RING-SHAPED TOUCH SENSOR and ASSOCIATED HUMAN-MACHINE INTERFACE

Abstract

A sensor assembly (R) in the form of a ring, comprising a substantially cylindrical annular element (ANN) and at least two strain sensors (SENS1, SENS2, SENS3) mounted on the annular element at levels of at least two radii (r1, r2, r3) angularly distinct from it, respectively, so as to be sensitive to a deformation of the annular element. Figure to be published with the abbreviation: Fig. 1

Description

Title of the invention: RING-SHAPED TOUCH SENSOR and ASSOCIATED HUMAN-MACHINE INTERFACE. TECHNICAL FIELD OF THE INVENTION

[0001] The invention relates to a ring-shaped sensor, and extends to human-machine interfaces, systems allowing a user to control interactive systems, and is therefore in the field of human-machine interactions sometimes referred to by the acronym HMI. TECHNOLOGICAL BACKGROUND

[0002] Human-machine interfaces come in many forms, allowing a user to interact with an electronic system, such as a computer, robot, television, sound reproduction system, or game console. These interfaces are designed to i) generate control signals from a user action, and then ii) transmit these signals to the electronic system.

[0003] The wide variety of possible actions for a user to control the electronic system has been used to design various types of interfaces: these can be a keyboard, a mouse, a touch screen, an optical detection system, a microphone or a joystick, all equipped with sensors (contact, infrared, magnetometer, optical, camera, accelerometer, gyroscope...) whose function is to detect a specific action of the user (movement, gesture, speech, actuation...).

[0004] Among these various types of interfaces, development efforts have focused on ring-shaped devices, generally intended to be worn around the user's finger, as described in US patent 9,582,076 B2 and US patent application 2017 / 0242496 AL. Such devices have the advantages of minimal bulk, allowing them to be worn without hindering the user, but require complex sensors such as cameras for analyzing the ring's immediate environment. These sensors generate signals whose processing involves computationally intensive tasks, necessitating high-performance electronics, which are expensive and reduce the battery life of these interfaces. Furthermore, this complexity negatively impacts control reliability.

[0005] Still with regard to ring-shaped devices, there are simpler design systems, such as a ring with eight outer faces described in the article by Hyunchul Lim et al., “OctaRing: Examining pressure-sensitive multi-touch input on a finger ring device”, UIST 2016 Adjunct - Proceedings of the 29th Annual Symposium on User Interface Software and Technology, Association for Computing Machinery, Inc., pages 223-224. Each of the eight faces is equipped with a pressure sensor intended to be directly pressed by the user. Various control signals can be generated depending on the pressure sensors pressed simultaneously.

[0006] Along the same lines, US patent application 2020 / 0150715 A1 describes ring-shaped systems intended to be worn on a finger. Each of these systems is equipped with one or more sensors that can be used to detect an action of the ring wearer performed by the hand wearing the ring or to detect an action applied to it by the wearer's other hand. The user's actions on a sensor attached to the ring act on the sensor itself to activate it, the ring serving as a passive support. This document also discloses the detection of the bending angle of an elongated element attached to the ring and capable of unfolding..

[0007] Another device, a ring intended to be grasped with two hands by the user at two diametrically opposed areas, is described in US patent application 2022 / 0212112 A1. It comprises a flexible ring mounted on a base equipped with a deformation sensor, the assembly being configured to generate a signal representative of the amplitude of deformation of the flexible ring when the user deforms the ring by bringing their hands closer together.

[0008] However, a human-machine interface having a general ring shape, of simple design but allowing the generation of varied control signals from equally varied user actions remains to be defined. Description of the invention

[0009] The applicant's objective is to propose, on the one hand, a ring-shaped deformation sensor, and, on the other hand, an application of this deformation sensor to a human-machine interaction device in the form of a ring sensitive to the deformation of an annular element of the ring caused by a user.

[0010] For the purpose of achieving this goal, a first aspect of the invention is a sensor assembly in the form of a ring, comprising an annular element of substantially cylindrical shape and at least two strain sensors mounted on the annular element at levels of at least two distinct radii angularly from it, respectively, so as to be sensitive to a deformation of the annular element.

[0011] Such a sensor array allows the detection of a wide variety of actions on the annular element forming its mechanical structure. More specifically, mechanical actions such as pressure, taps, or friction applied to the annular element itself cause its deformation, and, thanks to the presence of at least two sensors, this deformation can be characterized spatially and tracked over time. The element The annular surface is not simply a passive support for the sensors, but participates, through its deformation, in the detection of a user action.

[0012] Thus, this sensor assembly allows the detection and characterization of a large number of actions, in a manner not limited to a binary action detection logic or to the measurement of a one-dimensional deformation amplitude. This sensor makes it possible to track the spatiotemporal deformation imposed on the annular element, as will be developed in more detail later in the description.

[0013] It is noteworthy that the actions do not have to be applied directly to the sensors, but can be applied at any point of the annular element, possibly at a plurality of points of this element, regardless of whether or not sensors are present at that point or those points. Furthermore, the point(s) of application of the action on the annular element do not have to be fixed, but can be movable on the surface of this annular element.

[0014] The very structure of the sensor assembly, with essentially sensors permanently fixed on an annular element, remains simple and does not include moving parts, guaranteeing the reliability and robustness of the assembly.

[0015] According to additional non-limiting features of the sensor according to the invention, considered individually or in any technically feasible combination:

[0016] - the annular element may have a monolithic structure;

[0017] - the annular element can be made of metal;

[0018] - the at least two distinct rays can be angularly spaced apart from each other at an angle between 20° and 160°;

[0019] - the at least two strain sensors can be of type piezoelectric;

[0020] - the at least two strain sensors may comprise at least one single-crystal piezoelectric thin element in the form of a plate extending in a plane of extension defined by a first direction and a second direction normal to the first direction, of dimensions in the first and second directions each greater than 100 pm and of thickness less than 50 pm, a ratio of the thickness to the dimension in the first direction or the dimension in the second direction being less than 0.1.

[0021] A second object of the invention is a human-machine interface comprising the sensor assembly according to the invention, further comprising an electronic module including an acquisition module configured to generate a detection signal in response to a deformation of the annular element based on signals emitted by at least two deformation sensors, the human-machine interface being configured in order to generate control signals from a device external to the human-machine interface in response to the detection signal.

[0022] Such an interface allows for the simple, precise, and time-dependent characterization of a global or localized deformation of the annular element based on local deformations measured by sensors. Various control signals from an external device can be associated with the occurrences and variations over time of the deformation, providing considerable flexibility of use relative to the simplicity of the device. Thus, a wide variety of user actions on the annular element can be associated with a range of control signals from the external device.

[0023] This is an interface that does not infringe on the user's privacy since it does not rely on the acquisition and processing of video or audio signals that could contain personal information. This interface is very well suited to integration into jewelry with a classic appearance, including small pieces.

[0024] According to additional non-limiting features of the sensor according to the invention, considered individually or in any technically feasible combination:

[0025] - the interface can be configured to keep only a first part awake of the acquisition module associated with detection and only activate the second parts of the acquisition module when it is determined that respective activations of these second parts are necessary for an analysis of the signals emitted by the at least two strain sensors;

[0026] - the calculator can be configured to (i) classify represented events by the signals emitted by at least two strain sensors and (ii) generate control signals from a device external to the human-machine interface in response to this classification;

[0027] - the control signals can be representative of at least one chosen action among a tightening, a short tightening or a long tightening of the ring between two fingers or between three fingers, repeated or not, in a centripetal direction to the ring; a tightening of the ring, repeated or not, parallel to an axis of revolution of the ring; a closing of a hand with one finger wearing the ring or the closing of a finger wearing the ring; and a rotation of the ring between two fingers of a first amplitude belonging to an angular interval or of a second amplitude greater than an upper bound of the angular interval;

[0028] - the control signals can be representative of at least one chosen action including touching the ring with a finger; tapping a surface with a finger wearing the ring; tapping a surface directly with the ring; applying pressure to the ring at a point moving around the ring by a finger wearing the ring; sliding the ring across a surface; sliding a finger across a surface first hand on the ring in a direction parallel to a finger of a second hand wearing the ring; a localization of any of the preceding actions; and a snap of a finger wearing the ring;

[0029] - the acquisition module can be configured to detect a rotation of the element annular based on the signals emitted by the at least two strain sensors; and in response to the detection step, generate a signal representative of the occurrence of a rotation of the annular element;

[0030] - the signal is representative of a rotation that has exceeded an angular value incremental;

[0031] - the interface may further include a configured wireless transmission module to transmit signals generated by the acquisition module to the external device and a power supply for the wireless transmission module and the acquisition module;

[0032] - the interface may further include a configured feedback device to notify a user that a detection signal has been generated by the acquisition module; and

[0033] - the interface can be configured to be threaded onto a part of a user or a rod of an external device control unit.

[0034] The interface according to the invention can be compact and passive when piezoelectric strain sensors are used, and the processing of the signals generated by these sensors can be simple and computationally lightweight, resulting in high autonomy of the device. Furthermore, the ring strain detection can be highly sensitive and can therefore operate without difficulty even with very rigid rings such as thick metal rings, like those commonly worn in jewelry.

[0035] The invention extends to a kit comprising a human-machine interface according to the invention and an external device configured to be controlled by means of the human-machine interface. BRIEF DESCRIPTION OF THE FIGURES

[0036] Other features and advantages of the invention will become apparent from the detailed description of the invention which follows with reference to the accompanying figures in which:

[0037] [Fig.1] Fig.1 represents a human-machine interface according to the invention, taking the form of a ring equipped with sensors;

[0038] [Fig.2] Fig.2 represents modes of use of the ring;

[0039] [Fig.3] The [Fig.3] represents structural features of the ring;

[0040] [Fig.4] The [Fig.4] is a functional diagram of elements constituting the ring;

[0041] [Fig. 5] Figure 5 illustrates signals generated by the ring in response to pressure on a first sensor of the ring;

[0042] [Fig. 6] Figure 6 illustrates signals generated by the ring in response to pressure on a second sensor of the ring;

[0043] [Fig. 7] Figure 7 illustrates signals generated by the ring in response to pressure on a third sensor of the ring;

[0044] [Fig.8] Fig.8 illustrates signals generated by the ring in response to two successive pressures on the ring sensors;

[0045] [Fig.9] Fig.9 represents signals generated by the ring in response to long and short tightening of the ring;

[0046] [Fig. 10] The [Fig. 10] illustrates signals generated by the ring during rotational actions in opposite directions;

[0047] [Fig. 11] The [Fig. 11] is a representation of signals generated by the ring during a rotation action;

[0048] [Fig. 12] The [Fig. 12] illustrates signals generated by the ring in response to a tap of a finger wearing the ring on a table;

[0049] [Fig. 13] The [Fig. 13] illustrates signals generated by the ring in response to the ring being struck on a table;

[0050] [Fig. 14] The [Fig. 14] represents signals generated by the ring in response to taps at the sensors;

[0051] [Fig. 15] The [Fig. 15] represents a first and a second block of an algorithm for using the ring;

[0052] [Fig. 16] The [Fig. 16] illustrates the location of a point on the ring;

[0053] [Fig. 17] The [Fig. 17] represents a third block of the ring usage algorithm;

[0054] [Fig. 18] Figure 18 represents a fourth block of the ring usage algorithm; and

[0055] [Fig. 19] The [Fig. 19] represents variations in the structure of the human-machine interface of the [Fig.1].

