WO2020011298A1

WO2020011298A1 - Method and device for determining a 3d orientation and a 3d position on the basis of a single-axis coil and an marg sensor

Info

Publication number: WO2020011298A1
Application number: PCT/DE2019/000187
Authority: WO
Inventors: Puian TADAYON; Thomas FELDERHOFF; Gerhard STAUDE
Original assignee: Fachhochschule Dortmund
Priority date: 2018-07-13
Filing date: 2019-07-12
Publication date: 2020-01-16
Also published as: DE102018005649A1

Abstract

The invention relates to a method for determining the 3D orientation and 3D position of objects using a device, having a reference module with a single-axis coil and an MARG sensor. The coil is supplied with a sinusoidally modulated excitation current in order to generate a harmonic magnetic field signal with a constant frequency. The device also has a sensor module with an MARG sensor. The reference module is used as a spatial reference for determining the orientation and position of the sensor module. The 3D orientation of each module is determined algorithmically in a cyclical manner from the data of the MARG sensors in the reference module and the sensor module. The MARG sensor of the sensor module detects a mixed signal which is composed of the harmonic magnetic field generated by the coil of the reference module and all of the other magnetic fields in the surroundings of the device. Thus, not just the value information but also the phase information of the coil signal is used to extract the coil signal from the mixed signal ascertained by the MARG sensor of the sensor module, whereby the amplitudes of the magnetic field components of the coil signal are ascertained complete with sign, and the unique spatial position of the sensor module relative to the reference module is determined from the ascertained amplitudes in a cyclical manner on the basis of the measurement data solely of the MARG sensor on the sensor module.

Description

Method and device for 3D orientation and 3D position determination based on a uniaxial coil and one

MARG sensor

The invention relates to a method for 3D orientation and 3D position determination of objects according to the preamble of claim 1 and a suitable device according to claim 18.

In many technical application areas, a reliable and precise possibility for determining the orientation and position of an object relative to the orientation and position of a second (reference) object in three-dimensional space is required. In the field of biomechanics, for example, the objects can be specific parts of the body. For the medical-diagnostic assessment of the mobility of the cervical spine or the neck area, for example, it is necessary to determine the head inclination and head position relative to the inclination and position of the upper body in order to record rotational and translational movements of the neck. In the area of robotics, the objects can be movable elements (tools, grippers, etc.), the orientation and position of which must be recorded relative to the robot base or a reference joint of the robot for the purpose of motion control. There are numerous established technologies for determining the position in three-dimensional space, but for various reasons (such as weight, size, energy requirement) are only suitable to a limited extent for implementation by means of mobile, location-independent systems or applications, or they have the restriction that there is a gap between the system components Visual contact must exist. In recent years (especially in mobile applications) the use of inertial sensors and magnetometers has been established to determine the 3D orientation of an object [1]. The fusion of 3D acceleration data, yaw rates and magnetic field data enables drift-free, reliable and accurate 3D Orientation determination, which, in contrast to optical or ultrasound-based methods, does not require the use of a source and is therefore well suited for mobile, location-independent applications [2]

Standard technical methods that enable the acquisition of position data in 3D space according to the current state of technology are as follows:

• camera-based methods

With camera-based methods, the position is usually determined. based on a camera marker system. Here, active (light-emitting) or passive (reflecting) markers are applied to the object to be detected. One or more camera (s) capture images on which the markers are recognized using image processing algorithms. When using a single camera, usually several markers are used, which are arranged in special patterns. These patterns are recognized in the recorded images and tracked ("tracked") for the purpose of determining their position [3]. When using several cameras, a marker position is determined via triangulation using the known positions of the cameras relative to one another (e.g. systems from the "Vicon" company) [4]

• ultrasound-based methods

Ultrasound-based methods work on a similar principle as the camera-based methods with multiple cameras. A transmitter attached to the object emits an ultrasound signal that is registered by at least three receivers. The position of the transmitter is determined from the different transit times of the signal using triangulation [5]

• methods based on inertial sensors

With the interactive sensor-based methods, the position is determined by double numerical integration of acceleration data [6].

• Magnetic field based methods

Magnetic field-based standard systems (eg "FASTRAK" or "trackSTAR") usually consist of a magnetic source, which is composed of three orthogonal coils and is usually placed firmly in the room and a three-axis magnetic sensor, whose position and orientation are determined relative to the source (see [7] - [1 1]).

In order to enable the use of the established magnetic field technology for mobile detection of body movements, new approaches have emerged in recent years (see [12], [13]), in which the magnetic field source of the measuring system is worn on the body in addition to the sensors ( see Figure 1). In Figure 1a from [12], the magnetic source with the three orthogonal coils is integrated in a pyramid-shaped module. In Figure 1 b from [13] the coils are realized as windings in the form of a belt. Research is also working on solutions that enable position determination based on a uniaxial magnetic source, ie a single coil. Previous approaches to using a single coil require several magnetic field sensors at a fixed, known distance from one another in order to be able to conclusively conclude the 3D position. The positions of the sensors with respect to all three axes (X, Y, Z) must differ. For example, in [14] three magnetic field sensors are used at a fixed distance from each other in combination with an inertial sensor for determining the 3D orientation and 3D position. The approach in [15] does not require an inertial sensor, instead four magnetic field sensors are required at a fixed distance from each other. The previously available methods for 3D position determination have properties that are particularly disadvantageous for use in mobile applications. In camera and ultrasound-based methods, visual contact is required between the sensors and the source. If visual contact is interrupted, for example, by an obstacle, it is no longer possible to determine the 3D position.

