JPWO2019020962A5 - - Google Patents

Description

The present invention relates generally to magnetic inertial technology.

More precisely, the invention relates to determination of orientation by means of a magnetometer.

In particular, the invention is advantageously applied in the case of measurements in metropolitan areas or "indoors", ie inside buildings.

Conventionally, magnetometers are used for orientation calculations in implanted systems.

In this case the hypothesis is made that the magnetic field measured by the sensor is the earth's magnetic field pointing towards magnetic north due to the horizontal component. The difference between the direction of magnetic north and geographic north (called the magnetic declination) is known and tabulated. Without loss of generality, magnetic north and geographic north are hereinafter considered to be merged and therefore the magnetometer points to what is called north.

Typically, the determination method used to calculate orientation from magnetometer measurements is based on the following.
- Modeling associating the measurements supplied by the magnetometers with information about the orientation of the system - Characterization of the association of this relationship for the measurements to be made

In conventional methods, the calculations performed may be Kalman-type filtering with the magnetic field as the measurement, allowing readjustment of the orientations contained within the state along with the measurement noise, which is a function of the relevant properties.

によって、方位情報を含むと書くことにあり、Ｒは、回転行列であり、物体の基準フレームから地球の慣性座標系に進むことを可能にする。ψ、θ、φは、オイラー角であり、Ｍ_{ＥＡＲＴＨ}は、地球の磁場である。 Modeling assumes that the measured magnetic field M is, for example, the equation

contains orientation information, and R is a rotation matrix that allows one to go from the body's frame of reference to the earth's inertial frame of reference. ψ, θ, φ are the Euler angles and M _EARTH is the earth's magnetic field.

The relationship in this equation is characterized by the measurement variance, ie the error in this equation is assumed to be a Gaussian random variable with zero expected value.

This variance serves to automatically compute the Kalman gain weighting the readjustments , taking into account different noises (dynamic noise and magnetometer measurement noise related to the external environment).

In still other methods, computation with linear filtering is performed using the same type of modeling and characterization.

In this method, it is the relative adjustment of the gain magnitudes in the equation that characterizes the relationship. Generally, this gain is manually adjusted. Those skilled in the art will then know how to weight it as a function of modeling relevance.

Thus, in these two methods, the parameters characterizing the model association (variance of Gaussian noise for Kalman filtering, gain for linear filtering) are generally constant, regardless of the measured magnetic field. is a parameter.

Relevant characterizations of modeling that take account of the measured magnetic field while comparing it with a geomagnetic model of the earth's magnetic field have recently been proposed, for example in the following publications.
-W. T. Faulkner, R. Alwood, W. T. David andJ. Bohlin, "GPS-denied pedestrian tracking in indoor environments using an imu and magnetic compass," The Institute of Navigation International Technology 2010. Conference Proceedings, (San Diego, Calif.), p. 198-204, January 2010-M. H. Afzal, V. Renaudin and G. Lachapelle, "Magnetic field based heading estimation for pedestrian navigation environments", International Conference on Indoor Positioning and Indoor Navigation, 9 January 2011, (Guimarães, Portugal), 9 January 2011.

However, the characterization performed in this way does not give full satisfaction. First, because they are based on an a priori model of the magnetic field, which requires having a model that is at least as accurate as desired, and second, because they cannot reject all disturbances . It is from.

The document "Unscented Filtering for Spacecraft Attitude Estimation" (JL Crassidis and F. Landis Markley) describes a model for estimating spacecraft orientation. This model is based on applying a Kalman filter and does not consider any correlation. However, this model also has a bias that makes it less reliable for its part.

Proposing a simplified model is also known from document EP2264485, where the estimated magnetic field is equal to the sum of the measured magnetic field value, the magnetic disturbance value and the measurement noise value. However, this model of changes in disturbances only considers the temporal relationship of the disturbances , which does not add much reliability.

A general aim of the invention is to propose a solution that allows a better characterization of modeling relationships, especially those used in highly disturbed environments.

In particular, the invention proposes a method of determining orientation by means of a magnetic sensor, the magnetic field being measured by means of a magnetometer, the computing means performing recursive processing computations during a predetermined sampling time,
- estimating a heading prediction that is a function of the heading determined at the previous sampling time;
- Readjust the estimated and predicted orientation as a function of magnetic orientation determined from magnetometer measurements.

