METHOD FOR CONTROLLING AN EXTREMITY OF A VIRTUAL AVATAR BY MEANS OF MYOELECTRIC ACTIVITIES OF AN EXTREMITY OF A PERSON AND SYSTEM FOR THIS PURPOSE

Description

[0001] The present invention relates to a method of controlling a limb of a virtual avatar by the myoelectric activities of a limb of a subject.

[0002] The present invention is aimed in particular at the population of people who have suffered amputations, particularly upper limb amputations.

[0003] Some amputees have a myoelectric prosthesis, which is a contact prosthesis whose movements are controlled using electrical signals from the user's muscles.

[0004] One of the main problems encountered by patients wearing myoelectric prostheses is the cognitive load associated with the control and the permanent visual control required for its use.

[0005] The myoelectric control required to control prosthetics is based on the activation of two antagonistic muscles and the quality of the dissociation of the control of these muscles. Such control is time-consuming and requires a lot of training.

[0006] The action is made all the more difficult because patients have no information about the prosthesis's state other than that provided by visual feedback. They must remain focused on an object to be reached and monitor the action the prosthesis is performing. This is extremely costly in terms of cognitive resources, such as concentration and muscle control. Because of this, many patients abandon their prostheses, which become what are commonly referred to as "closet prostheses."

[0007] For these various reasons, prosthetics are currently underused in relation to their capabilities.

[0008] AARON TABOR et al.'s article, "Designing Game-Based Myoelectric Prosthesis Training," in Design and Cognitive Impairment, 2017, discusses a game developed to train amputees to use their myoelectric prosthesis. This paper describes the development of a visual interface ("The Falling of Momo"). The game character is controlled using an electrode wristband placed on the amputated limb, using proportional control identical to that used for myoelectric prosthetics. The intensity of the muscle contraction perceived by the electrodes is proportional to the movement of the game character, providing visual feedback to the user.

[0009] US 2014 / 198034 discloses an electrode wristband for use with glasses with an integrated screen that operates by gesture recognition. Muscle contractions perceived by the electrodes during predetermined gestures, for example extending the index finger and flexing or extending the wrist, allow actions to be performed on the glasses screen, for example scrolling through a drop-down menu. Gesture recognition can be confirmed by haptic feedback, by applying a discrete all-or-nothing signal at the end of the gesture.

[0010] The article by GAURAVKUMAR K PATEL et al: "Context-dependent adaptation improves robustness of myoelectric control for upper-limb prostheses" in Journal of Neural Engineering vol.14, no.5 (2017), describes a protocol to improve the control of a myoelectric forearm prosthesis using an electrode bracelet and according to a learning model. The signals obtained by the electrodes are processed to obtain the envelope. The envelopes of the muscle activity signals are used in a gesture recognition algorithm. The output signal of this algorithm is an estimate of the normalized activation level associated with a function of the prosthesis, for example, hand opening or wrist rotation. Based on this output signal, an activity threshold and a gain are calculated to define the control-command information sent to the prosthesis to set it in motion.The activity threshold is modified to avoid parasitic activities of other functions. The gain is set by a function.

[0011] An aim of the present invention is to propose a method for improving the learning of muscle dissociation and the fine work of muscle contractions.

[0012] To this end, the present invention relates to a method for controlling a limb of a virtual avatar by the myoelectric activities of a limb of a subject as defined in claim 1.

[0013] Advantageous aspects of the invention are defined in dependent claims 2 to 5.

[0014] The invention also relates to a system suitable for implementing the method of controlling a limb of a virtual avatar by the myoelectric activities of a limb of a subject as defined in claim 6.

[0015] Advantageous aspects of the invention are defined in dependent claims 7 to 11.

[0016] These characteristics and advantages of the invention will appear on reading the description which follows, given solely as a non-limiting example, and made with reference to the appended drawings, in which: there figure 1 is a schematic view of a system according to one embodiment of the invention; the figure 2 is a schematic view of a subject wearing a myoelectric prosthesis; and the figure 3 is a graph representing the distribution of deviations from the target for each movement during trials of back-and-forth movements between two targets using a virtual avatar for three types of sensory feedback and a configuration in the absence of any sensory feedback in non-amputee subjects.

[0017] On the figure 1, a system 1 suitable for implementing the method according to the invention is shown.

[0018] The system 1 comprises a device 4 for measuring myoelectric activity signals. The myoelectric activity signals come from electrical impulses transmitted by nerve fibers to a muscle and causing the contraction of this muscle.

[0019] The measuring device 4 comprises a plurality of electrodes 6. The electrodes 6 are for example three-phase surface electrodes adapted to be placed on a limb 8 of a subject 10 and to receive the myoelectric activity signals from the muscles of the limb 8 of this subject 10.

[0020] The electrodes 6 are typically configured to acquire myoelectric activity signals.

[0021] In particular, the electrodes 6 are configured to acquire a first and a second raw calibration myoelectric activity signal resulting from contractions of varying intensities of a first and a second antagonistic muscles, for example the biceps and the triceps.

[0022] The electrodes 6 are typically further configured to acquire a first and a second raw myoelectric activity signal resulting from contractions of the first and second antagonist muscles of the limb 8 of the subject 10, for a determined duration.