In the inertial sensor-based methods for position determination, the sensor noise is also integrated with the integration of the sensor data. As a result, the position determined contains an error that increases over time (so-called drift effect), so that the position becomes unusable after a short time without additional reference sensors or countermeasures [6]. With the magnetic field-based methods, the standard systems available to date rely on a stationary, three-axis magnetic source and are therefore only suitable for mobile applications to a limited extent.

The previous approaches from [12] and [13] to use the magnetic field-based position determination by a mobile or portable magnetic field source for mobile applications as well as the standard systems are based on the use of three orthogonal coils for magnetic field generation. This results in restrictions for use in long-term mobile applications:

• Operating three coils is associated with a high energy requirement: although the system from [12] also uses inertial sensors for position determination and the coils are only activated every 0.5-1.0 seconds for position determination and drift compensation, the four leave the pattern According to the authors, AA batteries only took about 30 minutes to measure. · The implementation form from [12] leads to a high volume and weight of the reference module; the module pattern described in [12] has a volume of approx. 800 cm ³ and a weight of 450 g.

• In the implementation form from [13], the exact, orthogonal alignment of the coil windings with respect to one another is a critical variable. If the coils are not exactly aligned with each other, this leads to inaccuracies in the position determination. In practice, however, a shift or deformation of the coil windings as a result of body movements cannot be ruled out.

With the previous approaches from [14] and [15] for the use of a single coil for magnetic field generation, there is the restrictive requirement that several magnetic field sensors have to be used and that these must also be positioned at a fixed distance from one another. This requirement leads to problems in applications in which sensors are to be attached to the body. If the sensors are attached to clothing, for example, slipping of the clothing can easily change the sensor distances and thus falsify the position determination. Attachment to the skin is also problematic since the sensor positions can change during body movements due to the elasticity of the skin. In principle, it is conceivable to accommodate the magnetic field sensors at a fixed distance from one another in a common module (or on a circuit board) and thus to ensure an exact, unchangeable positioning of the sensors with respect to one another. However, the distances must be large enough that the measured magnetic fields differ from one another and in relation to the measurement noise to a sufficiently high degree so that a robust position determination is possible. This means that the corresponding module must inevitably have a certain size, which makes it difficult to attach it to smaller parts of the body (for example on the fingertips).

The object of the present invention is therefore to realize a mobile and location-independent acquisition of 3D orientation and 3D position data, which does not require visual contact between the system components used. The achievement of the object according to the invention results from the characterizing features of claim 1 in cooperation with the features of the generic term. Further advantageous embodiments of the invention result from the subclaims.

The invention is based on a method for 3D orientation and 3D position determination of objects, with the aid of a device comprising a reference module with a uniaxial coil and its control electronics and a MARG sensor (Magnetic, Angular Rate, and Gravity sensor), wherein the coil for generating a harmonic magnetic field signal of constant frequency is subjected to a periodic excitation current, a sensor module with a MARG sensor, the reference module serving as a spatial reference for determining the orientation and position of the sensor module and from the If the data of the MARG sensors in the reference module and sensor module are algorithmically determined, the 3D orientation of the respective module is determined, the MARG sensor of the sensor module detects a mixed signal which is derived from the harmonic magnetic field generated by the coil of the reference module and all other magnetic fields in the environment of the facility. Such a generic method is further developed in the manner according to the invention in that not only the amount information but also the phase information of the coil signal is used to extract the coil signal from the mixed signal from the mixed signal determined by the MARG sensor of the sensor module, as a result of which the amplitudes of the magnetic field components of the coil signal have the correct sign are determined, and from the determined amplitudes of the magnetic field components of the coil signal, the unique spatial position of the sensor module relative to the reference module is determined on the basis of the measurement data of only the MARG sensor on the sensor module. This determination of the unambiguous spatial position of the sensor module relative to the reference module can take place cyclically or repeatedly, for example in order to be able to detect changes in the position of the sensor module.

The system presented in the context of the invention basically consists of the following two modules:

- A reference module consisting of the following main components: · Coil + control electronics

• MARG sensor (magnetic field and inertial sensors are offered by numerous manufacturers in a common chip housing, which is why the combination of magnetic field and inertial sensor is often viewed as a sensor that is known as MARG (Magnetic, Angular Rate, and Gravity) sensor is called)

• Battery pack

• microcontroller

• optionally a radio module (depending on the application, it may be necessary to provide a radio module in at least one of the modules for wireless transmission of the resulting data (3D orientation / position or data on rotational / translational movements))

- A sensor module consisting of the following main components:

• MARG sensor • Battery pack

• microcontroller

• optionally a radio module.

The reference module is used to generate the magnetic field and as a spatial reference for determining the orientation and position of the sensor module. The 3D orientation of the respective module can be determined using the data of the MARG sensors in the reference and sensor module using known fusion algorithms ([16], [17]). Joint inclination movements of both modules can be compensated for in this way. Using the known orientation of the sensor module, it is also possible to mathematically rotate the measured magnetic field signals into a defined reference position, in which the z-axis of the magnetic field sensor is aligned parallel to the coil axis of the reference module (this is for the use of the The dipole equations are required to determine the 3D position (see formulas (1) to (3).) For the position determination, the coil on the reference module (similar to the existing approaches from [14] and [15]) with a periodic excitation current , preferably stimulated with a sinusoidally modulated excitation current to generate a harmonic magnetic field signal of constant frequency The measurement signal detected by the MARG sensor of the sensor module is a mixed signal which is composed of the coil-generated harmonic magnetic field and all other magnetic fields in the vicinity of the system. These include, for example, the Earth's magnetic field and Magne fields of electronic devices (see Figure 2).

