WO2008146098A1

WO2008146098A1 - Method for determining the position of an excitation on a surface and device for implementing such a method

Info

Publication number: WO2008146098A1
Application number: PCT/IB2007/052720
Authority: WO
Inventors: Damien Berget; Rémi Duheille; Thomas Fillon; Jean-Michel Linotte; Sébastien Pelletier; Olivier Schevin
Original assignee: Sensitive Object
Priority date: 2007-05-28
Filing date: 2007-05-28
Publication date: 2008-12-04

Abstract

A method for determining the position of the origin of an excitation on a surface of an object (3) fitted with transducers (4) comprising the steps of : (a) receiving signals Si(t) originating from said transducers; (b) calculating a signature SIGs of these signals; (c) comparing this signature SIGs with predetermined reference signatures SIGrRj, by a calculus of a resemblance function RESj; (d) determining a reference point j0 nearest to the origin of the excitation, based on the resemblance function and on a computation of a validation criterion.

Description

Method for determining the position of an excitation on a surface and device for implementing such a method.

FIELD OF THE INVENTION The present invention relates to methods for determining the position of an excitation on a surface and to the devices for implementing such methods.

BACKGROUND OF THE INVENTION

The present invention concerns more precisely a method for determining a position of an origin of an excitation on a surface of an object, said object comprising N_REF predetermined reference points, N_REF being a natural integer at least equal to 1, N_TRANS transducers being fixed to said object for measuring mechanical parameters of said object, N_TRANS being a natural integer at least equal to 2.

Such a known method for determining the location of an excitation on a surface of an object is described in document WO-A- 03/107261. OBJECTS AND SUMMARY OF THE INVENTION

One purpose of the present invention is to improve the methods for determining the position of an impact on an object, in particular to obtain an even more precise, faster signal processing and adaptable to a large number of objects.

To this effect, a method of the kind in question comprises the following steps:

(a) receiving at least two signals s,(t) originating respectively from said transducers ; (b) calculating a signature SIG₅ of said signals;

(c) comparing said signature S1G_S with predetermined reference signatures SIG_R) , by a calculus of comparison functions chosen in the group consisting of resemblance functions RES₁ and difference functions DIFFi / said predetermined reference signatures SIG_R characterising the signals received by the transducers when an excitation is generated at reference point j of the surface, j being an index comprised between 1 and N_REF; (d) determining at least one reference point jθ nearest to the origin of the excitation, based on said comparison functions and on a computation of at least one validation criterion.

Thanks to these dispositions, it is possible to determine a more precise position of an impact on a surface of an object.

In various embodiments of the method, one may further use one and / or other of the following dispositions : - the signals received at step (a) are representative of an elastic wave generated in said object by said excitation; said mechanical parameter is chosen in the group consisting of: acceleration, speed, displacement and stress; said tranducers are chosen in the group consisting of: piezoelectric transducers, capacitive piezoelectric transducers, magnetostrictive piezoelectric transducers, electromagnetic piezoelectric transducers, acoustic velocimeters, accelerometers, optical sensors, microelectromechanical system sensors (MEMs); the method comprises, before step (b) , a step

(a' ) of detecting an excitation and the following steps of the method are continued only if an excitation is detected; - at step (a' ) , an excitation is detected if at least one of said signals S₁(t) has an amplitude which is greater than a threshold proportional to noise level;

- at step (a'} , a first energy E_p of the signals

S₁Ct) is determined on a first duration N_F and a second energy E of the signals is determined on a second duration N which is longer than said first duration, and an excitation is detected if the first energy of at least one of said signals S₁(t) is greater than a threshold proportional to the second energy E of said signal; - at step (a' ) , an excitation is detected if a variation of phase of at least one of said signals s,(t) is lower than a predetermined threshold;

- at step (a') , a resemblance parameter is computed between at least two different windows of at least one of said signals S₁(t) and an excitation is detected if said resemblance is lower than a predetermined threshold;

- at step (a') , signatures of said at least one signal S₁(t) are computed for said at least two windows, and said resemblance parameter is computed between said signatures;

- at step (a' ) , a signal signature of said at least one of said signals S₁ (t) is computed and a resemblance parameter is computed between said signal signature and a predetermined noise signature which is representative of noise, an excitation being detected if said resemblance parameter is lower than a predetermined threshold;

- after detection of a first excitation at step (a' ) , a detection of a second excitation is inhibited during a first delay; - said first excitation is detected using a first criterion at step (a' ) and wherein a second excitation is detected, during a second delay after said first delay, if a second criterion is met by the signals, said second criterion being different from said first criterion; - at step (b) , the signature functions SIG_s(ω) are computed using one of the following formulas:

where S₁(Co) and S,(co) are the respective Fourier transforms of S₁(t) and S₂(t) ;

- at step (c) , a correlation function COR₁(Co) is computed using one of the following formulas:

COR₎(ω)=|siG_R)(ω)-SIG_s(ω)| ,

COR₎(ω)=(siG_Rj(ω)-SIG_s(ω))² , and a resemblance function RES is computed using one of the following formulas:

∑COR,(co) RES=I- α.M

RES = ε+∑COR,(ω) '

said reference function RES₁ being intended to estimate the degree of similitude between the signature functions SlG_s(ω) of the received acoustic signals s,(t) and reference signature functions SIG_Rj(ω) characterising an excitation at references point j of the surface;