[0056] [Fig.20] Fig.20 illustrates a variance curve derived from signals emitted by the sensor of the [Fig.l];

[0057] [Fig.21] Fig.21 illustrates a step in determining the occurrence of a event based on sensor signals;

[0058] [Fig.22] The [Fig.22] illustrates a method of controlling the human-machine interface;

[0059] [Fig.23] Fig.23 represents a table compiling test results of the process of [Fig.22];

[0060] [Fig.24] Fig.24 represents a table compiling test results of the process of [Fig.22];

[0061] [Fig. 25] [Fig. 25] illustrates actions applied to the ring of [Fig. 1] and recognized by the process of [Fig.22];

[0062] [Fig. 26] Figure 26 illustrates the determination of the amplitude of a rotation of the ring;

[0063] [Fig.27] Figure [Fig.27] represents various variance profiles according to the type of actions applied to the ring;

[0064] [Fig. 28] Figure 28 represents a thin piezoelectric element; and

[0065] [Fig. 29] Figure 29 illustrates a strain sensor incorporating a thin element piezoelectric; DETAILED DESCRIPTION OF THE INVENTION

[0066] An embodiment of the present invention is described by means of Figures 1 to 19 and the associated passages below. Figures 1 to 4 and 19 provide information on the structure, use, and operating principle of the human-machine interaction ring.

[0067] Figures 5 to 18 detail a first option for exploiting the signals generated by the deformation sensors following manipulation of the ring by a user.

[0068] Figures 20 to 27 detail a second option for exploiting the signals generated by the deformation sensors following manipulation of the ring by the user.

[0069] Figures 28 and 29 illustrate a piezoelectric element that can be used to form a strain sensor integrated into the human-machine interface according to the invention.

[0070] Figure 1 illustrates the structure and general operating principle of the ring according to the invention, with (A) a perspective view of a sensor assembly R according to the invention, comprising an annular element ANN instrumented by three sensors SENS1, SENS2, and SENS3. The sensor assembly R may hereafter also be referred to as the ring R. The sensors may be conventional strain sensors, for example, resistive gauges, whose electrical resistance varies with their deformation, or piezoelectric sensors. The latter are strain sensors based on a piezoelectric measuring element, that is, they take advantage of the piezoelectric behavior of a material to transform into an electrical signal a deformation it undergoes under the effect of external forces, for example, when it is bonded to a surface and that surface deforms.The annular element ANN can be made of any material suitable for forming a self-supporting structure that can be passed around a finger or rod. The annular element may have a composite structure. Preferably, the annular element has a monolithic structure. That is to say, formed from a single piece and a single material. Thus, the annular element can be formed from a single piece of metal. This latter configuration has the advantages of a simple, robust, and reliable structure.

[0071] This ring instrumentation allows it to be coupled with an electronic module (EL) to form a human-machine interface for controlling, for example, an electronic device such as a computer, a smartphone, a television, or a media player. In this example, three piezoelectric strain sensors, SENSi to SENS3, are bonded to the annular element, on which the electronic module (EL) is mounted. This module is configured to measure the electrical charges or voltage generated by the sensors, analyze them, and transmit the information wirelessly to an external device such as those mentioned above.

[0072] As illustrated in (B) of [Fig. 1], when the ring R is held between two fingers F1 and F2, a pressure force F is applied on both sides of the ring by the fingers, creating an overall deformation of the ring that can be characterized using the three sensors SENS1 to SENS3. More generally, and as will be detailed below, the detection of this deformation, whether a single or repeated event, and the profile of this deformation, corresponding, for example, to pressure between the fingers or to a strike, can be used to generate a control signal for an external APP device. Indeed, the sensors make it possible to detect the locations of the points of contact on the ring element, the temporal evolution of these locations, as well as the temporal evolution of the amplitude of the applied force.

[0073] As illustrated in (C) of [Fig.1], when the ring is rolled between the fingers in a rotational movement Rot caused by the combination of the pressure force F and the displacements Dplt of the fingers Fl and F2, the position at which this force is applied moves substantially, and this displacement can be measured by comparing the signals from the three sensors.

[0074] Various user actions on the ring can thus be detected and associated with predetermined actions, such as (i) a single press by pinching between the fingers associated with a confirmation action, (ii) a double press associated with a back button action in a menu of a ring-controlled application, (iii) a rotation in one direction or the other associated with scrolling up or down in a list or with volume control of a music player, and (iv) tapping the finger wearing the ring on a hard surface such as a table or (v) directly tapping the ring on a hard surface such as a table associated with confirmation actions equivalent to a single or double click of a computer mouse. The intensity of the pinch or taps, or the rotation speed, can also be associated with specific commands to control the external device.Other actions can be considered, such as those listed. in Table 1 below. In this document, when rotations are mentioned, and unless otherwise indicated, they refer to a rotation of the ring about its axis of revolution Ax, illustrated in [Fig. 3]. A direction of rotation can be counterclockwise around this axis of revolution, or opposite to counterclockwise. An axis of revolution is considered here to be an axis of revolution that describes the general shape of the annular element, which extends parallel to the axis of revolution and has an overall shape possessing rotational symmetry about this axis, including the examples in Figures 1, 3, and 19.

[0075] It should be noted that the ring used as an example in this embodiment is equipped with three sensors, but a ring with two sensors, the minimum number required to spatially characterize the ring's deformation, would also work. Using three sensors provides redundancy and an additional measurement, enabling discrimination in ambiguous situations. It is also possible to use a larger number of sensors, with the disadvantages of increased integration and cost of the ring, as well as a more complex electronic processing system to be integrated onto the ring element, requiring more electrical contact traces and signal processing channels. The advantage, however, is improved robustness of event detection.An event is understood here as a deliberate action by the user on the ring in order to produce a control signal from an external device, with the sensors generating signals in response to this action.

[0076] Also, the SENSi, SENS2 and SENS3 sensors are shown located on the outer surface of the ring, but they could also be located on its inner surface, or even integrated into the volume of the annular element, for example by being sandwiched between two concentric rings.

[0077] In (C), [Fig.3] illustrates a ring equipped with two sensors SENSi and SENS2 arranged on the inner surface of the ring, substantially at 90° to each other (the respective radii rl and r2 make an angle of substantially 90° to each other).

[0078] The respective positions of the sensors on the ring are located by means of rays, the center of one of the sensors defining a reference ray, the positions of the other sensors being located by the angles that the rays passing through their respective centers (in a cross-sectional view along the axis of revolution Ax of the ring) make with the initial ray. In the example of [Fig. 3] in (A), the three sensors are arranged substantially homogeneously around the ring, at angular intervals of approximately 120°. In this case, the ray r1 associated with the SENS1 sensor can be considered the reference ray, the rays r2 and r3 associated respectively to the SENS2 and SENS3 sensors being located at approximately 120° and 240° respectively from the radius rE

[0079] There are no strict requirements regarding the relative positioning of the sensors, except that two sensors should not be placed in diametrically opposite positions if the intention is to spatially characterize the deformation caused by pressure applied at two diametrically opposite points, for example, between two fingers of the user: such symmetrical positioning of the two sensors limits the spatial characterization capabilities of the deformation. It is also preferable to space the two sensors sufficiently apart to facilitate this spatial characterization, which is all the easier when the deformations measured by the sensors have different amplitudes.

[0080] Thus, it is preferable that the radii defining the positions of the sensors be spaced angularly between 20° and 160°, more preferably between 40° and 140°. Of course, the angles used may depend on the number of sensors integrated on the ring.

[0081] The ring may also be equipped with a feedback device (FB), informing the user that their manipulation of the ring by tightening, tapping, or rotating it has been registered. The FB device may be of any type conventionally used for this purpose, such as a light, sound, or haptic device.

[0082] The ring can be manipulated while it is threaded onto a part of a user, and in particular onto a finger F6 of a hand H as illustrated in (A) of [Fig. 2], or threaded onto a rod or wand RD of an external device control system as in (B) of [Fig. 2]. It is of course possible to construct the ring so that it is worn, for example, on a wrist, arm, or leg of the user. Alternatively, the ring can be used "free" without being threaded onto any part, as illustrated in [Fig. 1].

[0083] The ring preferably has dimensions that allow it to be slipped onto a finger like a piece of jewelry such as a ring, with, for example, the annular element ANN having a small inner diameter, illustrated in (A) of [Fig. 3], of between 1 and 3 cm and a length L, illustrated in (B) of [Fig. 3], along its axis of revolution Ax of between 2 mm and 2 cm. The radial thickness Thck of the annular element, defined, for example, by (fin - fext) / 2 for a perfectly cylindrical annular element of outer diameter fext, as illustrated in (A) and (B) of [Fig. 3], can be greater than 0.1 mm, preferably 0.4 mm, and less than 3 mm, preferably 1 mm. Such thicknesses ensure sufficient rigidity for the annular element to be self-supporting while allowing for deformations imposed manually by the user. be detectable, even if the ring element is made of metal. Among the metals from which the ring element can be formed, a metal that is not very susceptible to corrosion and is hypoallergenic should preferably be chosen. This could, for example, be a metal conventionally used in jewelry such as gold, silver, stainless steel, brass, or titanium, either solid or plated.

[0084] Furthermore, the ring is not necessarily a closed ring, but can also be an open ring with a cut C, as illustrated in (C) of [Fig. 3]. Even without strict rotational symmetry, the ring is considered to have an oriented axis of rotation Ax, illustrated in (B) of [Fig. 3] and perpendicular to the page plane in (B) and (C) of [Fig. 1], located at the center of the ring and along which the ring extends. The orientation of the axis Ax is indicated by the arrowhead in the figure.

[0085] The ring element is considered to be substantially cylindrical in shape in the sense that it is suitable for slipping onto a finger. However, the substantially cylindrical shape of the ring element can encompass various shapes, the important point being that it can be comfortably worn on the finger by a user and that sensors can be integrated into it.

[0086] Figure 4 illustrates the elements integrated over a ring R and the relationships functional between the elements composing it and with an external device. Thus, the R ring comprises, in a two-sensor configuration, two SENSi and SENS2 sensors connected to an EL electronic module for processing the signals from the sensors.

[0087] The electronic module EL includes an ACQ acquisition module in which an analog-to-digital converter CONV receives analog signals from the SENSi and SENS2 sensors and converts them into digital signals processed by a CALC digital computer. The digital computer processes the digital signals, generates detection signals in response to the digital signals, generates SCon control signals for an external APP device based on the detection signals, and sends these control signals to a COM communication module. The COM communication module transmits, preferably wirelessly, the generated control signals to the external APP device, located at a distance from ring R.

[0088] The CALC computer is also configured to, if necessary, control the FB feedback device and trigger an action indicating to the ring user that their action on the ring has been taken into account to send a control signal to the external APP device.

[0089] A battery BAT provides the energy required for the operation of the analog-to-digital converter CONV, the digital computer CALC, the communication module COM, and the feedback device FB. If the sensors are resistive gauges, the battery can also be used to supply them with electrical current.

[0090] Optionally, the battery can be charged by means of an EH energy harvesting device, or “Energy Harvesting” in English terminology, such as a miniature electromechanical device configured to conventionally harvest energy from ambient vibrations or user movements of the ring R.

[0091] With regard to the structure of the sensor assembly, the invention is not limited to the examples mentioned above and the annular element and the sensors can be implemented in different ways, illustrated by Figures 1, 3 and 19 as described below.

[0092] Thus, the sensor assembly or ring R can include an open or closed annular element ANN formed from a block as illustrated in A) of [Fig. 19], an annular element formed from two coaxial elements ANNi / 2c and ANN2 / 2C as illustrated in B) of [Fig. 19], or formed from two angularly complementary parts ANNi / 2A and ANN2 / 2A connected by assembly elements AE, preferably rigidly to each other so as to transmit a deformation applied to one element to the other, as illustrated in C) of [Fig. 19].

