WO2005015257A1

WO2005015257A1 - System for determining the position of a mobile transceiver in relation to fixed transponders

Info

Publication number: WO2005015257A1
Application number: PCT/EP2004/051561
Authority: WO
Inventors: Martin Vossiek; Leif Wiebking; Klaus Pistor; Mark Christmann
Original assignee: Siemens Aktiengesellschaft
Priority date: 2003-08-06
Filing date: 2004-07-21
Publication date: 2005-02-17
Also published as: DE10336084A1

Abstract

The invention relates to a local position measuring system. In said system a base station is located on a mobile object, whose position is to be determined and a plurality of transponders are distributed around the locality.

Description

SYSTEM FOR DETERMINING THE POSITION OF A MOBILE TRANSCEIVER WITH REGARD TO FIXED TRANSPONDER

description

Local position measuring system

In the area of material tracking and logistics, there is a great need for systems that are able to determine the local position of means of transport such as cranes, vehicles, trolleys, pallet trucks, forklifts, AGVs (Automated Guided Vehicles). 10 Usually radio-based local position measuring systems work in such a way that the mobile object (crane, forklift, vehicle, etc.) is provided with a transponder and the area around the object - for example the walls of a workshop 15 - is equipped with several so-called base stations. The transponder then cyclically sends out signals that are received by all base stations with a characteristic delay difference. By calculating these running time differences, the local position 20 of the mobile object is then concluded. A disadvantage of these solutions is that for the exact determination of the running time differences in the base stations, it is absolutely necessary that either the clocks or the clock signals in the base stations run very synchronously, for example expensive frequency standards must be used, and / or means must be provided which synchronize the clocks of the base stations and / or additional reference transponders precisely measured with regard to their position are used. In addition to these expenditures, it is also disadvantageous that a measurement-related information gain can only be achieved by transmitting at least two measured values from, for example, two base stations to a central unit, where they have to be offset against one another. Accordingly, it is necessary in such systems that all base stations are connected to a central station via a wired or wireless network. Because in large measuring ranges and / or angled rooms mostly by many Such a network causes considerable costs for base stations. These costs can often be far higher than the costs of the actual measuring components.

Proceeding from this, the object of the invention is to provide arrangements of system components for distance measurement with associated system topology as well as methods which solve the position determination of a moving object in a particularly cost-effective and efficient manner with high measurement accuracy and measurement rate under the circumstances mentioned.

This object is achieved by the inventions specified in the independent claims. Advantageous refinements result from the dependent claims.

Accordingly, an arrangement for determining the position of a mobile object, in particular a means of transport in the form of a crane, vehicle, trolley, pallet truck, forklift or AGV, has a base station for transmitting a base signal and / or receiving

Transponder signals and a large number of transponders for receiving the basic signal and for the subsequent or consequent transmission of transponder signals. Furthermore, the arrangement has means for determining the position of the movable object, taking into account the transponder signals. The base station is arranged on the mobile object. The position of the transponders in the room is known, in particular the transponders are arranged fixed in space, for example in a factory hall.

By reversing the usual arrangement of the base station and transponders, numerous new possibilities and advantages arise, which are described below. The advantages alone make it clear that such a reversal was obviously not obvious to the person skilled in the art, since he has not yet carried it out. The transponders are preferably arranged in this way and the means for determining the position are set up such that the position of the object can be determined not only in the xy plane, for example in the factory building, but also the height of the object above the floor, that is to say the position of the object in all three spatial dimensions.

The means for determining the position are arranged in particular in the base station.

To make better use of the frequency bands available for the large number of transponders, the arrangement is advantageously operated in time and / or frequency multiplex. However, some of the transponders preferably transmit their transponder signal in the same frequency band. For this purpose, the transponders are arranged, for example, at least twice the distance from one another at which the transponder signal can just about be received. Alternatively, the transponders 20 _max are separated from one another, where D _{mβx is} the distance at which the transponder signal with a selected statistical confidence, which is, for example, approximately 90%, is no longer above a selected threshold. The values of the confidence and the threshold, in particular the noise threshold, are usually derived from measurement technology and from specifications. The threshold is determined, for example, as follows.

Default: 90% of the measured values in the entire working range should be at most 3 dB above the noise at D _max . Then the following steps are carried out:

- measure noise without transponder,

determine the statistical distribution of the amplitude A of the measurement signal as a function of the distance d,

- D _max can then be determined using the default (90%).

Instead of 90%, the desired confidence in other applications can also be 80% or 95%, for example. The means for determining the position of the object are preferably set up such that a position of the object in the past can be taken into account when determining the position of the object. This position of the

Saved object in the past and predicted a predicted position of the object with a model for the movement of the object. The position of the object predicted with the aid of the model and the position of the object measured with the aid of the transponders and the base station are then both taken into account for determining the position of the mobile object. An arithmetic mean can be formed from the predicted and the measured position. However, the predicted and the measured position are preferably weighted with the aid of a confidence measure which expresses how likely the correctness of the predicted or the measured position is assumed to be.

In a similar way, the value of the measured position can be improved by itself by a transponder signal, from which a position results, which is predefined from the positions resulting from the other transponder signals and / or from a predicted value Position deviates, is not taken into account for determining the position of the object.

The model for the movement of the object can be set up in such a way that an approximately constant acceleration of the object is assumed. In practice, this has proven to be a very good approximation for means of transport.

Most of the above-mentioned aspects for improving the accuracy of the position determination can be taken into account if a, in particular extended, Cayman filter is used to determine the position of the object. In terms of circuitry, the arrangement can be implemented particularly inexpensively if the transponders each have an oscillator which can be excited quasi-phase-coherently by the base signal to generate the transponder signal actively, that is to say with its own energy supply, by the base signal.

The base station is preferably connected to the means for determining the position of the object via a wireless interface, in particular WLAN.