- the method comprises, after step (d) , a further step (e) of interpolating the position of the origin of an excitation k if a reference point jθ has been determined at step (d) wherein an interpolation function F,_Mcφ0|_atl0n of the comparison function RES is matched with the values of said comparison function at said neighbouring reference points j , and wherein an interpolated position k of the origin of the excitation is determined as the position corresponding to a maximum of the interpolation function; the interpolation function F_{interpolatlon} is a polynomial function of position coordinates on the surface ;

- the interpolation function equals:

+ *■/ + C.X.y + d^,X + ^β.}^> + f , where x and y are the Cartesian coordinates on the surface, a, b, c, d, e and f are parameters of the interpolation function; the reference signature functions and the signature function are frequency domain functions SIG_R)(ω) and SIG_s(ω) respectively, where ω is the angular frequency, and the method comprises, at step (b) , a step of calculating an initial signature function of said signals and deforming said initial signature function by applying a predetermined deformation function to said initial signature function, to thus obtain said signature functions which are used at subsequent steps of the method;

- said deformation function is linear. said deformation function is non linear and depends upon the frequency. Another object of the invention is a device designed for implementing a method as defined above.

BRIEF DESCRIPTION OF THE DRAWINGS

Other features and advantages of the invention will be apparent from the following detailed description of several embodiments thereof, given by way of non-limitating examples, with reference to the accompanying drawings. In the drawings :

Figure 1 is a block diagram illustrating a device that implements the method according to one embodiment of the invention,

Figure 2 is a perspective diagrammatic view showing an example of the device of Figure 1, - Figure 3 is a diagram showing RESEMBLANCES RES₁ in a particular example,

Figure 4 is a diagram showing an example of a resemblance function in case of an excitation generated at a reference point k of a touch screen, - Figure 5 is a grid of a tactile surface of the device of figure 1, and

Figure 6 is a diagram showing one of the signals issued by the transducers of the device according to one embodiment of the invention, where the excitation is an impact on the surface of the MATERIAL, this diagram is showing an example of different threshold values used for detecting the excitation.

MORE DETAILED DESCRIPTION

In the different figures, the same references denote identical or similar elements.

The following definitions will be used hereafter in the detailed description. 1- A SENSOR refers herein to a sensor comprising:

- A MATERIAL that can be a solid object of any type, including without limitation: table, shelf, window pane, wall, door, window, monitor, TV monitor, display panel, interactive terminal, keyboard, keypad, toy, vehicle dashboard, seat back, floor, vehicle shock absorber, panels... - A set of TRANSDUCERS that transform the physical information generated by the EXCITATION into a electrical SIGNAL. TRANSDUCERS can be, including without limitation, piezoelectric sensors, capacitive sensors, magnetostrictive sensors, electromagnetic sensors, acoustic velocimeters , accelerometers , optical sensors, laser interferometers, laser vibrometers , microelectromechanical system sensors (MEMs) ; the number of TRANSDUCERS (N_TRANS) is a natural integer at least equal to 2 ; - POSITIONS that are points or areas on a surface of the MATERIAL that can be defined, but not limited to, by their coordinates x and y; - (optionally) a mesh grid (MESH) of POSITIONS that defines connectivity between POSITIONS. 2- An EXCITATION can be but is not limited to:

An impact generated by a tool or any part of a human or animal body on the MATERIAL; - The dragging of a tool or of any part of a human or animal body on the SENSOR; - A static pressure of a tool or of any part of a human or animal body on the SENSOR;

The action of an actuator on the surface of the MATERIAL .

3- The time SIGNAL is the information provided by TRANSDUCERS. s_t(t) denotes the SIGNAL transmitted by the i^th

TRANSDUCER; i is a natural integer at least equal to 1 ; i varies between 1 and N_TRAHS

4- The SIGNATURES (SIG) characterize the SIGNALS received by the TRANSDUCERS when an excitation is generated at a given POSITION on the MATERIAL.

5- The COMPARISON function is the comparison between two given SIGNATURES: the COMPARISON function may be a DIFFERENCE function or a RESEMBLANCE function.

6- The REFERENCES are POSITIONS where SIGNATURES have been measured and/or computed and then stored.

7- A DATABASE stores all the REFERENCES and their associated SIGNATURES ( SIG_R) ) ; The number of REFERENCES learned/stored (N_REF) is a natural integer at least equal to 1; j varies from 1 to N_REF Generally speaking, the present invention enables to determine the POSITION of the origin of an EXCITATION on a SENSOR, in particular to obtain a precise and fast signal processing and scalable to a large number of MATERIALS. To this effect, the method according to the invention generally comprises the following steps:

(a) Receiving at least two SIGNALS s,(t) and S₂(t) originating respectively from TRANSDUCERS and generated by an EXCITATION on the SENSOR; (a') Rejecting unwanted signals (UNWANTED

SIGNALS) including but not limited to SIGNALS that are not generated by an EXCITATION of the SENSOR;

(b) Calculating the SIGNATURE SIG₅ of the received SIGNALS; (c) Comparing this SIGNATURE SIG₈ with the

SIGNATURES of the REFERENCES SIG_R/ that belong to the DATABASE, by a computation of their RESEMBLANCE RES,, including but not limited to resemblance functions described in documents WO-A-03/107261, WO-A-06/015888 and WO-A- 06/069596 and in this document;

(dl) Determining a candidate position

(CANDIDATE), i.e. a position on the surface of the

MATERIAL, from which the EXCITATION probably originates, this determination being made by selecting the REFERENCE (jO) whose RESEMBLANCE gives the highest value;

(d2) Validating the CANDIDATE on the basis of some given validation parameters and acceptance criteria;

