different data. However, with linear polarization, even a slight deviation in polarization angle can leak into cross-polarization, creating interference waves and degrading the communication quality of cross-polarized communication lines. Therefore, the cross-polarization ratio during transmission is specified, and is a very strict value, for example, 30 dB or more (within ±1.8 degrees of the linear polarization angle). The azimuth and attitude of a moving earth station change with movement. In the case of linear polarization, even if the antenna outputs a constant polarization, if the antenna's azimuth and attitude change, the polarization as seen from the satellite will change, and it may no longer meet the standard. Therefore, earth stations for mobile devices are equipped with an IMU (Inertial Measurement Unit) that includes sensors such as gyroscopes and accelerometers, and it is necessary to estimate its own attitude from the sensor output and control it so that the polarization as seen from the satellite always stays within the standard.

[0004] One method for estimating direction and attitude is the AHRS (Attitude Heading Reference System). While many AHRS methods exist, when high-level attitude control is required, as mentioned above, the usable algorithms are limited. One such method is a complementary filter that uses a Kalman filter, which estimates the attitude angle, gyro bias, linear acceleration, and magnetic disturbance by iteratively processing with the filter. [Prior art documents] [Non-patent literature]

[0005] [Non-Patent Document 1] “Freescale Sensor Fusion Kalman Filter (fusion.c) Technical Note,” https: / / github.com / memsindustrygroup / Open-Source-Sensor-Fusion / tree / master / docs [Internet], accessed March 10, 2023. [Non-Patent Document 2] MATLAB® Sensor fusion and tracking toolbox, imufilter, https: / / jp.mathworks.com / help / fusion / ref / imufilter-system-object.html [Internet], accessed March 10, 2023. [Non-Patent Document 3] D. Roetenberg, et. al., "Compensation of Magnetic Disturbances Improves Inertial and Magnetic Sensing of Human Body Segment Orientation", IEEE Transactions on Neural Systems and Rehabilitation Engineering, Vol. 13, No. 3, September 2004 [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] The above techniques include a method that does not use magnetism to suppress the significant accuracy degradation caused by magnetic disturbances from metals. This method does not estimate orientation, only integrates raw sensor output data, and determines only attitude. This filter estimates three states—attitude angle error, gyro bias error, and linear acceleration error—from a single measurement, the difference in gravity vectors. How the single error is distributed among the three is determined by various covariance values ​​used in the Kalman filter. However, because the states change dynamically, correctly distributing the errors is difficult. As a result, when the movement changes drastically, not only the orientation accuracy but also the attitude accuracy deteriorates compared to algorithms that use magnetism.

[0007] Embodiments of the present invention solve these problems and, as an example of a problem that is not limited thereto, provide an attitude and direction estimation device that improves attitude angle estimation error. [Means for solving the problem]

[0008] According to one embodiment, the attitude and direction estimation device comprises an angle of arrival measuring unit, an inertial measurement unit, an absolute coordinate output unit, and an attitude and direction calculation unit. The angle of arrival measuring unit receives radio waves transmitted from a satellite and measures the angle of arrival of the received radio waves. The inertial measurement unit comprises at least an accelerometer and an angular velocity meter. The absolute coordinate output unit obtains the absolute coordinates of the current location. The attitude and direction calculation unit calculates a gravity direction measurement value by subtracting a previously estimated linear acceleration from the acceleration measured by the accelerometer, generates a first difference vector which is the difference between this value and the gravity direction estimated from the previously estimated attitude, generates a second difference vector which is the difference between the measured angle of arrival obtained by the angle of arrival measuring unit and the angle of arrival estimated from the absolute coordinates, the satellite's orbital information, and the previously estimated attitude, and updates the previously estimated attitude with a complementary filter based on the first difference vector and the second difference vector to obtain the estimated attitude. [Brief explanation of the drawing]

[0009] [Figure 1] FIG. schematically showing an example of an attitude and orientation estimation device according to an embodiment. [Figure 2] GRAPH showing an example of the output of an attitude and orientation estimation device according to an embodiment. [Figure 3] FIG. schematically showing an example of an attitude and orientation estimation device according to an embodiment. [Figure 4] FIG. schematically showing an example of a satellite communication earth station according to an embodiment. [Figure 5] FIG. showing an example of the relationship between polarization and attitude. [Figure 6] FLOWCHART showing an example of the processing of an attitude and orientation estimation device according to an embodiment.