[0023] In particular, the device 4 for measuring myoelectric activity signals comprises eight electrodes 6.

[0024] The acquisition frequency of the electrodes 6 is for example 200 Hz. Such a frequency is particularly suitable for obtaining the signal envelope.

[0025] However, other frequency values ​​are possible, for example between 100 Hz and 2000 Hz in order to obtain more precise signals.

[0026] The measuring device 4 is typically in the form of a flexible bracelet. Such a shape makes it possible to place the measuring device 4 around a limb 8, the limb 8 being for example an arm, the residual part of a forearm or stump, or an arm. The electrodes 6 are then arranged in a circle and towards the inside of the circle.

[0027] The electrodes 6 are connected together by any suitable material to form a bracelet. The bracelet is, for example, stretchable. According to one variant, the bracelet is available in several sizes adapted to different morphologies and to adapt to the upper limb as well as to the lower limb. The bracelet may further comprise a closing device, for example a self-gripping system, a buckle, snap fasteners or any other possible option.

[0028] Other forms of the measuring device 4 are conceivable, for example the electrodes are not connected to each other, or the measuring device 4 is integrated into a myoelectric prosthesis 11 as visible in the figure 2 A myoelectric prosthesis is a contact prosthesis that operates using myoelectric activity signals and is controlled by its user.

[0029] The system 1 also includes a remote communication device 12.

[0030] The remote communication device 12 is typically a radio communication device.

[0031] For example, the remote communication device 12 is a Bluetooth or Wi-Fi communication device.

[0032] The remote communication device 12 is configured to transmit the myoelectric activities recorded by the electrodes 6 to a device 14 for processing the myoelectric activities.

[0033] According to an advantageous embodiment, the system 1 also comprises a sensory feedback device 16.

[0034] The sensory feedback device 16 comprises, for example, a plurality of vibrating elements 18.

[0035] The sensory feedback device 16 is typically in the form of a wristband. For example, the sensory feedback device 16 is in the form of an elastic wristband. Preferably, the sensory feedback device 16 and the measuring device 4 are integral.

[0036] For example, the sensory feedback device 16 comprises sixteen vibrating elements 18. Alternatively, the sensory feedback device 16 comprises a different number of vibrating elements 18, for example six vibrating elements.

[0037] The sensory feedback device 16 makes it possible to provide sensory feedback to the subject 10 following a muscular contraction resulting in the detection of myoelectric activity by the electrodes 6. Its operation will be explained later.

[0038] The processing device 14 comprises at least one processor. The processing device 14 further comprises a memory capable of recording the myoelectric activities recorded.

[0039] The processing device 14 shown in the figure 1 includes three computers 20, 21, 22.

[0040] Preferably, the processing device 14 comprises at least one display device 23.

[0041] On the figure 1 , each computer 20, 21, 22 comprises a display device 23. The display device 23 is typically a screen.

[0042] The three computers 20, 21, 22 are configured to communicate with each other, for example using a switch configured to interact in a local network.

[0043] The myoelectric activity signal measuring device 4 is configured to communicate remotely with the processing device 14. In the example of the figure 1, the measuring device 4 is configured to communicate remotely with a first computer 20.

[0044] The first master computer 20 is configured to receive the raw myoelectric activity signals acquired by the electrodes 6 and transmitted by the remote communication device 12, process said raw myoelectric activity signals and transmit the processed myoelectric activity signals, in particular to a second computer 21 and to a third computer 22.

[0045] Advantageously, the first computer 20 is configured to identify two electrodes 6 of the measuring device 4 allowing maximum acquisition of the myoelectric activity signals of each of the two antagonist muscles.

[0046] The first computer 20 is configured to determine minimum and maximum myoelectric activities for each of the two antagonist muscles from an acquired raw calibration myoelectric activity signal.

[0047] To this end, the first computer 20 is also adapted to set the raw calibration myoelectric activity signal acquired by the electrodes 6 to a zero average value.

[0048] The first computer 20 is further configured to determine the absolute value of the raw calibration myoelectric activity signal set to zero mean value.

[0049] The first computer 20 is configured to filter the signal in absolute value to obtain the envelope.

[0050] For example, the first computer 20 includes a second-order low-pass Butterworth filter with a cutoff frequency of 1.5 Hz configured to filter the resulting signal to obtain its envelope.

[0051] The first computer 20 is adapted to determine a myoelectric activity threshold, called EMGx offset, below which the contraction of the muscle is not necessarily voluntary and must not be taken into consideration in order to avoid disturbing a movement produced.

[0052] The first computer 20 is configured to take into account only the values ​​recorded between the determined threshold and the maximum value of the envelope of the filtered signal.

[0053] The first computer 20 is configured to normalize the envelope values ​​of the filtered signal.

[0054] The first computer 20 is typically configured to assign the value 0 to the previously calculated myoelectric activity threshold, and to assign the value 1 to the maximum value of the envelope of the filtered signal.

[0055] In particular, the first computer 20 is configured to determine a coefficient for converting the normalized myoelectric activity into a component of movement speed of a limb of a virtual avatar 24.

