WO2025133375A1 - Contact force-dependent haptic feedback vibrotactile interface - Google Patents

Abstract

The invention relates to a method for controlling a haptic interface, wherein the haptic interface comprises - a pad (10), configured to be touched by one or more contact members (3) at one or more contact points; - actuators (12_p), arranged against the pad, wherein each actuator is configured to induce vibration of the pad when subjected to a control signal (V _p ); - detectors (14_i), coupled to the pad, wherein each detector is configured to produce a detection signal (Q _i ) dependent on the intensity of a force exerted on the pad at each contact point; and wherein the method comprises estimating the intensity of the force exerted by each contact member at each contact point and generating haptic feedback dependent on each intensity.

Description

Description

Title: Vibrotactile interface with force-dependent haptic feedback

TECHNICAL FIELD

The technical field of the invention concerns tactile interfaces of the vibrotactile type.

PREVIOUS ART

Vibrotactile interfaces are vibrating interfaces, allowing a user to perceive tactile sensations by simply applying a finger to the interface. A vibration on a rigid interface can in fact intuitively encode information to the user, through a sensation felt at the finger. This is an important issue when visual information is non-existent or degraded. This is the case, for example, of tactile interfaces intended for visually impaired users, or tactile interfaces located outside the user's field of observation, for example a vehicle driver. A vibrotactile interface allows the transmission of precise spatial and temporal information.

A vibrotactile interface exploits the sensitivity of skin mechanoreceptors to vibratory stimuli, the latter producing skin deformation. Mechanoreceptors are generally sensitive to vibrations in frequencies typically between 10 Hz and 1 kHz, with high sensitivity observed between 200 Hz and 300 Hz. The amplitude of the vibration is also an important parameter, the minimum amplitude to trigger tactile stimulation depending on the vibration frequency. It is considered that the lower the frequency, the higher the amplitude must be for the vibration to be perceived. The amplitude can, for example, be a few pm or a few tens of pm. Another important parameter, conditioning tactile stimulation, is the shape of the vibration wave, as well as the duration, the latter being able to be between 0.1 s and a few seconds. The delay between two consecutive vibrations can also influence the stimulation perceived by the user.

Thus, tactile interfaces provide haptic feedback through vibration patterns triggered by the tactile interface being touched by a finger. The vibration patterns are predetermined so that the tactile sensations created are easily detectable. These can, for example, mimic sensations induced by mechanical devices, such as a cursor, wheel, or button. Developments have also allowed the development of “tactons”, a contraction of the Anglo-Saxon terms “tactile” and “icons”, which are localized and structured sensory information.

Some touch interfaces allow measurement of the pressure force exerted by a finger. Such interfaces have for example been described in documents US5241308 or US55210813, the position and pressure force of a finger being estimated by capacitive effect. Document US201000053116 describes a touch screen combining capacitive detection of the position of a finger with specific force sensors, so as to discriminate an intentional contact of a finger, exerted with a certain force, from an accidental contact.

US2016313841 describes a haptic interface, of the touch screen type, for simulating pressure exerted on a “dome switch” type switch, without however detailing how the haptic effect is generated, which simulates the deformation and release of the switch.

US2021255751 describes a haptic interface comprising an array of pressure sensors (202A- 202P) and piezoelectric actuators (404A- 404P) intended to produce the haptic effect. Such a configuration assumes on the one hand pressure sensors, on the other hand piezoelectric actuators.

The inventor designed a more advanced touch interface, allowing the measurement of force exerted by several fingers simultaneously, and allowing haptic feedback to be controlled based on the force exerted.

STATEMENT OF THE INVENTION

A first object of the invention is a method for controlling a haptic interface, the haptic interface comprising:

- a slab, configured to be touched by one or more contact members, at one or more contact points,

- actuators, arranged against the slab, each actuator being configured to induce a vibration of the slab when it is subjected to a control signal;

- detectors, coupled to the slab, each detector being configured to produce a detection signal dependent on an intensity of a force exerted on the slab, at each point of contact; the method comprising: a) determining a position of each point of contact on the slab, at a measurement time; b) from the detection signal generated by different detectors, at the measurement instant; estimation of a force exerted on the slab, at each point of contact; c) from the position of each point of contact and the force exerted at each point of contact, definition of a haptic feedback of the slab, the haptic feedback comprising a target displacement assigned to each point of contact of the slab; d) from the haptic feedback of the slab, defined during c), determination of a control signal to be applied to the actuators, so as to obtain the target displacements assigned to each point of contact; e) addressing a control signal to each actuator, as a function of each control signal determined during step d); f) reiteration of steps a) to e) by incrementing the measurement instant, steps a) to e) being reiterated until a criterion for stopping the iterations is reached; steps a) to f) being implemented by a processing unit connected to a memory, the memory comprising response functions of a set formed by the slab, each actuator and each detector.