In previous approaches, the coil signal is filtered out of the superimposed signal by using a bandpass filter with a center frequency that is matched to the known frequency of the coil signal (FIG. 2b). The amplitude of the resulting sinusoidal signal is then determined by means of envelope detection, for example by means of window-wise maximum determination [14] or a Hilbert transformation [15] (FIG. 2c). This is carried out separately for all three sensor axes, so that after the steps described, the amplitudes associated with the coil signal are present in the x, y and z directions. This amp lituden depend on the distance between the coil and reference module with respect to the respective spatial axis. The greater the distance between the modules on the respective axis, the smaller the amplitude of the respective coil signal component. This is evident from dipole equations (1) to (3), which show the mathematical relationship between the components of a magnetic field vector B = [Bx By Bz] and a measurement position [xyz] relative to the position of the source (coordinates

[0 0 0]) (if the radius of the coil is significantly smaller than the distance between the coil and the measuring point of the magnetic field, the coil can be approximated as a magnetic dipole):

where k is a constant dependent on the coil parameters number of turns, current strength and cross-sectional area and r = y * ² + y ² -fz ^{2 is} the distance between the sensor and the coil.

There is no simple closed solution for the inversion of equations (1) to (3), which is required to calculate the position vector components x, y, z as a function of the magnetic field vector components Bx, By, Bz. However, the position vector components for a given magnetic field vector can be clearly determined with the aid of numerical optimization algorithms in which the deviation between the calculated and measured magnetic field vector is minimized [14],

[15]. The prerequisite for this, however, is that the magnetic field vector components have the correct sign, since different position solutions result for different sign combinations according to formulas (1) to (3). In the methods from [14] and [15], however, only the amounts | Bx |, | By | appear after the envelope curve detection and | Bz | of the respective magnetic field vector components, so that there are several possible position vector solutions for a sensor measured value, all of which have the same magnitude components | Bx |, | By | and | Bz | exhibit. This is the reason why these methods require several magnetic field sensors to determine the position when using a single coil: the fixed ones Spatial distances between the magnetic field sensors are used here in order to limit the possible solution vectors to the correct solution and thus to the correct 3D position.

The method according to the invention is based on an alternative procedure. In contrast to the existing approaches, not only the amount information but also the phase information of the coil signal is used to extract the signal from the superimposed measurement signal. This makes it possible to determine the amplitudes of the magnetic field components Bx, By, and Bz with the correct sign. This in turn allows an unambiguous solution and thus an unambiguous position to be determined on the basis of the measurement data of an individual magnetic field sensor using an optimization algorithm via the magnetic field form (1) to (3).

The essential core of the method according to the invention is a magnetic field-based system for 3D orientation and 3D position determination based on a single, uniaxial coil on the reference module and a single MARG sensor (consisting of a 3D acceleration, rotation rate, and magnetic field sensor) on the sensor unit.

The implementation of such a method is made possible by the fact that, in contrast to existing systems based either on a three-axis magnetic source or a combination of several magnetic field sensors, not only amount information but also phase information of the coil-generated magnetic field is evaluated. The new approach put forward in the context of the method according to the invention regards the coil signal as an amplitude-modulated signal, so that an existing coherent demodulation method based on a Hilbert transformation and a Costas loop can be used to determine the coil signal amplitude in the correct phase.

In the present application, the amplitude of the useful signal depends on the distance between the sensor module and the coil on the reference module, so that it can assume any values. In order to avoid the dependence of the phase difference term on the amplitude of the useful signal, the fact is used that under the given application criteria (position axis z> 0) it is impossible that all three magnetic field components have the value Ό 'at the same time can have. This approach is not part of the classic demodulation process or Costas Loop, but new.

The invention presented here offers a new magnetic field-based solution for determining the 3D orientation and 3D position using a single coil on the reference module and a single MARG sensor on the sensor unit. In contrast to optical and ultrasound-based methods, the method presented does not require visual contact between the reference and sensor modules, so that the function of the system is not disturbed by objects that are located between the units (an exception is made of ferromagnetic objects Materials because they influence the magnetic field measurement). In contrast to the inertial sensor-based method, there is no drift effect in the method presented, since no data has to be integrated to determine the position.

The invention also offers numerous advantages over previously known magnetic field-based approaches:

- Using a magnetic source with a single coil instead of a three-axis source with three coils enables:

• easy attachment to the body

• a low energy requirement · a small size (small volume)

• light weight

- Using a single magnetic field sensor on the sensor module instead of multiple sensors at a fixed distance enables:

• simple and targeted application to the body low energy requirement small size (small volume) • light weight

In contrast to the previous approaches, which are based on the use of a uniaxial magnetic source ([14], [15]), the sensor unit does not require several magnetic field sensors at a fixed distance from each other, but only a single MARG sensor. This makes it easier to implement miniaturized sensor modules that can also be attached to small objects (such as fingertips). Accordingly, the new method makes it possible, for example, to develop a system for recording body postures and body movements that is more comfortable to wear and has a lower energy requirement and therefore facilitates applications in the mobile environment and in the long-term area.

The invention described here thereby enables not only a 3D orientation determination but also a mobile, location-independent 3D position determination which does not require visual contact between the objects. The invention thus opens up new application options, among other things. in the fields of biomechanics, robotics and virtual reality. The system components are so scalable in terms of size, weight and energy requirement that the invention is also suitable for applications in which the components are intended to be worn on the body over a longer period of time (for example in medical technology applications for long-term movement detection). The drawing shows a particularly preferred embodiment of the method according to the invention and the corresponding device.