[0093] The sensors can be mounted on the same surface, either internal or external to the ring, as illustrated in Figures 1 and 3, or they can be mounted on different surfaces of the ring. Figure 19 illustrates in A) a configuration where the SENS1 and SENS3 sensors are mounted on an external surface Sext of the annular element and the SENS2 sensor on an internal surface Sint of this annular element; in B) a configuration where the SENS1 and SENS3 sensors are mounted on an external surface of the annular element and the SENS2 sensor is positioned between a first coaxial element ANN1 / 2C and a second coaxial element ANN2 / 2C of the annular element; in D) a configuration similar to the configuration illustrated in A), except that the sensors are mounted in recesses or notches formed on the annular element.Placing the sensors in recesses has the advantage of protecting them from external damage when they are directly accessible from the outside, or making the ring more comfortable for the user to wear when they are located on the inner surface in contact with the skin.

[0094] The sensors can also be mounted on parts of the annular element other than its outer or inner surfaces. Figure 19 illustrates this by E) a configuration where the sensors are mounted on a surface St of the edge of the ring, so as to be positioned along a plane intersecting the longitudinal axis of the ring.

[0095] The sensors are mounted on the annular element so as to be fixed to its deformations and thus generate signals representative of these deformations. They can be glued to a surface of the annular element, crimped between two elements of the annular element, or integrated in any other way considered suitable by those skilled in the art.

[0096] Any combination of the sensor implementations illustrated in Figures 1, 3 and 19 is acceptable and falls within the scope of the invention described herein.

[0097] I. Analytical option for event detection

[0098] Several types of user actions can be applied to the ring, each type of action being associated with a given command from the external device used to control it. Each action results in a deformation event of the ring, which is reflected in the generation of signals by the sensors. Each signal generation event corresponds to an event that we seek to characterize and which allows us to determine the type of action exerted on the ring: tightening, rotation, tapping, or other.

[0099] Analytical methods for detecting these types of action using the signals generated by the sensors are explained below, using the ring illustrated in [Fig. 1], which is equipped with three sensors evenly distributed around the periphery of the ring. Similar results would be obtained with a different number of sensors or sensors distributed differently. The explanations below are intended to illustrate the principles on which the practitioner can base their implementation of the ring according to the invention with various sensor configurations in terms of number and location.

[0100] In the figures mentioned below, the references SI, S2, and S3 are associated respectively with the signals generated by the three sensors Sensi, Sens2, and Sens3 in response to an action S00 applied to the ring element by a user. The reference "S" represents the signals S2 and S3, and the reference T represents time. The references S, SI, and T are expressed in mV, mV, and s, respectively.

[0101] 1.1 Tightening

[0102] The action of squeezing the ring between two fingers, as illustrated in [Fig. 1] in (B), is referred to here as "tightening". This figure illustrates a situation in which the SENS2 sensor undergoes elongation and the SENS1 and SENS3 sensors undergo compression.

[0103] A clamping action can be detected by exceeding a detection threshold THR0 by the amplitudes of the signals generated by the sensors.

[0104] It is also possible to detect not only the occurrence of the application of a clamping force, but also its angular location on the ring. Indeed, depending on the location of the applied clamping force, the three sensors will provide different signals. An analysis of the relationships between the signals makes it possible to determine the location of the applied clamping force, either by ab initio simulation of the ring, or by calibrating it. The clamping force location is understood to mean the determination of an angle between 0° and 180° defining the position of one of the two fingers applying the clamping pressures relative to a reference angle, for example defined by the radius rl of the SENSp II sensor. It is assumed that the second finger applies pressure at a location diametrically opposite to that of the first finger.

[0105] Figures 5, 6, and 7 illustrate the results of three experimental situations in which the ring is tightened such that one of the two fingers applying the tightening action is located at the level of sensors SENS1, SENS2, and SENS3, respectively, with the other finger located diametrically opposite the first finger. Figure (A) shows graphs of the amplitudes of signals S2 and S3 generated by sensors SENS2 and SENS3, respectively, as a function of the amplitude of signal S1 generated by sensor SENS1. Figure (B) shows graphs of the amplitudes of the signals generated by sensors SENS2 and SENS3, based on the same data as the corresponding graphs in (A), but expressed this time as a function of time T.

[0106] When, as in [Fig. 5], one of the fingers applying the clamp is located on the SENS1 sensor, the two curves illustrated in (A) trace approximately straight lines of the same slope. This is due to the symmetry of the ring, which causes the mechanical load to be distributed symmetrically at the levels of the SENS2 and SENS3 sensors.

[0107] When, as in [Fig.6], one of the fingers performing the clamping is located on the SENS2 sensor, the curve corresponding to the SENS2 sensor forms approximately a straight line of negative slope while the curve corresponding to the SENS3 sensor forms approximately a straight line of positive slope.

[0108] When, as in [Fig.7], one of the fingers performing the clamping is located on the SENS3 sensor, the curve corresponding to the SENS2 sensor forms approximately a straight line of positive slope while the curve corresponding to the SENS3 sensor forms approximately a straight line of negative slope.

[0109] The fact that the amplitudes of the signals generated by sensors SENS2 and SENS3, expressed as a function of the amplitude of the signal generated by sensor SENSi, form essentially straight lines is representative of the fact that the ratios of the amplitudes of the signals generated by sensors SENS2 and SENS3 to the amplitude of the signal generated by sensor SENSi are essentially constant. This characteristic is due to the fact that The pressure application zones on the ring remain stationary during the application of pressure.

[0110] Also, as illustrated in (B) of figures 5, 7 and 6, the maximum amplitude among the signals generated by the three sensors is in the signal generated by the sensor at the level of which one of the user's fingers is located while tightening, indicating that this location of the ring has a greater deformation than those of the locations of the other two sensors.

[0111] A clamping action produces two peaks in absolute value in the signals generated by the sensors: a first peak caused by the application of clamping pressure to the ring, and a second peak due to the release of this pressure, identified respectively by Maxl and Max2 in (B) of [Fig.5] for the signal SI. The first peak and the second peak have opposite signs.

[0112] The analysis of curves such as those in figures 5, 6 and 7 makes it possible to trace back to the occurrence of a clamping action as well as to a characterization of this clamping action, a characterization which is a function of the areas of application of the clamping forces.

[0113] For example, a tightening action without significant displacement of the pressure application areas can be considered to have been applied to the ring if a criterion is met, according to which the ratio between the amplitudes of two of the signals is constant to within ±20% of the average value of the ratio during the duration of the action. The percentage can be adjusted by the user. Detecting such an action eliminates the possibility that an action involving significant displacement of the pressure areas on the ring, such as rotation or sliding, may have been applied. In other words, meeting this criterion implies that a pressure action with substantially constant localization has been applied to the ring by a user.To improve the reliability of the application of the criterion, it can be applied only to signal amplitude values ​​exceeding a certain level, in order to avoid the effect of relative noise which becomes very important for small amplitude values ​​and / or adjust the percentage.

[0114] Thus, it is important to note that an analysis of the ratios between the signals generated by the different sensors makes it possible to characterize the clamping action according to a criterion linked to the location of the fingers applying the clamp, not only at the level of a particular sensor, but also at an arbitrary position varying continuously between two sensors. The straight lines obtained by clamping actions gradually shifting from the situation in [Fig. 5] to that in [Fig. 6] would show a corresponding continuous variation in the slopes of the curves. To each slope or group of slopes, and therefore to each ratio of amplitudes of signals generated by two sensors, it is possible to associate a particular location of the action exerted on the ring and to associate with this location a particular command of the external device.

[0115] 1.2 Double tightening

[0116] Double clamping is a variant of the clamping action described above, in which the clamping action is repeated within a predefined time interval, such as the double-click of a computer mouse.

[0117] This action is detected by counting, for each sensor, the number of maxima occurring during the predefined time interval. Figure 8 illustrates in (A), (B), and (C) the signals S (in mV) generated by the sensors in response to double-tightening actions at sensors SENS1, SENS2, and SENS3, respectively, expressed as a function of time T (in s). The amplitude maxima (in absolute value) clearly show four maxima identified as Max1 to Max4 in (A) of Figure 8 for the signal SL. The occurrence of a double-tightening action can thus be associated with the detection of four absolute maxima in the signal generated by a sensor over a determined time interval, preferably on the order of a second or less.

[0118] 1.3 Long tightening

[0119] A long clamping differs from a clamping that can be described as short, such as those illustrated by figures 5 to 7, in that the clamping is maintained for a certain duration exceeding a predetermined threshold value of duration THR4.

[0120] During a tightening action, the signals generated by the sensors each include a first maximum Maxl in the form of a peak at the beginning of the tightening and a second maximum Max2 of opposite sign to Maxl in the form of a peak with the opposite orientation when the tightening is released. An event corresponds to a long tightening action when two maxima are detected, each exceeding a predetermined threshold level, and separated by a duration exceeding the predetermined threshold value THR4, as illustrated in B) of [Fig. 9].

[0121] The profile of the curves in B) is explained by the fact that the voltage generated by a piezoelectric sensor will gradually return to zero after application of the deformation, even without the sensor returning to its initial physical state, according to a characteristic time depending on the sensor itself and the acquisition system, i.e. about 0.2 s for the system which generated the signals represented here.

[0122] 1.4 Rotation - Method 1

[0123] Herein, "rotation" refers to the action of rotating the ring between two fingers while simultaneously applying pressure with those same fingers. This action by the user results in a deformation of the ring, which moves with the rotation. It is a rotation of the ring according to this principle that allows the transition from configuration (B) to configuration (C) in [Fig. 1]. The sensors can characterize this deformation and its movement, and therefore the rotation imposed on the ring, as described below.

[0124] [Fig.10] shows graphs representing signals SI, S2 and S3 (in mV) experimentally generated by the piezoelectric sensors SENSi, SENS2 and SENS3 of the ring in [Fig.1], respectively, as a function of time T (in s), in (A) for a rotation Rot in one direction from sensor SENSi to sensor SENS2 then sensor SENS3 (direction 1-2-3) and in (B) for a rotation in the opposite direction from sensor SENS3 to sensor SENS2 then sensor SENSi (direction 3-2-1), by rotating the ring between two fingers.

[0125] It is observed that the sensor signals oscillate in an approximately sinusoidal pattern. The maxima of the oscillations, which correspond to maxima of the deformations at the three sensors, appear in an order determined by the direction of rotation. This direction can therefore be deduced from the analysis of the signals generated by the sensors by finding the maxima and determining their order of appearance. Thus, it is possible to characterize the rotation in terms of direction, angular amplitude, and speed.

[0126] 1.5 Rotation - Method 2

[0127] The first method for characterizing a ring rotation is based on the order in which maxima appear in the deformation amplitude. A second method, based on the same signals generated by the sensors, can be considered.

[0128] Fig. 11 shows in A) and B) graphs in which the signals S (in mV) generated by the sensors SENS2 and SENS3 are expressed as a function of the signal SI (in mV) generated by the sensor SENSi, forming two distinct curves in the plane (Si, S) defined by the signal SI and one or the other of the signals S2 and S3. In these two graphs, the progress of the curves over time is followed by moving along them in the directions indicated by the arrows.

[0129] With regard to the graph in A), in a ZSq-Start zone of the graph corresponding to the beginning of the application of the clamping force using the two fingers of [Fig. 1], it can be observed that the two curves move away from the origin of the graph along straight lines until they exceed a threshold level Thr represented by a circle centered on the origin of the graph. The ratios of the signal amplitudes remain essentially constant, due to the fact that the pressure application zones on the ring remain stationary: this is currently a simple clamping action, already described above. The distance of each curve from the origin of the graph is defined by the square root of the sum of the squares of the amplitudes of the signals concerned: either a first pair of signals consisting of the signals from sensors SENS1 and SENS2, or a second pair of signals consisting of signals SENSi and SENS3 sensors. The distance from each point of these curves to the origin point (0,0) of the graph represents the root of the quadratic sum of the coordinates of the curve at that point.