According to one possibility, step b) comprises an inversion of a direct detection model, the direct detection model determining, for each detector, a detection signal as a function of a force exerted at each point of the slab. The direct detection model can take into account the control signal addressed to each actuator.

Step b) may include:

- bi) extraction, from the memory, of a detection matrix, each term of which corresponds to a detection signal generated, in a detector, by a unitary support exerted in a position on the slab, the unitary support being a predetermined support force, the detection matrix being stored in the memory;

- bii) formation of a detection vector, comprising the detection signal resulting from the detectors;

- biii) from the detection matrix and the detection vector, estimation of a force vector, each term of which corresponds to an estimation of the force exerted at each point of contact.

Step b) may include taking into account a time delay applied to the detection matrix.

According to one possibility, the detection matrix is stored, in the memory, in the frequency domain, in different frequency bands. According to one possibility:

- sub-step bi) comprises an extraction, from the memory, of a charge transfer matrix, each term of which corresponds to a charge generated, in a detector, by a unit voltage applied to an actuator, the unit voltage being predetermined.

- sub-step bii) comprises a formation of a control vector, comprising the actuation signal of each actuator of a previous iteration;

- sub-step biii) comprises an estimation of the force vector from the load transfer matrix and the control vector.

Step c) may comprise an inversion of a direct control model, the direct control model determining, for each actuator, the control signal as a function of each target displacement. The direct control model may take into account the force exerted on the plate at each contact point.

According to one possibility, step c) comprises:

- ci) extraction, from the memory, of a displacement matrix, each term of which corresponds to a displacement of the slab induced, at a point, by a unit voltage applied to an actuator, the unit voltage being a predetermined voltage, the displacement matrix being stored in the memory;

- cii) formation of a displacement vector, comprising the target displacements;

- ciii) from the displacement matrix and the displacement vector, estimation of a control vector, each term of which corresponds to a control signal to be applied to each actuator.

Step c) may include taking into account a time delay applied to the displacement matrix.

The displacement matrix can be stored, in memory, in the frequency domain, in different frequency bands;

According to a possibility

- sub-step ci) comprises an extraction, from the memory, of a charge transfer matrix, each term of which corresponds to a charge generated, in a detector, by a voltage applied to an actuation transducer;

- sub-step cii) comprises a formation of a force vector, comprising the force estimated at each point of contact during step b); - sub-step ciii) comprises an estimation of the control vector from the load transfer matrix and the force vector.

Following step e) of an iteration and step b) of a subsequent iteration, the method may comprise an estimation of an impedance of the contact member.

A second object of the invention is a haptic interface comprising:

- a slab, configured to be touched by one or more contact members, at one or more contact points,

- actuators, arranged against the slab, each actuator being configured to cause a vibration of the slab when it is subjected to a control signal;

- detectors, coupled to the slab, each detector being configured to produce a detection signal dependent on an intensity of a force exerted on the slab, at each point of contact;

- a processing unit, configured to implement steps a) to f) of a method according to the first subject of the invention.

The invention will be better understood by reading the description of the exemplary embodiments presented in the remainder of the description, in conjunction with the figures listed below.

FIGURES

Figure 1 shows a general view of a haptic interface allowing an implementation of the invention.

Figure 2A shows a configuration of a haptic interface, in which each actuator and each detector forms the same piezoelectric transducer.

Figure 2B shows another configuration of a haptic interface, in which each actuator and each detector forms the same piezoelectric transducer.

Figure 2C shows a configuration of a haptic interface, in which the actuators and detectors are piezoelectric transducers different from each other and are two by two concentric.

Figure 2D shows another configuration of a haptic interface, in which the actuators and detectors are different piezoelectric transducers.

Figure 2E shows a configuration of a piezoelectric transducer, which can operate as an actuator or as a detector.

Figure 2F shows another configuration of a piezoelectric transducer, which can function as an actuator and as a detector. Figure 3A diagrams the main components of an example of a haptic interface according to the invention.

Figure 3B represents the main steps implemented by the haptic interface described in connection with Figure 3A.

Figure 3C is a detail of a force estimation step, described in Figure 3B

Figure 3D is a detail of a step of determining a control signal, described in Figure 3B.

Figure 4A illustrates an extraction of a reduced detection matrix, by extracting certain columns from a tensor stored in a memory, in the frequency domain.

Figure 4B illustrates an extraction of a displacement matrix, by extracting certain lines from a tensor stored in memory, in the frequency domain.

Figure 4C illustrates an extraction of a reduced displacement transfer matrix, by extracting certain columns from a tensor stored in memory.