MODE FOR CARRYING OUT THE INVENTION

[0010] Hereinafter, embodiments will be described with reference to the drawings. In the following embodiments, only the parts that are essentially necessary for the configuration of the present embodiment are shown, and the parts not related to the operation of the present embodiment may be omitted from illustration and description.

[0011] In the Kalman filter, the parameter estimated within the filter loop is called "state", and the estimation proceeds in two steps: prediction and update. The state estimated by this algorithm is not the estimation target of the algorithm itself such as the attitude angle or gyro bias, but the error parameters such as the attitude angle error, gyro bias error, linear acceleration error, and magnetic disturbance error. Since it is an error, in the prediction step, the state prediction value is set to 0.

[0012] The "measured values" of the Kalman filter, that is, the measurement data used to calculate the state "update" value by subtracting the predicted value from the measured value and multiplying it by the Kalman gain, are the gravity vectors due to the acceleration measured by the accelerometer. In conventional methods, the magnetic vectors measured by the magnetic sensor are also used. Specifically, the "measured values" are two vectors: (1) the difference vector obtained by subtracting the predicted linear acceleration from the measured acceleration and subtracting the gravity vector corresponding to the downward direction predicted from the predicted attitude, and (2) the difference vector between the north direction vector calculated by subtracting the predicted magnetic disturbance from the measured magnetism and the north direction vector predicted from the predicted attitude direction. As mentioned above, the predicted state value is 0, so the estimated state value is calculated by directly multiplying these measured values ​​by the Kalman gain. Note that this filter is a complementary filter that uses the error vector as the measured value.

[0013] The updated error state is used to update the attitude angle, gyro bias, linear acceleration, and magnetic disturbance values. The predicted values ​​for the next step are calculated as follows: The attitude angle is predicted from the gyro measurement data, the gyro bias is predicted directly from the estimate in the previous step, and the linear acceleration and magnetic disturbance are predicted by multiplying the estimates from the previous step by a conventionally fixed coefficient between 0 and 1. In a normal Kalman filter, the "state" prediction value is created from the updated values ​​in the previous step to form a loop, but in this filter, since the "state" prediction value is 0, multiple vectors that form the "measured values" are generated from the predicted values ​​such as attitude angle to form a loop.

[0014] While the system utilizes the Earth's magnetic field measured by magnetic sensors, the magnetic vectors measured by these sensors often do not point to the correct north. There are two main reasons for this. First, the magnetic field emitted by nearby magnetized iron objects interferes with the measurement. Second, the direction indicated by the Earth's magnetic field is not necessarily true north. When magnetized metals are nearby, it is not uncommon for the magnetic field to point in a direction far removed from north. Therefore, algorithms that use magnetism are unsuitable when you want to determine the attitude angle with high precision to obtain a cross-polarization ratio of 30 dB or higher.

[0015] There are algorithms that determine only the attitude without using magnetism and disregarding directional accuracy. One method removes magnetic vectors from the "measured values" and estimates the attitude angle error, gyro bias error, and linear acceleration error, which are the state of the object, from only the difference in gravity vectors.

[0016] (First Embodiment) Figure 1 shows one typical implementation according to one embodiment. The attitude and direction estimation device 100 comprises an angle of arrival measuring unit 2, an inertial measurement unit (hereinafter referred to as IMU 3), an absolute coordinate output unit 4, an antenna 5, and a received radio wave processing unit 6.

[0017] The angle of arrival measurement unit 2 processes the radio waves received by the antenna 5 in the received radio wave processing unit 6, performing RF and baseband processing such as down-conversion and filtering to convert them into a signal that can measure the angle of arrival, and then estimates and outputs the angle of arrival. The IMU 3 includes at least a 3-axis gyroscope and a 3-axis accelerometer, and outputs the measured angular velocity and acceleration. The absolute coordinate output unit 4 includes, for example, a GPS (Global Positioning System), and outputs the absolute coordinates on Earth of the mobile object to which the GPS is attached, such as latitude, longitude, and altitude. This could be a GNSS receiver other than GPS, or if the mobile station's range of movement is extremely narrow, it could be a memory that stores fixed values.

[0018] The attitude and orientation calculation unit 1 receives input from these outputs and estimates the attitude using a filter. The filter receives input and performs iterative estimation in discrete time, and has the updated value from the previous step and the predicted value for the new step. This is used along with the new data input to proceed with the estimation. Hereafter, the updated and predicted values ​​may be referred to as "estimated" or "estimated value" without distinction. Specifically, the estimation process is represented by the following steps.