[0130] In a second step, the ring is rolled between the two fingers. The amplitudes of the remaining signal pairs are such that the square roots of their quadratic sums remain above the threshold value, if not constant. This behavior is representative of the pressure exerted on the ring by a human: it is difficult to maintain constant pressure on the ring while rolling it between two fingers, but it is possible to maintain a certain level of pressure. The rolling results in displacements over time in opposite directions of rotation along the two curves, clockwise for sensor SENS2 and counterclockwise for sensor SENS3, these two directions depending on the direction of rotation imposed on the ring.The ratios between the amplitudes of the signals generated by the SENS2 and SENS3 sensors and the amplitude of the signal generated by the SENSi sensor change continuously due to the ongoing movement of the pressure zones applied to the ring. Detecting these changes indicates the occurrence of a rotational action. The amplitude of these changes is representative of the angle of rotation. The curve can be followed, and an angular amplitude of rotation W can be assigned to it between two given instants, either through calibration or ab initio calculations.

[0131] At the end of the rotation, the finger grip is released, the signal amplitudes return to zero, which means that the curves simultaneously return to the perimeter defined by the threshold level Thr and then fall back to the origin point of the graph, at the levels of the ZSq-End2 and ZSq-End3 grip release zones, for the curves resulting from the signals generated by the Sens2 and Sens3 sensors, respectively. At the moment of grip release, the rotation is stopped, the pressure application points are fixed, and the curves form straight lines, as in the case of a grip described above using Figures 5 to 7.

[0132] Analysis of the graph in A) of [Fig. 11], whether one considers a single curve or both curves, allows us to trace back to the detection of a rotation action of the ring and to the characteristics (angle, duration) of this rotation by correspondence between the points of crossing the threshold level by these curves at the beginning and at the end of the rotation action and a rotation of the ring, correspondence determined by simulation calculations or calibrations of the ring.

[0133] In practice, it may be advantageous to initiate an analysis of the curve as soon as the threshold level Thr is reached, to analyze the curve continuously, and to generate a rotation detection signal each time the rotation angle exceeds one of a series of thresholds (for example spaced at intervals of 10° or any other predetermined value), and stop the analysis when the curves fall back below the Thr threshold.

[0134] This is illustrated in [Fig. 1 1] in B), with the curve representing the signal S2 expressed as a function of the signal SI, signals generated respectively by the sensors Sens2 and Sens1. The curve representing the signal S2 exits the threshold display circle Thr at a point A, rotates clockwise around the origin point of the graph while remaining outside the circle, and then crosses it at a point B. Points A and B are separated angularly by a total rotation angle Wtot. During rotation, the increase in the rotation angle exceeded several thresholds indicated by the values ​​Wi and W2.

[0135] 1.6 Finger tap

[0136] Tapping the finger wearing the ring, preferably on a hard surface such as a table, generates a signal with a higher frequency content than squeezing or rotating the ring. Figure 12 illustrates the S1, S2, and S3 signals detected when the finger wearing the ring is tapped. The frequency components of the signals generated during a finger tap thus make it possible to distinguish this action from squeezing or rotating the ring.

[0137] A method for detecting a tap of the finger wearing the ring may include, as a first step, summing the signals generated by the sensors during an event to obtain a sum signal. A high-pass filter is applied to the sum signal, and then a first RMS value (RMSH) of this filtered sum signal is calculated. A low-pass filter is applied to the same sum signal, and then a second RMS value (RMSL) of this filtered sum signal is calculated. The ratio of these two RMS values ​​is then compared to a predetermined threshold value (THRi). The cutoff frequencies of the applied filters may be, for example, 10 Hz. If the RMSH / RMSl ratio is greater than or equal to the threshold value (THRi), then the ring deformation event is considered to have been generated by a tap of the finger. If the RMSh / RMSl ratio is less than the threshold value (THRb), ​​then the event is classified as a squeeze.The sum of the signals can be a weighted sum, with each signal associated with a given weighting coefficient. This allows the coefficients to be chosen to select the sensor(s) used for detection. An effective value can be defined as the square root of the average of the squares of the signal value over a period of time.

[0138] 1.7 Ring Tap

[0139] The procedure used for tapping a finger can be applied to the ring when it is tapped directly on a surface, preferably a hard one, such as a table. The difference is that the signals generated by the sensors have a content of components High frequencies are more significant than when only the finger wearing the ring is tapped on the hard surface. For example, the waveforms of the SI, S2, and S3 signals in [Fig. 13], generated when the ring is directly tapped on a table, can be compared to those in [Fig. 12]. This allows us to compare the RMSH / RMSL ratio to a threshold value THR2, preferably greater than THRi. If the RMSH / RMSL ratio is greater than or equal to the threshold value THR2, then the event is considered to have been generated by tapping the ring.

[0140] 1.8 Tap on a sensor

[0141] Tapping a sensor is an action that consists of directly tapping the ring at the location of one of the sensors with a finger. The difference with squeezing is that the force is applied to only one side of the ring, and not to both sides. In the case of the configuration illustrated in [Fig. 1], the sensors are tapped directly and the measured amplitudes of the signals S (in mV) generated in response by the sensors are illustrated by means of the graphs in [Fig. 14], with (A), (B) and (C) the measurement results for taps on sensors SENS1, SENS2 and SENS3, respectively.

[0142] For each of the situations (A), (B) and (C), it is observed that the maximum amplitude among all the signals generated by the sensors is that corresponding to the sensor at the level of which the tap is made, and that it corresponds to the maximum amplitude of a peak.

[0143] Thus, a tap event on a given sensor can be identified by identifying the sensor whose generated signal has a peak with the highest amplitude, calculating the ratio of this amplitude to the maximum amplitudes of each of the other sensors, and, if these ratios are all greater than a threshold value THR3, considering the event as having been generated by a tap on the sensor which generated the signal having the peak with the highest amplitude, therefore the maximum amplitude.

[0144] 1.9 Example of an algorithm

[0145] As illustrated above, several types of ring deformation events can be detected by analyzing the signals from the sensors mounted on it.

[0146] Figures 15 to 18 illustrate a method of using a ring R equipped with at least two sensors SENSi and SENS2, according to an algorithm allowing the detection of several types of events each characterized by the signals generated by the sensors following an action by the user on the ring.

[0147] During a user action on the ring, two digital signals SI and S2 are generated by the two sensors SENSi and SENS2, respectively, and the analog-to-digital converter CONV, according to the algorithm illustrated by [Fig. 15].

[0148] A first calculation block B1 serves to discriminate background noise or accidental or purposeless manipulations from a deliberate action by the user. In the first calculation block Bb, the CALC computer performs the quadratic summation of the amplitudes of the signals SI and S2 during a summation operation BImo. During a test operation B1m5, the computer determines whether the quadratic summation Q of the signal amplitudes exceeds a predetermined threshold value Thr, indicating that an action has been deliberately performed on the ring. This test operation has a function of detecting an event measured by the ring's sensors.In response to this determination, if the sum exceeds the threshold value, an action detection signal SAct is generated, and in response to this signal, a second calculation block Bl2 is implemented, and the signals SI and S2 are recorded in a computer memory MEM in functional communication with the CALC computer during an operation B1i 2o-. The recorded signals will be used in subsequent operations. Recording is stopped when the quadratic sum falls below the threshold value THR.

[0149] More generally, the amplitude of a signal representing a sum of unaveraged signals generated by the sensors, or the amplitude of a single signal, averaged or unaveraged, representing a signal generated by a single sensor, could be used as the amplitude of the Q signal and compared to the threshold value to determine the occurrence of a deliberate action. In all cases, the amplitude of a signal or a combination of signals, processed or unprocessed, representing a deformation of the ring, is compared to a predefined threshold, and it is determined that a deliberate action has been exerted on the ring when this amplitude exceeds this predefined threshold, such as the threshold level Thr shown in [Fig. 11]. Averaging over a combination (in particular, a sum) of several signals representing signals generated by the sensors provides greater reliability in this determination.

[0150] Alternatively, in order to discriminate background noise or accidental or purposeless manipulations from a deliberate action by the user, the BImo test operation could focus on a variance in the amplitudes of the signals emitted by the sensors. This variance can, for example, be calculated over a rolling period of a signal or of the weighted or unweighted sum of several signals generated by the sensors, and can be considered an event detection signal when it exceeds a predetermined threshold, set, for example, by a user: the variance of the signals becomes indeed greater when the user performs an action on the ring.

[0151] The function of test step B1m5 is thus to detect an exceedance of a threshold by a quantity representative of the amplitude of a signal or a combination of signals each representative of a deformation of the annular element and generated by one or more of the deformation sensors mounted on the annular element in response to an action of the user, whether the quantity in question is derived directly from the amplitudes of the signal(s) considered, from a variance of this or these signals, or is representative of any other information derived from the signal(s) considered.

[0152] The Bli block is advantageous from the standpoint of human-machine interface autonomy: except when an event exceeding a certain threshold level is detected, only a portion of the ACQ acquisition module associated with the BU block requires a power supply, which significantly limits the overall power consumption of the device. Indeed, a comparator, an electronic device consuming very little energy, can be sufficient to implement the Blb block, particularly when piezoelectric sensors are used, as the voltages they generate are sufficient to operate the comparator without requiring amplification.

[0153] The second calculation block Bl2 is responsible for performing an initial analysis of the SI and S2 signals, intended to detect whether or not a rotation is applied by the user to the ring and, if so, to characterize this rotation. When the Bl2 block is implemented, the CALC computer performs the operations described below. In preliminary steps, (i) a counter in the CALC computer initializes a rotation increment count variable n to 0, at a step Bl21 0, and the CALC computer determines and assigns, to a rotation angle variable Wod, the exit angle of the circle of radius Thr in the (SI, S2) plane by the S2 signal as determined according to method 2 for characterizing a rotation based on the SI and S2 signals, at a step Bl2 15. For illustration, refer to the ZSq-Start area in A) of [Fig. 11]. In the graph in B) of the [Fig.11], this exit angle corresponds to the orientation of the line passing through the origin of the graph and the point A located on the circle of radius Thr. .

[0154] A test operation Bl2 20 ensures that as long as the quadratic sum of the signals remains above the threshold Thr, as indicated by the filled test condition indicator Y, the following computation loop is continuously implemented by the computer. A current rotation angle W is calculated based on the signals SI and S2 according to method 2 during an operation Bl2 25. A difference W-Wo is calculated in an operation Bl2 30, and the result of this difference is compared during two test operations Bl2 35 and Bl2 45 to a positive angular threshold value ThrAng and its opposite -ThrAng, respectively. If the test operation Bl2 35 returns a positive result, indicated by Y in the figure, then an event determination signal ROT- is generated, representative of the fact that i) there has been a rotation of the ring by an angle defined by the value of ThrAng, and ii) this rotation is in a direction defined as negative. Conversely, if the Bl2 45 test operation returns a positive result, indicated by Y in the figure, then a ROT+ event determination signal is generated, representing that i) the ring has rotated by an angle defined by the ThrAng value, and ii) this rotation is in a direction defined as positive. ROT- and ROT+ signals can be generated sequentially as the user continues rotating the ring. Each of these signals is generated in response to an incremental increase in the rotation imposed on the ring by the user, and each can be associated with a control signal from the external APP device.