PRESENTATION OF SPECIAL METHODS OF IMPLEMENTATION

Figure 1 represents a haptic interface 1, intended to be touched by an external organ 3, for example a finger, to control a device 2. The device may be, in a non-limiting manner, a consumer appliance, the dashboard of a vehicle, a device for identifying a person, or a device for assisting a visually impaired or disabled person.

In the examples shown in this application, the external member 3 is a finger, which corresponds to most of the applications envisaged. Alternatively, the external member 3 may be a stylus, or any other means making it possible to act on the interface 1.

The haptic interface comprises a slab 10, rigid, transparent or opaque. The slab 10 may for example comprise glass or a metal, or an organic compound, for example plexiglass. The slab 10 is delimited by a contact surface 11, intended to be touched by an external organ, in particular a finger. The thickness e of the slab 10 is preferably less than 10 mm, or even less than 5 mm. The thickness e is adjusted according to the dimensions of the slab, and the mechanical properties of the material forming the slab (rigidity, solidity). It is for example between 1 and 5 mm for glass or a material such as plexiglass.

The panel can form a touch screen, including light sources, for example OLED (organic light-emitting diode) type. The slab 10 is coupled to actuators 12 _p , for example piezoelectric actuators, p being an integer designating each actuator. Each actuator is configured to move the slab, by generating a vibration of variable amplitude and frequency, preferably in a frequency range between 10 Hz and 1 KHz, preferably including a frequency band between 200 Hz and 300 Hz, which is a frequency band at which the tactile sensation is particularly felt. Each actuator 12 _p is powered by a control signal V _p , so as to generate a vibration whose characteristics (amplitude, frequency) are configured to produce haptic feedback from the interface. The control signals V _p are transmitted, via a control circuit 13 to each actuator 12 _p .

In this example, each _12p piezoelectric actuator comprises a piezoelectric material, for example AIN, ZnO or PZT, arranged between two electrodes. Each actuator can be connected to the slab by gluing. Alternatively, the _12p actuators can be electromagnetic, electromechanical or piezoresistive actuators.

The slab 10 is coupled to detectors 14 _p for example piezoelectric transducers, i being an integer designating each detector. Each detector 14j is configured to form a detection signal Q _t as a function of an intensity of a pressing force exerted by the finger. The detection signal Q _t is collected by a detection circuit 15. The detection signal Q _t is formed by charges induced by the pressure of the finger on the detector 14j. Each detector 14j can be formed of a piezoelectric material as described in connection with the actuators 12 _p .

The touch surface 11 comprises a position sensor 17, for example of the capacitive type. The position sensor 17 is connected to conductive tracks 16, arranged in a two-dimensional network. The conductive tracks 16 are adjacent to the contact surface 11. The conductive tracks extend parallel to the slab 10, or in the slab 10, below the contact surface 11. When the slab 10 is opaque, the conductive tracks 16 may be made of a standard conductive material, for example a metal. When the slab 10 is transparent, the conductive tracks 16 are preferably made of a transparent conductive material, for example a conductive oxide, a standard material being ITO (Indium Tin Oxide). The conductive tracks 16 extend, for example, in rows and columns. The conductive tracks may be polarized according to a polarization voltage. The interface comprises an insulating layer extending between the conductive tracks 16 and the contact surface 11, so as to allow detection, by capacitive effect, of contact with the finger 3 of a user, or any other type of external organ. The position sensor 16 is configured to generate a position signal S _m comprising the number and position of contact points on the detection surface 11. The conductive tracks 16 make it possible to form a mesh on the slab, defining N mesh points, each point of the mesh corresponding to a position likely to be detected by the position sensor. The number N of positions is determined as a function of the spatial resolution of the position sensor. It is possible to take into account a number of positions lower than that determined by the spatial resolution of the position sensor. This involves taking into account the support surface of a finger, extending along a diameter of a few mm, for example 5 mm. Thus, the spatial resolution of the position sensor and the conductive tracks makes it possible to define a maximum number of positions N _max . The haptic interface takes into account a number N of different positions, likely to be occupied by the finger on the contact surface 11, with N < N _max .

The device comprises a processing unit 20, supplied, at different measurement times, by the detection signals Q _t generated by the detection circuit 15 and by the position signal S _m generated by the position sensor 17. The processing unit 20 is intended to address a control signal V _p to the control circuit 13, so that the latter transmits each control signal V _p respectively to each actuator 12 _p .

The processing unit 20 is configured to determine a haptic pattern as a function of the position of each finger 3 contacting the contact surface 11, but also as a function of the pressing force exerted by each finger 3. The processing unit 20 comprises a microprocessor, connected to a memory 25 in which instructions are stored allowing implementation, by the microprocessor, of certain steps described in connection with FIG. 3A.