A mobile object has a base station for transmitting a base signal and / or receiving a transponder signal to and / or from a multiplicity of transponders, the position of which in the room is known, the position of the object being determinable taking into account the transponder signals.

A large number of transponders are arranged in a room in order to receive a base signal from a base station arranged on a mobile product and then to send transponder signals, taking into account the position of the object.

In a method for determining the position of a mobile object, in particular a means of transport, a base station for transmitting a base signal and / or receiving transponder signals, a plurality of transponders for receiving the base signal and thereby sending out transponder signals and means for determining the position of the Object used taking into account the transponder signals, the base station being arranged on the object and the position of the transponders in the room being known.

Advantageous configurations of the object, the large number of transponders and / or the method result analogously to the advantageous embodiments of the arrangement and vice versa.

Further features and advantages of the invention result from the description of exemplary embodiments with reference to the drawing. It shows:

1 shows an arrangement of transponders and a base station arranged on a mobile object in a hall;

Figure 2 shows a base station and a transponder that communicate with each other;

3 shows a demodulator for a base station according to FIG. 2;

FIG. 4 shows an alternative demodulator for a base station according to FIG. 2;

Figure 5 spectral components of an arrangement with a base station and a plurality of transponders;

FIG. 6 a distance determination via edge detection;

FIG. 7 shows a channel division in the frequency domain;

F gur 8 a main channel with center frequencies of subchannels;

FIG. 9 shows a possible transponder arrangement when using subchannels;

FIG. 10 shows a transformation of radial distances into Cartesian coordinates; FIG. 11 geometric quantities in a trigonometric target position calculation;

FIG. 12 simulation of the measurement noise model with a Gaussian noise level;

FIG. 13 shows an error ellipse and a validity area of the distance measurements calculated therefrom;

FIG. 14 shows a flow diagram of a sequential expanded Cayman filter;

FIG. 15 shows a visualization of the sequential expanded Cayman filter;

Figure 16 shows an application scenario for an intelligent factory;

Figure 17 shows an application in an automatic indoor crane;

Figure 18 is a block diagram of a base station;

Figure 19 is a block diagram of a transponder.

Properties of an exemplary embodiment of an arrangement according to the invention consist in that a) the base station is able, based on a radar measuring principle, to directly determine the absolute distance to a transponder, b) the base station is on a mobile object and the transponders, the For example, are attached to the room walls, act as place / way markers, c) the transponders send the base signal (measurement signal) from the base station as a transponder signal, coded or modulated, so that a transponder has a characteristic identifier that is assigned by the base station recognized and can be assigned to a placemark or a spatial point.

The advantages of such a system topology are that, in the simplest version, the transponders do not have to be networked with one another or with a central station. On the other hand, each measurement from a base station to a transponder leads to an information gain - i.e. a radial distance value to the placemark. Furthermore, synchronization of the clock signals of several

Transponders and reference transponders are completely dispensed with

The description of exemplary transponders and base stations which are particularly advantageous for such a system topology is explained further below. Further advantageous properties of the arrangement then result from the properties of these elements.

The basic embodiment of the topology according to the invention is shown in FIG. 1 using an exemplary arrangement.

For example, on the walls and / or the ceiling of a hall, the transponders Ti, T ₂ , ..., T4 are mounted at defined locations whose coordinates have to be measured in the global coordinate system. The transponders are preferably attached in such a way that the base station Bi can find as many and as far as possible always as many and on average approximately the same distance away in each area of the measurement area in their detection range. A uniform but preferably 50% offset arrangement of the transponders on the opposite walls, as indicated in FIG. 1, advantageously solves this task, for example for narrow rectangular areas, for example for typical factory halls. The transponders are only to be supplied with energy and receive their characteristic modulation frequency / coding, for example through a programming process or also through a programming plug. As has already been described, there is no measurement data on the transponders and there is therefore no need for a complex background network. The transponders practically act as placemarks. The distance information is generated exclusively at the base station mounted on the measuring object, which has the advantage that controls based on the measurement data can be carried out directly on the moving object. If you want to have the data available at a stationary point, for example in a central station Z on the edge of the hall, the data can be sent there from the base station, for example via a LAN interface (Local Area Network) LAN, preferably via a conventional WLAN (wireless LAN) radio interface to get redirected. Since, in the simplest case, the transponders do not have to contain an interface or means for signal processing, they can be implemented very inexpensively and in a power-saving manner, and furthermore they can be easily assembled and installed. As a result, it is reasonable to use a relatively large number of such placemarks in terms of costs. Because a base station is always able to determine a large number of placemarks, there are some important advantages. For one, each increases radial

Distance measured value to a placemark the accuracy of the spatial location determination. In particular, a high degree of diversification of the measurement paths can very effectively reduce the physically unavoidable problem of measurement inaccuracies due to multipath reflections.

Multipath reflections, i.e. the property that radio signals not only get the shortest way from the base station to the transponder and vice versa, but are also reflected on objects and walls in a roundabout way, are the main cause of measurement errors in radio-based distance measuring systems indoors. To the others, a large number of placemarks results in a highly redundant arrangement. The failure of individual transponders with a suitable system design leads to only a very slight loss of accuracy, but the basic functionality of the arrangement is retained in any case. A transponder identified as defective can also be replaced during operation. The latter feature can lead to considerable cost savings, particularly in production plants, since not every malfunction leads to a production stop. Suitable algorithms, which, for example, check whether all the placemarks to be expected can be found at the currently determined position, can be used to identify the lack or faulty operation of a transponder by the base station. An error message to the control room can then trigger the necessary repair measures.

So far, only one topology with only one base station was shown in the versions. Of course, the arrangement is also on topologies with several

Base stations can be expanded, one of which, for example, is arranged on one of several movable objects.

In the following, particularly advantageous designs of the basic elements of the arrangement will now be discussed.

As can be seen from FIG. 2, an exemplary arrangement of a base station BS and a transponder TR of the multiplicity of transponders has a multiplicity of individual components. The base station BS and the transponder TR communicate with one another.