[0155] In response to positive results from test operations Bl2 35 and Bl245, the rotation increment count variable n is incremented by 1 during an operation Bl2 55, and the current rotation angle W is assigned to the rotation angle variable Wod. The count value n thus indicates the number of times the ring has been rotated by a value ThrAng during the rotation being analyzed and is therefore representative of a total rotation amplitude, corresponding to the sum of the n rotation increments. By repeating these operations, rotations can be tracked in both directions, according to amplitude increments defined by the value of ThrAng.

[0156] This algorithm implements method 2 for characterizing a rotation of the ring, but method 1 could be applied, or any other suitable method.

[0157] If the test operation Bl2 2 0 returns a result N, indicating that the quadratic sum of the signals SI and S2 has fallen below the value Thr, then another test operation Bl2 70 is performed to check if the count value n is 0, indicating that the analyzed event is not representative of a rotation. If so, then a rotation-absence detection signal SNo.ROt is generated and, in response, a third computation block Bl3 is implemented during a step Bl2 75.

[0158] Alternatively, the test operation Bl2 70 could be implemented independently of the count value n, by implementing the test principle described in the "Tightening" section. Thus, the test operation Bl2 70 could consist of testing whether a ratio between the signals S1 and S2 is substantially constant over the duration of the analyzed event, i.e., within an interval defined by XX% around the average value of the ratio over the duration of the recorded event, XX% representing a percentage of the average value defined by the user, for example 20%, which constitutes a value allowing for a relevant discrimination of the constancy of the ratio.

[0159] The third computing block, Bl3, illustrated in [Fig. 17], has the function of analyzing the signals generated by the sensors when it has been determined that this event is not generated by a rotation. More specifically, the function of the Bl3 block is to discriminate The type of action among the ring tightening, finger tap, and ring tap actions, and to characterize the tightening actions. For this purpose, the computer implements the operations of block Bl3, the principles of which are explained in the Tightening, Finger Tap, and Ring Tap sections described above.

[0160] First, the digitized signal SI is filtered in parallel using two filtering operations Bl3 xo and Bl3 i5 employing a high-pass filter and a low-pass filter, respectively, to obtain two filtered signals. Each of the two signals filtered by the filtering operations Bl310 and Bl215 is averaged during averaging operations Bl315 and Bl3 2s to obtain the RMSH and RMSH means of the two averaged filtered signals, respectively. The averaging period can be selected by the user according to their preferences.

[0161] According to the principle explained above in the sections "Finger Tap" and "Ring Tap", a first test operation Bl3 35 checks whether the value of the RMSh / RMSl ratio is greater than a high threshold value ThrH. If a positive response, indicated by Y in the figure, is returned, then a detection signal Hit(R) is generated, representing that the event was caused by tapping the ring on a rigid surface.

[0162] If a negative response to test operation Bl3 35, indicated by N in the figure, is returned, then test operation Bl3 40 is implemented, checking whether the value of the RMSH / RMSL ratio is greater than a low threshold value ThrL, with ThrL < ThrH.

[0163] If a positive response, indicated by Y in the figure, is returned, then a Hit(F) detection signal is generated, representative of the fact that the event was caused by a finger tap action (F6) bringing the ring onto a rigid surface.

[0164] If a negative response to the test operation Bl3.40, indicated by N in the figure, is returned, then it is determined that the event was caused by a clamping action, and a fourth computing block Bl4 is implemented during an operation Bl3 45.

[0165] The operations Bl340 to Bl340 together constitute a B13 FA operation Analysis of the frequency content of the signals emitted by the sensors, based on which the actions of finger tapping (Hit(F)) and ring tapping (Hit(R)) are detected, and whether the user's action is a tightening action, the latter being characterized more precisely by block Bl4 described below. The principle employed is that the energy of a signal or a sum of signals is distributed more predominantly in the higher frequencies for the action Hit(R) than for the action Hit(F), and more predominantly in the lower frequencies for ring tightening actions than for the action Hit(F). Block Bl3 of [Fig. 13] illustrates a particular implementation of this principle, but other implementations may be considered by a person skilled in the art to implement this principle.

[0166] Fundamentally, this involves analyzing the energy distribution of the detected signals according to frequencies. For example, it would be possible to perform a Fourier transform on the signals and analyze the energy distribution in the frequency domain, for instance by comparing the energy contained in the signals for frequencies above 10 Hz with the energy contained in the same signals for frequencies below 10 Hz.

[0167] The fourth processing block, Bl4, illustrated in [Fig. 18], analyzes the signals generated by the sensors when it has been determined that the event detected by block BU is not caused by rotation, nor by either a ring tap or a finger tap, as determined in blocks Bl2 and Bl3. More specifically, block Bl4 characterizes the type of tightening applied to the ring by the user: short tightening, long tightening, double tightening, and the location of the tightening. To this end, the processor performs the operations of block Bl4 described in the Tightening, Finger Tap, and Ring Tap sections above.

[0168] Operations B14_i0, Bl4_i5, and BI4.20 of block Bl4 allow for the spatial characterization of the event being analyzed. Operation Bl4_i0 consists of calculating the ratio of the SI and S2 signals, then averaging this ratio during an averaging operation BI4.20 over the entire duration of the detected event. This average is compared to a pre-prepared calibration table that establishes a correspondence between averaged ratio values ​​and the location of a clamping force applied to the ring. The calibration table can be stored in MEM memory. Based on this comparison, a Loc signal representative of the location of the event being analyzed is generated.

[0169] Figure 16 illustrates the local location of the application of pressure (tightening, tap) by a finger Fl on a ring R equipped with sensors SENSi and SENS2. A table prepared in advance by calibrating the ring can indicate a correspondence between the ratio of the amplitudes of the signals generated respectively by the SENSi and SENS2 sensors and a Local location marked by a radius rLoc angularly offset with respect to a reference radius rref. The same table, or a similar table taking into account the two diametrically opposed contact areas existing during a squeeze between two fingers, can be used to locate the beginning and end of the squeeze illustrated by [Fig. 11] and thus evaluate the angular amplitude of the rotation applied to the ring during a rotational action. In general, any action (squeezing, rotation, tap) can be associated with a Local signal representing a Local location of the action considered, based on a ratio of amplitudes of two signals generated simultaneously by two sensors located on two distinct radii of the ring.

[0170] Operations Bl4 30 to Bl4 50 of block BI4 allow the type of clamping applied to the ring and which caused the generation of signals SI and S2 to be determined.

[0171] The BLno operation consists of detecting the number of peaks, that is, the number of amplitude maxima, in absolute value, for each of the signals SI and S2, and averaging the number of peaks detected. Alternatively, one can simply count the number of peaks detected for a signal. However, averaging over several signals makes the detection more robust.

[0172] Based on operation BI4.30, test operation Bl4 35 returns a positive response when the number of detected peaks is close to 4, for example, within a range of 3.5 to 4.5 encompassing four to account for the averaging effect, as indicated by Y in the figure. In response to a positive response to test operation Bl4 35, a detection signal D.Sq is generated, representative of a double clamping action (2 peaks per clamping action, therefore 4 peaks for a double clamping action, as explained in the "Double Clamping" section above).

[0173] Based on operation Bl4 30, test operation Bl4 40 returns a positive response when the number of detected peaks is 2, as indicated by Y in the figure. In response to this detection, the time interval AT separating the two detected peaks is determined during operation Bl4 45. During a test operation BI4 40, the time interval AT is compared to the predetermined duration threshold value THR4 as explained in the "Long Clamping" section above.

[0174] If it is determined that AT is greater than THR4, as indicated by Y in the figure, then a signal L.Sq representative of a long clamping action is generated.

[0175] If it is determined that AT is not greater than THR4, as indicated by N in the figure, then a signal S.Sq representative of a short clamping action, or clamping, is generated.

[0176] Blocks B11 and B12 directly process the SI and S2 signals generated by the sensors. Blocks B13 and B14 process the version of these signals recorded in MEM memory during step BI1.20.

[0177] One advantage of this algorithm is that it consumes only a minimum of energy, the parts of the ACQ acquisition module associated with blocks Bl2 and Bl3 of the algorithm being activated only when it has been determined that it is necessary to use their respective functionalities.

[0178] When averaging operations are mentioned in this description, it will be understood that, unless otherwise indicated, this refers to averaging over time, over a period that can be chosen by the user or the manufacturer of the human-interface. machine according to the desired behavior for this interface. However, these durations may be on the order of a few tenths of a second.

[0179] Also, when we talk about a signal generated by the sensors, it may be a converted or processed signal and not only one directly from the sensor, as long as it can be considered representative of a deformation applied to the ring, that is to say, as being able to be used to characterize this deformation.

[0180] Of course, a person skilled in the art will be able to develop similar algorithms to enable the user to benefit from all the types of action described in this description.

[0181] In addition, it is entirely possible to detect other actions than those described above, such as a three-finger press, a finger slide on the ring without rotating it ("swipe" in English terminology), a finger flexion, hand extension, fist clenching, rotation around the finger without moving the pressure areas applied to the ring, etc.

[0182] Furthermore, a person skilled in the art would understand that it is possible to implement algorithms other than those described here to exploit the signals generated by the sensors, including an approach called artificial intelligence, allowing the detected signals to be classified into different predefined categories by carrying out machine learning based on a database of signals measured during realistic actions by a user wearing the ring.

[0183] In response to the detection of events and thus to the generation of signals representative of particular actions applied to the ring, the computer can be configured to transmit these signals to the external device via the COM communication module in the form of an SCon control signal. Alternatively, the SCon control signal can be a signal of a command associated with the type of action applied to the ring.

[0184] II. Learning option for event detection

[0185] The preceding section presents a first option for the detection of actions applied to the ring, of an analytical type: each action applied to the ring causes a deformation of the latter, a deformation resulting in the generation of signals by the sensors, which can be analyzed in order to determine the action exerted and therefore the control signal to be generated.

[0186] A second option for determining the type of action exerted on the ring is based on the classification of the event generated by this action by a computer system that has benefited from automatic learning, often referred to as "machine learning", or "deep learning" in the case of a neural network, in English terminology.

[0187] This approach can, for example, rely on the use of a neural network trained to classify events detected by sensors into different categories defined by a user or an agent responsible for preparing and configuring the ring. Alternatively, methods other than neural networks could be used, such as the k-nearest neighbors algorithm (sometimes abbreviated as k-NN) or decision tree forests, respectively known as "k-nearest neighbors" and "random decision forest" in English terminology.

[0188] For illustrative purposes, this description will be based on the use of a neural network for processing data from sensors. For the use of neural networks for gesture detection, see the following references: (1) Nguyen-Trong, K., Vu, HN, Trung, NN, & Pham, C. (2021), Gesture Recognition Using Wearable Sensors With Bi-Long Short-Term Memory Convolutional Neural Networks. IEEE Sensors Journal, 21(13), 15065-15079; (2) Kim, M„ Cho, J., Lee, S., & Jung, Y. (2019), IMU Sensor-Based Hand Gesture Recognition for Human-Machine Interfaces. Sensors (Basel, Switzerland), 19(18), 3827; (3) Chuang, W.-C., Hwang, W.-J., Tai, T.-M., Huang, D.-R., & Jhang, Y.-J. (2019), Continuous Finger Gesture Recognition Based on Flex Sensors (Basel, Switzerland), 19(18), 3986.

[0189] Figures 20, 21, and 22 illustrate the implementation of such an approach based on the use of a neural network, for the specific case where three sensors, SENS1, SENS2, and SENS3, are used and deliver signals S1, S2, and S3, respectively. The approach consists of (i) detecting the occurrence of an event and (ii) classifying this event from a pre-prepared list of event classes. The classification is handled by the CALC computer in [Fig. 4], which includes a neural network that has been previously trained to classify events according to the profile of the signals delivered by the sensors.