Preferably, the actuators and detectors are distributed under the slab, advantageously forming a regular mesh. The actuators and detectors can be arranged in different configurations, shown diagrammatically in Figures 2A to 2D. These figures represent a top view of the actuators and detectors, the slab being shown in transparency.

Figure 2A shows a configuration in which each transducer acts as an actuator and a detector. Each upper electrode is connected, via a resistor R, to the output of a voltage amplifier A. The resistor R and the voltage amplifier are part of the actuation circuit and the measuring circuit. Under the effect of a voltage V _p , forming the control signal, applied to the terminals of the amplifier A, a voltage is established between the upper electrode and the lower electrode, which induces vibration of the actuator. Applying a pressing force to the upper electrode compresses the transducer, which generates a detection signal by integrating a charge Q _i=p resulting from the compression. However, in such a configuration, the charge Q _i=p resulting from the compression can be small compared to the electric charge generated by the polarization of the amplifier by the control signal V _p .

A first alternative, shown in Figure 2B, consists of using a switch C, the switching of which allows a connection of the upper electrode either to the voltage amplifier previously described, in which case the transducer operates as an actuator, or to a charge amplifier A', to form the detection signal. The switch allows a physical separation between the control signal and the detection signal.

Figures 2C to 2D show alternatives, in which the actuators 12 _p are separated from the detectors 14 j. In Figure 2C, a configuration is shown comprising pairs formed by an actuator 12 _p and a detector 14 i = _p arranged concentrically. The upper electrode of the actuator is annular, and extends over a surface greater than the upper electrode of the detector, the latter being disc-shaped. A voltage amplifier A is arranged to allow the excitation of the upper electrode of the actuator by the control signal V _p . A charge amplifier A' makes it possible to form a detection signal from charges formed at the upper electrode of the detector.

In Figure 2D, a configuration is shown comprising actuators 12i separated from the detectors _14p .

Figures 2E and 2F show possible transducer configurations. Figure 2E shows a transducer comprising a PZT piezoelectric material, on either side of which extend two electrodes E1 and E0. Electrode E0 extends between plate 10 and the PZT material. Electrode E0 can be connected to ground while a potential is measured or applied to electrode E1. Alternatively, electrode E0 and electrodes E1 are connected to two opposite potentials.

In Figure 2E, two electrodes E1 and E2, electrically insulated from each other, are opposite the electrode E0. The electrode E1 is annular and the electrode E2 is arranged in the center of the electrode E1. An electrode, for example E1, can be used to apply a potential to the PZT, while the electrode E2 allows a measurement to be carried out. This corresponds to the configuration shown in Figure 2C. Regardless of the embodiment, it is preferable for the upper electrode of a 12 _p actuator to extend over a surface area greater than that of a 14 j detector. The diameter of the upper electrode of the 12 _p actuator may be of the order of 10 mm to 30 mm, so as to allow sufficient deformation of the slab to be felt by the finger. Since the production of electrical charges by compression is easier to measure, the upper electrode of the detectors may be smaller, for example a diameter of the order of or less than 5 to 10 mm.

Preferably, the _12p actuators extend, between the upper electrode and the lower electrode, with a thickness of between 100 pm and 500 pm. The 14j detectors are thicker, the thickness being for example greater than 500 pm or greater than 1 mm, so as to limit their intrinsic capacitance.

The actuation 13 and detection 15 circuits are connected to the processing unit 20, the latter being configured to generate control signals, addressed to the actuation circuit 13, so that the actuators produce vibrations forming a haptic pattern providing predetermined haptic feedback. Each haptic pattern depends on the position of the contact on the contact surface and the force exerted on the latter.

The invention takes advantage of an independent measurement of the pressure exerted at the level of each finger contacting the slab 10, and of a possibility of modulating the vibration pattern so that the haptic feedback depends on the pressure exerted.

Figure 3A shows the main components of the haptic interface. Panel 10 includes:

P actuators 12 _p , P being an integer greater than or equal to 2, controlled by the control circuit 13, the latter transmitting a signal to each actuator as a function of a control signal V _p addressed by the processing unit 20; the detectors 14j, connected to the detection circuit 15, the latter producing a detection signal i dependent on a pressing force F _m exerted by each finger applied against the slab. the conductive tracks 16, connected to the position sensor 17, the latter producing a position signal S _m dependent on the position of each contact m on the slab.