(e) Optionnally, improving the accuracy of the POSITION of the origin of the EXCITATION by an interpolation process if needed;

(f ) Optionnally, using this recognition information to improve the validation parameters used for future recognitions . A schematic drawing of a device 1 using the above method is shown on Figure 1. This device may comprise for instance: at least one SENSOR 2, including a MATERIAL 3 and TRANSDUCERS 4, at least an external surface of said MATERIAL being exposed to an EXCITATION; a sensor signal conditioning unit (SSCU) 5 that may perform for instance the following tasks: o filter the SIGNALS to remove unwanted characteristics and to reduce the influence of the electronic and electromagnetic noises and increase the signal-over-noise ratio; o amplify the SIGNALS to achieve a good numerical quantification; o digitalize the SIGNALS in order to make them usable by a Central Processing Unit (CPU) ; a Central Processing Unit (CPU) 6 that receives the digitized SIGNALS processes the METHOD described above and that delivers the action associated to the recognized POSITION. This CPU can consist in but is not limited to: o a Computer Central Processing Unit; o a Digital Signal Processing Unit; o a microcontroller; o a PLD (programmable logic device: CPLD, FPGA) . The central processing unit 6 may be in communication with a memory 7 including a DATABASE and with one or several interfaces 8 for communication with external systems .

The sensor signal conditioning unit 5, central processing unit 6, memory 7 and interfaces 8 may be integrated in a single electronic central unit 9 such as, for instance, a microcomputer. Alternately, they may be integrated into several packages, communicating together.

Figure 2 represents a particular example of a device 1 designed to implement the present invention, which includes a central processing unit 9 such as a microcomputer, or a digital signal processing unit. Said central processing unit includes said sensor signal processing unit 5 (which might otherwise be an external device) , said central processing unit 6, said memory (e.g. a hard disk or any other type of memory) and said interface 8. In the example shown on Figure 2, a monitor 3 may be connected to the central processing unit 2 ;

A SENSOR 5 has at least two TRANSDUCERS 4 which are connected to the central processing unit 9, so that said central processing unit 9 can identify the position of an EXCITATION on an external surface of the MATERIAL 3 of the

SENSOR.

The MATERIAL 3 is a table in the example shown on Figure 2, but could be any of the other objects mentioned above in the definition of the MATERIAL. The TRANSDUCERS 4 are fixed to the MATERIAL 5. They are linked to the sensor signal conditioning unit 5 using any analogical transmission channel that can be for instance:

- a cable 10 that connects directly the outputs of the TRANSDUCERS to the inputs 11 of the SSCU; - a wireless connection based on radio technologies, or else .

The sensor signal conditioning unit 5 is also linked to the CPU using any digital transmission channel that can consist in a wire or wireless connection. The TRANSDUCERS may be of any type defined above and are designed to measure for instance:

- the displacement or/and velocity or/and acceleration of vibrations due to the EXCITATION,

- variations of static strain due to a pressure maintained on the surface of the MATERIAL,

- variations of stresses inside the MATERIAL.

The memory 7 has a DATABASE including the SIGNATURES for

N_REF POSITIONS 13 on the surface 12 of the MATERIAL 3 (each

POSITION is identified by an index j from 1 to N_REF) . This DATABASE allows the system to localize the POSITION of the origin of an EXCITATION on the SENSOR 2.

To constitute such DATABASE, the method according to the present invention may include a preliminary training process or pre- learning process during which EXCITATIONS are generated at each given POSITION j of the surface 12.

This pre-learning process is usually done by a calibration system, comprising:

- an electronic controller,

- a multi axis robot, for example a two-axis robot, the displacements of each axis of the robot being controlled by the electronic controller,

- a support member for fixing the SENSOR under the robot,

- a vibration exciter, for example an amplified shaker with a force transducer, fixed on an axis of the robot, - an impact tool fixed on a moving part of the vibration exciter and adapted to come into contact with the SENSOR when a predetermined impact signal is given by the electronic controller to the vibration exciter.

During the pre-learning process, the TRANSDUCERS of the SENSOR are connected to the electronic controller of the calibration system.

The electronic controller is driving the robot and consequently the displacements of the vibration exciter so that the vibration exciter can produce impacts on the MATERIAL of the SENSOR at all the POSITIONS j from 1 to

During the pre-learning process, at each reference position, the electronic controller is:

- generating a predetermined impact signal, - storing signals coming from the TRANSDUCERS,

- computing a SIGNATURE S1G_R) , and optionally repeating these three steps to compute a mean value for the SIGNATURE SIG_R| .

All the SIGNATURES SIG_R) are stored in the DATABASE with the REFERENCES.

Acquiring SIGNALS

The first step (a) of the method consists in acquiring the SIGNALS and converting them into a digital format. This acquisition stage impacts strongly the efficiency of the method: if the SIGNALS are not correctly converted because of phenomena such as, for instance, electronic saturations or quantification noises, the robustness of the method may be decreased. The characteristics of the SIGNALS emitted by the TRANSDUCERS are modified from one SENSOR to another for at least the following reasons:

- the VIBRATIONS inside the MATERIAL propagate with different velocities and damping factors, - the processes used to fix the TRANSDUCERS to the MATERIAL modify the characteristics of the SENSOR itself,

- the dispersion of the TRANSDUCERS responses.

The SSCU 5 is in charge of this task. It includes a set of components including: o A set of amplifiers; o An Analog-to-Digital Converter; o An Output serial interface; o A Control Serial interface,- o A Signal Conditioning Unit. The SSCU combined with the method of the present invention is able to automatically adapt its parameters to the SENSOR, using the rules defined in its Signal Conditioning Unit.