In a preferred embodiment, during said readjustment , the computing means determines the magnitude of the readjustment as:
- estimated as a function of a model of change in the magnetic disturbance between two sampling times, and
- estimated as a function of an a priori estimate of the disturbance magnitude .

In this way, the process takes into account the change in magnetic orientation perturbations between two sampling times and the potential spatial correlation of perturbations due to the environment for consecutive sampling times.

The resulting readjustment is more reliable than the prior art.

Advantageously, said readjustment is a function of the measured magnetic field gradient. In this way, the fact is taken into account that the more disturbed the environment, the higher the gradient and therefore the more corrupted the orientation measurement.

を計算することによって、関連する方位擾乱を推定し、

であり、

は、分散ａ［ｋ］^２（１－α［ｋ］^２）のガウス確率変数であり、
－ｋは、以前のサンプリング時間であり、
－ａ［ｋ］は、磁気擾乱の演繹的な大きさを表すパラメータであり、
－ｕ＾およびσ_ｕは、擾乱における変化の期待値および分散の推定として計算手段によって計算される２つのパラメータである。なお、

は、ｕ＾として表す。 In particular, for a given sampling time k+1, the computing means

Estimate the associated azimuth disturbance by computing

and

is a Gaussian random variable of variance a[k] ² (1−α[k] ² ),
-k is the previous sampling time,
−a[k] is a parameter representing the a priori magnitude of the magnetic disturbance ,
−u and σ _u are two parameters calculated by the computing means as the expected value of the change in the disturbance and an estimate of the variance. note that,

is represented as u.

Advantageously, the computing means estimate the parameter a[k] as a linear function of the norm of the magnetic field gradient.

The parameter u[k] is calculated by the computing means as the difference between the magnetic orientations determined directly from the magnetometer outputs for times k+1 and k, for example for that part, and from this difference, the predicted Changes in orientation (ω[k]dt) are subtracted.

The parameter σ _u may be estimated as a function of displacement rate or displacement between two consecutive sampling times.

The processing, for example, performs Kalman filtering where the states have at least the actual orientation and magnetic orientation perturbations (φ,φ ^(d) ) as parameters.

The heading prediction may be determined as a function of measurements of one or more sensors of the inertial unit.

A change model is predetermined to consider the spatial or temporal correlation of changes in the disturbance .

The invention further relates to a device for determining orientation by means of a magnetic sensor, comprising a magnetometer and computing means for computing the orientation from the magnetic field measured by said magnetometer, said computing means comprising: For the sampling time set, the processing described above is performed.

The invention also proposes a magnetic inertial navigation system comprising at least one azimuth measuring device of this kind.

Advantageously, a system of this kind is used in an urban environment or inside a building.

The present invention also provides
- a computer program product containing code instructions for the execution of a method of the type described above when the program is run on a computer - a computer program product containing code instructions for the execution of a method of this type, readable by a computer device It concerns possible storage means.

Other features and advantages of the invention will become more apparent from the following description, which is merely illustrative and non-limiting and should be read in conjunction with the accompanying drawings.

1 is a diagram of an apparatus for an embodiment of the method according to the invention; FIG. An example of a case for implementation of the method according to the invention is represented in more detail. 1 shows the main steps of a method according to an embodiment of the invention;

と記載される周囲磁場（典型的には、近くの金属物体によってわずかに変化しうる地球の磁場）内で移動する物体１の運動の推定のために用いられる。既に知られているように、磁場は３次元ベクトル場であり、すなわち、３次元のベクトルを、物体が移動可能な各３次元の点に関連付ける。 General remarks - measuring device With respect to Fig. 1, the proposed measuring device is for example:

is used for estimating the motion of an object 1 moving in an ambient magnetic field (typically the earth's magnetic field, which can be slightly altered by nearby metallic objects). As is already known, the magnetic field is a three-dimensional vector field, ie it associates a three-dimensional vector with each three-dimensional point to which the object can move.

This object 1 may be any movable object whose position is required to be known, for example vehicles, drones etc. and pedestrians.

の投影を測定できる要素を意味するものである。 The object 1 comprises a plurality of magnetic measurement sensors 20, namely axial magnetometers 20, in a case 2 (support). An axial magnetometer measures the component of the magnetic field, i.e. the magnetic field vector along its axis at the level of the magnetometer 20

means the element from which the projection of the can be measured.

More precisely, magnetometer 20 is integral with case 2 . The magnetometer 20 has substantially the same motion with respect to the case 2 and with respect to the object 1 in the ground frame of reference.