The processing unit 20 is connected to the detection circuit 15 and to the position sensor 17, from which it receives the detection signal Q _t and the position signal S _m respectively. The processing unit 20 comprises three processing blocks: a processing block 21 makes it possible to estimate a force F _m applied at each contact point m from the position signal S _m and each detection signal Q _ir , possibly taking into account a control signal V _p ; a haptic block 22, configured to define a target displacement U _m at the level of each finger contacting the slab 10. The haptic block comprises a haptic memory, in which the vibration patterns are stored for different configurations of each finger touching the slab. Each configuration is defined according to parameters such as: number of fingers in contact with the slab, position of each finger contacting the slab and intensity of the pressing force exerted by each of the fingers contacting the slab. a control block 23, configured to generate a control signal V _p addressed to the control circuit 13 of the actuators 12 _p , so as to obtain, in each finger contacting the slab 10, a displacement U _m conferring the haptic feedback defined by the haptic block 22. The displacement U _m of the slab, at each point of contact, must be as close as possible to the target displacement U _m . The control signal V _p is established from the target displacements U _m defined by the haptic block 22, possibly taking into account the force exerted by each finger contacting the slab.

Each block may be implemented by one or more microprocessors. Each block may be formed of functions programmed in a specific computing environment, for example Matlab or Python. The processing block 21 and the control block 23 may implement inverse filtering, as described below.

Figure 3B illustrates the main steps of an iterative process implemented by the haptic interface 1:

Step 100: At a measurement time, one or more fingers m touch the slab. M is an integer designating the number of fingers simultaneously touching the plate.

Step 110: The position of each finger touching the slab is determined by the position sensor 17. This results in the formation of a position signal S _m for each position.

Step 120: The pressure exerted by each finger is converted into a detection signal Q _t by the detectors 14j. The detection signal corresponds, for example, to a quantity of charges generated by each detector. Step 130: this step is implemented by the processing block 21 of the processing unit 20. This involves estimating a support force F _m exerted at the measurement time by each finger m. This step is described in more detail below, in connection with figure 3C.

Step 140: this step is implemented by the haptic block 22 of the processing unit 20. This involves determining a target displacement U _m for each finger, so as to form haptic feedback from the interface and provide sensory information to the user which depends on the number of fingers, their positions on the panel and the pressure force they exert.

Step 150: this step is implemented by the control block 23 of the processing unit 20. This involves defining a control signal V _p for each actuator 12 _p , so as to obtain the displacement U _m of the slab in each finger, as close as possible to the target displacement U _m . This step is described in more detail below, in connection with figure 3D.

Step 160: Activating each actuator 12 _p according to the control signals defined in step 150, to produce haptic feedback resulting from the movement of the plate at each contact point. The haptic feedback may be associated with other types of stimulation, for example the emission of an audible or visual signal. When the slab forms a touch screen, the image formed on the screen may vary according to the haptic feedback.

Following step 160, steps 100 to 160 are repeated, with the measurement time being incremented.

Steps 130, 140 and 150 are carried out by the processing unit 20, preferably in the time domain. They assume taking into account response functions of the panel, some of which are stored in the memory 25 in the frequency domain.

A first frequency response function is a detection matrix H _c , each term H _c (i,ri) of which corresponds to a complex detection signal generated, in a detector 14i, by a unit harmonic support exerted at a position n on the slab. The unit harmonic support, of unit amplitude, is a support force exerted at a predetermined frequency whose intensity is predetermined, for example 1 N. The detection matrix H _c is stored in the memory 25 of the processing unit 20. The detection matrix H _c has a dimension (I,N), where I corresponds to the number of detectors and N corresponds to the number of positions considered. All or part of the detection matrix H _c , and more particularly its inverse H _c ~ is used in step 130, H _C denotes the inverse or a pseudo inverse, for example a Moore-Penrose pseudo inverse. A second frequency response function is a displacement matrix H _a , each term H _a (n,p) of which corresponds to a displacement of the slab induced, at a point, by a unit voltage exerted on an actuator 12 _p . The unit voltage is a predetermined voltage, for example IV. The displacement matrix H _a has a dimension (N, P), where P corresponds to the number of actuators and N corresponds to the number of positions considered. The displacement matrix H _a is stored in the memory 25 of the processing unit 20. All or part of the displacement matrix H _a , and more particularly its inverse H _a ~ is used in step 150. When the displacement matrix H _a is not invertible, H _a ^-1 denotes a pseudo inverse, for example a Moore-Penrose pseudo inverse.

A third, optional, response function is a charge transfer matrix G, each term G(p, j) of which corresponds to a charge generated, in a detector 14j, by a unit voltage applied to an actuator 12 _p . The charge transfer matrix is stored in a memory 25 of the processing unit 20. The charge transfer matrix G has a dimension (P, I). The charge transfer matrix G is stored in the memory 25 of the processing unit 20. The charge transfer matrix G can be used in step 130.