Another object of the invention is to increase the number of types of EXCITATIONS recognized by the DEVICE; the differences between hitting the SENSOR using a finger pulp and hitting the SENSOR using another type of EXCITATION can be characterized for instance with the following indicators: - the envelop of the SIGNALS received by the TRANSDUCERS , the energy distribution in the spectrum.

The SSCU receives the SIGNALS from the SENSOR and is in charge of converting into a format easily analyzable by the CPU 6. Analog-to-Digital Converters such as computer sound card devices can be used to realize this task.

However to improve the efficiency of the method according to the present invention, the implementation of devices described in document PCT/IB2006/002988 may be used. Using such devices, the SSCU is able to adapt its parameters using information coming from the other components of the device .

Triggering

The second step (a' ) of the method according to the present invention consists in rejecting UNWANTED SIGNALS, for instance in order to: reduce the CPU load, for instance for implementing devices using low-consumption CPUs or for allowing to share CPU resources with other applications; - compute information relative to external environment influences.

These UNWANTED SIGNALS will be considered as SIGNALS generated by other means than an EXCITATION on the SENSOR. The UNWANTED SIGNALS may be generated for instance by. - airborne sounds, vibrations transmitted by the support of the SENSOR, electromagnetic fields, electronic noises. When using commonly used methods based on fix threshold values, this detection of an EXCITATION may be difficult if SIGNALS are noisy or if several EXCITATIONS occur in a short time. This is why several variants have been implemented to detect if the SIGNALS are likely to be generated by an EXCITATION or should be considered as UNWANTED SIGNALS. Four of these variants are described hereafter. Any combinations of these four variants together or with other variants, may be implemented in order to improve the performances of the present invention. 1- Triggering on time-domain characteristics

In a first variant of the method according to the invention, we consider that the studied SIGNALS are generated by an EXCITATION when one of the temporal characteristics of these SIGNALS is over the estimated characteristics of the UNWANTED SIGNALS. The studied characteristics can be for instance: the amplitude of the SIGNALS, the partial energy of the SIGNALS, the instant power. When the first variant is based on the amplitude of the SIGNALS, the studied SIGNALS will be considered as being generated by an EXCITATION when at least one of the amplitudes of the SIGNALS is greater than a given threshold

Thi depending on an estimated noise level A_M)1S1, such as: Th₁ —k_jm(;.A_xυJSF , Where k_ιma is a predetermined factor.

The UNWANTED-SIGNAL level A_XOISF can be estimated, for instance, using moving average methods over the absolute amplitudes of the SIGNALS Si(t) SIGNALS or can be based on the maximum value of the absolute amplitudes of the last SIGNALS considered as an UNWANTED SIGNALS.

When the first variant is based on energy, partial energy E₁, is computed for shorter windows having a time length of N_p signal samples, which are may possibly overlap one another. An EXCITATION is detected if partial energy E₁, becomes greater than an energy threshold calculated with the value of the total energy E in a larger window. For

example, an EXCITATION may be detected if E₁, ≥ k_ιm .E ,

N wherein : k_TRiG is a predetermined factor, E is the energy of the SIGNAL,

N is the length of the SIGNAL, - N_p < N .

2- Trigerring on frequency domain characteristics

In a second variant, the detection of an EXCITATION uses the evolution of the phases of the SIGNALS versus time. With noise or random signal (noise) such as for instance white noise, pink noise, the phases of the SIGNALS are not stable and always changing; on the contrary, when the SIGNALS are generated by an EXCITATION on the SENSOR, a phase pattern tends to be stable versus time.

To simplify the notation, the prefix Φ will be used below to indicate the phase of a complex variable. For example, ΦS^ω) is the phase of S₁(G)). CO is the index of pulsation and is proportional to the angular frequency.

The respective Discrete Fourier transforms (DFT) S₁(G)) of the sampled windowed SIGNALS S₁(t) are computed (t represents the index of samples and ω represents the index of the pulsation coefficients) . These Discrete Fourier transforms may be Fast Fourier Transforms (FFT) , permitting thus to obtain quick results, without a high calculation power. The window applied to the SIGNALS can be (but is not limited to this types of windows) a rectangular window or a Hanning window or a Blackman window.

Consequently, in this variant, the first option is based on the comparison of the phases of DFT coeeficients of at least one SIGNAL ΦS^G)) for at least two successive windows wl and w2 of the SIGNAL. A variation of phase ΔΦ, between first and second window will be computed by a following formula:

where :

ΦS_iwl(ω) is the phase in the first window of the DFT of the SIGNAL, - ΦS_lvv2(ω) is the phase in the second window of the DFT of the SIGNAL, ω, and G)₂ are respectively the low and high limits of the indexes of pulsation coefficients for whom the variation is evaluated. The second option is based on the comparison of the phases of DFT coefficients of at least one SIGNAL for at least one window. A variation of phase ΔΦ, will be computed by one of the following formula:

where :

ΦS_t(ω) is the phase of the DFT of the SIGNAL,

CO₁ and ω₂ are respectively the low and high limits of the indexes of pulsation coefficients for whom the variation is evaluated. δ is an integer value different from 0.

Then, an EXCITATION is detected if the variation of phase is lower than a predetermined threshold Th: ΔΦ, <Th .