Preferably, the reference frame of the object 1 comprises an orthonormal Cartesian reference point and the magnetometer 20 has a predetermined position at this reference point.

In FIG. 2 the case 2 is fixed to an object 1 (eg a leg of a pedestrian) by attachment means 23 . These attachment means 23 consist, for example, of bracelets, for example with self-gripping straps that grip the limb and allow an integral connection.

Clearly, the invention is not limited to estimating motion of pedestrians, but is particularly advantageous in this type of use. Because it can greatly reduce the volume, which is necessary for a person to be able to carry the case in an ergonomic way.

Case 2 may comprise computing means 21 (typically a processor) for performing the processing operations of the method directly in real time, or alternatively the measurements may be sent to an external device, e.g. (smartphone) 3 or to a remote server 4 via communication means 25, or alternatively the measurements are stored in local data storage means 22 (e.g. flash type memory).

The communication means 25 may also implement short-range wireless communication, such as Bluetooth or Wi-Fi (especially in embodiments with a mobile terminal 3), or a cellular network for long-range communication (typically may even be a means for connecting to UMTS/LTE). It should be noted that the communication means 25 may also be a wired connection (typically USB), eg for transferring data from the local data storage means 22 to the mobile terminal 3 or the server 4 .

If it is a mobile terminal 3 (or server 4) with "intelligence", it includes computing means 31 (or 41), such as a processor, for performing the processing operations of the method described below. . When the computing means used are the computing means 21 of Case 2, Case 2 may further comprise communication means 25 for transmitting the estimated position. For example, the owner's location may be transmitted to the mobile terminal 3 to display the location in the navigation software interface.

The respective data computing means 21, 31, 41 of case 2, smart phone 3 and remote server 4 may perform all or part of the steps of the method independently and according to the application.

They each comprise storage means for this purpose in which sequences of code instructions for execution of the method are stored wholly or partly.

The prediction and readjustment computation means performs on the one hand (step 101) a filtering 100 that computes an estimate of the orientation value by prediction, and on the other hand (step 102) performs a readjustment as a function of the error estimate (FIG. 3).

In particular, if the angular velocity ω is known, for example given by a gyrometer, step 101 calculates the orientation ψ _(k+1) at time k+1 as ψ _(k+1) = ψ _k +ω . Calculated to be equal to Δt, ψ _k is the orientation of the previous time k, and Δt is the period separating these two sampling times.

Readjustment 102 takes into account measurements made by magnetometer 20 .

In the following, the magnetic orientation measurement (derived from magnetometer measurements) at a given time k is denoted as z _ψ [k], where z _ψ [k]=ψ _k .

によって与えられ、ＭｙおよびＭｘは、地上の基準フレームにおける磁場の２つの水平成分であり、これらの２つの成分は、その慣性ユニットによって物体１のために決定される姿勢の関数として計算される。 Typically, the magnetic orientation is

, and My and Mx are the two horizontal components of the magnetic field in the ground reference frame, and these two components are calculated as a function of the attitude determined for body 1 by its inertial unit.

を、この時点で推定された方位ψ_{（ｋ＋１）}の合計に等しいように計算し、有利には、再調整は、
－前の時間ｋに関して計算または調整されるゲインＫ_ｋの関数であり、
－予測された方位ψ_{（ｋ＋１）}と測定値ｚ_ψとの間の誤差Ｅｒｒ（ψ_{（ｋ＋１）}，ｚ_ψ）の推定の関数である。 The calculation means calculates the readjustment heading corresponding to time k+1.

is calculated to be equal to the sum of the currently estimated orientations ψ _(k+1) , and advantageously the readjustment is
- a function of the gain K _k calculated or adjusted with respect to the previous time k;
- is a function of the estimate of the error Err(ψ _(k+1) , z _ψ ) between the predicted heading ψ _(k+1) and the measured value z _ψ .

を計算することによって方位を再調整する。 Thus, the computing means use the magnetic orientations derived from the magnetometer measurements to readjust the states and, in particular,

Readjust the azimuth by calculating

Typically, the error Err(φ _(k+1) , z _φ ) is a function of the predicted orientation estimated at time k+1 (before readjustment ) and the magnetic orientation z _φ derived from magnetometer measurements. , can be a simple difference between

Nevertheless, other error functions are possible, especially for non-linear filtering.

In particular, the readjustment K _k . Err(φ _(k+1) , z _φ ) and in particular the gain K _k may advantageously depend on the magnetic field gradient measured by the magnetometer 20 .