The base station BS comprises in particular an oscillator OSZ _B for generating an oscillating signal s _tx (t) which is output or can be tapped at an oscillator output. The oscillator output is with a transmit antenna connected, which can optionally also be used as a receiving antenna ANT _B , as shown here, so that the signal s _tx (t) can be emitted via the antenna ANT _B.

In the present case, a directional coupler RK is connected in the base station BS between the oscillator output and the antenna ANT _B. This has an additional output that leads to a mixer RXMIX and other components.

The transponder TR has an antenna ANT _T , with which the signal of the base station BS, which was generated with the oscillator OSZ _R and sent out via the antenna ANT _B , can be received as the reception signal e _rxt (t). In the present example, the antenna advantageously also serves as a transmitting antenna ANT _T.

In addition, the transponder TR has an oscillator OSZ _τ connected to the antenna ANT _T. A clock control CKL / Sw is also provided to excite the oscillator 0SZ _τ . The oscillator OSZ _τ is cyclically switched on and off with the clock control CLK / Sw with a frequency f _m . The signal sosz (t) generated by the oscillator OSZ _{τ is} generated quasi-phase coherently with the comparison signal sigiN.

It is crucial that the oscillator OSZ _B and the oscillator 0SZ _{τ are} quasi-phase coherent (quasi-phase-locked). This is done as follows: Switch on the oscillator 0SZ _τ , while the signal sent by 0SZ _B , i.e. the transponder receive signal e _rx t (t), couples over to 0SZ _τ - which happens because 0SZ _τ transmits the signal from OSZB via the Transponder antenna, which is connected to OSZ _τ , receives - so OSZ _τ swings with a phase predetermined by the signal from OSZB. The property that an on-phase oscillator tries to follow the phase of a stimulating signal is a basic physical property of any oscillator, which is here However, according to the invention is used to couple two oscillators quasi-phase coherently with one another or to operate the oscillator 0SZ _τ quasi-phase coherently to the oscillator OSZ _B or to the signal e _rxt (t).

If the frequencies of the oscillators OSZ _B and 0SZ _{τ are} exactly the same, then s _0S7 , (t) corresponds exactly to the received signal e _rxt (t) apart from the switching pauses. If the frequencies between the oscillators OSZ _B and OSZ _τ deviate from one another, the phase difference between the two signals changes during the duty cycle - that is, after the coherent oscillation. However, if the duty cycle is very short, the resulting phase difference is also very small, so that the phase difference between the base signal and the generated comparison signal is small, the term being small in relation to the intended communication or measurement task. For example, the value π / 10, that is about 20 °, is often used as the limit for a small phase deviation. Such signals with only small phase deviations are referred to below as quasi-phase coherent and the period of time in which this coherence exists is called the coherence period. The two vibrations can therefore be regarded as almost coherent (here called "quasi-phase coherent") for this very short time. If you switch on OSZ _τ cyclically for a very short period of time, the quasi-phase coherence between the two oscillators remains permanently, since OSZ _τ swings coherently with the phase of the signal generated by OSZ _B , _i.e. the received signal e _rxt (t), each time it is switched on. In effect, this means that the signal Sosz (t) almost corresponds to a signal that would result if the received signal e _rxt (t) were _{blanked out} and sent back with a switch in the same cycle. The decisive difference is that Sosz (t) was newly constructed and therefore has an amplitude that is several orders of magnitude larger than a signal that is only reflected back modulated. In the transponder TR, a more or less large part of a received signal e _rxt (t) is coupled to the oscillator OSZ _τ . The received signal e _rxt (t) excites the oscillator OSZ _τ quasi-phase coherently to oscillations, as a result of which it generates an oscillator signal which is coupled out of the oscillator as the signal sosz (t) and is derived via an output. The input for the received signal e _rxt (t) and the output for the oscillator signal B can be completely or partially identical. But they can also be implemented separately.

The signal s ₀ s7 (t) generated in the transponder TR is sent back to the base station BS by means of the antenna ANT _{T of} the transponder TR and received by the base station BS with the antenna ANT _B.

The signal received in this way is separated from a currently transmitted signal in the base station BS via the directional coupler RK and mixed in the mixer RXMIX with part of the signal currently generated by the oscillator OSZ _{B of} the base station.

Mixing components of no interest are suppressed with a FLT filter connected downstream of the RXMIX mixer. This filter FLT of the base station BS is preferably designed as a bandpass filter, the center frequency of the clock rate of the clock controller CLK / Sw of the transponder TR being adapted.

The exemplary base station is thus designed like a common FMCW radar device, the topology shown being only an example, but in principle any conventional designs of CW (continuous wave) radars with or without frequency modulation can be used. The frequency modulation can take place continuously linearly or else in arbitrary courses or in discrete steps. Advantageously, only the components behind the reception mixer RXMIX and the signal evaluation have to be adapted according to the modulation in the transponder TR.

The function of the advantageous method for

Distance measurement can be derived as follows:

A monofrequency signal of the form is initially used as the transmission signal s _tx (t) of the base station

s _a (t) = sm (ω _c + ω _m ,) t + φ ₀

assumed, where ω _{c is} the center frequency, ω _{sw is} an initially fixed modulation frequency, t is time and φ _{0 is} any phase offset. This signal is from the

Base station sent to the transponder and hits by the transit time τ / 2, with τ / 2 = dist / c, dist as the distance between base station BS and transponder TR and c as the speed of light, delayed with transponder TR as transponder reception signal e _rx (t) = s _x (t-τ / 2). As described above, the oscillator 0SZ _{T of} the transponder TR is cyclically switched on and off. The period with which the oscillator 0SZ _{τ is switched} on or off is referred to below as Ts, where Ts = 1 / (2f _rak ).