[0019] The direction of gravity is calculated by subtracting the predicted linear acceleration from the input acceleration data, and the direction of gravity is calculated from the predicted attitude, and the difference vector (first difference vector, gravity vector difference) is calculated.

[0020] Next, the satellite's direction, azimuth, and elevation are calculated from the device's absolute coordinate data and pre-stored satellite orbit information. The calculated satellite direction is the direction defined on the ground plane (hereinafter referred to as the global frame), not the direction defined on the antenna plane. The satellite direction vector defined on the antenna plane is calculated from the calculated satellite direction and estimated attitude, and the difference vector (second difference vector, arrival angle vector difference) between this and the satellite direction vector calculated from the input arrival angle data is calculated.

[0021] The two difference vectors, the first and second difference vectors, contain information about the difference between the current estimated attitude and the actual attitude. However, since these are generated from measurements that contain a lot of noise, it is necessary to apply an appropriate filter. A filter that estimates the response from the difference between two series of variables with different frequency responses is called a complementary filter.

[0022] This algorithm processes data in discrete time. Furthermore, the pose and other parameters to be estimated cannot be directly measured. A typical filter for estimating such unobservable states in discrete time is the Kalman filter. The following explanation will use the Kalman filter as an example, but the filter used in this disclosure is not limited to the Kalman filter; other filters with similar functionality may be used.

[0023] In a Kalman filter, the states are defined as attitude angle error, gyro bias error, and linear acceleration error. The arrival angle is not affected by disturbances like magnetism, which are influenced by previous states; it is only affected by noise errors or outliers similar to fading fluctuations. However, if there is enough foresight to determine the arrival angle, the errors are basically random. Therefore, states equivalent to magnetic disturbances, such as arrival angle deviation, are not necessary, and the states are the same three errors as in the case of no magnetism. The equation of state is the same as the equation of state listed in, for example, the MATLAB Sensor fusion and tracking toolbox, imufilter, https: / / jp.mathworks.com / help / fusion / ref / imufilter-system-object.html, and therefore the state covariance is also the same. On the other hand, the various equations related to measured values ​​are different. Only the differences are described below. For other parts, please refer to the above-mentioned literature. Note that the notation for variables and other elements is similar to that used in the document “Freescale Sensor Fusion Kalman Filter (fusion.c) Technical Note,” https: / / github.com / memsindustrygroup / Open-Source-Sensor-Fusion / tree / master / docs, but this does not affect the scope of rights in this disclosure.

[0024] In the following formulas, the + in the upper right corner represents the updated value, the - represents the predicted value, and the k in the lower right corner represents the sample number. The G in the upper left corner represents the global frame, and S represents the sensor frame (frame defined on the antenna plane). T represents the transpose.

[0025] Measurement value S z ε,k This can be expressed as follows:

number

[0026] S g^ G,kis the gravity direction vector obtained from the current posture, S g^ A,k is the gravity direction vector obtained by subtracting the linear acceleration from the acceleration measurement value. S u^ G,k is the satellite direction vector obtained from the current posture orientation, absolute coordinates, and satellite orbit information, S u^ SL,k is the satellite direction vector calculated from the measured arrival angle. The unit of gravity is [g]. Also, the units of angles and each velocity are basically degrees and dps. Let the sample interval be δt, then the observation matrix for converting the predicted value to the state can be expressed as follows.

Number

[0027] α = π / 180, I3 is the 3 × 3 identity matrix, O3 is the 3 × 3 zero matrix, and (ω×) is expressed as follows. However, ω = [ω x , ω y , ω z .

[0028]

Number

[0029] The observation noise covariance can be expressed as follows.

[0030]

Number

[0031] [[ID=**50**]] [[ID=**51**]]Q[[ID=**52**]] vA [[ID=**53**]]is the acceleration noise variance [g[[ID=**54**]] 2 [[ID=**55**]]], Q[[ID=**56**]] wA [[ID=**57**]]is the linear acceleration noise variance [g[[ID=**58**]] 2 [[ID=**59**]]], Q[[ID=**60**]] vG [[ID=**61**]]is the gyroscope noise variance [(deg / s)[[ID=**62**]] 2 [[ID=**63**]]], Q[[ID=**64**]] wb [[ID=**65**]]is the gyroscope drift noise variance [(deg / s)[[ID=**66**]] 2 [[ID=**67**]]], Q[[ID=**68**]] vSL It should be noted that in the original text, the tags and vA etc. seem to be some kind of specific identifiers in a certain system. If there are specific requirements or meanings for these tags in the context, the translation may need to be adjusted accordingly. Also, the text contains some mathematical notations and expressions which are translated as accurately as possible while maintaining the original format.This is the noise dispersion of the angle of arrival.