[0190] IL 1 Detection and isolation of an event

[0191] Before even considering the classification of an event, considered here as resulting from a deliberate action by the user on the ring in order to generate a control signal from an external device, it is necessary to detect and isolate this event in a continuous stream of signals generated by the sensors in response, or not, to a deliberate action by the user.

[0192] Figure 20 is a graph representing the time evolution of the sum Var of variances over a sliding window of the three signals S1, S2, and S3 in the form of a curve: the x-axis represents time and the y-axis the magnitude of the sum of variances. In this case, the signals and the sum of their variances represent a deformation of the ring in response to A double tightening action, that is, two brief tightening actions of the ring performed close together in time, for example by the right hand of a user wearing the ring on their left hand. Each peak PI and P2 of the curve represents an individual tightening action.

[0193] The graph shows two variance threshold values: Vs and VE, which may be the same or different. Here, Vs is greater than VE, but Vs could be less than VE.

[0194] The first variance threshold Vs is used to determine the onset of the event: when the Var curve reaches and then exceeds the threshold value Vs, an event is considered to have already started. The time To indicates the moment when the curve reaches the threshold Vs. The event is considered to have started at a time Ts = T0 - dTs, where dTs represents a safety margin determined by the practitioner to include a portion of the curve with an amplitude smaller than the threshold value Vs, but which may be significant for classifying the event.

[0195] Symmetrically, the second variance threshold VE is used to determine the end of the event: when the Var curve falls below the threshold value VE, the event is considered to have ended. Time T2 indicates when the curve reaches the threshold VE. The event is considered to have ended at a time TE = T0 + dTE, provided that the curve remains below the threshold value for a duration dTE. dTE represents a safety margin determined by the practitioner to include a portion of the curve with an amplitude lower than the threshold value VE, but which may be significant for classifying the event. This margin also allows for the inclusion of two parts of a Var curve separated by a portion of the curve with an amplitude lower than the threshold VE within the same event.

[0196] Thus, in the case where the curve falls below the threshold value VE, at time T1 on [Fig. 20], for a duration less than the duration dTE, we consider that there is only one single event. The application here is to consider an event consisting of a double squeeze, with two peaks of variance amplitude close in time but separated by a brief period where the variance of the sum of the signals is low.

[0197] One advantage of choosing Vs > VE is to combine a good level of selectivity in the detection of an event, selectivity conditioned by a relatively high value of Vs, with consideration of the whole of the event, consideration favored by a relatively low value of VE.

[0198] Of course, this principle applies to any event including brief phases of ring release, during which no mechanical force is applied to the ring, or a mechanical force too weak for the curve of the sum of the variances of the signals to exceed the threshold VE.

[0199] From a practical standpoint, the sensor signals can be stored in a buffer, possibly a portion of the MEM memory, for a duration at least equal to dTs. Then, if an event is detected, the signals generated by the sensors from the beginning to the end of the event can be stored in the MEM memory for processing. The width of the variance calculation window, dTs and dTE, can take identical or distinct values ​​adjustable by the practitioner, for example, 50 ms.

[0200] By proceeding as explained above, the start and end times of an event are determined from the signals generated by the sensors.

[0201] Figure 21 represents a diagram summarizing the above explanations. The diagram represents a step S00 of user action applied to a ring R, triggering an event at the detection signals S1, S2, and S3 generated by the sensors integrated into the ring. The signal generation step S00 is followed by a step S10 of determining the start and end times of the event; this step includes steps S12, S14, S16, and S18.

[0202] At a step S12, the variances Varl, Var2 and Var3 of the three signals SI, S2 and S3 generated respectively by the three sensors Sensl, Sens2 and Sens3 are calculated on a sliding window.

[0203] At a step S14, the variances are summed to give the summed variance Var illustrated by [Fig.20].

[0204] At step S16, the computer determines that the summed variance Var exceeds a given threshold and generates a Trig command to trigger the recording of sensor signals in MEM memory. Data recording prior to the generation of the Trig command can be achieved by using a buffer that records data according to a sliding window. Step 16 also serves to wake up parts of the acquisition module from a standby phase, similarly to test step B1m5 described above in relation to the algorithm in [Fig. 15].

[0205] At a step S18, the calculator determines that the event has ended, and returns the start Start and end End times of that event.

[0206] II.2 Actions to be detected

[0207] Preparing a training program for a neural network involves listing in advance the classes in which this network will have the task of classifying the data submitted to it.

[0208] Table Tab. 1 below lists the types of action that may be of interest to identify from the recorded data of ring deformation events. Action Description Direction Variations Squeeze Squeeze the ring between two fingers Single, Double, Triple Short, Long 3 Fingers squeeze Squeeze the ring between three fingers Single, Double, Triple Short, Long Lat Squeeze Squeeze the ring between two fingers along its axis of rotation Single, Double, Triple Short, Long Rotate Rotate the ring Positive, Negative Short, Long, Small, Large Fast, Slow Touch Touch the ring worn with a finger Single, Double, Triple Short, Long Toc Tap a surface with a finger wearing the ring Single, Double, Triple Ring Toc Tap a surface directly with the ring Single, Double, Triple Slide Apply pressure to a point moving on the ring around the finger Positive, Negative Fast, Slow Surf Slide Slide the ring placed on a surface Right, Left Fast, Slow Up, Down Short, Long Swipe Slide a finger on the ring in a parallel direction on the finger wearing the right ring,Left hand closure. Close the hand with the ring on one finger. Single, double, triple finger only. Finger Snap, Finger Snap, Trash

[0209] Tab. 1

[0210] The “Squeeze” action SQ corresponds to a radial tightening of the ring between two fingers, as illustrated in (B) of [Fig.1].

[0211] The “3 Fingers Squeeze” action 3FS corresponds to a radial tightening of the ring between three fingers, as illustrated in (B) of [Fig.25].

[0212] The "Lat Squeeze" action LS corresponds to a squeezing of the ring R between two fingers Fl and F2 which apply forces F in a direction parallel to its axis of revolution, as illustrated in (C) of [Fig.25].

[0213] The action “Rot” ROT corresponds to a rotation of the ring around the finger wearing it, as illustrated in (C) of [Fig.1]. To determine the direction of rotation, it is considered that, when the ring is observed with the oriented axis Ax directed towards the observer, the trigonometric direction, that is to say the counterclockwise direction, is the positive direction of rotation.

[0214] The "Touch" action TCH corresponds to the pressure of a finger Fl of one hand of the user exerting a force F on the ring R worn around a finger of the other hand of the user, as illustrated in (A) of [Fig.25].

[0215] The "Toc" action corresponds to striking a rigid surface with a finger wearing the ring.

[0216] The "Ring Toc" action corresponds to tapping a rigid surface directly with the ring worn around a finger.

[0217] The “Slide” action SL corresponds to a radial centripetal force F applied at least partially towards the inside of the ring R at a point P moving on the periphery of the ring along a trajectory Traj so as to rotate around the finger F6 carrying the ring, as illustrated in (D) of [Fig.25].

[0218] The action “Surf Slide” SS corresponds to the sliding of the ring R placed on a rigid surface Surf, such as the top of a table, for example in four directions Left, Right, Up and Down parallel to the rigid surface, opposite in pairs, the Left and Right directions being normal to the Up and Down directions, as illustrated in (E) of [Fig.25].

[0219] The “Swipe” action SW corresponds to a friction of the ring along a direction parallel to the finger wearing the ring. This situation can be described as the displacement of a point of application of a centripetal force F to the ring and / or the finger wearing it along a trajectory Traj substantially parallel to the axis Ax of revolution of the ring, as illustrated in (F) of [Fig.25].

[0220] The "Hand closure" action corresponds to the formation of a fist with one hand, one finger of which is wearing the ring. Alternatively, only the finger wearing the ring can be bent. The ring is deformed by the expansion of the finger wearing it due to the latter's muscular contraction, which results in a centrifugal force acting from the inside of the ring outwards.

[0221] The "finger snap" action corresponds to a snap of the finger wearing the ring.

[0222] The "Trash" action corresponds to actions by the user not related to the will Execution of a command by the external device: this could be, for example, a gesture such as picking up a pen or a cup. These actions can be considered parasitic actions that should not be taken into account for the control of the external device, and the events they unintentionally generate can be classified as such so as not to cause the CALC computer to generate control signals.

[0223] Some of the actions listed above may exhibit characteristic directions or senses and / or variations.

[0224] The Slide and Rot actions, involving an angular displacement around the axis of revolution of the ring, can be applied in the trigonometric direction (positive direction) or in the opposite direction (negative direction).

[0225] The Rot action can have two variants: Small and Large, which are characterized by the angular amplitude α of the rotation imposed on the ring, illustrated in [Fig. 26]. For example, a rotation by an angle α within an angular interval between 20° and 40° can be considered a small-amplitude rotation, thus corresponding to the Small variant. Similarly, a rotation by an angle α greater than the upper limit (40°) of the angular interval, for example greater than 90°, can be considered a large-amplitude rotation, thus corresponding to the Large variant.

[0226] The Swipe and Surf Slide actions can be performed by a movement (of a finger applying pressure for slide or of the ring itself for Surf Slide) from left to right (right direction) or from right to left (left direction), from the user's point of view.

[0227] The Surf Slide action can be performed in an arbitrary direction, but the right and left directions described above and the directions towards Pavent (Up) and towards the back (Down) can be considered in the first place.

[0228] The Slide, Surf Slide and Rot actions can respectively exhibit higher or lower speeds of movement of a finger on Panel, of Panel on a surface, or of rotation, corresponding to relatively slow or relatively fast movements corresponding respectively to Slow and Fast variants.

[0229] Furthermore, with the exception of the intrinsically brief actions (Toc, Ring Toc, Swipe, Finger Snap), each of the actions in the table can have a relatively short or a relatively long duration. These durations can be illustrated, for example, by means of the variances of the corresponding events, as shown in graphs (A) and (C) of [Fig. 27] (see the explanations concerning the graph in [Fig. 20]). A short action corresponds to an applied force amplitude with a peak and will therefore generate only a peak in the variance Var, as in (A), whereas a long action will generate a variance that remains of relatively large amplitude over a certain period before falling close to zero at the end of the event, illustrated in (C) by a plateau.

[0230] Also, the actions Touch, Toc, Ring Toc, Squeeze, 3 Fingers Squeeze, Lat Squeeze, Click, and Hand Closure can be repeated during the same event (see the example in [Fig. 20], which represents a double action), thus creating variants of the basic event, as illustrated in (A), (B), and (D) of [Fig. 27] (see the explanations concerning the graph in [Fig. 20]). One, two, and three occurrences of a basic action correspond respectively to the Single, Double, and Triple variants of that action; for example, repeating the Toc action three times in rapid succession corresponds to the Triple variant of that Toc action.

[0231] The curves in Figures 20 and 26 are, of course, simplified curves intended to explain the principle. Curves obtained during actual measurements will generally have much less regular, or even very irregular, profiles.

[0232] Of course, each action, direction of action and variant of action can be associated with a respective control signal to an external device to be controlled by means of the human-machine interface constituted by the ring R. A high speed or a high amplitude of an action can be used to generate a control signal of the same type as the same action having a lower amplitude or speed, or a control of another type.

[0233] A first application example is navigation within a drop-down menu of a computer interface using rotation actions. A small-amplitude rotation (Rot) can move to an item in the menu immediately adjacent to or after the current item, depending on the direction of rotation, while a large-amplitude rotation will move to a more distant item in the menu.

[0234] A second application example, still concerning rotations, is that of listening to a piece of music in a playlist. A small rotation can be associated with volume control, while a large rotation can be associated with choosing which piece of the playlist to play: the next or preceding piece in the playlist, depending on the direction of the rotation.