With each switch-on, the oscillator 0SZ _τ oscillates with its preferred arrangement exactly with the current phase of e _rxt (t) at its oscillation frequency ω _0S z. If, for example, the oscillator is switched on at time t = -τ / 2, it oscillates with the phase

.phi..sub.i

^{= ar} S ( ^s tχ (- ^τ )} ⁼ ( ^ω c- ^| - ^ω sw) (- ^τ ) - ^, -Φo

on and the oscillator signal s _os? , (t) therefore corresponds to: ^• Osz (t) ^: sin (ω _osz • I t + - | - (ω _c + ω _sw ) - τ + φ ₀ )

The oscillator _signal s _OS7 , (t) then arrives, again delayed by the transit time τ / 2, as the reception signal s _rx (t)

^s rx (*) = s _0Sz (t ~) = sin (ω _0S7 ^• t - (ω _c + ω _sw ) ^• τ + φ ₀ )

to the base station BS and is mixed in it with the current transmission signal s _tx (t). If one neglects the high-frequency mixed products and one simply assumes that ω _0S7 . = ω _c , which is possible with a suitable choice of ω _sw without restricting generality, the mixed signal s _mLX (t) results

^s mix) = ^cos (t ^{• ω} sw + ^{τ •} ( ^ω c + ^ω sw)) ^•

In the following it is now assumed that behind the reception mixer RXMIX electronic components / means DEMOD are provided, which lead to the fact that the change in voltage over time in the time interval between switching on and off, ie from 0..Ts, in the sense of averaging is to eliminate. A simple envelope demodulator according to the prior art, in which the signal is rectified and then low-pass filtered, would e.g. work in this sense. An embodiment of such a simple demodulator DEMOD with a rectifier GR and the low-pass filter TP is shown in FIG. 3.

An advantageous variant of a demodulator DEMOD, outlined in FIG. 4, consists in mixing the mixed signal down to a low frequency, preferably with a frequency close to or equal to the cyclic frequency f _{mk of} the clock control CLK / Sw, and then using a filter TP which has at least one low-pass behavior to filter. A possible execution of this Variant has a local oscillator LOZF, a mixer ZFMIX and a filter FLT. If the frequency of the local oscillator LOZF is designed so that negative mixing frequencies can arise, the mixer ZFMIX, as is generally known, as an IQ mixer (IQ: in-phase and quadrature phase, ie 90 ° phase-shifted) is the real one - and supplies the imaginary part. The filter FLT is preferably designed as a low-pass filter or a band-pass filter.

When viewed in the time interval 0 to TS, the means shown have the effect that a kind of effective value of the voltage is determined by the mixed signal s _mιx (t). This effective value then forms the actual measurement signal s _meSs (t) _below . Constant amplitude factors are neglected in the following presentation without loss of generality. The effective value of s _mιx (t) in the time interval 0 to TS, _i.e. s _mΘSS (t), is calculated as follows:

(\ cos ω _c • τ + ω, sw ^{τ +} 2 ' ^T s' ^ω sw ^• sin V s ^{' ω} sw s _mess (t) = J _o ^S s _mιx (t) = const. - V ^ω sw

Since the measuring system is preferably operated frequency-modulated, the case is considered below in which the modulation frequency ω _{sw is} modulated as a function of time. If ω _{sw is} detuned linearly from -B / 2 to + B / 2 over the bandwidth B during a period of T, the following applies

_2πBt

the result of s _mess (t) for the resulting FMCW measurement signal s _meSsfmc w (t): Sm ^m e ^e s ^s s ^s fmcw-- ( ^t ) ^{= c0s ω} c ^{"τ +}

As was shown above, in the derivation with the signal s _me s _sf m _cw (t) only the signal is exactly reproduced during a switch-on period. The fact that this signal is additionally modulated by the periodic modulation in the transponder results in a frequency shift of s _meSs m _c (t) or additional spectral components. Since this effect of modulation is generally known and is described, for example, in M. Vossiek, R. Roskosch and P. Heide: "Precise 3-D Object Position Tracking using FMCW Radar", 29th European Microwave Conference, Munich, Germany, 1999, in In the following, only a single spectral _{component is} considered as an example, or the signal s _meS sfmcw (t), initially as if it had not been cyclically modulated.

This measurement signal Sm _es sfm _c w (t) now has two decisive and very advantageous differences from signals from standard FMCW transponder systems.

On the one hand is the measuring frequency f _mesΞ , that is, the derivation of the phase of the cos argument

f _mess = dist - - + - - = f _beat + Δb, T c 2-T

corresponds to the frequency component Δb = B * Ts / (2 T) shifted. The frequency fea corresponds to the normal FMCW measurement frequency and contains the actual measurement information, namely the distance between the base station BS and

Transponder TR with τ = 2 dist / c. On the other hand, the signal S _me ssf _mc w (t) is amplitude-weighted with a trigonometric, in particular Si function (Si (x) = sine (x) / x). If one transforms this amplitude-weighted signal with the Fourier transformation in the frequency domain, the outstanding metrological properties of this signal become clear. The Fourier transform of the present Si function yields a rectangular function, the width Δp of the rectangular

A ^BT S

is.

Similar to M. Vossiek, R. Roskosch and P. Heide: "Precise 3-D Object Position Tracking using FMCW Radar", 29th European Microwave Conference, Munich, Germany, 1999, presented separately, starting from peak values it is advantageous to evaluate the frequency or phase spacing, Δf or ΔΦ, between the two sidebands to calculate the distance dist, since then the modulation frequency of the transponder TR, which is not exactly known a priori, is not included in the evaluation comes with. The frequency spacing, Δf corresponds to fbeat twice, as can be seen in FIG. 5. If the measurement signal, as has already been shown above, is demodulated with a classic envelope demodulator, or mixed with other means as precisely as possible to the frequency 0 or another possible known frequency, the frequency spacing from a side band, for example fbe _a t _/ - as it is customary in normal FMCW systems, is of course also sufficient to determine the distance.