[0032] Since the state is the same as in the MATLAB document mentioned above, the state covariance can also be the same as in the MATLAB document. However, the state covariance differs slightly depending on the document; in the MATLAB document, each 3 × 3 block in the matrix contains only diagonal elements, but it is also acceptable to include off-diagonal elements as in the Freescale document mentioned above.

[0033] The rest of the code is basically the same as in the MATLAB documentation, so it will be omitted. Since we defined the error of the parameter we want to estimate as the state, the update formula for the Kalman filter estimates the error of the state. We then use these to correct the predicted values ​​of attitude, gyro bias, and linear acceleration to obtain the updated values. The method for obtaining the next predicted values ​​for these values ​​is basically the same as in the MATLAB documentation mentioned above.

[0034] The attitude and orientation calculation unit 1 outputs the attitude or attitude and orientation calculated in this manner.

[0035] The results in this disclosure are shown in the simulation results. The route is as follows: The first third of the journey is a 5.14 degree downhill, the last third is a 5.14 degree uphill, and the middle section is flat, a distance of approximately 3000m, traveled at 40km / h. There are stops at the beginning and end, and a 90-degree right turn almost in the middle, but the vehicle stops briefly before starting again.

[0036] Figure 2 shows the estimated changes in X and Y rotations, which represent attitude, within the attitude and orientation. In the upper figure, the dashed line represents the correct value, and the solid line represents the algorithm used in the MATLAB literature as a comparative example. Although the comparative example works well, an error of more than 1 degree appears in the Y rotation. If the polarization tolerance is within ±1.8 degrees, considering errors generated by the antenna, an error of more than 1 degree on one rotation axis cannot be said to be sufficiently small. The lower figure plots the estimation results of the comparative example in the upper figure and the embodiment. The thick line represents the embodiment, and the thin line represents the comparative example, and it is clear that the attitude error is reduced in this embodiment.

[0037] In this embodiment, the linear acceleration prediction value is obtained in accordance with MATLAB literature. S a^ - k+1 The coefficient c is applied to the updated value from the previous step. a A low-pass filtering model is employed that predicts the frequency by multiplying it by a factor.

[0038]

number

[0039] In MATLAB literature, the coefficient value is fixed and defaults to 0.5. However, in this model, when the actual linear acceleration increases due to acceleration, deceleration, or rotation, the linear acceleration estimate will not keep up with the actual linear acceleration. Therefore, it is advisable to increase the coefficient value when it is determined that the linear acceleration is increasing. Specifically, monitor the gyro measurements and, for example, if significant rotation is detected by threshold determination, or if the linear acceleration estimate also exceeds the threshold, or if it can be determined that acceleration, deceleration, or rotation is occurring by combining the measured angular velocity and acceleration, then increase the coefficient value. a It's best to increase it to a value close to 1.

[0040] The measurements obtained from the two difference vectors are ultimately allocated to attitude angle error, gyro bias error, and linear acceleration error. However, if the arrival angle measurement happens to be close to the angle before the turn, such as in the case of a sharp turn with a small radius, the algorithm may not be able to keep up with the turn. Since the algorithm can operate even with some absence of arrival angle measurement, in such cases, it is advisable to reduce the contribution of the difference vector of the arrival angle. Specifically, this involves reducing the Q of the observation noise covariance. vSL It would be good to make it larger.

[0041] In this embodiment, there are no particular restrictions on the method for estimating the satellite's angle of arrival. Any commonly used method is acceptable. In mobile phones, since the number of elements in a phased array antenna is only a few, it is not uncommon to estimate the angle of arrival by detecting the phase and amplitude of all elements. However, in satellite communications, especially communications with geostationary satellites, the beam width is very narrow, resulting in a large antenna aperture, and the frequency often exceeds 10 GHz. In the case of a phased array, the element spacing is narrow, resulting in several thousand elements, making it difficult to detect the phase and amplitude of all elements. In most cases, the angle of arrival is estimated by detecting the beam by electronically or mechanically shaking it, by simulating dividing the entire antenna into several parts, or by simultaneously forming multiple received beams. Any of these methods are applicable. Since there are differences in accuracy and measurement frequency depending on the method, it is advisable to change the filter constants and other parameters according to the method used.