[0235] Among the actions detailed above, some will tend to deform the annular element globally; this is particularly true of the so-called "global" actions of gripping, rotating, and closing a hand or finger wearing the ring. The detection of these actions by sensors distributed on the annular element will be particularly effective.

[0236] Other actions will tend to deform the annular element more locally, less globally than the actions listed in the preceding paragraph, such as the so-called "local" actions of touching, pressing, tapping, and sliding. However, the detection of these local actions remains possible, and this with the same sensor system as for global actions, which allows for greater integration and simplification of the sensor and the human-machine interface described in this document.

[0237] Of course, it remains possible to combine (i) the sensor assembly described here in order to detect global actions and (ii) contact sensors (piezoelectric, capacitive or other) or actuation buttons in order to generate control signals sensitive to global deformations of the ring, local actions on it, and / or combinations thereof.

[0238] II.3 Training the neural network

[0239] Training the neural network consists of providing it, during a learning phase, with a set of training data corresponding to events of identified classes. The classes are transmitted to the neural network along with the corresponding data. Based on the data and associations provided, the neural network then "learns" to recognize the events and associate each one with a given class, which corresponds to a user action on the ring. Ideally, each user action on the ring corresponds uniquely to a particular class of events that can be determined from the signals generated by the sensors and stored in memory.

[0240] Once the actions generating the events that the neural network must be able to process and classify have been defined, it is necessary to produce the corresponding training data. The actions can be chosen from Table 1 shown above. One solution for producing the training data for the chosen actions is to use the ring, applying these actions to it and recording the events generated in response.

[0241] Training data can thus be obtained by recording the signals generated by the sensors for, for example, 100 repetitions of a ring deformation event caused by a given action of the user, and performing this operation for each type of action to be identified.

[0242] Steps S00, S10, S20 and S30 of the diagram in [Fig.22] can be used to illustrate this process of producing training data.

[0243] At step S10, detailed above, the CALC computer determines the start and end of an event defined by the signals SI, S2 and S3 generated in response to an action by the ring user at step S00.

[0244] At step S20, the signals generated during the event are sampled and recorded in MEM memory during a Rec recording operation. It is preferable that the same number of measurements be associated with each detected event. If, for example, the measurement signals are sampled at a frequency of 100 Hz and the number of samples for each of the signals S1, S2, and S3 is set to 100 for a given event, the signals are recorded for a duration of one second. If the event lasts less than one second, the data is padded with zeros to maintain the number of 100 samples.

[0245] At a step S30, the recorded data are preferably normalized by a Cond conditioning implemented by the computing unit: the amplitude of each signal averaged over the duration of the event can be subtracted from each sample, and then the result can be divided by the variance of the signal during the event.

[0246] This data conditioning allows for datasets to be in the same format, facilitating their processing and classification using the neural network. In this example, each event is associated with a matrix of 3x100 samples.

[0247] The data conditioning can be adapted to the number of sensors, the use of the ring which can influence the length of the events to be recorded, the desired sampling rate, or the processing capacity of the computer integrated into the ring.

[0248] Steps S00 to S30 are repeated until event records are obtained in a number and diversity considered sufficient by the practitioner.

[0249] The recorded data here are the training data for the neural network: each data recording is associated with the corresponding action. Part of the data can be used for the actual training, and another part for validating the training; the latter will be referred to as validation data.

[0250] Training a neural network conventionally involves providing the neural network with training data and the expected results, including the class of each event in the training data. At the end of the training phase, the neural network's data processing is expected to be sufficiently reliable for use in real-world conditions. The level of reliability is assessed using validation data, which allows testing the network's performance.

[0251] The tables in Figures 23 and 24 illustrate the results of training a neural network. For each of these examples, a neural network was trained to classify events into a number of classes corresponding to given user actions. Following training, a test was performed, and the table allows for a comparison between the classifications made by the neural network and the known classes from the validation data. The training and validation data consisted of several dozen recordings of each action, collected from about ten different users.

[0252] The table in [Fig.23] shows the test results of the classification for the following actions: Touch, Toc, Squeeze, Slide_pos and Slide_neg (Slide actions in a positive direction and an opposite negative direction, respectively), Rot_pos and Rot_neg (Rot actions in the counter-clockwise direction and in the opposite direction, respectively), Surf_slide_rl and Surf_slide_lr (Surf Slide actions to the left and to the right, respectively), and Swipe_rl and Swipe_lr (Swipe actions to the left and to the right, respectively).

[0253] The table in [Fig.24] shows the test results of the classification for the following actions: Single_squeeze and Double_squeeze (occurrence and two occurrences of the Squeeze action during an event, respectively), Rot_neg_small and Rot_neg_large (Rot actions in the opposite direction to the counterclockwise direction for small and large amplitude rotations, respectively), Rot_pos_small and Rot_pos_large (Rot actions in the counterclockwise direction for small and large amplitude rotations, respectively), and Trash.

[0254] The rows of the tables correspond to the actual (known) classes of the validation data, and the columns to the classifications of this data by the neural network. In an ideal case, the main diagonal of the tables would contain only "1"s, and the other cells would contain only "0"s. Preliminary results show classification reliability of 86% and 91%, respectively, for the tables in Figures 23 and 24. These scores largely validate the relevance of this approach for determining user actions on the ring based on sensor measurement signals.

[0255] IL 4 Event classification by neural network

[0256] Once the neural network is trained, it can be used for a use practical human-machine interface in the form of a ring illustrated for example by [Fig.1].

[0257] The [Fig.22] is a diagram summarizing the control procedure of an external APP device by means of ring R.

[0258] Steps S00 to S30 proceed in the same way as for the production of training data in the previous section.

[0259] Following the data conditioning in step S30, the conditioned data associated with an event are provided as input to the CALC computer's neural network, which performs a Class classification of the event in step S40. This classification determines the type of action that triggered the event and generates a Sciass signal representative of the event's class and the type of action that triggered it. During this step, the neural network determines to which class the event belongs. Step S40 can be considered a detection step for a particular event, and the Sciass signal is also a detection signal for a given event and therefore for the action that triggered it.

[0260] At step S50, the CALC processor of the ACQ acquisition module generates a control signal SCon in response to the SCiass signal and thus to the data classification by the neural network. This operation can be performed, for example, by searching a table for a command associated with the event class contained in the control signal. Alternatively, the SCiass signal can be used as the SCon signal. The SCon signal can be sent to the external APP device via the communication module Com.

[0261] One advantage of using a neural network is that the user, provided they train the network, can adapt the interface's response to their own gestures and even add new detection classes to the "vocabulary" understood by the interface. Thus, unlike the analytical Option I, any type of action applied to the ring that leads to reproducible events in the deformation measurement signals of the ring element can be used to control the external device. A user could therefore "train" the ring R to recognize a given manipulation of the ring, regardless of whether or not this manipulation was intended by its designer.

[0262] III. Piezoelectric strain sensors

[0263] Strain sensors make it possible to estimate the surface deformation that an element of any mechanical system undergoes when it is subjected to external forces (force and moments of force that are applied to it by external elements).

[0264] A sensor particularly suited to the ring-shaped human-machine interface described above is a thin, single-crystal piezoelectric element in the form of a plate extending in a plane of extension defined by a first direction and a second direction normal to the first direction, with dimensions in the first and second directions each greater than 100 pm and a thickness less than 50 pm, a ratio of the thickness to the dimension in the first direction or the dimension in the second direction being less than at 0.1. The piezoelectric element can exhibit a first sensitivity Sx to deformation along the first direction and a second sensitivity Sy to deformation along the second direction, a crystalline orientation of the element being such that abs(Sy / Sx) <0.1, corresponding to a so-called "unidirectional" sensitivity, abs((Sy+Sx) / Sx) <0.1, corresponding to a so-called "bidirectional" sensitivity, or for at least two first directions of the extension plane making an angle between them between 30° and 60°, abs((Sx-Sy) / Sx) <0.1, corresponding to a so-called "omnidirectional" sensitivity.

[0265] Such a thin piezoelectric element is suitable for forming the basis of a passive strain sensor, which can combine accuracy, sensitivity, conformability, flexibility, lightness, stability, linearity, directivity and applicability to wide ranges of deformations, as described in detail in French patent application FR2303635.

[0266] Thus, this piezoelectric element makes it possible to measure deformations greater than 5000 micrometers per meter with a resolution on the order of 1 nanometer of deformation per meter. These figures can be compared with those of conventional resistive gauges, which can measure deformations up to 12000 micrometers per meter but with a much lower resolution, on the order of 1 micrometer of deformation per meter, or with those of piezoelectric gauges in a housing, which are only capable of measuring deformations limited to about 300 micrometers per meter with a resolution of 1 nanometer per meter.

[0267] This piezoelectric thin element can be provided with a pair of electrically conductive layers located respectively on two opposite faces of the piezoelectric thin element.

[0268] A strain sensor may include at least one thin piezoelectric element as described above, located on a flexible sheet.

[0269] According to additional features, considered individually or according to any technically feasible combination:

[0270] - at least one thin piezoelectric element can be encapsulated between the foil flexible and another flexible sheet;

[0271] - the sensor may include at least one charge amplifier connected to the minus a thin piezoelectric element;

[0272] - at least one charge amplifier can be integrated on the flexible sheet;

[0273] - the sensor may comprise a plurality of thin piezoelectric elements as described above, oriented in different directions with at least 30° of difference between them;

[0274] - the sensor may comprise a first, a second and a third thin element piezoelectrics, each exhibiting the characteristic abs(Sy / Sx) < 0.1, the first direction of the second piezoelectric thin element can make an angle of 90° with the first direction of the first piezoelectric thin element, the first direction of the third piezoelectric thin element can make an angle of 45° with the first direction of the first piezoelectric thin element;

[0275] - the sensor may comprise a first, a second and a third thin element piezoelectric, each of which can exhibit the characteristic abs(Sy / Sx) < 0.1, the first direction of the second piezoelectric thin element being able to make an angle of 120° with the first direction of the first piezoelectric thin element, the first direction of the third piezoelectric thin element being able to make an angle of 240° with the first direction of the first piezoelectric thin element;

[0276] - the sensor may comprise a plurality of charge amplifiers each connected to one of the respective thin piezoelectric elements; and

[0277] - a sensor can combine at least two sensors according to the invention connected electrically in parallel.

[0278] Each of the sensors integrated into the ring R, such as the Sens1, Sens2 and Sens3 sensors, can consist of a sensor combining one or more of the characteristics listed above.

[0279] Indeed, the ring-shaped human-machine interface, equipped with sensors sensitive to the deformation of the annular element as a whole, can advantageously employ deformation sensors whose sensing element is the thin, single-crystal piezoelectric element described above. Its advantages are numerous; in particular, one can mention the sensitivity and, if desired, the directivity of a sensor based on such an element, but also the precision, conformability, flexibility, lightness, integrability and small dimensions, stability, linearity, directivity, and applicability to a wide range of deformations.

[0280] The ring R will particularly benefit from the integrability and sensitivity of the thin piezoelectric element. The sensitivity allows, for example, the annular element ANN to be formed in materials and dimensions conventional for manufacturing rings: lower sensitivities would necessitate the use of more deformable and / or thinner materials, reducing their robustness and wearing comfort.

[0281] Indeed, a classic ring, made of metal, is rigid and deforms little under the action of forces applied by fingers, (usually of amplitudes less than 10 N. Such forces generate deformations on the order of 5 pdef, which are difficult to measure using conventional strain gauges.