In FIG. 5, all are summarized once again

Spectral components of the measurement signal of the arrangement according to the invention are shown. The baseband frequency components result from direct reflections on the object, as is usual with an FMCW radar. Centered around the modulation frequency f _mk are the spectral components /

Modulation components of a transponder. By determining The distance from the base station to the transponder can then be deduced from, for example, f _beat using the formulas shown above.

The range spectrum of a transponder consists of two spectral blocks symmetrical to the center frequency, the range information preferably being determined on the basis of the inner edges of the spectral blocks. These edges are preferably determined by differentiating the spectrum. As indicated in FIG. 6, the maxima Extr.2 and Extr.3 of the differentiated spectrum correspond to the turning points of the inner edges. The determining distance is designated by DIST, SCHW is the threshold value above which the detected extremes must lie, and the dashed line DELTA shows the derivation of the spectrum.

Determination of the radial velocities via the Doppler effect

As is known for FMCW radar, these can also be used for

Determination of relative speeds can be used. For this purpose, the distance measurement to each transponder takes place via a rising and falling frequency _ramp , so that from the two measured frequency differences ΔE _ώwι and ΔF _U via the known FMCW

Doppler relationship also results in the radial speed of the target in the direction of the current transponder.

V = - ^ 2- /, HF

The radial distance results from the relationship when the target is moving

Procedure for addressing the transponders

As already shown above, the transponders thus act as placemarks. Accordingly, it is necessary for a base station to be able to unambiguously assign each transponder signal that it receives to a transponder and thus to a place of origin during a measurement. This is solved according to the invention based on the arrangements according to FIGS. 2 to 4 as follows.

As has already been indicated in FIG. 5, the spectral components that originate from a transponder are centered around its modulation frequency f _mk . If the modulation frequency f _{mk is} selected differently for each transponder, each signal component in the measurement signal is the

To clearly assign the base station to a transponder. The following boundary conditions must be observed: The modulation frequencies f _mk , the N (i = 1 ... N) transponder should be selected so that the signal components of different transponders in the measurement signal of the base station do not overlap or interfere with each other. Consequently, the modulation frequencies f _ml must be sufficiently far apart.

The required minimum distance ΔE _m mm-_ between two transponder modulation frequencies can be estimated as follows. In general, it can be assumed with the arrangement that after a maximum system measuring range D _max at the latest, a transponder signal is so strongly attenuated by the normal propagation attenuation that it has completely disappeared in the measurement signal of the base station. If the modulation frequencies f _mk i are then selected such that they are further apart than the frequency range ΔJ0 specified by the measuring range D _max , mutual interference is largely excluded. The minimum required channel bandwidth results from the equations shown above 4 B - d "

ΔF = c - T

B describes the frequency ramp bandwidth, T the ramp duration.

The spectral channel division described in the baseband is illustrated graphically in FIG. 7. Now the maximum evaluation bandwidth and thus the number of channels that can be implemented is limited in a practically realizable base station. If the arrangement e.g. used in an elongated hall, while at the same time a high transponder density is required on the walls, the number of channels may not be sufficient. In this case, the channels can be used multiple times by stations that are very far apart. According to the invention, the always necessary

Differentiate between different transponders transmitting in a channel by introducing so-called subchannels whose center frequencies are slightly shifted by the amount Af _{sub in} relation to the main channel center frequency (see FIG. 8).

It is very important that transponders with center frequencies of the same main channel are so far apart from each other that the target to be measured is achieved by a maximum of only one transponder per channel. The first index hereinafter denotes the main channel, the second the subchannel of the associated main channel.

As an example, FIG. 9 shows a conceivable arrangement of the transponders for a system with six main channels, each with three subchannels. D _max is the distance at which a transponder is just visible from the mobile base station. It is important that transponders of the same channel are at least 2D _max apart. This assembly ensures that with each measurement pro Main channel only one transponder is visible in one channel.

The modulation frequencies f _mki of each transponder can be determined very precisely, since it is known that the spectral _{components of} a transponder are each mirror-symmetrical about the modulation frequencies f _mki . In the simplest case, the mean value of the frequency of the right and left modulation components can simply be formed to determine the modulation frequency. A further improvement in the accuracy and reliability in the determination of the respective modulation frequency can be achieved by correlation or folding operations or other known operations for the detection of symmetry points such as center of gravity or

Achieve torque calculations. The operations mentioned are each to be applied individually to the area of a channel.

If mobile units move in very large areas, this addressing space, which is formed by the channels and subchannels, may not be sufficient to find a unique transponder identifier at every point in the area. This case is solved according to the invention as follows: Since a base station always has several during a measurement

Transponder 'sees ¹ , several neighboring transponders can form address groups that are unique in the entire area. It is only necessary to ensure that the sequence of neighboring transponder addresses is not repeated in different areas.

Frequency multiplexing of the individual transponders ensures that the signals of other transponders in the immediate vicinity are not disturbed. Another advantage of this is the time saved with the

Frequency division multiplexing. All transponders transmit at the same time and are also recorded together. Neither arise Time shifts between individual measuring radii still require a communication layer above the measuring layer. The transponder is therefore an extremely simple and at the same time inexpensive and robust system component.

Another advantage of this invention is that several base stations are located in the area to be measured and can also measure. The transponders oscillate in a quasi-phase coherent manner on external signals. For example, if two base stations send their ramp simultaneously to a transponder in the immediate vicinity, this will swing statistically distributed to one of the stations, namely the one that couples the greatest instantaneous amplitude to the oscillator 0SZ _τ at the time the transponder is switched on. Consequently, a weaker base station also has a statistical chance, for example, of having full amplitude at the zero crossing of the oscillation of the stronger station and of allowing the transponder to oscillate coherently on its signal. If the transponder swings coherently onto one base station, this is completely uncorrelated for the other station with its own signal (there is no phase relation) and only has a damping effect. This provides sufficient signal levels and few

Base stations between which a certain distance is not a problem.