A fourth, optional, response function is a displacement transfer matrix K, each term K(n, ri) of which corresponds to a displacement induced, at a position n of the slab 10, by a unit force exerted at a contact point located at another position. The displacement transfer matrix is stored in the memory 25 of the processing unit 20. The displacement transfer matrix K has a dimension (N, N). The displacement transfer matrix K is stored in the memory 25 of the processing unit 20. All or part of the displacement transfer matrix K can be used in step 150.

In this example, the matrices H _a , H _c previously described are stored in the memory 25 in frequency form for W frequency bands w. The matrices K and G are stored in the memory 25 in time form.

The operation of the haptic interface 1 is governed by expressions, describing relationships between the displacements generated at the level of each finger in contact with the panel, the force exerted on the fingers, as well as the measured detection signals and the control signals addressed to the actuators.

où : The displacement of the slab U _n at a point n is such that:

Or :

- H _anp is a term of the displacement matrix H _a addressing the point of the slab of rank n and the actuator of rank p

- V _p is the control signal addressed to the actuator 12 _p .

- K _nm is a term of the matrix K addressing the point of the slab of rank n and the finger in contact with the slab of rank m.

F _m corresponds to the force applied by the finger at position m on the slab.

Expressed in matrix form, (1) becomes:

U = H _a V + KF (2)

V is a control vector, each term of which corresponds to the control signal V _p addressed to each actuator 12 _p . V is of dimension (P,l).

The detection signal Q, detected by a detector 14, is such that:

G _ip is a term of the matrix G addressing the actuator 12 _p and the detector 14 j.

H _Cin is a term of the detection matrix H _c addressing the detector 14i and each point of the rank n mesh.

Expressed in matrix form, (3) becomes

Q = GV + H _C F (4)

Expressions (1) to (4) define a direct model, translating the effect, on the slab, of constraints exerted on the latter, the constraints being able to include both an electrical constraint (the control signals V addressed to each actuator), and a mechanical constraint (the force F exerted on the slab by each finger in contact).

The matrices H _a , H _c , K, G determine the responses of the slab, both the mechanical response, in this case a displacement of the slab at different points, and the electrical response, in this case the detection signal of each detector. Expressions (1) and (2) correspond to a direct control model, because it allows the displacement of the slab to be determined in response to a control signal. Expressions (3) and (4) correspond to a direct detection model, because it translates the detection of loads under the effect of a force exerted on the slab. The direct models defined in connection with expression (1) to (4) use matrices H _a , H _c , K defined for the points of the mesh corresponding to each detected contact point. Knowing the position of the contact points, the matrices H _a , H _c , K are formed from tensors H _a , H _c , K, defined for each point of the mesh and stored in memory 25.

The memory 25 stores all the tensors H _a , H _c , K for each point of the mesh. During steps 130 and 150, terms of these tensors which correspond to each position m detected by the position sensor 17 can be extracted from the memory 25. The extractions from the tensors H _a , H _c , K are described below, in connection with FIGS. 4A to 4C.

Figure 3C describes the sub-steps of step 130. During step 130, the direct detection model is inverted, so as to estimate the force F exerted by each finger in contact with the slab, the number of which M and the position on the slab is provided by the position sensor 17. F is a vector, of dimension (M, 1) of which each term F _m is an estimate of the force F _m applied by the finger at the position m.

The inversion of the direct detection model can be established by determining H _c ^-1 . For this, during a substep 131, columns of the tensor H _c corresponding to each position m contacted by a finger are extracted from the memory 25. Figure 4A represents such an extraction. In the example shown in Figure 4A, the tensor H _c of dimensions (I, N, W) is represented. W detection matrices H _c of dimensions (I, M) are extracted from the tensor H _c , each column of which is associated with a position m detected by the position sensor 17. In other words, a detection matrix H _c formed from the column or columns whose index n corresponds to a detected position m is used in each frequency band.

In a sub-step 132, a pseudo inverse H _c ~ ¹ of each matrix H _c is calculated, in each frequency band w.

During a step 133, the delay term e ^-JÛJT is applied, translating, in the frequency domain, a time delay between the application of the force and its measurement, T corresponds to a time delay, of the order of a few milliseconds. Working in the frequency domain facilitates the formalization of the time delay T.

During a sub-step 134, a charge vector Q is formed, of dimension (/,1) of which each term is the detection signal Q _t of a detector 14j. During a sub-step 135, the mechanical effect, on the slab, of the control signal V applied to the actuators is taken into account. During the first iteration, an initial control signal, resulting from an initialization, is taken into account. The initial control signal may for example be predetermined, or zero in all frequency bands. During the following iterations, the control signal V resulting from the previous iteration is taken into account. Sub-step 135 is optional. Preferably, it is not implemented during the first iteration.