3- Triggering on the RESEMBLANCE between two successive SIGNALS

In a third variant, the detection of the origin of an EXCITATION uses the RESEMBLANCE of the SIGNATURES of at least two successive windows of SIGNALS. When two successive windows of noise are compared, the RESEMBLANCE has a mean value equal to the probability that the same noise occurs on the two windows. Therefore the RESEMBLANCE decreases when an EXCITATION occurs . On the contrary to variants using thresholds on the CHARACTERISTICS of the SIGNALS for the detection of the origin of an EXCITATION, this variant is able to detect an EXCITATION even if the signal-to-noise ratio is very low. The RESEMBLANCE can be computed by using various types of formula, given in prior art.

Consequently, an EXCITATION is detected if the resemblance between the two windows wl , w2 is lower than a predetermined threshold : RES_vUl2≤Th_Rn

Wherein:

RES_wlv2 is the RESEMBLANCE of the SIGNATURES of two successive windows wl and w2 of SIGNALS.

Th_RFS is a given threshold value for the RESEMBLANCE .

4- Triggering on the RESSEMBLANCE between the studied

SIGNALS and an estimated SIGNATURE of the UNWANTED SIGNALS.

In a fourth variant, the detection of the origin of an EXCITATION uses a RESEMBLANCE of the studied SIGNALS with a specific SIGNATURE SIG_NOISI that is representative of the UNWANTED SIGNALS. The SIGNATURE SIG_NO1S, can be estimated by using only the SIGNATURE of the last UNWANTED SIGNALS acquired or by combining the SIGNATURES of several UNWANTED SIGNALS. The combination can consist in computing the mean value of the SIGNATURES of a given number of last UNWANTED SIGNALS but more complex operations can be realized such as for instance by means of phases or methods used in the pre- learning process. Then, the RESEMBLANCE calculated between the SIGNATURES of the current SIGNALS and SIG_NOISt is low if the current SIGNALS are generated due to an EXCITATION. Consequently, an EXCITATION is detected if:

RE^_{S NO}I_Sh - Tl¹R_hS where :

RES_SNOlst is the RESEMBLANCE between the SIGNATURE of the studied SIGNALS and the SIGNATURE that is representative of the UNWANTED SIGNALS,

Th_RES is a given threshold value of RESEMBLANCE. Analyzing the UNWANTED SIGNALS

The DEVICE will analyze from time to time some UNWANTED SIGNALS and will extract their characteristics.

These characteristics can be (but are not limited to this list) : - the maximum of the amplitude of these UNWANTED SIGNALS; the energy contained in the studied UNWANTED SIGNALS; the modules of the DFT coefficients of the UNWANTED SIGNALS.

Once these characteristics have been computed, the system computes the average of these characteristics using several samples of UNWANTED SIGNALS.

These average values will be used to adapt factors such as (but not limited to this list) : threshold values such as Th, Thi, Th_RES as defined above ; bandwidth of the analogical or digital filters to apply on the SIGNALS. Determining the CANDIDATE

The third step (b) of the METHOD consists in computing the SIGNATURE of the SIGNALS and finding the SIGNATURE of the REFERENCE present in the DATABASE that best fits with the SIGNATURE of the SIGNAL. This REFERENCE will be called the CANDIDATE.

This research is based on the computation of resemblance criteria, corresponding for instance to those described in documents WO-A-03/107261 , WO-A-06/015888 and WO-A-06/069596.

According to a new variant, when two SIGNALS are transmitted by the SSCU to the CPU, the SIGNATURE of the SIGNALS can be computed using the information present in the DFT of the SIGNALS according to one of the following formula :

SIG₅(ω)= ^S ₁ ^(O)H^S ₂W :io:

S,(ω|+|S,(ω

Where :

are respectively the absolute values

(modulus) of the DFT of the SIGNALS 1 and 2.

By extension, the variant can be applied to more than 2 SIGNALS.

The values of the SIGNATURE computed with formula (10) and (11) are bounded between -1 and +1. The formula (12) and (13) are not bounded but require less processor instructions than formula (10) and (11) .

Then, in a subsequent step (c) of the method according to the invention, the DIFFERENCES between the SIGNATURE of the SIGNALS SIG_s(ω) and the SIGNATURE of a REFERENCE is measured with the calculus of one of the following functions DlFFλώ): DIFF_l(ω)=\siG_Rl(ω)-SIG_s(ω)\ (14) or,

Here, the bigger the differences between the SIGNATURE of the SIGNALS and the SIGNATURE of REFERENCE are, the higher the results of these formulas are.

In order to compute RESEMBLANCE instead of DIFFERENCE criteria, it is possible to use one of the following formulas (but the method of the invention is not limited to these formulas) :

RES₁ = 1 - ω ( 16 ) a.N_ω

1

RES₁ ( 17 ) ε + Σ DIFF₁ (ω)

where : α is a constant value higher than 1 ;

N_ω is the number of discrete frequencies of each SIGNATURE; ε is a positive decimal constant value, to avoid that RES₁ becomes too high.

When formula (10) and (11) are used, each value returned by formula (16) is a number between 0 and 1. Consequently, it is possible to consider the result returned by the formula (16) as a normalized RESEMBLANCE.

This variant, used together with the methods presented in documents WO-A-03/107261 , WO-A-06/015888 and WO-A-

06/069596, increases the performance to determine the POSITION of the origin of an EXCITATION on an SENSOR, at step (dl) of the method of the invention.

Reducing the influences of the external disturbances

According to a second variant, the information previously computed on UNWANTED SIGNALS will be used to reject the noisy frequencies of the studied SIGNALS. A noisy frequency is a discrete frequency having a DFT coefficient the module of which is less than a given threshold. This threshold is a given factor of the estimated value of the DFT coefficients of the UNWANTED SIGNALS at the same frequency. RES₃ will be computed by selecting only not-noisy frequencies. This variant of the method can be applied whatever the RESEMBLANCE formula is.