Thus, realignment correction is all the more important when the magnetic field varies strongly and is therefore likely to induce significant errors in orientation measurements.

The readjusted heading values thus obtained are stored by said computing means 21, 31, 41 and/or for the rest of the processing and magnetic inertial navigation information (linear velocity, angular velocity, position, heading etc.) is used by said means for the calculation of .

The readjusted heading values may also be transmitted by the computing means to the interface means and displayed, for example, on the screen of the telephone.

Model of Change in Environment-Related Disturbances Below is detailed an example of a possible readjustment calculation in which the measured values of the magnetic orientation z _ψ are corrected by an estimate of the environment-related magnetic field disturbances .

－ψは、決定されるのを望まれる実際の方位に対応し、
－

は、ガウス測定誤差に対応し、
－ψ^（ｄ）は、環境に関連する磁方位擾乱に対応する（「ｄ」は擾乱を意味する）（典型的には、擾乱は金属基盤に関連し、都市環境または建物内の電気ケーブルに関連する）。 The magnetic orientation z _ψ derived from the magnetometer measurements can in fact be considered to be decomposed as follows.

- ψ corresponds to the actual orientation desired to be determined,
-

corresponds to the Gaussian measurement error and
−ψ ^(d) corresponds to the magnetic azimuth disturbance (“d” means perturbation ) associated with the environment (typically perturbations are associated with metal substrates, electrical cables in urban environments or buildings). related).

Environmentally related magnetic orientation disturbances are highly spatially correlated. The magnetic fields associated with environmental perturbations are in fact continuous vector fields, and when two given points A and B are adjacent in space, the magnetic fields of these two points A and B are closer together.

In one embodiment, the environment-related azimuth perturbation φ ^(d) is estimated by the computational means using a formula that takes into account the variation correlation, which is due to the magnetic field perturbation between two consecutive sampling times: Predictable (temporal, spatial or more complex correlations).

This estimate is constructed to enable a Markov model for changes in ψ _d (which can be used in a recursive filter),
- allows to consider models of changes in magnetic orientation perturbations (temporal, spatial correlations or more complex correlations),
- Allows building a filtering model in which the orientation is observable (otherwise any hope of building an orientation estimate is lost).

In the proposed estimation, the readjustment magnitude is estimated as a function of a model of the change in the magnetic disturbance between the two sampling times and as a function of an a priori estimate of the disturbance magnitude .

－ａ［ｋ］は、磁気擾乱の演繹的な大きさを表すパラメータであり、
－ｕ＾およびσ_ｕは、２つのサンプリング時間の間の磁方位擾乱の変化のモデル（ガウス確率変数の形）である確率変数の期待値および分散であり、
－

は、磁力計のノイズを考慮する推定である。 The inventors have discovered (and verified mathematically) the following. That is, in the case of spatial correlation (for metropolitan or indoor disturbances ), a good estimate of the change in the disturbance ψ ^(d) between two sampling times separated by a time step is advantageously: That's right.

−a[k] is a parameter representing the a priori magnitude of the magnetic disturbance ,
−u and σ _u are the expectations and variances of random variables that are models (in the form of Gaussian random variables) of changes in magnetic orientation disturbances between two sampling times;
-

is an estimate that takes into account magnetometer noise.

This estimate is subtracted from z _ψ for the error calculation, which can be calculated directly between ψ _(k+1) and (z _ψ −ψ ^(d) [k+1]), or ψ ^{(d )} and ψ of that state.

Determining Parameter a[k] The parameter a[k] represents the a priori magnitude of the magnetic orientation disturbance . It characterizes the orientation and is calculated, for example, as a linear function of the norm of the magnetic field gradient.

であり、ａ_０およびａ_１は、フィルタリング処理の実施前にあらかじめ固定される２つのパラメータであり、Ｂ＾［ｋ］は、時間ｋの磁場である。 for example,

where a ₀ and a ₁ are two parameters that are pre-fixed before the filtering process is performed, and B^[k] is the magnetic field at time k.

Thus, as noted above, the readjustment terms K _k . Err(φ _(k+1) , z _φ ) is a function of the magnetic field gradient measured by magnetometer 20, which allows effective correction.

Other methods for determining the parameter a[k] are possible.