For this new arrangement, the simple structure of the transponders and the use of frequency division multiplexing means that neither protocol nor network are required in the system. The transponders act as simple waypoints that only need to be supplied with voltage. This principle eliminates the need for a complex network. The unique identification of each individual transponder is ensured via its characteristic modulation frequency, a protocol including that otherwise the necessary communication layer is completely eliminated in the arrangement.

Due to these outstanding properties, a local position radar (LPR) according to the described arrangements represents the simplest possible position measuring system based on microwave technology.

Coordinate transformation of the radial distances

Only the radial distances and speeds between the target object and the respective transponders are obtained as measurement results from the FMCW evaluation. However, since the Cartesian coordinates of the target object are of crucial importance in practice, the information obtained in a radial system must be transformed into a Cartesian system (cf. FIG. 10).

There are two ways to do this:

- Transformation via trigonometric relationships.

- Transformation via an extended Kaiman filter with pre-sorting adapted to the environment.

The transformation via trigonometric relationships has the advantage that the method is straightforward to implement in terms of programming and does not require any initial conditions.

However, there are areas in which the algorithm described below is poorly conditioned. As a result, the same error of a measured radial distance, depending on the target position, can have different effects on the error in the Cartesian values after the transformation. The problem of poor physical condition tends to occur especially in the "peripheral areas" Transformation algorithm, which leads to an undesirable heterogeneity in the measurement error of the LPR.

Furthermore, a reliable evaluation of the results with regard to their variances is almost impossible.

For this reason, a transformation has been transferred and applied to this task via an expanded Kaiman filter described below. This algorithm offers the possibility of evaluating the measurement results and makes almost complete use of the available information. In addition, the problem of poor condition does not arise with this method.

A reliable initial value is required to initially initialize the Cayman filter. This is determined using a trigonometric transformation.

Calculation of an initial value using trigonometric relationships

In order to be able to transform the measured radial distances into Cartesian coordinates with a recursive caiman filter, an initial value is required at system start, which must be as close as possible to the actual location of the target object.

For this purpose, a geometric method described below is used, in which the best radial distances are selected by presorting and the target coordinates are calculated with these triples and then averaged.

Target coordinate calculation from a triple of transponders:

According to the geometric arrangement of the three transponders B1, B2 and B3 shown in FIG. 11, the target vector z - z _a ~ z _ß - z _y can be calculated in three different ways:

The angles o, ß, γ can be calculated using the cosine theorem from the respective partial triangle. D _a , D ~, D are the right-hand rotating matrices by the indicated angular amount. However, the selected three transponders can have a different geometric arrangement (eg lie on a straight line) or the target can lie outside the large triangle, for example. Depending on the constellation (a total of five different geometrical arrangements are possible), the rotation matrices D _α and D may have a clockwise or a counterclockwise rotation. D _ß always acts clockwise. The following rotating matrix combinations are possible for any transponder-target constellation:

D _a , D _ß , D _r

where the arrow here implies turning right or left. Since you do not know the location of the destination, but want to measure it, the comparison of z _a and z _γ (turning clockwise and anti-clockwise once) with z _ß helps. In this way you can determine the correct direction of rotation of D _a and D _γ and calculate the target position z from the mean of z _a , z _ß and z _γ . At the same time, a possible error in one of the three measured distances d, d ₂ , d ₃ is weakened by the averaging. Are the deviations between z _a , z _ß and z _γ larger than 2m, the triple transponder is completely excluded from the total calculation.

Target coordinate calculation by averaging the results of different triples

The significant error in the radial position data caused by multipath propagation is almost Gaussian distributed over the entire distance range. The resultant advantage is the possibility of achieving a significant improvement by averaging by including the results of different triplet triples. Since the measurement error primarily depends on the level of the received transponder signal P _t , it is sufficient to preselect the n stations for which ^ ≥ E ^ _ιn applies, where P _{mm describes} the smallest empirically determined level at which the raw signal evaluation still provides usable distance values.

If you get n stations through this pre-sorting, follow

Values for the target coordinates that can be averaged. In the following it is checked whether individual coordinate values deviate from the mean value by more than 5 m. If this is the case, these are excluded so that the overall result is not adversely affected. A new calculation of the mean value from the remaining data follows.

Position calculation using a Cayman filter

A caiman filter (KF) is a model-based recursive filter. It is ideally suited for estimating an object position, as the filter can also be used for variable

Observation interval the past measured values can be taken into account on the basis of the movement model of the object. In addition to the estimated object position, the KF also provides a statistical measure of the reliability of the estimate. With the help of the movement model and the reliability measure can predict a space in which the object to be tracked is located. According to the invention, it is thus possible to reject the measurements disturbed by multipath propagation. Because of the non-linear relationship between the measured

An extended KF (Extended Kaiman Filter, EKF) is used for distances and the Cartesian position to be estimated.This transforms the measured radial distances directly into Cartesian target coordinates and optimally estimates the system status, based on all information measured up to the current time. The problem of the poor condition of certain target areas described in the trigonometric method does not occur here in principle, which leads to a significantly higher homogeneity in the measurement accuracy. In addition to the motion model, the covariance matrix of the measurement noise also offers the possibility to improve the filter effect through a priori knowledge. Studies have shown that the measurement noise can be modeled using the following equation:

0 λ R ^{0 r} 22 j ^ύ pred dB ^{+ ύ} max, <ZB

d, muhιpatk r ₂₂ = i + k ₂ ( ₂ I ⁵ "- ^S - I - 1) + k ₃ C _v + n _{v, mulnpath}

The variances of the errors from distance and speed measurement, rn and r _?? are, in accordance with the first term of the above equations, proportional to the square of the measured distance d _actual - The second term _takes into account the dependence of the measurement variance on the measured signal level, S _actual . It becomes zero when the level S _S0 n determined on the basis of the radar equation corresponds to the measured level, and it grows exponentially with the amount of Difference. The third, constant term takes into account the variance of the family of transponders and the variance of the assembly errors. The errors in distance and speed measurement caused by multipath propagation can be found in n _d and n _v . The measurement noise model implemented in the Kaiman filter does not take into account the errors caused by multipath propagation. The simulation in FIG. 12 shows the variance of the measurement noise according to the above model for a signal level noisy with σ = 2. The lower curve shows the level-dependent variance, the middle one the distance-related variance and the upper one the sum of the first two curves plus a constant.