[0042] As in this embodiment, the method of inputting measurement data from multiple systems as measured values ​​into a filter to estimate the state is called tight coupling. The method of estimating the state individually from each system and then combining the multiple estimated states to estimate the total state is called loose coupling, but tight coupling is known to have smaller errors compared to loose coupling.

[0043] Furthermore, compared to using magnetism for the second difference vector, the angle of arrival of satellite radio waves is extremely accurate, allowing for a very small attitude error.

[0044] (Second Embodiment) The second embodiment is a modification of the first embodiment. Figure 3 shows an implementation according to one embodiment. The attitude and direction estimation device 100 includes a line-of-sight determination unit 7 in addition to the configuration of the above-described embodiment. Note that other configurations and other data flows are the same as in Figure 1, so some parts are omitted from the illustration.

[0045] Satellite communications cannot receive radio waves from the satellite if there are obstacles such as buildings between the satellite and the earth station. Furthermore, even if the radio waves are not completely blocked, partial obstruction by street trees or other obstacles may prevent a sufficient signal-to-noise ratio (SNR) from being achieved. In such cases, the angle of arrival may not be measurable, or the measured angle of arrival may be an outlier. These conditions are described as insufficient line of sight to the satellite, or line of sight lost, making it impossible to determine the angle of arrival.

[0046] Furthermore, while IMU measurements are generally taken once every few milliseconds to tens of milliseconds, the frequency of arrival angle measurements is often less frequent. While it is possible to match the frequency of data acquisition from the IMU to the arrival angle estimation frequency, if the arrival angle estimation frequency becomes as high as one second, the movement of the object is considerably faster, making it highly likely that necessary rotations and accelerations will be missed. Therefore, it is desirable to continue processing data from the IMU even when the arrival angle estimation result has not been input.

[0047] In this case, the difference vector (second difference vector) that uses the angle of arrival vector to the satellite is not used, and processing continues using only the first difference vector. The behavior in this case is the same as in the MATLAB documentation.

[0048] The line-of-sight determination unit 1 also receives input from the line-of-sight determination unit 7 as input to the attitude and direction calculation unit 1. The line-of-sight determination unit 7 receives input from the received radio wave processing unit or the angle of arrival estimation unit (not shown) and determines whether there is a line of sight. Specifically, it receives input from the received radio wave processing unit, such as power, SNR, CNR (Carrier to Noise Ratio), reception success or failure based on error detection when receiving packets, and estimated attitude. It verifies the angle difference with the angle of arrival of the previous sample, and whether there is a significant discrepancy between the change in attitude and the change in the angle of arrival. It then determines whether the angle of arrival is reliable and whether there is sufficient line of sight to give the estimated angle of arrival sufficient accuracy, and outputs the result. At the same time, it receives input of the angle of arrival from the angle of arrival estimation unit and stops outputting the angle of arrival if there is not sufficient line of sight. If the reliability is determined to be insufficient, a threshold can be set to switch the output on or off, or the angle of arrival may be output simultaneously with reliability information (such as SNR). In a filter, the reliability of the arrival angle may be weighted according to factors such as SNR to adjust its contribution to the updated value of the arrival angle. In a Kalman filter, processing such as increasing the observation noise covariance with respect to the arrival angle when the reliability is low is applied.

[0049] The angle of arrival error tends to be relatively close to a normal distribution when the SNR is high, but it often produces outliers when the line of sight is unstable. Since the Kalman filter does not work well with error distributions other than a normal distribution, it is desirable to either not use the angle of arrival data when it is clear that there are outliers, or to use it with reduced weight. Therefore, this embodiment enables continuously accurate attitude calculations.

[0050] The attitude and orientation calculation unit 1 receives input regarding line of sight. If there is a line of sight and a good angle of arrival is obtained, it performs the same processing as in Figure 1. However, if there is no line of sight, or if there is no angle of arrival data due to timing issues with IMU data acquisition, it processes the data using only the difference vector related to gravity.

[0051] Specifically, the measured values ​​in equation (1) are as follows.