During a sub-step 136, the matrix G is taken into account.

In a sub-step step 137, the direct control model is inverted. According to one possibility, the inversion of the direct control model is carried out by estimating the inverse H” ¹ of the matrix H _c resulting from step 134, for example a pseudo inverse as previously described. Expression (5) is a way of inverting the direct control model, according to a first approach

Alternatively, a simplified direct control model is implemented, neglecting the influence of the actuators on each detection signal. The simplified electrical model is thus:

Q = H _C F (6) and its inversion is:

F = e ^-Jûrr H _c ^-1 Q (7) Expression (5) is considered preferable, as it is less subject to error.

Relative to the actual force exerted at each point, forming a vector F, F = e~ ^ja>T F. In other words, the estimation error is a time error

The implementation of expression (7) leads to a larger error, because:

The matrix product GV corresponds to an estimation error of F.

Each exerted force can be expressed in the time domain by an inverse Fourier transform, for example an inverse fast Fourier transform (IFFT), f = IFFT(F) (9) The transition to the time domain is optional. It corresponds to sub-step 138.

Figure 3D describes the sub-steps of step 150. During step 150, the direct detection model is inverted, so as to estimate the control signal V _p to be addressed to each actuator to obtain the target displacement U _m determined by the haptic block 22.

The inversion of the direct detection model can be established by determining H _a ^-1 . For this, during a substep 151, rows of the tensor H _a corresponding to each position m contacted by a finger are extracted from the memory 25. Figure 4B represents such an extraction. In the example shown in Figure 4B, the tensor H _a of dimension (N, P, W) is represented, defined for different spectral bands w. W matrices H _{a of} dimensions (M, P) are extracted from the tensor H a , each row of which is associated with a position m detected by the position sensor 17. In other words, a matrix H _a formed from the row or rows whose index n corresponds to a detected position m is used in each frequency band.

In a sub-step 152, a pseudo inverse H _a ~ ¹ of each matrix H _a is calculated, in each frequency band w.

During a step 153, the delay term e ^-JÛJT is applied, translating, in the frequency domain, a time delay between the generation of the control signal and the formation of the vibrations on the slab. The delay T is as defined in step 133.

During a sub-step 154, a vector of target displacements U, of dimension (1, M), is formed, each term of which corresponds to a target displacement to be applied in position m.

During a sub-step 155, the electrical effect on the slab of the force applied in each contact position m is taken into account, in the form of the force vector F defined in step 137. Sub-step 156 is optional.

During a step 156, the terms corresponding to each position m are extracted from the tensor K, as illustrated in Figure 4C. This allows the formation of a matrix K.

Selon une variante, on met en oeuvre un modèle direct de commande simplifié, en négligeant l'influence des actionneurs sur chaque signal de détection. Le modèle électrique simplifié est ainsi : In a sub-step 157, the direct control model is inverted. According to one possibility, the inversion of the direct control model is carried out using the inverse H _a ^] of the matrix H _a , for example a pseudo inverse as previously described. Expression (10) is a way of inverting the direct control model, according to a first approach.

Alternatively, a simplified direct control model is implemented, neglecting the influence of the actuators on each detection signal. The simplified electrical model is thus:

U = H _a V(ll) and its inversion is:

V = H^e-J^Ù (12)

Expression (10) is considered preferable because it is less subject to error. Compared to the determined target displacement, the displacement actually exerted is U = e~ ^JùyT U. In other words, the estimation error is simply a delay.

The implementation of expression (12) leads to a larger error, because

U = e~^ ^T U + KF (13). KF corresponds to an error term.

Finally, the signals are expressed in the time domain by an inverse Fourier transform, for example an inverse fast Fourier transform (IFFT). v = IFFT(V) (14)

The transition to the time domain corresponds to sub-step 138. This makes it possible to address control signals to each actuator, defined in the time domain.

The invention can be applied for identity control type applications. As soon as at least one finger is detected on the contact surface, the vibratory pattern (or haptic pattern) emits vibrations of certain amplitudes. The force exerted by the finger, in reaction to this vibration, can be considered as a mechanical impedance, which depends on the finger. The invention allows an estimation of the mechanical impedance simultaneously on several fingers. Such control can also allow discrimination between a finger and another object contacting the interface.