[0056] For example, the conversion coefficient is determined as follows for the avatar's arm movement. Typically, the very rapid rotation of the elbow is 4 rad / s. This value is taken as a starting point, and subject 10 gives his feeling regarding this speed. The conversion coefficient is increased or decreased so that the speed feeling is consistent with the muscular activity of subject 10. This coefficient is specific to each muscle.

[0057] The first computer 20 is typically configured to assign, for each muscle, a zero velocity component when the normalized value assigned to the value of the envelope of the filtered signal is 0, and to assign a maximum velocity component when the value assigned to the normalized value of the envelope of the filtered signal is 1.

[0058] In particular, the first computer 20 is capable of attributing, for each muscle, to each normalized value of the envelope of the filtered signal, a speed component between the zero speed component and the maximum speed component.

[0059] The first computer 20 is further configured to convert the normalized myoelectric activity into a movement speed component of the limb of the virtual avatar 24 by applying the conversion coefficient.

[0060] The first computer 20 is configured to determine a speed and a direction of movement of the limb of the virtual avatar 24 displayed by the second computer 21 by subtracting the speed components obtained for each of the muscles at a given instant. For example, in the case of the rotation of an elbow, the subtraction of the speed values ​​makes it possible to determine a flexion or extension speed of an elbow of a virtual avatar 24 displayed by the display device 23 of the second computer 21.

[0061] The display device 23 of the second computer 21 is configured to display the virtual avatar 24 representing at least one limb corresponding to the movement which is to be controlled by the myoelectric activities. This may be the arm or a leg, a whole body or a half-body depending on the chosen application.

[0062] The cessation of muscle contractions, for example, causes the movement to stop, the positioning of the elbow of the virtual avatar 24 is then in a fixed position which corresponds to the last position reached.

[0063] According to variants, the calculated speed is a hand opening or closing speed of a virtual avatar, or a wrist rotation speed of a virtual avatar, or a knee or ankle flexion / extension speed of a virtual avatar, or any other conceivable option.

[0064] The first computer 20 is configured to transmit a movement command to the second computer 21.

[0065] Motion control typically includes determining the speed and direction of movement.

[0066] Preferably, the first computer 20 is configured to determine a sensory parameter of the member of the virtual avatar 24 to be associated with at least one vibrating element 18 and to be transcribed into vibrations.

[0067] The sensory parameter is for example chosen from a flexion angle of an elbow of the virtual avatar 24, an anterior elevation angle of a shoulder of the virtual avatar 24, an abduction / adduction angle of the shoulder of the virtual avatar 24, a medial or lateral rotation angle of the shoulder of the virtual avatar 24, a flexion angle of a wrist of the virtual avatar 24, a radial or ulnar rotation angle of the virtual avatar 24, a degree of opening of a hand of the virtual avatar 24, a degree of opening of a digital pincer grip involving the thumb and one or more other fingers of the hand of the virtual avatar 24, a gripping force of a hand of the virtual avatar 24, a flexion angle of a knee of the virtual avatar 24, a flexion angle of the ankle of the virtual avatar 24, a speed of rotation of one of the movements mentioned in this list, or any other sensory modality capable of being substituted by vibrations,such as heat or the configuration of a myoelectric prosthesis 11 of the subject 10.,

[0068] The configuration of the myoelectric prosthesis 11 corresponds to the choice of one of the modes of the myoelectric prosthesis 11 in use to know on which joint the muscular activation will act. For example, if the prosthesis is in a mode called "wrist pronation-supination", the vibrations will signal the wrist pronation-supination movement. Similarly, in a mode called "elbow", the vibrations will signal the elbow movements.

[0069] The second computer 21 is dedicated to the visual interface with subject 10.

[0070] According to a variant not shown, the second computer is directly integrated into the first computer 20. This makes it possible to reduce the number of interfaces.

[0071] The second computer 21 is configured to receive the movement command transmitted by the first computer 20.

[0072] In an advantageous embodiment, the third computer 22 is configured to activate the sensory feedback device 16.

[0073] According to a variant not shown, the third computer is directly integrated into the electrode bracelet 6 by means of a microcontroller and a pulse width modulation control system.

[0074] The first computer 20 is configured to send to the third computer 22 the myoelectric activities processed and converted into an instantaneous value of the sensory parameter by applying the conversion coefficient.

[0075] The third computer 22 is typically configured to determine ranges of values ​​of the sensory parameter to be associated with at least one vibrating element 18.

[0076] The third computer 22 is typically configured to activate at least one element 18 corresponding to a previously determined range of values ​​of the sensory parameter.

[0077] The third computer 22 comprises, for example, a reading program configured to establish the correspondence between the instantaneous value of the sensory parameter and the activation of the corresponding vibrating element 18.

[0078] The third computer 22 is then configured to activate the corresponding vibrating element 18.

[0079] The sensory feedback device 16 is connected to the third computer 22, for example using cables.

[0080] In the variant where the third computer 22 is directly integrated into the electrode bracelet 6, the electrodes 6, the sensory feedback device 16 and the third computer 22 are grouped together.

[0081] Preferably, the third computer 22 is configured to activate the vibrating elements 18 selectively and discreetly.

[0082] For example, the third computer 22 is capable of determining the number and position of the vibrating elements 18 to be activated according to ranges of values ​​of a chosen sensory parameter.

[0083] Preferably, in each range of values ​​of the sensory parameter, only one determined vibrating element 18 is activated.