[0052]

number

[0053] The observation matrix is ​​as follows:

[0054]

number

[0055] The observed noise covariance is defined as follows:

[0056]

number

[0057] These cases involve variations in matrix size depending on the chosen option, but the processing algorithm remains unchanged. In other words, instead of switching algorithms, the entire set of algorithms can be kept the same, and the changes can be made simply by altering the size of the parameters used.

[0058] This approach allows for continuous attitude updates even when line of sight to the satellite is lost or when new arrival angle data is unavailable due to timing issues.

[0059] As explained in the section on conventional technology, the accuracy of attitude gradually deteriorates due to the absence of the second difference vector. The main cause of this deterioration is angular drift due to gyro bias, and since the angle of arrival is estimated about once per second, if the gyro is a vehicle-grade calibrated IMU, significant attitude errors usually do not occur. On the other hand, in cases where line of sight to the satellite is lost for a long period, such as in a long tunnel, the attitude error becomes a significant amount. However, the purpose of correctly detecting attitude is to transmit radio waves to the satellite with the correct polarization, so in places where radio waves cannot reach the satellite at all, such as inside a tunnel, and therefore no transmission is performed, a slight deviation in attitude is not a problem. Since tracking the satellite direction using the angle of arrival also becomes impossible, the satellite direction will be continuously updated using the estimated change in attitude angle. In this case, it is sufficient if the deviation in attitude and azimuth is small enough that the search for the satellite direction can be completed in a short period of time when the line of sight to the satellite is restored after exiting the tunnel.

[0060] (Third embodiment) Figure 4 is a schematic diagram showing an example of a satellite communications earth station 200 according to one embodiment. This is a satellite communications earth station including the attitude and direction estimation device of the present invention. Parts common to the first embodiment will not be described. The satellite communications earth station 200 includes, for example, a receiving antenna 8, a transmitting antenna 9, a beam control unit 12 constituting the antenna control unit 11, a polarization calculation unit 13, and a filter 14, in addition to the configuration of the first embodiment.

[0061] The satellite communications earth station 200 according to this embodiment includes a radio wave transmission section to the satellite. According to this embodiment, the transmission polarization can be calculated using an attitude estimated with high accuracy. The antenna control unit 11 receives attitude input from the attitude and direction calculation unit 1. In addition, it receives absolute coordinates from the absolute coordinate output unit 4 and the angle of arrival from the angle of arrival measurement unit 2.

[0062] Since the output of the angle of arrival measurement unit may have a large error due to noise, the angle of arrival may be filtered by filter 14 to suppress the noise. The filter should preferably be one that allows real-time prediction, such as a Kalman filter or an α-β filter.

[0063] The polarization calculation unit 13 calculates the polarization in the global frame, i.e., the polarization as if the antenna plane were perfectly horizontal, from the satellite's orbital information, the polarization angle to be transmitted, and the absolute coordinate information of the earth station. Since the earth station is almost never perfectly horizontal, it uses the input attitude and azimuth to convert it into the polarization defined on the antenna plane (sensor frame) and outputs it to the beam control unit 12. The beam control unit 12 receives the input of the highly accurate angle of arrival and polarization with suppressed noise, and controls the transmitting antenna 9 so that the azimuth, elevation angle and polarization of the transmitted beam match them. If the transmitting antenna is a phased array antenna, it controls the phase and amplitude of each element.

[0064] The transmitting antenna 9 receives a transmission signal that has been modulated to an appropriate shape and shaped in the RF section to have the appropriate frequency, spectrum, and power, and is transmitted from the transmitting antenna 9 in the appropriate direction and polarization.

[0065] By applying this embodiment, polarization can be estimated with higher accuracy. As a result, when an acceptable polarization error is specified, the range of acceptable errors in other parts, such as the antenna control unit and the antenna itself, increases, and the difficulty of design and manufacturing is reduced.

[0066] Furthermore, from the satellite's orbital information and the Earth station's absolute coordinates, not only polarization but also the satellite's direction in the global frame, i.e., azimuth and elevation, can be calculated. By combining this with the estimated attitude, it can be converted into azimuth and elevation in the sensor frame. On the other hand, filtering the arrival angle measurement results also yields the estimated arrival angle in the sensor frame, but in many cases, the azimuth and elevation of the two do not perfectly match. In this case, it is desirable to calculate the polarization according to the azimuth and elevation that is thought to be more accurate. However, the estimated arrival angle does not include all information about attitude. As shown in Figure 5, even if the transmitting antenna is rotated around the line connecting the transmitting antenna 9 and satellite 300 as the axis of rotation, the measured azimuth and elevation do not change. On the other hand, if the antenna is outputting fixed linear polarization and is rotated in this way, the polarization as seen from the satellite will change in accordance with the rotation. Therefore, the polarization cannot be determined solely from the arrival angle estimation result.