This variant improves the efficiency of the device of the invention when the device is placed in a noisy environment; it also improves the efficiency of the method of the invention for determining that EXCITATIONS with low energy occur.

Acceptance criteria

Once all the RESEMBLANCES RES₁ are computed, step (d2) of the method of the invention consists in computing some acceptance criteria to carry out the validation step.

Acceptance criteria can be computed based on formulas including (but not limited to) :

MAX = max(RES)

- MEAN = .> RES

^1N RF 1- I=I

CONTRAST₀ = MAX-MEAN

- CONTRAST, = ^MAX

MEAN MAX²

CONTRAST, =

MEAN Figure 3 gives a graphical representation of RESEMBLANCES RES₁ showing a maximum of RESEMBLANCES for

REFERENCE j 0.

Thus, it is possible to validate that POSITION (x_k,y_k) of an EXCITATION k is corresponding to REFERENCE jθ, if it passes the following criteria: MAX≥Th_ms andIor CONTRAST ≥Th_(OXIR,,, where :

CONTRAST is one of the above contrast parameters CONTRASTo , CONTRAST₁ , CONTRAST₂ ,

MAX = RES₁₀ - Th_{ιi[ K} and Th_{c (ΛJK4SJ} are some given parameters of the system .

Lots of criteria can be built to define the validation of the recognition of REFERENCE jθ.

Affining the estimated POSITION of the origin of the EXCITATION

As soon as the REFERENCE jθ is recognized as the nearest REFERENCE (CANDIDATE) from the POSITION of the origin of the EXCITATION, it is possible to increase the precision of the estimated POSITION of the origin of the EXCITATION on the SENSOR using some interpolation techniques, at step (e) of the method according to the invention.

This is very useful for applications including (but not limited to such applications) touch screens or touch tablets as SENSORS, for which the POSITION of the origin of an EXCITATION on the SENSOR needs to be estimated with a high accuracy.

Figure 4 gives a perspective representation of a resemblance surface of the RESEMBLANCES RES₁ concerning a touch screen. Each REFERENCE of the DATABASE is represented at the Cartesian coordinates ( x , y ) where x , y are the coordinates of the j^th REFERENCE. The value on the z-axis represents RES₁. Thus, this resemblance surface has some minima and maxima, with a maximum located at the point k near the POSITION jθ.

To improve the localisation of the origin of an EXCITATION on the surface of the SENSOR, an interpolation function F_{11 oK)tK)n} is used. Any coordinate system can be used for this function. Parameters of the interpolation function fi_nterpolati_on ^can k>^e obtained using a process of minimization of the difference between the actual values RES₁ and those of the function F_{1111erp0latl011} computed for the POSITION of the REFERENCE jO and its close neighbours of known coordinates

(X₁, Y_j)/ by minimizing the following formula: d θparameters Σ(F_1I1(e,po.a_tlo_n (x ₎'yrP^arameterS)- ^RES _J )^{2 ( 2} ° )

For example, the interpolation function can be a polynomial. One of the simplest choices is to use a parabolic equation:

^_nI,_rpolau_on (*,^«■*>,C.a) = d.X^' + 6./ + C.X.y + d.X + β.}^>+ f (21)

In that case, the parameters a, b, c, d, e and f can be calculated with at least five REFERENCES: jO and four other REFERENCES closed by jO. A first technique is to use jO and the nearest horizontal and vertical REFERENCES, i.e. jl, j3, j5, j7 on figure 5. In that case, the equation (20) has a unique solution. But, a common technique is to use jO and the eight REFERENCES around jO. These are the REFERENCES jO to j8 on figure 5. In that case, the solution is an approximate solution and the solution of the nine equations 20 can be found easily with well known linear algebra formula including, but not limited to it, least- square regression.

Then, the POSITION of the origin of the EXCITATION can be estimated by calculating the position of the maxima of the equation 21, obtained by solving:

=_Q

In the case of the interpolation function of equation 21, the previous system of equation writes:

2.a.x+c.y+d^~0 2.b.v+c.x+e = 0

giving an exact position (X_k-Yi₄) for the maxima of the interpolation function. Consequently, it is possible to estimate the POSITION of the origin of the EXCITATION on the SURFACE with a better accuracy than the pitch of the MESH of the DATABASE. This method allows also the device to work with a smaller set of REFERENCES and by the way it reduces the processing resources required to determine the POSITION of the origin of the EXCITATION without decreasing the accuracy of the POSITION.

Correcting drifts of the SENSORS

Environmental variations that can slightly change the size of the SENSOR, like temperature (coefficient of thermal expansion) , ageing or hygrometry may influence elastic waves propagation in the SENSOR, inducing therefore some difficulties for the method to determine where an

EXCITATION occurs. In fact, the computed RESEMBLANCES become lower and the RESEMBLANCES RES₁ have less contrast, implying no recognition of the POSITION of the origin of the EXCITATION or errors in the recognition.

In the same way, multiple copies of a given SENSOR (including but not limited to the glass of a touch screen) can have slight size variations (thickness, length, width) as well as MATERIAL properties variations (density, elasticity,...) which may also influence significantly elastic waves propagation in the SENSOR. In that case, these variations can be critical especially if a unique DATABASE of the SENSOR is used for multiple copies of this SENSOR. It has been found by the inventors, that contraction/expansion methods can be applied to the SIGNATURES to minimize the impacts of these variations.