In particular, a[k] may also be determined by comparison with a model of the Earth's magnetic field, see, for example, the publication Assessment of Indoor Magnetic Field Anomalies using Multiple Magnetometers Assessment of Indoor Evaluation of indoor magnetic field anomalies using )”-M. H. Afzal, V. Renaudin, G.; It may be determined by implementing techniques such as those proposed in the Proceedings of the 23rd International Conference of the Lachapelle-The Satellite Division of the Institute of Navigation (IONGNSS2010) September 21-24, 2010.

However, the solution in which the parameter a[k] is a function of the magnetic field gradient allows better correction and does not require a model of the earth's magnetic field.

The decision parameter u[k] of parameter u represents the most probable value for the change in disturbance .

Ｅ［．］は数学的期待値であり、ω［ｋ］はジャイロメータの回転速度であり、ｄｔはサンプリングピッチである。 This parameter is calculated, for example, using a gyrometer, which is considered short-term correction. The most likely change in disturbance is then given by the difference between the change in magnetic orientation measurement and the change in gyrometer orientation.

E[. ] is the mathematical expectation, ω[k] is the rotation speed of the gyrometer, and dt is the sampling pitch.

Thus, the parameter u[k] is calculated by the computing means as the difference between the magnetic orientations determined directly from the magnetometer outputs for times k+1 and k, and from this difference the predicted rotation ω[ k]dt is subtracted.

The decision parameter σ _u of the parameter σ _u represents the image of the correlation between the perturbation of step k and the perturbation of step k+1.

Advantageously, one can choose to index it on the rate of displacement (or the displacement between two similar consecutive sampling times) to take into account the spatial correlation of the magnetic disturbance .

をとることができ、ｃは、調整係数である。
for example,

where c is an adjustment factor.

Claims (14)

A method for determining orientation with a magnetic sensor, wherein a magnetic field is measured by a magnetometer, a computing means performs recursive processing computations for a predetermined sampling time,
- estimating the predicted heading as a function of the heading determined at the previous sampling time;
- correcting said predicted orientation estimated as a function of magnetic orientation determined from magnetometer measurements;
During the correction , the calculation means determines the magnitude of the correction by :
- as a function of the model of the magnetic disturbance change between two predetermined sampling times to take into account the spatial correlation of the magnetic disturbance change , and
- as a function of an a priori estimate of said magnitude of said magnetic disturbance ,
presume,
A method characterized by: For a given sampling time k+1, said calculating means:

Estimate the azimuth disturbance associated with the environment by computing , where

and

is a Gaussian random variable of variance a[k] ² (1−α[k] ² ),
-k is the previous sampling time;
-a[k] is a parameter representing the magnitude of the correction of the magnetic disturbance of the a priori estimation ;
- u^ and σ _u are two parameters calculated by the computing means as the expected value of the change in the magnetic disturbance and an estimate of the variance;
The method of claim 1. the magnitude of the correction is a function of the measured magnetic field gradient;
3. A method according to claim 1 or 2. the computing means estimates the parameter a[k] as a linear function of the norm of the magnetic field gradient;
4. A method according to claim 3, with reference to claim 2. The parameter u[k] is calculated by the computing means as the difference between the magnetic orientations determined directly from the magnetometer outputs for times k+1 and k; the change (ω[k]dt) is subtracted,
3. The method of claim 2. the magnetic orientation determined is the orientation of a moving object equipped with the magnetometer;
the parameter σ _u is estimated as a function of the speed of displacement of the moving object or the displacement of the moving object between two successive sampling times;
3. The method of claim 2. The method performs Kalman filtering, where the state has at least the actual orientation and the orientation perturbation (ψ, ψ ^(d) ) as parameters.
3. The method of claim 2. the predicted orientation is determined as a function of measurements of one or more sensors of the inertial unit;
8. A method according to any one of claims 1-7. the model of change is predetermined to consider the temporal correlation of the change in disturbance;
The method of claim 1. A device for determining orientation with a magnetic sensor, said device comprising a magnetometer and computing means for calculating orientation from the magnetic field measured by said magnetometer,
said computing means perform the processing of the method according to any of claims 1 to 9 for different successive sampling times;
device. A magnetic inertial navigation system comprising at least one device for determining orientation according to claim 10. 12. Use of the system according to claim 11 applied in urban environments or inside buildings. A computer program,
said program comprising code instructions for execution of the method of any of claims 1 to 9 when run on a computer;
computer program. A storage means readable by a computing device, comprising:
A computer program comprising code instructions for execution of the method according to any of claims 1 to 9,
memory means.