[0084] More preferably, a single vibrating element 18 is assigned to a range of values ​​of the sensory parameter for each range of values.

[0085] According to the example where the sensory parameter is the elbow flexion angle, when the subject 10 contracts the biceps and / or the triceps, the electrodes 6 are configured to acquire a first and a second raw myoelectric activity signal resulting from contractions of the biceps and the triceps, respectively;for a determined duration, the remote communication device 12 is configured to transmit the first and second raw myoelectric activity signals to the first computer 20, the first computer 20 is configured to process the first and second raw myoelectric activity signals and to transmit to the second computer 21 a command to move the arm of the virtual avatar 24 and to the third computer 22 an instantaneous position of the arm of the virtual avatar 24, the second computer 21 is configured to display the movement of the arm of the virtual avatar 24 and the third computer 22 is configured to activate each vibrating element 18 in steps of 20° of elbow flexion angle. It is understood that the step may have a value other than 20°, for example 10°, 15°, 25°, 30°.;

[0086] Preferably, the third computer 22 is configured to determine the duration and sequence of the vibrations.

[0087] For example, the third computer 22 is configured to activate a determined vibrating element 18 for a duration of 100 ms alternating with a stop for a duration of 100 ms.

[0088] Preferably, the vibration intensity is adaptable to improve the comfort of the subject 10.

[0089] Alternatively, the processing device 14 comprises a single computer, or a smartphone, a television or a tablet, or any conceivable combination of these elements. For example, all of the functions of the first, second and third computers described above are capable of being implemented by this single computer.

[0090] According to another variant, the processing device is at least partially embedded in the measuring device. For example, the embedded processing device part is configured to implement all of the functions of the first and second computers described above.

[0091] A method for controlling a limb of a virtual avatar by the myoelectric activities of a limb of a subject will now be described. The method is implemented by system 1.

[0092] The device 4 for measuring myoelectric activity signals is placed on a limb 8 of the subject 10 so that the electrodes 6 are in contact with a first and a second antagonistic muscle of this limb.

[0093] For example, subject 10 is a person with a limb amputation. Subject 10, for example, has an arm amputation and has a transhumeral amputation, i.e., above the elbow. The device 4 for measuring myoelectric activity signals is placed around the amputated arm of subject 10 so that the electrodes 6 are in contact with the biceps and triceps, two antagonistic muscles.

[0094] Alternatively, the subject does not have an amputation and the device for measuring myoelectric activity signals is placed around the subject's arm so that the electrodes are in contact with the biceps and triceps. The non-amputee subject has the arm immobilized in a device. This makes it possible to reproduce the muscular situation of an amputee subject, i.e., an isometric contraction of the biceps and triceps characterized by an absence of displacement.

[0095] The method includes a first calibration step.

[0096] The electrodes 6 acquire a first and a second raw calibration myoelectric activity signal, the first and the second raw calibration myoelectric activity signal resulting from contractions of varying intensities of the first and the second antagonist muscles of the limb 8 of the subject 10, respectively, for a determined duration.

[0097] The myoelectric activity signal measuring device 4 communicates remotely with the treatment device 14 by means of the remote communication device 12. In the example of the figure 1 , the measuring device 4 communicates remotely with the first computer 20.

[0098] Advantageously, the method further comprises the identification of two electrodes 6 of the measuring device 4 allowing maximum acquisition of the myoelectric activity signals of each of the two antagonist muscles. To do this, the subject 10 performs dissociated contractions of the two antagonist muscles wearing the measuring device 4, for example the biceps and the triceps.

[0099] Each electrode 6 acquires a raw myoelectric activity calibration signal which it transmits to the treatment device 14 by means of the remote communication device 12.

[0100] The acquisition frequency of electrodes 6 is for example 200 Hz.

[0101] Once the contractions have been carried out, the acquisition is stopped and the processing device 14 selects the two electrodes 6 which have each acquired their own significant activity during the activation of one of the antagonist muscles and which have acquired little or no co-contraction activity.

[0102] Co-contraction activities correspond to a simultaneous contraction of the two antagonistic muscles.

[0103] In particular, the first computer 20 selects the two electrodes 6 allowing maximum acquisition of the raw myoelectric activity calibration signals of each of the two antagonist muscles.

[0104] After the identification of two electrodes 6 allowing maximum acquisition, the method comprises the determination, in particular by the processing device 14, of the minimum and maximum myoelectric activities for each of the two antagonist muscles from the acquired raw calibration myoelectric activity signal.

[0105] In particular, the first computer 20 determines minimum and maximum myoelectric activities for each of the two antagonist muscles from the acquired raw calibration myoelectric activity signal.

[0106] To do this, subject 10 performs, for example, the strongest possible muscular contraction of the first muscle, for example the biceps, for a determined duration, then releases. The determined duration is between one second and three seconds, preferably two seconds. Then subject 10 performs the strongest possible muscular contraction of the second muscle, for example the triceps, for a determined duration, then releases. The determined duration is, for example, two seconds.

[0107] A first and a second raw calibration myoelectric activity signal resulting from these contractions is acquired by the electrodes 6.

[0108] The remote communication device 12 transmits the raw calibration myoelectric activity signal to the processing device 14.