It shows:

1 variants for mobile use of the conventional, magnetic field-based positioning based on a three-axis source, a): source in the form of a pyramid-shaped module [12], b): source in the form of three coil windings [13],

FIG. 2 - magnetic field sources (a), superimposed measurement signal (b) for a sensor axis and bandpass-filtered signal (c),

FIG. 3 - 2D view of two alternative sensor positions relative to the coil, FIG. 4 magnetic field components Bx and By for two different position coordinates in correlation to the coil signal,

FIG. 5 coherent AM demodulation by means of Hilbert transformation and multiplication with complex reference oscillation, FIG. 6 ideal amplitude frequency response of the filtered signal x (nT)

FIG. 7 amplitude frequency response of the Hilbert transformed signal,

FIG. 8 amplitude frequency response to the signal y (t),

FIG. 9 demodulation method with Costas loop,

FIG. 10 processing block for compensating the useful signal term from the

Signal e (t),

1 1 loop filter implementation (1st order low-pass filter), FIG. 12 determination of the initial phase position by means of a synchronization signal, FIG. 13 example for synchronization via known position position of the sensor module, FIG. 14 temporal sampling to avoid Eddy Field influences

Using the DC method with a coil.

The method according to the invention is based on the fact that when processing the sensor measurement data not only the amount information but also the phase information of the periodic, possibly rectangular, but preferably excitation current sinusoidal coil signal is used for 3D position determination. In the following, the relationship between the signs of the magnetic field components and the phase position of the coil signal for a periodic, preferably a sinusoidal, coil signal is shown.

Provided that the sensor unit in the application is always located above the coil on the reference module, ie in the positive z range (z> 0), the following dependency can be determined from formulas (1) to (3) formulate the sign of the magnetic field components and the position of the sensor module relative to the reference module:

(4)

(5)

(6)

For the sake of illustration, a simplified 2D example is considered below, in which the x and y components of the magnetic field components (Bx and By) are schematically sketched for two different positions of the sensor module. Figure 3 shows the two sensor positions in the 2D position coordinate system, the origin of which is defined by the position of the coil.

In Figure 4 the associated magnetic field components Bx and By are shown together with the coil signal. The DC component essentially results from the earth's magnetic field (see left partial image a in FIG. 4). According to formulas (1) to (3), the amplitude of the sinusoidal signal components depends on the distance of the sensor module from the reference module with respect to the three spatial axes.

The magnetic field components for sensor position 1 show that they have the same phase position as the coil signal (FIG. 4a). This is because both the x and y components of the position coordinates are positive. At the second sensor position, the y component is also positive, while the x component is negative. As a result, the signal has a phase shift of 180 ° compared to the coil signal (FIG. 4b). The present invention takes advantage of this relationship between the phase position of the coil signal and the sign of the measured magnetic field component in order to extract the signal component correlated with the coil signal from the overall signal with the correct phase and sign. Based on the signal generated in this way, the 3D position can be clearly determined using formulas (1) to (3) without the need for additional magnetic field sensors. The procedure developed for this is explained in the following sections.

As already described, the amplitude of the coil signal contained in the measurement signal according to formulas (1) to (3) depends on the distance between the coil and reference module with respect to the three spatial axes x, y and z. The coil signal component can be regarded as an amplitude-modulated signal in which the useful signal can be interpreted as a distance-dependent amplitude and the sinusoidal excitation signal as a carrier signal. Accordingly, an amplitude demodulation method can be used to obtain the useful signal from the modulated signal. The basic principle of the coherent demodulation method used here is shown in FIG.

The input signal x (nT) is the measurement signal, which is composed of the high-frequency carrier signal x _c (nT), the low-frequency useful signal s (nT), a noise component w (nT), and a DC component b _e (for explanation - The basic principle is first simplified from a one-dimensional measurement signal):

x (nT) = s (nT) - x _c (nT) + w (nT) + b _e . (7)

The carrier signal x _c (nT) corresponds to the coil excitation signal and is defined as follows:

x _c (nT) - cos (2 f _c nT + <p _c ). (8th) where f _{c is} the coil excitation and thus the carrier frequency of the modulated useful signal, and ^< Pc represents the phase position of the coil excitation signal.

The parameters of the bandpass filter are to be designed in such a way that the noise component w (nT) and the direct component b _e (which essentially results from the superposition of the static earth magnetic field) are largely filtered out and the resulting signal ^x ( ^nT ) im Essentially reduced to the frequency range of the band-limited, modulated useful signal, so that

x (nT)% s (nT) cos (2 f _c nT t <p _c ) (9) applies. FIG. 6 shows the idealized amplitude frequency response of the Fourier transform X (f) to the bandpass-filtered output signal. The frequency f _{max represents} the maximum frequency of the band-limited useful signal s (nT).

The filtered signal is then transformed on the one hand by means of a Hilbert FIR filter [18], and on the other hand delayed by half the filter order q of the Hilbert filter in order to adjust the phase position of this signal branch to that of the Hilbert filter branch (see figure 5).

The resulting signals ^x i ( ^nT ) and ^x r ( ^nT ) represent the imaginary and real part of an analytical signal ^x ( ^nT ) ^{= x} r ( ^nT ) + J ^x i ( ^nT ), which is characterized by that it has no frequency components in the negative frequency range. The idealized amplitude frequency response of the Fourier transform is shown in FIG. 7 for illustration

shown for the signal ^x (f ⁾ .