In a first implementation of the METHOD, a contraction/expansion coefficient a is determined to adapt the SIGNATURES of the REFERENCES stored in the DATABASE and/or the SIGNATURE of the SIGNALS to get better

RESEMBLANCES. Optimal α coefficient is the one which maximize the RESEMBLANCE between the SIGNATURES of the REFERENCES and the SIGNATURE of the SIGNALS.

To this end, a DATABASE of contracted/expanded

SIGNATURES of the REFERENCES SIG'_K)(cϋ) is created and used in the method of the invention instead of the initial DATABASE of SIGNATURES of the REFERENCES SIG_Rj(ω) with the following formula:

Siσ_Rl(ω)= SIG_Rl(f_a(ω)) f_σ can be but is not limited to a multiplicative function of the frequency axis. f_a can be defined for instance by the following formula: f_a(ω) = ceil((] +a)*ω)

Where ceil (x) is a function that rounds x to the nearest integer.

In that case, α equals zero if there is no drift. By extension, another implementation can consist in applying to the SIGNATURES a function depending on the contraction/expansion coefficient and on the index of the pulsation coefficients. The previous formula becomes: SlG'_Rι(ω)=g_a(ω,SIϋ_Rι) Where g_(/(ύ),SIG_Rl) can be but is not limited to a function combining the values of the SIGNATURE SIG₁^ for successive indexes ύ) of the pulsation coefficients. The aim of this function is to approximate the contraction/expansion of the SIGNAL on the time domain (respectively expansion/contraction in the continuous frequency domain) by an appropriate interpolation function of the SIGNATURE in the discrete frequency domain.

In this extension, the contracted/expanded SIGNATURES of the REFERENCES SlG'_R/ {ώ) may be computed for instance by the following algorithm:

ω'=(l + α)*ω ω, =floor(ω') to, = co, +1

IFco <ω, +Δ THEN

ELSEIF ω >ω;-Δ THEN

SIG'_Rj(co)= SIG_R)(ω₂) ELSE

SIG_Rj(ωl)*(ω -ωi-Δ)+SIG_R)(ω^")*((ω₂-Δ)-(rf)

ISICJ _R, IG)I = ^■

^{RjV ;} 1-2*Δ

END

Where :

- Floor (x) is a function that rounds x to the nearest integer less than or equal to x;

- Δ is a value between 0 and 0.5 (excluded) . In this algorithm, if the drift provides frequencies G)' near the known frequencies CD₁' or G),' (first and second case in the algorithm) , the contracted/expanded SIGNATURE SIG'_Rj (ω) directly uses the values in the SIGNATURE SIG_Rj(ω), else (third case in the algorithm) the contracted/expanded SIGNATURE SIG'_R| (ω) is computed with an interpolation function.

Thanks to this algorithm, a precise and fast calculation of contracted/expanded SIGNATURE SIG'_R| (ω) is done .

In the same way, the contraction/expansion coefficient α can be applied on the SIGNATURE of the SIGNALS using the same process so that SIG\ [O)) is used instead of SIG_s(ω) . Moreover, after the step of having determined that the REFERENCE JO is a CANDIDATE for the POSITION of the origin of the EXCITATION, the method of the invention comprises a step of determining correction parameters. The contraction/expansion coefficient α is one of these parameters.

Treating Rebounds

After having determined the POSITION of the origin of an EXCITATION, the process is inhibited during a predetermined delay DTh₁, for example corresponding to 10 ms, depending on the MATERIAL of the SENSOR. This inhibition avoids the detection of an unwanted rebound due to the fact that the EXCITATION continues to propagate on the SENSOR even after having being determined by the method. After the delay DThI, the detection process is reactivated. But, as presented on figure 6, the threshold value is changed from its initial state (Thi) to a higher or lower level (Th₂), depending for instance on:

- the variant of the method of the invention used to detect the EXCITATION; the POSITION determined for the EXCITATION. This threshold value will come back to its initial state after a delay DTh₂ using a pre-determined function, for instance : - step-by-step function, exponential function.

Optionally, the current value of the threshold can be transferred to the SSCU so that the SSCU will adapt its parameters to the current state of the DEVICE. The preferred embodiments of this inhibition of the process and of this adaptation of the threshold value are: to improve the detection of EXCITATIONS generated by the movement of a tool or a part of a human or animal body on an SENSOR; to suppress the false detections due to the release of a tool or any part of a human or animal body from the SENSOR.

Claims

1. A method for determining a position of an origin of an excitation on a surface (12) of an object (3), said object comprising N_REF predetermined reference points (13) , N_RUF being a natural integer at least equal to 1, N_TRANS transducers (4) being fixed to said object for measuring at least one mechanical parameter of said object, N_TRANS being a natural integer at least equal to 2, said method comprising the steps of:

(a) receiving at least two signals s,(/) originating respectively from said transducers (4) ;

(b) calculating a signature S1G_S of said signals,-

(c) comparing said signature SIG_S with predetermined reference signatures SIG_R) , by a calculus of comparison functions chosen in the group consisting of resemblance functions RES₁ and difference functions DIFF₃, said predetermined reference signatures SIG_R) characterising the signals received by the transducers (4) when an excitation is generated at reference point j of the surface, j being an index comprised between 1 and N_REF;

(d) determining at least one reference point jθ nearest to the origin of the excitation, based on said comparison functions and on a computation of at least one validation criterion.

2. The method as claimed in claim 1, wherein the signals received at step (a) are representative of an elastic wave generated in said object (3) by said excitation.