[0109] In particular, the remote communication device 12 transmits the raw calibration myoelectric activity signal to the first computer 20.

[0110] The processing device 14 then sets the raw calibration myoelectric activity signal to zero mean value.

[0111] In particular, the first computer 20 sets the raw calibration myoelectric activity signal to zero mean value.

[0112] The processing device 14 determines the absolute value of the raw calibration myoelectric activity signal set to zero mean value.

[0113] In particular, the first computer 20 determines the absolute value of the raw calibration myoelectric activity signal set to zero mean value.

[0114] The processing device 14 filters the signal in absolute value to obtain the envelope.

[0115] In particular, the first computer 20 filters the signal in absolute value to obtain the envelope.

[0116] For example, the filter used is a second-order low-pass Butterworth filter with a cutoff frequency of 1.5 Hz.

[0117] From the envelope of the filtered signal, the processing device 14 determines for each of the two antagonistic muscles the minimum value and the maximum value of the envelope of the filtered signal.

[0118] In particular, the first computer 20 determines for each of the two antagonistic muscles the minimum value and the maximum value of the envelope of the filtered signal.

[0119] The minimum and maximum value of the filtered signal envelope correspond to the minimum and maximum myoelectric activities.

[0120] Advantageously, the processing device 14 stores the minimum value and the maximum value of the envelope of the filtered signal.

[0121] Then the processing device 14 determines a myoelectric activity threshold, called EMG offset x , below which the contraction of the muscle is not necessarily voluntary and must not be taken into consideration in order to avoid disturbing the movement produced. Indeed, values ​​located below this threshold correspond to residual activity of the muscles following muscular fatigue, and could disrupt the movement.

[0122] In particular, this threshold is determined by taking into account the perception of subject 10 and is calculated as follows, for each of the two antagonist muscles: offset EMG x = perc x ∗ EMG x max − EMG x min + EMG x min , with EMG offset x the myoelectric activity threshold; perc x the percentage over the range of muscle activity; EMG x max the maximum value of the filtered signal envelope, and EMG x min the minimum value of the filtered signal envelope.

[0123] The percentage of the extent of the muscular activity of the subject 10 is determined so that the speed of movement of the limb of the virtual avatar is zero when the subject feels relaxed.

[0124] Typically, this percentage is between 4% and 10%.

[0125] This percentage has an impact when the subject wishes to generate a movement speed of the virtual avatar's member but also to cancel this speed.

[0126] This percentage is refined for each muscle.

[0127] The myoelectric activity threshold value is typically set by default to 5% of the muscle activity range.

[0128] Then, this myoelectric activity threshold is typically tested and adjusted with subject 10 in order to assess whether it does not disrupt the control of the virtual avatar 24.

[0129] Preferably, the value of the myoelectric activity threshold can be adjusted at any time and in particular after several minutes of work, where muscle fatigue can be felt and residual muscle activity can become bothersome.

[0130] In particular, the first computer 20 determines the myoelectric activity threshold.

[0131] From the determination of the myoelectric activity threshold, the first computer 20 will only take into account the values ​​recorded between the determined threshold and the maximum value of the envelope of the filtered signal.

[0132] Advantageously, the processing device 14 memorizes the myoelectric activity threshold.

[0133] Then the processing device 14 normalizes the determined threshold and the maximum value of the envelope of the filtered signal.

[0134] In particular, the first computer 20 normalizes the determined threshold and the maximum value of the envelope of the filtered signal.

[0135] The first computer 20 typically assigns the value 0 to the previously calculated myoelectric activity threshold, and the value 1 to the maximum value of the envelope of the filtered signal.

[0136] The processing device 14 determines, for each muscle, a conversion coefficient making it possible to convert any normalized value of the envelope of the filtered signal between the determined threshold and the maximum value of the envelope of the filtered signal into a speed component. The conversion coefficient is adapted to the subject for each of the muscles in order to make the perception of movement as faithful as possible. The conversion coefficient typically makes it possible to assign, for each muscle, a zero speed component when the normalized value of the envelope of the filtered signal is 0, and to assign a maximum speed component when the normalized value of the envelope of the filtered signal is 1.

[0137] Advantageously, the processing device 14 stores the conversion coefficient.

[0138] Such calibration of the myoelectric activity signals recorded allows fine adjustment of the control of the virtual avatar 24 adapted to the subject 10, particularly when the myoelectric activities are very unequal between the two antagonistic muscles.

[0139] After this calibration phase, the method includes a step of moving the limb of the virtual avatar 24.

[0140] System 1 is ready to acquire new myoelectric activity signals and process them.

[0141] The two selected electrodes 6 acquire a first and a second raw myoelectric activity signal resulting from contractions of the two antagonist muscles of the limb 8 of the subject 10, for a determined duration.

[0142] For each of the first and second raw myoelectric activity signals, the remote communication device 12 transmits the raw myoelectric activity signal to the first computer 20.

[0143] The first computer 20 sets the raw myoelectric activity signal to zero mean value.

[0144] The first computer 20 determines the absolute value of the raw myoelectric activity signal set to zero mean value.

[0145] The first computer 20 filters the signal in absolute value to obtain the envelope.