Claims

1. Method for controlling a haptic interface, the haptic interface comprising - a slab (10), configured to be touched by one or more contact members (3), at one or more contact points, - actuators (12 _p ), arranged against the slab, each actuator being configured to induce a vibration of the slab when it is subjected to a control signal (l^,); - detectors (14j), coupled to the slab, each detector being configured to produce a detection signal (Ç depending on an intensity of a force exerted on the slab, at each point of contact; the method comprising: a) determining a position of each point of contact on the slab, at a measurement instant; b) from the detection signal generated by different detectors, at the measurement instant; estimating a force (F _m , F ) exerted on the slab, at each point of contact; c) from the position of each point of contact and the force (F _m ) exerted at each point of contact, defining a haptic feedback of the slab, the haptic feedback comprising a target displacement (U _m , U ) assigned to each point of contact of the slab; d) from the haptic feedback of the slab, defined during c), determining a control signal (V _p , 7) to be applied to the actuators, so as to obtain the target displacements assigned to each contact point; e) addressing a control signal to each actuator, as a function of each control signal determined during step d); f) repeating steps a) to e) by incrementing the measurement time, steps a) to e) being repeated until a criterion for stopping the iterations is reached; steps a) to f) being implemented by a processing unit (20) connected to a memory (25), the memory comprising response functions of a set formed by the slab, each actuator and each detector; the method being characterized in that step c) comprises an inversion of a direct control model, the direct control model determining, for each actuator, the control signal (l^,) as a function of each target displacement. 2. Method according to claim 1, in which step b) comprises an inversion of a direct detection model, the direct detection model determining, for each detector, a detection signal (Q _t , Q) as a function of a force exerted at each point of the slab. 3. Method according to claim 2, in which the direct detection model takes into account the control signal addressed to each actuator. 4. Method according to any one of claims 2 or 3, in which step b) comprises: - bi) extraction, from the memory (25), of a detection matrix (H _c ), each term of which corresponds to a detection signal generated, in a detector, by a unitary support exerted in a position on the slab, the unitary support being a predetermined support force, the detection matrix being stored in the memory, the detection matrix forming the direct detection model; - bii) formation of a detection vector (Q), comprising the detection signal resulting from the detectors; - biii) from the detection matrix and the detection vector, estimation of a force vector (F), each term of which corresponds to an estimate of the force exerted at each point of contact. 5. Method according to claim 4, in which step b) comprises taking into account a time delay applied to the detection matrix. 6. Method according to claim 5, in which the detection matrix is stored, in the memory, in the frequency domain, in different frequency bands. 7. Method according to any one of claims 4 to 6, in which: - sub-step bi) comprises an extraction, from the memory, of a charge transfer matrix (G), each term of which corresponds to a charge generated, in a detector, by a unit voltage applied to an actuator, the unit voltage being predetermined. - sub-step step bii) comprises a formation of a control vector (V), comprising the actuation signal of each actuator (12 _p ) of a previous iteration; - sub-step step biii) includes an estimation of the force vector from the load transfer matrix and the control vector. 8. Method according to any one of the preceding claims, in which the direct control model takes into account the force exerted on the plate at each point of contact. 9. Method according to any one of the preceding claims, in which step c) comprises: - ci) extraction, from the memory, of a displacement matrix (H _a ), each term of which corresponds to a displacement of the slab induced, at a point, by a unit voltage applied to an actuator, the unit voltage being a predetermined voltage, the displacement matrix being stored in the memory, the displacement matrix forming the direct control model; - cii) formation of a displacement vector (U), comprising the target displacements - ciii) from the displacement matrix and the displacement vector, estimation of a control vector (V), each term of which corresponds to a control signal (V _p ) to be applied to each actuator. 10. Method according to claim 9, in which step c) comprises taking into account a time delay applied to the displacement matrix. 11. A method according to any one of claims 9 or 10, wherein - the displacement matrix is stored, in the memory, in the frequency domain, in different frequency bands. 12. Method according to any one of claims 9 to 11, in which: - sub-step ci) comprises an extraction, from the memory, of a charge transfer matrix, each term of which corresponds to a charge generated, in a detector, by a voltage applied to an actuation transducer; - sub-step cii) comprises a formation of a force vector, comprising the force estimated at each point of contact during step b); - sub-step ciii) comprises an estimation of the control vector from the load transfer matrix and the force vector. 13. Method according to any one of the preceding claims comprising, following step e) of an iteration and step b) of a following iteration, an estimation of an impedance mechanics of the contact member from an estimate of the force exerted by the finger in response to a vibration of the contact surface. 14. Haptic interface comprising: - a slab (10), configured to be touched by one or more contact members (3), at one or more contact points, - actuators (12 _p ), arranged against the slab, each actuator being configured to cause a vibration of the slab when it is subjected to a control signal (1^,); - detectors (14j), coupled to the slab, each detector being configured to produce a detection signal (Ç depending on an intensity of a force exerted on the slab, at each point of contact; - a processing unit (20), configured to implement steps a) to f) of a method according to any one of the preceding claims.