Since ^ max is always smaller than the carrier frequency ^ c by system design, the Hilbert transform of the signal ^x ( ^nT ) can be analytically represented as follows according to [21]: C; (hT) = s (nT) sin (2nf _c nT + <p _c ). (10)

With the formulas (9) and (10) the analytical signal

thus also be formulated as

The NCO (Numerically Controlled Oscillator) shown in FIG. 5 generates a cosine signal whose frequency ^f c and phase position represent an estimate for the frequency and phase position of the carrier signal. Together with a 90 ° phase-shifted version of the signal, this signal forms the complex signal v (nT) with v (nT) = v _r (nT) + jV _j (nT)

The signal ^v ( ^nT ) is multiplied by the signal - ( ^nT ), so that the complex signal y ( ^nT ) with real part ⁱ ( ^nT ) and imaginary part i ( ^nT ) according to y (nT) = x- (nT) v (nT)

= i (nT) + jq (nT) (13) results.

If the frequency and phase of the NCO signal almost correspond to the actual frequency and phase of the carrier signal contained in the input signal (ie

* ^f c and <Pc ^< Pc or ^Dί * ⁰ and ^{D (} R °), the real part i (nT) (which is often also referred to as an in-phase component) is reduced to the useful signal s (nT) , The imaginary part ^< i ( ^nT ) (quadrature component) is zero in this case. In the frequency domain, the multiplication of the input signal by the NCO signal corresponds to a so-called down conversion, in which the frequency components of the input signal are shifted by the frequency ^{- f} c (see FIG. 8). As already mentioned, this method only works if the frequency and phase relationship between the NCO and the carrier signal match. In practice, however, this is generally not the case, among other things, due to component tolerances in carrier signal generation. Even a very slight deviation of the reference frequency fc generated by the NCO from the actual frequency fc of the carrier signal leads to a drift effect, which causes an error in the demodulation that increases with time. This can be problematic - at least for long-term measurements.

In order to avoid such effects due to frequency or phase deviations, the use of a phase locked loop is suitable. An established form of a digital phase locked loop is the so-called Costas loop [19, 20]. FIG. 9 shows the demodulation method expanded by a digitally implemented Costas loop.

The phase position of the NCO signal is no longer constant as in formula (12), but rather time-variable, so that v (nT) = v _r (nT) + jv _j (nT)

e -j (2 f _c nT + <? _c i_nrj) r 141 giit. The phase-locked loop has the task of continuously adapting the phase position of the NCO to that of the carrier signal, so that the phase difference Lf ( ^h T ^' ) <p _c - <p _c (nT) is minimized. Due to the general relationship f (t) = <p (t) (15) between frequency f and phase <, continuous phase adjustments of the NCO also compensate for frequency deviations to a certain extent. It is therefore assumed below that fc ⁼ fc and thus f ⁰ , and that frequency deviations via the phase differences in the phase-locked loop are taken into account or compensated for.

The instantaneous phase difference in the Costas loop is determined by the product e (nT) from the multiplication of the in-phase component by the quadrature component

estimated. With small phase differences, the sine term in (16) and thus also the total product e (nT) tends to zero. This makes e (nT) suitable in principle as a control deviation variable for controlling the phase position of the NCO signal. All s ² (nTj

However, the term i contained therein is problematic since it is dependent on the useful signal amplitude. To compensate for this term, the processing block shown in FIG. 10 is called “Ampi. -Comp. ”(Abbreviation for“ amplitude compensation ”).

It gives an estimate of the useful signal component s ² (nT) _from the in-phase and the quadrature components via the relationship

derived.

The useful signal or amplitude compensated output signal e _c (nT) results as for Df «1. (18)

The signal ^e c ( ^nT ) is only dependent on the phase difference ^D Y for

¹ even ^e c ( ^nT ) ^D <R. Before the signal is fed to the NCO as a phase deviation variable, it goes through a so-called loop filter. This filter on the one hand filters out noise components, on the other hand influences the dynamic behavior of the controller. The loop filter determines, among other things, the response times and the overshoot of the controller in the event of sudden changes in the phase difference. It is usually implemented as a first-order recursive low-pass filter (see FIG. 11).

The parameters gO and g1 determine the filter characteristics and the desired dynamic behavior of the loop filter. A detailed description for the optimal parameter design can be found, for example [22].

In principle, the useful signal can be extracted from the respective measurement signal using the demodulation method described. However, there are two problems to be considered according to the status described so far:

- If the useful signal component s (nT) contained in the input signal goes to zero, the signal e _c (nT) goes through the division in the amplitude compensation

Formula (18) against infinity. Since the useful signal corresponds to the amplitude of the coil signal contained in the measurement signal and this depends on the distance from the source module to the coil module with respect to the respective spatial axis (x / y / z), zero crossings in the useful signal according to formulas (4) to (6) are possible. The following shows how the division by zero at zero crossings in the useful signal can be avoided.

- The Costas Loop has the property that a minimization of the control deviation can be achieved either by converging the phase difference towards zero or towards ± p depending on the initial phase position of the carrier signal. After the “locking” of the phase-locked loop, the output signal i (nT) follows the amplitude profile of the useful signal and thus also detects phase changes (ie change of sign). In this method, however, the initial phase position is not clear or is only known with a phase uncertainty of 180 °. An additional mechanism is therefore required that phase position (and thus the initial sign of the respective magnetic field component) is determined. Possible solutions for this are shown below.