[0146] The first computer 20 normalizes the values ​​of the envelope of the filtered signal, for example by a rule of three from the myoelectric activity threshold previously calculated and stored, and the maximum value of the envelope of the filtered signal previously determined and stored to obtain a value between 0 and 1.

[0147] The first computer 20 typically assigns, for each muscle, to each normalized value of the envelope of the filtered signal, a speed component between the zero speed component and the maximum speed component, by applying the previously determined and stored coefficient.

[0148] The first computer 20 determines a speed and a direction of movement of a member of the virtual avatar 24 displayed by the second computer 21 by subtracting the speed components obtained for each of the muscles.

[0149] For example, in the case of rotation of an elbow, subtraction of the speed values ​​makes it possible to determine a flexion or extension speed of an elbow of a virtual avatar 24 displayed by the second computer 21.

[0150] The first computer 20 transmits a movement command to the second computer 21.

[0151] The motion command typically includes the determined speed and direction of movement of the virtual avatar limb 24.

[0152] The second computer 21 receives the movement command transmitted by the first computer 20.

[0153] The display device 23 of the second computer 21 displays the virtual avatar 24 representing at least one limb corresponding to the movement which is to be controlled by the myoelectric activities. This may be the arm or a leg, a whole body or a half-body depending on the chosen application.

[0154] The member of the virtual avatar 24 moves according to the movement command transmitted by the first computer 20.

[0155] For example, stopping muscle contractions causes the movement to stop, the positioning of the elbow of the virtual avatar 24 is then in a fixed position which corresponds to the last position reached.

[0156] Thus, when the subject 10 contracts at least one of the muscles on which the electrodes 6 are placed, the virtual avatar 24 flexes or extends its elbow according to the settings made adapted to the subject.

[0157] According to variants, the calculated speed is a hand opening or closing speed of a virtual avatar, or a wrist rotation speed of a virtual avatar, or a knee or ankle flexion / extension speed of a virtual avatar, or any other conceivable option.

[0158] Subject 10 thus has visual feedback of the contractions he performs with the two antagonistic muscles, by the virtual avatar 24 whose movements are adapted to the subject's muscular contraction capacities. The learning of muscular dissociation and the fine work of muscular contractions are thus improved.

[0159] In an advantageous embodiment of the method, after having determined a speed of movement and a direction of movement of the limb of the virtual avatar 24 by acquiring the first and second myoelectric activity signals, the method comprises activating the sensory feedback device 16, in particular by the processing device 14.

[0160] The vibrating elements 18 of the sensory feedback device 16 are arranged on the limb 8 of the subject 10 comprising the two antagonistic muscles whose myoelectric activity is measured.

[0161] The vibrating elements 18 are for example sixteen in number and arranged in two bands of eight on either side of the electrodes 6.

[0162] The activation of the vibrating elements 18 is for example carried out in the following manner.

[0163] For each myoelectric activity signal acquired by the electrodes 6, the first computer 20 determines an instantaneous value of a sensory parameter, for example an instantaneous position of the limb of the virtual avatar 24 from the speed and direction of movement determined previously.

[0164] The first computer 20 sends to the third computer 22 the instantaneous value of the sensory parameter of the member of the virtual avatar 24.

[0165] The third computer 22 determines ranges of values ​​of the sensory parameter of the member of the virtual avatar 24 to be associated with at least one vibrating element 18.

[0166] The third computer 22 typically activates at least one element 18 corresponding to a previously determined range of values ​​of the sensory parameter.

[0167] In particular, the reading program of the third computer 22 makes the correspondence between the instantaneous value of the sensory parameter and the activation of the corresponding vibrating element 18.

[0168] Preferably, the third computer 22 activates the vibrating elements 18 selectively and discreetly.

[0169] For example, the third computer 22 determines the number and position of the vibrating elements 18 to be activated according to ranges of values ​​of the sensory parameter.

[0170] Preferably, within each range of values ​​of the sensory parameter.

[0171] More preferably, a single vibrating element 18 is assigned to a range of values ​​of the sensory parameter for each range of values ​​of the sensory parameter.

[0172] According to the example where the sensory parameter is the elbow flexion angle, when the subject 10 contracts the biceps and / or the triceps, the electrodes 6 acquire a first and a second raw myoelectric activity signal resulting from contractions of the biceps and the triceps, respectively, the remote communication device 12 transmits the first and the second myoelectric activity signal to the first computer 20, the first computer 20 processes the first and the second myoelectric activity signal and transmits to the second computer 21 a command to move the arm of the virtual avatar 24 and to the third computer 22 an instantaneous position of the arm of the virtual avatar 24, the second computer 21 displays the movement of the arm of the virtual avatar 24 and the third computer 22 activates each vibrating element 18 in steps of 20° of elbow flexion angle. It is understood that the step may have a value other than 20°, for example 10°, 15°, 25°, 30°.

[0173] This allows for the most natural control possible, i.e. both rapid in the execution of the movement but also precise to allow stopping on a target without oscillation.

[0174] Preferably, the third computer 22 determines the duration and sequence of vibrations for each instantaneous position.

[0175] For example, for a given instantaneous position, the third computer 22 activates a determined vibrating element 18 for a duration of 100 ms alternating with a stop for a duration of 100 ms.

[0176] Preferably, the vibration intensity is adaptable to improve the comfort of the subject 10.