[0282] Resistive gauges, for example, have insufficient sensitivity for the application intended here. Even if they were mounted on an annular element flexible enough to measure user-induced deformations, their Their power consumption would disqualify them for the intended application. Furthermore, reducing the noise of the gauges by filtering to retain only the low frequencies of the measurement signals would introduce a prohibitive delay for a control interface.

[0283] It is preferable to consider sensors with low energy consumption, or even passive sensors. Piezoelectric sensors meet this criterion.

[0284] A first type of piezoelectric strain sensors is based on the use of a crystal of piezoelectric material with relatively good sensitivity to deformation and stable over time, but thick and rigid, often housed in a metal casing to which it is mechanically attached, these latter characteristics making it difficult to integrate into a structure having the dimensions of the ring R and poorly suited to installation on a curved surface such as that of the annular element ANN.

[0285] A second type of piezoelectric strain sensors is based on the use of composite structures comprising PZT (or lead titanium zirconate) bars located between sheets of polymer materials, or on polymer piezoelectric films called "PVDF" for poly(vinylidene fluoride) in English terminology.

[0286] These structures are relatively flexible, but too unstable over time and sensitive to temperature for an application such as those targeted for the R ring, intended among other things to be worn around a user's finger.

[0287] In contrast, the thin, single-crystal piezoelectric element described above meets all the criteria necessary for good integration into the ring and for good functionality of the ring: dimensions, flexibility, and sensitivity. The integrability of the piezoelectric element (ease of placing it in close contact with a rounded structure such as the ring element) is paramount here, and its sensitivity makes it possible to detect deformations even when simply applied to the finger of a ring element with a rigidity comparable to a traditional metal ring.

[0288] The piezoelectric element can be formed from lithium tantalate LiTaO3 in single-crystal form, which belongs to space group 3m, but also from lithium niobate LiNbO3 (group 3m), lead and magnesium niobate MgNb2(PbO3)3 (group PI), aluminium nitride AIN (group P63mc), barium titanate BaTiO3, potassium niobate KNbO3 or lead titanate TiPbO3 (all three of group P4mm). Sensors

[0289] Figure 29 illustrates in (A) a cross-sectional view of a SENS piezoelectric sensor based on a PIEZO thin piezoelectric element of extension plane chosen to present A particular behavior—unidirectional, omnidirectional, or bidirectional—responds to a unidirectional deformation applied to its plane of extension. The sensitivity behavior of the piezoelectric thin element (unidirectional, omnidirectional, or bidirectional) is transferred to the sensor incorporating this piezoelectric thin element. Such sensors may be particularly well-suited to specific situations, as illustrated by Figures 10 to 12 discussed below, but they can also be used in more general applications, as will become apparent later.

[0290] To take advantage of the thinness, and therefore the flexibility and conformability, of the PIEZO thin element, the SENS sensor comprises a flexible SH1 sheet to which the PIEZO thin element is attached. The sheets are preferably made of flexible materials chosen according to the intended application and may, for example, be made of metal, polyvinyl chloride (PVC), polyimide (PI), polyethylene terephthalate (PET), biaxially oriented polyethylene terephthalate (Mylar®), or a composite material of epoxy resin and glass fibers. The PIEZO thin element can be attached to the SH1 sheet using a flexible adhesive such as an anisotropic conductive film (ACF), which also allows for electrical contact as described in patent document FR 3 122 985. In use, the SENS sensor can be attached to a surface to be characterized using an adhesive, for example, cyanoacrylate glue or epoxy resin.

[0291] In addition to the thin PIEZO element, in the example of [Fig. 9], a charge amplifier C.AMP is also fixed to the SH1 sheet and functionally connected to two conductive layers ELI and EL2 acting as electrodes, formed respectively on two opposite faces of the thin PIEZO element. The charge amplifier's function is to produce a voltage corresponding to the input load, which corresponds to the charge generated by the PIEZO element during its deformation, for the purpose of electronically processing the generated electrical potential and effectively measuring the deformation of the PIEZO element. Although not shown, a wired connection element, such as a ribbon cable, is connected to the charge amplifier to connect the sensor to an external measuring device.

[0292] Fig. 9 illustrates in (B) a sensor similar to that of the configuration illustrated in (A), but also comprising a second sheet SH2, which may be of the same nature as the sheet SH1, which allows the thin PIEZO element and C.AMP charge amplifier to be encapsulated by sandwiching between the sheets SH1 and SH2.

[0293] The thin piezoelectric element PIEZO preferably has a thickness less than 50 pm, more preferably less than 25 pm, and even more preferably less than 10 pm. If we consider a piezoelectric element defined as illustrated in [Fig. 3], a ratio of the thickness of the PIEZO element to its dimension Lx in a first direction of its extension plane, and / or of a dimension Ly in a second dimension of its extension plane normal to the direction Lx, is less than 0.1, preferably less than 0.05, more preferably less than 0.01.

[0294] The SH1 sheet and, where applicable, the SH2 sheet, may have a thickness of between 5 and 300 pm.

[0295] However, it is preferable that the SENS piezoelectric sensor, considered as a whole, be sufficiently flexible to conform to the surface of the annular element to which it is to be attached and capable of following its deformations. The practitioner can decide, for each application, on the characteristics of the thin piezoelectric element PIEZO, its support, and other elements such as the electrode layers or the means for making electrical contacts.

[0296] The flexibility of the thin piezoelectric element is advantageously exploited to fix one of its faces in close contact with the curved surface of the annular element ANN (via an electrode and optionally an adhesive film). In this way, the piezoelectric element is bonded to the annular element, undergoes the same deformation as the annular element where the piezoelectric element is fixed, and its deformation is therefore representative of that of the annular element.

[0297] The SH1 sheet can be made of, or replaced by, a flexible substrate such as a flexible printed circuit board, known as a "flex PCB," composed of layers of electrically insulating polymer and copper layers, allowing signals to be routed between the different components of an electronic circuit. Each sensor assembly can include a plurality of piezoelectric elements, each with its own substrate, as can the electronic module controlling them. Alternatively, a single flex PCB substrate can accommodate all the piezo elements and the electronic module controlling these elements. Alternatively still, a first flex PCB substrate can be common to all the piezo elements and a second flex PCB substrate can be dedicated to the electronic control module.

[0298] The exact type of sensor(s) to be integrated into the annular element ANN will depend on the type of action to be detected and the constraints imposed (autonomy, sensitivity...), according to the intentions of the designers.

[0299] The invention is not limited to the embodiment described above and alternative embodiments may be made without departing from the scope of the invention as defined by the claims.

Claims

Demands

1. A sensor assembly (R) in the form of a ring, comprising an annular element (ANN) of substantially cylindrical shape and at least two strain sensors (SENS1, SENS2, SENS3) mounted on the annular element at levels of at least two radii (rl, r2, r3) angularly distinct from it, respectively, so as to be sensitive to a strain of the annular element, the assembly being characterized in that the at least two strain sensors (SENS1, SENS2, SENS3) comprise at least one single-crystal piezoelectric thin element (PIEZO) in the form of a plate extending in an extension plane (xy) defined by a first direction (x) and a second direction (y) normal to the first direction, of dimensions (Lx, LY) in the first and second directions each greater than 100 pm and of thickness (Lz) less than 50 pm,a ratio of the thickness to the dimension (Lx) in the first direction or the dimension (Ly) in the second direction being less than 0.

1.

2. Sensor assembly (R) according to claim 1, wherein the annular element (ANN) has a monolithic structure.

3. Sensor assembly (R) according to claim 2, wherein the annular element (ANN) is formed of metal.

4. Sensor assembly (R) according to any one of claims 1 to 3, wherein the at least two distinct rays (rl, r2, r3) are angularly spaced from each other by an angle between 20° and 160°.

5. Human-machine interface (R) comprising the sensor assembly according to any one of claims 1 to 4, further comprising an electronic module (EL) comprising an acquisition module (ACQ) configured to generate a detection signal (ROT-, ROT+, Loc, S.Sq, L.Sq, D.Sq, Hit(F), Hit(R), SCiass) in response to a deformation of the annular element on the basis of signals (SI, S2, S3) emitted by at least two deformation sensors (SENSi, SENS2, SENS3), the human-machine interface being configured to generate control signals (SCon) from a device (APP) external to the human-machine interface (R) in response to the detection signal.

6. Human-machine interface (R) according to claim 5, configured to keep awake only a first part of the acquisition module (ACQ) associated with the detection (B1m5) and to activate second parts of the acquisition module only when it is determined that respective activations of these second parts are necessary for an analysis of the signals (SI, S2) emitted by the at least two strain sensors (SENSi, SENS2, SENS3).

7. Human-machine interface (R) according to claim 5 or 6, wherein the electronic module (EL) includes a computer (CALC) configured to (i) classify (S40) events represented by signals (SI, S2, S3) emitted by at least two strain sensors (SENSi, SENS2, SENS3) and (ii) generate (S50) control signals (SCon) from a device (APP) external to the human-machine interface (R) in response to this classification.

8. Human-machine interface (R) according to any one of claims 5 to 7, wherein the control signals are representative of at least one action (S00) selected from a clamping (SQ, 3SQ), a short clamping or a long clamping of the ring between two fingers (F1, F2) or between three fingers (F1, F2, F3), repeated or not, in a centripetal direction to the ring (R); a clamping (LS) of the ring, repeated or not, parallel to an axis (Ax) of revolution of the ring; a closing of a hand with one finger wearing the ring or the closing of a finger wearing the ring; and a rotation (ROT) of the ring between two fingers (F1, F2) of a first amplitude (a) belonging to an angular interval or of a second amplitude greater than an upper bound of the angular interval.

9. Human-machine interface (R) according to any one of claims 5 to 7, wherein the control signals (SCon) are representative of at least one action (S00) selected from a touch (TCH) of the ring by a finger (F1); a tap of a surface by a finger wearing the ring; a tap of a surface directly with the ring; the application (SL) of pressure on the ring at a point moving on the ring around a finger (F6) wearing the ring; a slide (SS) of the ring on a surface (Surf); a slide of a finger of the first hand on the ring in a direction parallel to a finger of the second hand wearing the ring; a location of any of the previous actions; and a snap of a finger wearing the ring.

10. Human-machine interface (R) according to any one of claims 5 to 9, wherein the acquisition module is configured to: - detect (Bl2 35, BI2-45, S40) a rotation (Rot, Rot_pos, Rot_neg, Rot_pos_small, Rot_pos_large, Rot_neg_small, Rot_neg_large) of the annular element on the basis of the signals (SI, S2) emitted by the at least two strain sensors (SENSi, SENS2, SENS3); and - in response to the detection step, generate a signal (ROT-, ROT+, Sciass) representative of the occurrence of a rotation of the annular element (ANN).

11. Human-machine interface (R) according to claim 10, wherein the signal (ROT-, ROT+) is representative of a rotation having exceeded an incremental angular value (-ThrAng, ThrAng).

12. Human-machine interface (R) according to any one of claims 5 to 11, further comprising a wireless transmission module (COM) configured to transmit signals generated by the acquisition module (ACQ) to the external device (APP) and a power supply (BAT) for the wireless transmission module (COM) and the acquisition module (ACQ).

13. Human-machine interface (R) according to any one of claims 5 to 12, further comprising a feedback device (FB) configured to signal to a user the generation of a detection signal (ROT-, ROT+, Hit(R), Hit(R), Loc, S.Sq, L.Sq, D.Sq, SCiass) by the acquisition module (ACQ).

14. Human-machine interface (R) according to any one of claims 5 to 13, configured to be threaded onto a part (F6) of a user or a rod (RD) of an external device control device (APP).

15. Kit comprising a human-machine interface (R) according to any one of claims 5 to 14 and an external device (APP) configured to be controlled by means of the human-machine interface (R).

Images