As described in the previous section, zero crossings in the useful signal by dividing the amplitude compensation in formula (18) are problematic. To avoid the problem case, the fact is used that the useful signal components in the three measurement signals By, By, and Bz are all based on the same coil-generated magnetic field or the same carrier signal. The amplitudes of the three measurement signals differ according to the respective distance between the reference and coil modules. As stated above, the phase positions either do not differ at all (with the same sign) or by 180 ° (with different signs). A phase difference of 180 ° makes no difference for the estimation of the phase position by means of the phase locked loop: the control deviation goes towards zero for both ^D F ° and ^D F ^p . It therefore does not matter which of the three measurement signal components (By / By / Bz) are used to determine the control deviation or to estimate the phase position of the carrier signal (only the initial phase position can differ by 180 °. As explained above, the Starting phase position without any additional measures an uncertainty of 180 ° anyway, so the difference in the starting phase position by using any measurement signal components is irrelevant at this point). The problem of zero crossings in the useful signal can thus be avoided by always using that measuring signal component for regulating the phase position which has the greatest amount. Provided that the sensor module is always spatially above the coil module (ie in the position range z> 0), according to formulas (1) to (6) at least one measurement signal component is always non-zero. The selection of the largest measured signal component for the phase control prevents the division by zero in the useful signal or amplitude compensation.

As described above, there is a phase uncertainty of 180 ° with respect to the initial phase position of the carrier signal contained in the measurement signal. Thus, the initial signs of the coil-dependent magnetic field components can either be correct or inverted compared to the actual signs depending on the direction of convergence of the phase locked loop (direction ^D F ^{= 0} or ^D F ⁼ ± ^p ). In In this section, methods are presented with which the initial phase position can be determined.

A simple possibility for determining the initial phase position is the use of a synchronization signal which is generated by the source module when the coil control signal begins to be in a positive phase position and is transmitted by radio or cable, depending on the communication interface (FIG. 12). When the synchronization signal is received by the sensor module, the phase position of the NCO signal is reset so that the phase position of the NCO is synchronized with that of the carrier signal and the correct signs for the components of the magnetic field signal are thus automatically obtained.

A slight time delay between the synchronization signal and the initial phase position of the carrier signal in the measurement signal (e.g. due to latency in radio transmission) is not critical, provided that it is less than ^f s / ^f o ^' ° · ⁵ (i.e. half the samples of a sine period). Since an interface (cable / radio) must be available for the transmission of the determined orientation and position data in most applications, this synchronization method can be implemented in practice with very little additional effort.

If the position of the sensor module relative to the coil module is sufficiently known at a defined point in time (for example, directly when the sensor module is switched on), the initial position can also be determined without a synchronization signal. The position is known to be sufficiently known if the sign of the x or y axis of the position vector - which describes the position of the sensor module relative to the coil module - is known. Knowledge of the sign of the x or y axis can be used to correct the sign of the demodulated magnetic field components after the phase loop has snapped into place. Since the sign of the Bx or By component according to formulas (4) to (6) is only dependent on the sign of the x or y axis in the position space, the sign of the corresponding magnetic field component can be directly via the sign of the respective position coordinate be derived. If the sign If the magnetic field component does not correspond to the expected sign immediately after the phase locked loop has snapped in, the reference signal generated via the NCO has a phase deviation of 180 ° compared to the carrier signal, so that the signs of all magnetic field components have to be inverted once in order to obtain the correct sign. Otherwise there is no phase deviation, so that a correction of the sign is not necessary.

In practice, for example, the condition could be set that the sensor module must be "to the right of the coil" before it is switched on. In the case of a coordinate system as in FIG. 13, a positive sign of the Bx component would be derived from this.

This method has the advantage that neither radio communication nor a cable connection between the modules is required for the synchronization. It is therefore also suitable for applications in which the data are either only saved for later offline processing or are processed directly online, for example to control actuators or output signals depending on the data (for example output of an acoustic or visual warning signal for signaling certain body malpositions in biofeedback applications in the field of medical technology or biomechanics).

If there is neither a cable or radio interface in the application, nor can it be ensured that the position is known at a certain point in time, the initial phase can also be determined by transmitting additional information about the generated coil signal. For this purpose, characteristic signal components can be encoded in the coil signal, which can be identified via the signal processing on the part of the sensor module in order to determine the time of synchronization.

In the simplest case, a rectangular signal rise can be provided in the coil signal shortly before the start of the sine signal (with a known starting phase position). In order to ensure a more robust identification with respect to interference, an upstream, sinusoidal signal component with a frequency ^ ^f c before the actual coil signal (with frequency ^f c) can also be provided in the coil signal, so that the sensor module can identify the synchronization time by recognizing the frequency jump.

In the case of magnetic field-based systems with a three-axis magnetic field source, in addition to the AC method, in which an alternating magnetic field signal is generated at a constant frequency, the DC method is also used, in which the coil is excited by a pulsed signal. In the conventional DC method, the three coils of the magnetic field source must be activated sequentially (sequentially) in order to ensure that the coil magnetic fields can be distinguished and thus that the measured magnetic field belongs to the source axes (C, U, Z) on the part of the receiver or the magnetic field sensor , For this purpose, a signal section with the coil switched on and a signal section with the coil switched off is recorded in the sensor unit for each coil activation. To determine the magnetic field components, the difference between the two signal components is formed so that magnetic field components that are contained in both signals (for example the earth's magnetic field) are compensated for and ideally only the coil-dependent component results. For each of the three successively activated coils, a vector with three magnetic field components is determined in this way. Since the acquisition of a data set consisting of X, Y and Z components in the existing three-axis systems requires the duration of three coil excitation periods, the latency of the position calculation with the DC method is generally higher in these systems than in the AC method.