The implemented model thus describes the variance of the measurement noise more cheaply than it actually does.

This is permissible, since the preselection according to the invention precludes measurements which are very faulty from the position measurement. The variance of the predicted speed and position estimated by the EKF serves as a decision criterion. If a distance or speed measurement deviates from the predicted distance by more than 3sigma, this measurement is excluded from the position calculation. The error ellipse formed by the variances of the position error is shown in FIG.

An EKF variant for two-dimensional position determination from LPR data is shown in FIG. Since the LPR measurements on the transponders take place at the same time, it makes sense to apply the EKF sequentially to the measurement data and to make a prediction based on the motion model only from measurement to measurement.

In the Cayman filter equations in FIG. 14, a roof stands over a variable for an estimate and a superscript minus for a prediction. The EKF corrects its 2D estimate based on i measurement data at time k, z ^. At the start A starting value of the state vector x ₀ and its covariance matrix P ₀ must therefore be specified for the measurement. The filter amplification matrix K _{kl is} calculated with the aid of the measurement noise R calculated for i measurements at time k and the measurement matrix H _kl linearized at the predicted position. This is followed by the correction of predicted state vector x ^ by the i-th measurement, z _k] . The non-linear function h () is used here to calculate the predicted distance to transponder i and the speed component in the direction of the transponder on the basis of the state vector xj ^. The covariance matrix, P _{ta of} the state vector is then calculated from the prediction of the covariance matrix P ^~ . As long as the i measurements are processed, x _k] and P _k ^{~ are} equal to the filtered values. This leads to a sequential correction of the

Positionsschatzung. The state vector is transformed with the linearly approximated equations of motion in F _k from measuring cycle k to measuring cycle k + 1. This transformation also takes place for the covariance matrix, with Q taking into account the errors that are possible due to the model in relation to reality.

Exemplary embodiment e

In general, the use of arrangements according to the invention is suitable for all contactless local location tasks in daily and industrial life. Using these arrangements, centimeter-precise position tracking of objects with a high update rate is possible in real time for the first time.

FIG. 16 shows the scenario of an "intelligent" factory using LPR. The three masts represent the mounting locations of the transponders, while base stations are schematically fixed on the objects to be measured. Mobile goods, conveyors, cranes, forklifts and belt conveyors are ideal for position measurement etc. as well as people. Direct coordinate measurement in an absolutely defined system is possible directly, the measured values can be transferred to a connected database or logistics software and processed there.

In addition to the pure coordinate measurement, event-based controls, regulations and reactions based on the results of the position measurement are also possible. It is conceivable e.g. the definition of certain local

Zones in the database. If a certain object enters or exits a certain zone, a defined action can be triggered. Ride e.g. a forklift into a forbidden zone (e.g. marshalling area of a crane) can trigger an alarm to prevent accidents, e.g. can consist of a warning message to the crane driver.

Various similar scenarios for LPR use in industrial plants are conceivable. In the following, a concrete LPR application example for industrial tasks is described using the "automatic hall crane".

Example: Determining the position of cranes and crane lifting devices

FIG. 17 shows a crane 1 with a base station and measuring computer (evaluation unit, means for determining the position), a left hall ceiling side 2 with active transponders and a right hall ceiling side 3 with active transponders.

FIG. 17 shows the architecture of LPR on an indoor crane. Transponders are set up on the walls / sides of the production hall / driving area at positions measured in the defined measuring coordinate system. The LPR crane is measured on the movable crane trolley to be measured. Base station fixed so that as many transponders as possible can be reached via the omnidirectional antenna at every possible location. Temporary obstruction of individual transponders by obstacles is due to the redundancy of the system to a certain degree

(at least three arbitrary stations must be in sight at all times).

With the coordinate values achieved, various processes and treatises can now take place. For example, material flow tracking and positioning in a warehouse using LPR is easily possible. If the crane operator has the task of stacking goods at a certain point, he is able to use LPR to go to the exact location and unload the goods there. Furthermore, the position at the unloading point can be saved separately in the database, so that it can be tracked how many and which goods are in a particular storage location in which order. The particular advantage of this - in addition to the possibility of precise positioning - is the fact that it can be centrally determined at any point in time where a particular product is and how large the inventory is.

In addition to material flow tracking, LPR can optimize transport processes. For example, if the crane driver has to run a complex stacking and re-stacking program, the database can use an optimization algorithm to calculate an optimally timed stack program based on the exact position knowledge of the stacked goods and transmit this information directly to the driver. In this way, the decisive period of time for the material flow in the warehouse or in production can be minimized.

In addition to the logistical tasks such as the described material flow tracking, with LPR there is automatic control or position control of the crane due to the determined coordinates possible without any problems. No human labor is required for this, the loading and unloading process can be carried out fully automatically using LPR. In addition to control processes, LPR can also be used to implement a highly precise and dynamic position control of an object. The LPR data act as manipulated variables, the control algorithm can be carried out, for example, on the central evaluation computer or directly on the DSP of the base station.

System concept and structure of the base station

The task of the base station is, on the one hand, to generate a highly linear frequency ramp over the 5.8 GHz ISM band and, on the other hand, to process the useful signal carrying the distance information in analog and digital form.