3. The method as claimed in claim 2, wherein said mechanical parameter is chosen in the group consisting of: acceleration, speed, displacement and stress.

4. The method as claimed in claim 3, wherein said tranducers are chosen in the group consisting of: piezoelectric transducers, capacitive piezoelectric transducers, magnetostrictive piezoelectric transducers, electromagnetic piezoelectric transducers, acoustic velocimeters , accelerometers , optical sensors, microelectromechanical system sensors .

5. The method as claimed in anyone of the preceding claims, comprising, before step (b) , a step (a') of detecting an excitation and wherein the following steps of the method are continued only if an excitation is detected.

6. The method as claimed in claim 5, wherein at step (a' ) , an excitation is detected if at least one of said signals s,(t) has an amplitude which is greater than a threshold proportional to noise level.

7. The method as claimed in claim 5, wherein at step (a') , a first energy E_p of the signals S₁ (t) is determined on a first duration N_p and a second energy E of the signals is determined on a second duration N which is longer than said first duration, and an excitation is detected if the first energy of at least one of said signals S₁(t) is greater than a threshold proportional to the second energy E of said signal.

8. The method as claimed in claim 5, wherein at step (a' ) , an excitation is detected if a variation of phase of at least one of said signals S₁(t) is lower than a predetermined threshold.

9. The method as claimed in claim 5, wherein at step (a' ) , a resemblance parameter is computed between at least two different windows of at least one of said signals s, (t) and an excitation is detected if said resemblance is lower than a predetermined threshold.

10. The method as claimed in claim 9, wherein at step (a' } , signatures of said at least one signal S₁(t) are computed for said at least two windows, and said resemblance parameter is computed between said signatures.

11. The method as claimed in claim 5, wherein at step (a' ) , a signal signature of said at least one of said signals S₁(t) is computed and a resemblance parameter is computed between said signal signature and a predetermined noise signature which is representative of noise, an excitation being detected if said resemblance parameter is lower than a predetermined threshold.

12. The method as claimed in any one of claims 5 to 11, wherein after detection of a first excitation at step (a' ) , a detection of a second excitation is inhibited during a first delay.

13. The method according to claim 12, wherein said first excitation is detected using a first criterion at step (a¹) and wherein a second excitation is detected, during a second delay after said first delay, if a second criterion is met by the signals, said second criterion being different from said first criterion.

14. The method as claimed in any one of the preceding claims, wherein at step (b) , the signature functions SIG_s(co) are computed using one of the following formulas :

SIG₈(CO) = M⁰

S₂ (ω)

1S₁ (OO

SIG₅ (ω) =

S₂ (ω|² where S₁(ω) and S₂(ω) are the respective Fourier transforms of S₁(t) and S₂(t) .

15. The method as claimed in claim 14, wherein at step ( c) :

- a correlation function COR₁(Cu) is computed using one of the following formulas:

COR,(ω)=|siG_Rl(ω)-SIG_s(ω)| ,

COR₎(ω)= (siG_Rj(ω)-SIG_s(ω))² , and

- a resemblance function RES₁ is computed using one of the following formulas:

RES₁=I- α.R

said reference function RES_j being intended to estimate the degree of similitude between the signature functions SIG₈M of the received acoustic signals s,(t) and reference signature functions SIG_R)(co) characterising an excitation at references point j of the surface.

16. The method as claimed in any one of the preceding claims, comprising, after step (d) , a further step (e) of interpolating the position of the origin of an excitation k if a reference point jθ has been determined at step (d) wherein an interpolation function F_mRrpolallon of the comparison function RES is matched with the values of said comparison function at said neighbouring reference points j , and wherein an interpolated position k of the origin of the excitation is determined as the position corresponding to a maximum of the interpolation function.

17. The method as claimed in claim 16, wherein the interpolation function E_{1n old(Ioπ} is a polynomial function of position coordinates on the surface.

18. The method as claimed in claim 17, wherein the interpolation function equals:

^F _meΨok1u<A^x _^y^-b-cd) = a.x² +h.y² +c.x.y+d.x+e.y+f , where x and y are the Cartesian coordinates on the surface, a, b, c, d, e and f are parameters of the interpolation function.

19. The method as claimed in any one of the preceding claims, wherein the reference signature functions and the signature function are frequency domain functions SlG_R)(ω) and SIG_s(ω) respectively, where ω is the angular frequency, and wherein the method comprises, at step (b) , a step of calculating an initial signature function of said signals and deforming said initial signature function by applying a predetermined deformation function to said initial signature function, to thus obtain said signature functions which are used at subsequent steps of the method.

20. The method according to claim 19, wherein said deformation function is linear.

21. The method according to claim 20, wherein said deformation function is non linear and depends upon the frequency.

22. The method according to claim 20, wherein said deformation function is given by the following formula :

ω'=(l +α)*ω

G)₁ =floor(ω') ω₂ = ω, +1

IFco <ω_] +A THEN

SIG'_R»=SIG_Rl(ωi) ELSElF ω >ω,-Δ THEN

SIG'_R»= SIG_Rl(ω₂) ELSE

_CIP1 , _Λ SIG_R)(ω;)*(ω^'-ω;-Δ)+SIG_R,(ω^' ₂)*((ω;-Δ)-ω^I)

SlO _R (ω)= n^Λ ' 1-2*Δ

END

Where :

Floor (x) is a function that rounds x to the nearest integer less than or equal to x, and Δ is a value between 0 and 0.5.

23. A device adapted to implement a process according to any one of the preceding claims.