[0067] The angle of arrival filtering by filter 14 allows for parameter optimization specifically for estimating the angle of arrival, making it highly likely to have a smaller error than the azimuth and elevation angles calculated by the attitude-azimuth calculation unit 1 as part of the attitude-azimuth. Of course, if the parameters of the attitude-azimuth calculation unit can be completely optimized and the azimuth and elevation angles calculated from the attitude-azimuth can be calculated with very high accuracy, then those azimuth and elevation angles can be trusted.

[0068] Assuming that the filtered arrival angle estimation result has higher accuracy, a method for absorbing these differences will be described as a variation of the third embodiment. First, the attitude and elevation angles are calculated from the attitude and direction calculated by the attitude and direction calculation unit 1, the absolute coordinates, and the satellite's orbit information. The attitude and direction are corrected so that the roll angle of the antenna surface on the global frame does not change, so that this matches the filtered arrival angle estimation result. That is, the direction and elevation angles of the attitude and direction on the global frame are corrected. First, the attitude and direction are converted to the global frame, and the difference in direction between this and the arrival angle, which has also been converted to the global frame, is calculated. The attitude and direction are corrected so that this difference disappears. Next, the difference in elevation angles is similarly calculated, and the attitude and direction are corrected so that this difference disappears. The polarization that the antenna should output is calculated from the corrected attitude and direction, absolute coordinates, satellite's orbit information, and the polarization used.

[0069] This approach enables more accurate polarization estimation.

[0070] Figure 6 is a flowchart showing the processing in the first embodiment as an example of the processing of the attitude and direction estimation device 100 or satellite communication earth station 200 in each of the embodiments described above. The processing flow for other embodiments can also be read from the drawings as shown in the figures.

[0071] The attitude and orientation calculation unit 1 predicts the attitude angle, gyro bias, and linear acceleration (S100). For example, the attitude and orientation calculation unit 1 predicts the attitude angle, gyro bias, and linear acceleration based on the angular velocity obtained from IMU 3 using an algorithm from MATLAB literature.

[0072] The attitude and orientation calculation unit 1 calculates the gravity vector difference (first difference vector) and the arrival vector difference (second difference vector) (S102). The attitude and orientation calculation unit 1 calculates the first and second difference vectors based, for example, the acceleration obtained from the IMU 3, the absolute coordinates obtained from the absolute coordinate output unit 4, and the arrival angle obtained from the arrival angle measurement unit 2.

[0073] The attitude and orientation calculation unit 1 calculates a first difference vector, which is the difference vector between the gravity direction obtained by subtracting the linear acceleration predicted in S100 from the input acceleration data, and the gravity direction calculated from the predicted attitude.

[0074] The attitude and orientation calculation unit 1 also calculates a second difference vector, which is the difference vector between the satellite direction corrected based on the attitude predicted from the absolute coordinate data of the device itself and the satellite's orbital information held in advance, and the satellite direction calculated from the input angle of arrival data.

[0075] The attitude and orientation calculation unit 1 updates the attitude angle error, gyro bias error, and linear acceleration error (S104). This error update is implemented, for example, by estimation using the update formula for the state of the Kalman filter when a Kalman filter is used.

[0076] The attitude and orientation calculation unit 1 updates the attitude angle, gyro bias, and linear acceleration using the respective errors obtained in S104 (S106).

[0077] If the operation is to be terminated (S108: YES), the process will be completed after appropriate processing. If the operation is to be continued (S108: NO), the process from S100 will be repeated. In this case, the values ​​calculated in the previous iteration may be appropriately converted using coefficients or the like.