[0177] For example, a first intensity level is defined by the subject's feeling of a first vibration when the vibration intensity is increased from an absence of vibration. A second intensity level is defined by a transition from the subject's feeling of a vibration to an absence of vibration when the vibration intensity is decreased from a strong vibration.

[0178] A vibration intensity level sufficient to stimulate the subject is calculated using the average between the first and second vibration intensity levels.

[0179] According to another example, the vibration intensity varies according to the myoelectric activities recorded. In the case of a gripping force of a hand of the virtual avatar 24, the higher the force, the higher the vibration intensity applied.

[0180] Vibration sensory feedback allows for the addition of congruent sensory information to vision by reducing the subject's attentional load, as illustrated in the example below.

[0181] An example of implementation of the method according to the invention will now be described. EXAMPLE Materials and participants

[0182] Participants include subjects with transhumeral arm amputations, i.e. above the elbow, and non-amputee subjects, known as healthy subjects.

[0183] The tool used to recover the electrical activities of the biceps and triceps is the MyoArm Band ® from the company Thalmic. It is a bracelet comprising 8 surface electrodes, initially designed to work using a muscle activity pattern algorithm at the forearm level. It is used here to recover the electrical muscle activities of the biceps and triceps at the level of the participants' arms.

[0184] A 6-vibrator wristband, 7 mm in diameter and 2 mm thick, is used to send vibrotactile sensory feedback to the participant.

[0185] Three computers are connected to each other, each with a specific role. The main computer, the master computer, supports the programs of the entire protocol for the execution of the different exercises, and also has the role of (1) receiving the myoelectric activities sent from the MyoArm Band via Bluetooth, (2) processing them, (3) sending to the second computer the movement command translated into a speed (resulting from this processing) so that the elbow of a virtual avatar displayed on the screen of the second computer moves and (4) sending the instantaneous position calculated from the myoelectric activities to the third computer which is responsible for activating the vibrators. The exchanges between the master computer and the other two computers are done using a switch configured to interact in a local network.

[0186] The second computer is dedicated to the visual interface and is placed in front of the participant. Animatlab software is used on this computer to visualize the avatar's movement in real time. Additionally, during the exercises, all instructions are displayed on this computer. Participants simply focus on the screen and follow the various instructions.

[0187] Finally, the third computer (pi-top) is responsible for managing the activation of the vibrators. It receives instructions from the master computer indicating the position calculated from the myoelectric activities. A reading program on the pi-top matches the position with the activation of the corresponding vibrator. This computer consists of a Raspberry Pi3 to which the vibrator bracelet is connected. Calibration

[0188] Before any test begins, each participant goes through a necessary calibration phase. During this phase, the muscle activities from the 8 electrodes of the bracelet are displayed on a graphical interface. Simply put on the bracelet and watch the muscle activities recorded by the electrodes directly. The acquisition frequency of the MyoArm Band is 200Hz. The raw activity as well as the filtered signal trace are visible. During this phase, participants are asked to perform dissociated contractions of the biceps and triceps.

[0189] This allows the identification of the two electrodes with which the signal is clearest for the rest of the experiment. Once the contractions are performed, the acquisition is stopped and the two electrodes that had (i) significant own activity during biceps or triceps activation and (ii) no or little co-contraction activity are selected.

[0190] Once the electrodes are selected, the minimum and maximum myoelectric activity is determined on each of the electrode channels. To do this, participants are asked to perform a near-maximal contraction of the biceps for 2 seconds, followed by a relaxation and a 2-second contraction of the triceps.

[0191] An avatar with a right or left arm is available depending on the laterality of the participants' amputation. Then, the signal is normalized. A muscle activity threshold is defined (often between 5% and 7.5% of the normalized signal range) below which no movement would be generated. This prevents the appearance of residual muscle activity following muscle fatigue that could set the avatar in motion without the subject's real will. This threshold is adjustable during the experiment, giving some leeway to adjust the threshold depending on the participants' condition. Finally, a gain is established between muscle activity and elbow rotation speed. This allows for fine-tuning of avatar control, especially when muscle activity is very uneven.

[0192] Spatial markers are defined in order to activate the vibrations with a rapid stimulation made of alternation (activated for 100ms then stopped for 100ms and so on). After a presence (in the specific case of more than 2 seconds) at one of these markers, the stimulation stop time is increased in order to less solicit the participant but while continuing to inform him every 500ms by an activation of 100ms.

[0193] When participants contract their muscles, the virtual avatar's arm moves based on the calibration data and the vibrators activate in 20° angle increments. Experimentation

[0194] Participants were asked to perform back-and-forth movements between two targets to assess control stability over time. The second computer screen displayed the two targets representing the range of motion in which the avatar's arm was to be moved in relation to the initial target. The ranges of motion ranged from 20° to 100°. This training exercise included a series of 10 movements to be performed with visual sensory feedback, then 10 movements with visual and vibration sensory feedback, and 10 movements with vibration alone. When performing the movement, subjects were asked to pause on the target before moving on to the other target. A sound signaled the stable position of the avatar when it had not moved for 500 ms. However, it did not signal that the subjects were on the correct target.Target validation could therefore only be done with vision and / or vibration. At the end of each trial, the movement trace was shown to the participants. This allowed them to judge the speed of movement and the precision over the entire trial. The duration for the trial was proportional to the amplitude of the back and forth movements to be performed and ranged from approximately 20 seconds to 45 seconds for the trial with 100° amplitude. Once the 3 series of 10 movements were completed, the participants moved on to the test phase. This exercise lasted approximately 45 minutes.