When using the system concept presented with a MARG sensor and a magnetic field source with a single coil, this disadvantage does not apply when using a pulse-shaped control signal, since only one coil has to be activated here and the acquisition time required compared to triaxial measuring systems is therefore a factor of three reduced. The main advantage of DC-based amplitude determination is that it is less susceptible to interference due to Eddy Fields: the magnetic field signal is only sampled when the Eddy Fields, which result from the current induction when the coil is activated in nearby conductors result, have essentially subsided, as shown in FIG. 14. To determine the coil-dependent magnetic field components, a magnetic field vector (B. (T)) at time T with the coil switched off and a magnetic field vector {B _h. T _t )) recorded at 7 ^ _h with the coil switched on. If the pulse rate is high enough that the movement-dependent change in the magnetic field between the two measurement data sets is negligible, the magnetic field vectors can be as

B _lo JO = b _e (T,) + w (T _l ) (19) and

are represented, where b _e represents the direct component resulting mainly from the earth's magnetic field, H>, and w ₂ high-frequency noise components, and B represents the coil-dependent signal component.

By forming the difference between the two magnetic field vectors follows

By using a low-pass filter, the high-frequency noise component can be eliminated, so that the resulting signal

corresponds to the coil-dependent magnetic field component sought. The orientation of the two modules known from the sensor fusion of the MARG data can be used to compensate for the influence of the rotation in the magnetic field components of B (T '), so that the dipole equations (see formulas (19) - (21)) the determination of the relative 3D position between the two modules is possible. literature

[1] Norhafizan Ahmad, Raja Ariffin Raja Ghazilla, and Nazirah M. Khairi, “Reviews on Various Inertial Measurement Unit (IMU) Sensor Applications”, International Journal of Signal Processing Systems, Vol. 1, No. 2, Dec. 2013 [2] Welch, G.F., Foxlin, Έ. Motion tracking: no silver bullet, but a respectable arsenal ”, IEEE Comput. Graph. Appl., Nov./Dec. 2002

[3] Weinhandl JT, Armstrong BS, Kusik TP, Barrows RT, O'Connor KM, “Validation of a single camera three-dimensional motion tracking system”, J Biomech. 2010; 43 (7): 1437-1440 [4] US patent US 8786680 B2, motion capture from body mounted cameras,

DISNEY ENTERPRISES, INC., CALIFORNIA, Aug. 5, 201 1

[5] Bour P., Ozenne V., Marquet F., Senneville B., Dumont E., Quesson B., “Real-time 3D ultrasound based motion tracking for the treatment of mobile organs with MR-guided High Intensity Focused Ultrasound ”, Int J Hyperthermia, Jan 29, 2018

[6] N. Pires; AP Moreira, “3-D position estimation from inertial sensing: minimizing the error from the process of double integration of accelerations”, Industrial Electronics Society, IECON 2013 - 39th Annual Conference of the IEEE, 2013. [7] US patent US20140002063, SYSTEM AND METHOD FOR MAGNETIC

POSITION TRACKING ASCENSION TECHNOLOGY CORPORATION, Jan. 2, 2014

[8] US Patent US7835785, DC MAGNETIC-BASED POSITION AND ORIENTATION MONITORING SYSTEM FOR TRACKING MEDICAL INSTRUMENTS, Nov. 16, 2010

[9] US Patent US8450997, ELECTROMAGNETIC POSITION AND ORIENTATION SENSING SYSTEM, Brown University, Providence, RI (US), May 28, 2013

[10] F. H. Raab, E.B. Blood, T. O. Steiner, H.R. Jones, “Magnetic Position and

Orientation Tracking System ”, IEEE Transactions on Aerospace and Electronic Systems, vol. AES-15, No. 5, Sept. 1979, pp. 709-718. [1 1] J. Kuipers, “SPASYN - an electromagnetic relative position and orientation tracking system”, IEEE Trans. Instrumentati

[12] D. Roetenberg, P.J. Slycke and P.H. Veitink, “Ambulatory Position and

Orientation Tracking Fusing Magnetic and Inertial Sensing ”, IEEE Transactions on Biomedical Engineering, Vol. 54, No. May 5, 2007.

[13] US Patent US20090278791A1, MOTION TRACKING SYSTEM, XSENS

TECHNOLOGIES B.V., Enschede (NL); UNIVERSITEIT TWENTE, Enschede (NL), Nov. 12, 2009

[14] P. Tadayon, T. Felderhoff, A. Knopp, G. Staude, Fusion of Inertial and

Magnetic Sensors for 3D Position and Orientation Estimation, Conf. Proc. of the IEEE Eng. in Medicine and Biology Society, Orlando, USA, 2016

[15] Ke-Yu Chen, Shwetak Patel, Sean Keller, “Finexus: Tracking Precise Movements of Multiple Fingertips Using Magnetic Sensing”, CHI’16, San Jose, CA, USA, 2016 [16] S. O.H. Madgwick, A. J.L. Harrison, R. Vaidyanathan, “Estimation of IMU and MARG Orientation using a gradient descent algorithm,” IEEE International Conference on Rehabilitation Robotics, pp. 1-7, June 201 1.

[17] H. J. Luinge, P.H. Veltink, C.T.M. Baten, “Estimation of Orientation with Gyroscopes and Accelerometer,” Proceedings of The First Joint BMES / EMBS Conference Serving Humanity, Advancing Technology, Vol 1, pp. 844, Oct.

1999th

[18] Hahn, S., "Hilbert transforms in Signal Processing", Artech House, Inc.,

Norwood, 1996.

[19] J.P. Costas, "Synchronous Communications", Proc. Of IRE, vol. 44, pp.

1713-1718, Dec. 1956th

[20] I. S. Xezonakis, M. S. Sangriotis, "An all digital Costas Ioor-like PLL Circuit with a wide locking ränge", Int'nat'l Journal of Electronics, vol. 71, no. 6, pp. 951-965

[21] Bedrosian, E., "A Product Theorem for Hilbert Transforms", Rand Corporation Memorandum (RM-3439-PR), Dec. 1962, 1991. [22] Nicoloso, Steven P., "An Investigation of Carrier Recovery Techniques for PSK Modulated Signals in CDMA and Multipath Mobile Environments", Master Thesis, June 1997.