[0078] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These novel embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents. [Explanation of symbols]

[0079] 100: Posture orientation estimation device, 1: Attitude and direction calculation section, 2: Arrival angle measurement section, 3: Inertial measurement unit, 4: Absolute coordinate output unit, 5: Antenna, 6: Received radio wave processing unit, 7: Foresight / visibility determination unit, 8: Receiving antenna, 9: Transmitting antenna, 11: Antenna control unit, 12: Beam control unit, 13: Polarization calculation unit, 14: Filter, 200: Satellite Communications Earth Bureau 300: Satellite

Claims

1. An angle of arrival measuring unit that receives radio waves transmitted from a satellite and measures the angle of arrival of the received radio waves, An inertial measurement unit comprising at least an accelerometer and an angular velocity meter, A unit that obtains the absolute coordinates of the current location, and an absolute coordinate output unit, An attitude and orientation calculation unit calculates a gravity direction measurement by subtracting a previously estimated linear acceleration from the acceleration measured by the accelerometer, generates a first difference vector which is the difference between this measurement and the gravity direction estimated from the previously estimated attitude, generates a second difference vector which is the difference between the measured arrival angle acquired by the arrival angle measurement unit and the arrival angle estimated from the absolute coordinates, the satellite's orbital information, and the previously estimated attitude, and updates the previously estimated attitude with a complementary filter based on the first and second difference vectors to acquire an estimated attitude. A posture and orientation estimation device equipped with the following features.

2. The complementary filter estimates attitude, linear acceleration, and angular velocity bias from the first and second difference vectors using a Kalman filter. The attitude and orientation estimation device according to claim 1.

3. The complementary filter estimates attitude error, linear acceleration error, and angular velocity bias error as states, and estimates the attitude, linear acceleration, and angular velocity bias based on the states. The attitude and orientation estimation device according to claim 2.

4. The attitude and direction calculation unit continues processing using the first difference vector when line of sight to the satellite is lost. The attitude and orientation estimation device according to claim 1.

5. When the intensity of the radio waves received from the satellite, the signal-to-noise ratio, or the carrier-to-noise ratio falls below a threshold, it is determined that the line of sight to the satellite has been lost. The attitude and orientation estimation device according to claim 4.

6. The absolute coordinate output unit is a GNSS (Global Navigation Satellite System) receiver. The attitude and orientation estimation device according to claim 1.

7. An angle of arrival measuring unit that receives radio waves transmitted from a satellite and measures the angle of arrival of the received radio waves, An inertial measurement unit comprising at least an accelerometer and an angular velocity meter, A unit that obtains the absolute coordinates of the current location, and an absolute coordinate output unit, An attitude and orientation calculation unit calculates a gravity direction measurement by subtracting a previously estimated linear acceleration from the acceleration measured by the accelerometer, generates a first difference vector which is the difference between this measurement and the gravity direction estimated from the previously estimated attitude, generates a second difference vector which is the difference between the measured arrival angle acquired by the arrival angle measurement unit and the arrival angle estimated from the absolute coordinates, the satellite's orbital information, and the previously estimated attitude, and updates the previously estimated attitude with a complementary filter based on the first and second difference vectors to acquire an estimated attitude. An antenna that emits radio waves to the aforementioned satellite, An antenna control unit determines the direction of the output beam of the antenna based on the angle of arrival measured by the angle of arrival measuring unit, and determines the polarization of the output beam using at least the estimated attitude and the direction of the output beam. A satellite communications earth station equipped with these features.

8. The antenna control unit determines the direction of the output beam by applying a filter to the angle of arrival. The satellite communications earth station according to claim 7.

9. If the absolute coordinates and the beam direction estimated from the estimated attitude do not coincide with the direction of the output beam, the antenna control unit estimates a corrected attitude by correcting the yaw angle and pitch angle of the estimated attitude, and determines the polarization of the output beam from the corrected attitude and the direction of the output beam. The satellite communications earth station according to claim 7.

10. The angle of arrival measuring unit receives radio waves transmitted from the satellite and measures the angle of arrival of the received radio waves. The absolute coordinate output unit obtains the absolute coordinates of the current location, The attitude and orientation calculation unit calculates the gravity direction measurement by subtracting the previously estimated linear acceleration from the acceleration measured by the accelerometer. The attitude and orientation calculation unit generates a first difference vector, which is the difference between the previously estimated attitude and the gravity direction estimated from that attitude. The attitude and direction calculation unit generates a second difference vector, which is the difference between the measured angle of arrival obtained by the angle of arrival measurement unit and the angle of arrival estimated from the absolute coordinates, the satellite's orbital information, and the previously estimated attitude. The attitude and orientation calculation unit updates the previously estimated attitude using a complementary filter based on the first difference vector and the second difference vector to obtain an estimated attitude. Attitude orientation estimation method.

Images