The block diagram for such a base station is shown in FIG. A synthesizer (here in the form of a DDS-PLL) creates a highly linear triangular

Frequency changing FMCW radar signal. A PLL-stabilized VCO is tuned linearly over time over the entire ISM band (5.725-5.85GHz). The synthesizer preferably consists of a DDS controlled by the computing unit and a subsequent low-pass filter. The voltage signal generated in this way flows into the phase locked loop (PLL). This consists of a PLL module, a low-pass filter (loop filter) to suppress harmonics in the digitally generated control signal, a VCO and a feedback to close the control loop. It is also conceivable to use various other PLL arrangements (eg analog PLL, fractional PLL). The linear frequency signal can then optionally be amplified in an RF amplifier. Preferably via a hybrid mixer with low isolation, other known mixer arrangements or also arrangements with two separate antennas for transmitting and receiving are also conceivable, this signal with the received transponder signal mixed, bandpass filtered and shifted to a lower frequency range with an XO signal. A "LO Power Level 7" mixer is preferably used for this. The signal is then digitized in an AD converter via an anti-aliasing low pass and in the

Connection processed and evaluated by a digital signal processor (DSP) or another computing unit. The calculated coordinates and a diagnostic status word e.g. available in the TCP / IP data protocol. However, any other protocols such as e.g. RS232.

System concept and structure of the transponders

The transponder implemented in the LPR system with switching

Oscillator swings to that arriving from the base station

Signal is phase coherent and is with the characteristic modulation frequency _mod with a certain

Divider ratio switched on and off.

The essential components of the transponder circuit are shown in the block diagram of FIG. The central component is the 5.8 GHz VCO, which is excited to oscillate in phase synchronization with the signal arriving through the antenna in each switching cycle. A PLL ensures that its oscillation frequency remains constant even with large temperature fluctuations. The feedback is preferably implemented via a directional coupler, which feeds the VCO output signal to the antenna on the one hand and to the PLL on the other. The power of this HF-VCO is switched via a further low-frequency VCO (eg 40..60MHz) with a modulation frequency characteristic of this one transponder (also PLL stabilized). Furthermore, an asymmetrical switching on and off of the amplifier for supplying the 5.8 GHz VCO is caused by a microcontroller (preferably a simple PIC), which results in an asymmetrical switching clock ratio. The switch-on time is to be set via this switching cycle ratio so that the main lobe of the emitted si (x) spectrum fills the ISM band as optimally as possible and thus optimally uses the available bandwidth. The central process control is carried out by a microcontroller (PIC), which, in addition to determining the switching cycle ratio, also has the task of changing the divider ratio of the PLLs for generating the VCO signals. The clocking of the PIC and the PLL modules is accomplished via a quartz crystal (eg 4MHz). Overall, for example, there is an address space of

2 ⁹ = 512 bits before, ie 512 different modulation frequencies can be set via jumper coding. Conversely, this means that 512 different transponders can be clearly differentiated. When using other components, however, an extended address space is also conceivable.

Claims

claims

1. Arrangement for determining the position of a mobile object, in particular a means of transport, with a base station for transmitting a base signal and / or receiving transponder signals,

a large number of transponders for receiving the base signal and for transmitting transponder signals,

- Means for determining the position of the mobile object taking into account the transponder signals, characterized in that

the base station is arranged on the mobile object,

- The position of the transponders in the room is known.

2. Arrangement according to claim 1, characterized in that the transponders are arranged fixed in space.

3. Arrangement according to one of the preceding claims, characterized in that the transponders are arranged and the means for determining the position are set up in such a way that the position of the mobile object can be determined in all three spatial dimensions.

4. Arrangement according to one of the preceding claims, characterized in that the arrangement is operated in multiplex, in particular in frequency division multiplex.

5. Arrangement according to one of the preceding claims, characterized in that some of the transponders have an identical identifier, in particular by emitting their transponder signal in the same frequency band.

6. Arrangement according to claim 5, characterized in that the transponders, which have an identical identifier, are each at least 20 ^ apart, where D _{max is} the distance at which the transponder signal with a selected statistical confidence, for example about 90% is no longer above a selected threshold.

7. Arrangement according to one of the preceding claims, characterized in that the means for determining the position of the mobile object are set up such that a position of the mobile object in the past can be taken into account when determining the position of the mobile object.

8. Arrangement according to claim 7, characterized in that the means for determining the position of the mobile object are set up such that a predicted position of the mobile object can be predicted from the position of the mobile object in the past using a model for the movement of the mobile object and the predicted position can be taken into account when determining the position of the mobile object.

9. Arrangement according to claim 8, characterized in that the model for the movement of the mobile object is set up so that approximately a constant acceleration of the object is assumed.

10. Arrangement according to one of the preceding claims, characterized in that the means for determining the position of the mobile

Object are set up in such a way that a transponder signal, from which a position results, which is above a predetermined Value from the positions that result from other transponder signals and / or deviates from a predicted position is not taken into account when determining the position of the mobile object.

11. Arrangement according to one of the preceding claims, characterized in that the means for determining the position of the mobile object are set up in such a way that a, in particular expanded, Cayman filter can be used when determining the position of the mobile object.

12. Arrangement according to one of the preceding claims, characterized in that the transponders each have an oscillator which can be excited quasi-phase-coherently to the base signal for actively generating the transponder signal by the base signal.

13. Arrangement according to one of the preceding claims, characterized in that the base station is connected via a wireless interface to the means for determining the position of the mobile object.

14. Mobile object with a base station for transmitting a base signal and / or receiving transponder signals to and / or from a plurality of transponders, the position of which is known, for determining the position of a mobile object, taking into account the transponder signals.

15. A large number of transponders which are arranged in a room in order to receive a base signal from a base station arranged on a mobile object and in each case to transmit transponder signals, taking into account the position of the mobile object.

16. A method for determining the position of a mobile object, in particular a means of transport, in which

a base station is used to transmit a base signal and / or to receive a transponder signal,

a large number of transponders are used to receive the base signal and to transmit transponder signals,

the base station is arranged on the mobile object,

- The position of the transponders in the room is known.