Claims

claims

1. Method for 3D orientation and 3D position determination of objects, with the aid of a device comprising a reference module with a uniaxial coil and its control electronics and a MARG sensor (Magnetic, Angular Rate, and Gravity-

Sensor), wherein the coil is generated with a periodic excitation current to generate a harmonic magnetic field signal of constant frequency, a sensor module with a MARG sensor, the reference module as a spatial reference for orientation and

The position of the sensor module is used and the data of the MARG sensors in the reference module and sensor module are used to determine the 3D orientation of the respective module cyclically and algorithmically, the MARG sensor of the sensor module detects a mixed signal which is generated from the harmonic magnetic field generated by the coil of the reference module and all other magnetic fields in the vicinity of the device, characterized in that from the mixed signal determined by the MARG sensor of the sensor module not only the amount information but also the phase information of the coil signal is used to extract the coil signal from the mixed signal, whereby the Amplitudes of the magnetic field components of the coil signal are determined with the correct sign, and cyclically from the determined amplitudes of the magnetic field components of the coil signal the unique spatial position of the sensor module relative to the reference module based on the measurement data of the MARG sensor alone on the Se nsormodule is determined.

2. The method according to claim 1, characterized in that the 3D orientation of the respective module is determined on the basis of the data from the MARG sensors in the reference module and sensor module.

3. The method according to any one of claims 1 or 2, characterized in that based on the known orientation of the sensor module, the measured mixed signal is rotated mathematically into a defined reference position in which the z-axis of the MARG sensor of the sensor module is aligned parallel to the coil axis of the reference module is.

4. The method according to any one of the preceding claims, characterized in that the amplitudes of the magnetic field components are determined with the correct sign, after which the unique position of the MARG sensor on the sensor module relative to the reference module is determined on the basis of the measured mixed signal via the dipole equations can.

5. The method according to claim 4, characterized in that the connection between the phase position of the coil signal and the sign of the magnetic field component of the measured mixed signal measured by the MARG sensor on the sensor module is used to correlate the one with the coil signal To extract the signal component in the correct phase and thus with the correct sign from the measured mixed signal.

6. The method according to any one of the preceding claims, characterized in that the amplitude of the coil signal contained in the measured magnetic field signal is determined according to the dipole equations depending on the distance between the reference module and the sensor module with respect to the three spatial axes x, y and z.

7. The method according to any one of the preceding claims, characterized in that the portion of the coil signal of the reference module has an amplitude-modulated signal in which the useful signal can be viewed as a distance-dependent amplitude and the sinusoidal excitation signal as a carrier signal.

8. The method according to any one of the preceding claims, characterized in that an amplitude demodulation method is used to determine the useful signal required for the position determination from the mixed signal determined by the MARG sensor of the sensor module.

9. The method according to claim 8, characterized in that a coherent demodulation method is used as the amplitude demodulation method for determining the correct phase of the amplitude of the coil signal contained in the mixed signal.

10. The method according to claim 9, characterized in that a Hilbert transformation is carried out in the coherent demodulation method.

11. The method according to claim 10, characterized in that a phase locked loop, preferably a Costas loop, is used in the coherent demodulation process to avoid drift effects or other long-term effects.

12. The method according to any one of the preceding claims, characterized in that is used to avoid zero crossings in the useful signal during the amplitude compensation that never all spatial magnetic field components in the measured magnetic field signal can become zero at the same time.

13. The method according to claim 12, characterized in that in order to avoid zero crossings in the useful signal during the amplitude compensation, the largest component in terms of amount in the measured magnetic field signal is used for the phase control.

14. The method according to any one of the preceding claims, characterized in that the initial phase position of the coil-dependent magnetic field components in the measured magnetic field signal is synchronized by a synchronization signal, in particular a synchronization signal that is wired or transmitted by radio between the reference module and the sensor module.

15. The method according to claim 14, characterized in that the synchronization signal is generated at the beginning of the positive or negative phase position of the coil control signal.

16. The method according to any one of claims 1 to 13, characterized in that the initial phase position of the coil-dependent magnetic field components in a known relative spatial position between the reference module and the sensor module, preferably when the device is switched on.

17. The method according to any one of claims 1 to 13, characterized in that the initial phase position of the coil-dependent magnetic field components is synchronized by transmitting additional information with the aid of the generated coil signal.

18. Device for 3D orientation and 3D position determination of objects, comprising a reference module with a uniaxial coil and its control electronics and a MARG sensor (Magnetic, Angular Rate, and Gravity sensor), the coil for generating a harmonic magnetic field a periodic excitation current can be applied as a constant frequency, a sensor module with a MARG sensor, the reference module forming a spatial reference for the orientation and position determination of the sensor module and algorithmically the 3D from the data of the MARG sensors in the reference module and sensor module. Orientation of the respective module is determined, with the MARG sensor of the sensor module, a mixed signal can be detected, which is composed of the harmonic magnetic field generated by the coil of the reference module and all other magnetic fields in the vicinity of the device, characterized in that that only one MARG sensor is on the sensor module rdnet, which receives the mixed signal generated by one coil of the reference module for generating a harmonic magnetic field signal of constant frequency, and from this the unique spatial position of the sensor module relative to the reference module can be determined on the basis of the measurement data of the MARG sensor alone on the sensor module.

19. Device according to claim 18, characterized in that the reference module and / or the sensor module has a microcontroller for evaluating the measured values of the MARG sensor.

20. Device according to one of claims 18 or 19, characterized in that the sensor module and optionally the reference module has a radio module for data transmission.