[0195] Test phase: The test phase consisted of a total of 8 blocks of 3 trials each, with 1 trial per type of sensory feedback. Every two blocks (i.e., every 6 movements), a trial without any sensory feedback was performed. This trial served as a baseline to evaluate performance with the other types of sensory feedback. A total of 28 movements were therefore performed in this test phase. Similar to the training phase, a pause time on the targets to be reached was required. The objective was not to make as many back and forth movements as possible but to be precise and on the right targets. At the end of the trial, the trace of the movement performed was displayed on the screen to show participants the quality of their performance. This test phase lasted approximately 40 minutes. Results

[0196] There figure 3illustrates the behavior of non-amputee subjects and represents the distribution of deviations from the target for each movement during the round-trip trials represented for the four types of sensory feedback. The x-axis represents the number of stabilizations made by the subjects during the round-trip movements. The y-axis represents the deviation from this target. One round-trip is equivalent to two iterations. The solid or dotted lines represent the average deviations from the target for all trials using the same type of sensory feedback.

[0197] The plots show that for sensory feedback including vision, the subjects remain on average very precise with a deviation of less than 10° for more than 3 round trips. When we look at the plot of the condition with vibrotactile sensory feedback alone, it is stable over time with values ​​higher than the plots including vision but below the limit of 20° which corresponds to the discretization of 2 vibrators. We clearly observe that the last condition without any sensory feedback is different from the other 3 conditions with an error that increases over time. This attests that the subjects having no sensory feedback on the performance of their action lost their bearings and moved away from the targets over time. This plot therefore illustrates the positive effect of vibrotactile sensory feedback in the sense that it allows the subjects to situate themselves in space. The information transmitted by the vibrations is therefore correctly used and efficient.

Claims

1. Method for controlling a limb of a virtual avatar (24) by the myoelectric activities of a limb (8) of a subject (10), the method for controlling being implemented by a system for controlling a limb of a virtual avatar (24) comprising a device (4) for measuring myoelectric activity signals and a processing device (14) for processing the myoelectric activities comprising at least one display device (23) configured for displaying the virtual avatar (24), the method for controlling comprising a first calibration step comprising: - acquiring first and second raw calibration myoelectric activity signals by the device (4) for measuring myoelectric activity signals, the first and second raw calibration myoelectric activity signals resulting from contractions of variable intensities of first and second antagonistic muscles of the limb (8) of the subject (10) respectively, during a given period, and then: - determining the envelope of each of the first and second signals by the processing device (14), - determining the minimum and maximum myoelectric activities from the envelope of each of the first and second signals by the processing device (14), then, for each of the two antagonistic muscles: - determining a myoelectric activity threshold by the processing device (14), - normalizing the maximum myoelectric activity and the myoelectric activity threshold by the processing device (14), - determining a coefficient of conversion of the normalized myoelectric activity into a component of speed of movement of a limb of the virtual avatar (24) by the processing device (14), and the method for controlling comprising a second step of moving the limb of the virtual avatar (24) comprising: - acquiring first and second raw myoelectric activity signal by the device (4) for measuring myoelectric activity signals, the first and second raw myoelectric activity signals resulting from contractions of the two antagonistic muscles of the limb (8) of the subject (10), during a given period, and then, for each of the first and second raw myoelectric activity signals: - normalizing the myoelectric activity resulting from the signal by the processing device (14), - converting the normalized myoelectric activity into a component of speed of movement of a limb of a virtual avatar (24) by applying the coefficient of conversion by the processing device (14), - determining a speed and a direction of movement of the limb of the virtual avatar (24) by subtracting the movement speed components obtained for each of the muscles by the processing device (14), and - moving the limb of the virtual avatar (24) with the speed and direction of movement determined by the processing device (14).

2. Method for controlling according to claim 1, wherein, simultaneously with the step of moving the limb of the virtual avatar (24) with the determined speed and direction of movement, the method comprises the activation of a sensory feedback device (16) disposed on the limb (8) of the subject (10), the sensory feedback device (16) being activated selectively and discretely according to predetermined ranges of values of a sensory parameter.

3. Method for controlling according to claim 2, wherein the sensory parameter is selected from a bending angle of an elbow of the virtual avatar (24), an anterior elevation angle of a shoulder of the virtual avatar (24) an abduction / adduction angle of the shoulder of the virtual avatar (24), a medial or lateral rotation angle of the shoulder of the virtual avatar (24), a bending angle of a wrist of the virtual avatar (24), a radial or ulnar rotation angle of the virtual avatar (24), a degree of opening of a hand of the virtual avatar (24), a degree of opening of a finger grip involving the thumb and one or more other fingers of the hand of the virtual avatar (24), a gripping force of a hand of the virtual avatar (24), a bending angle of a knee of the virtual avatar (24), a bending angle of the ankle of the virtual avatar (24), a speed of rotation of one of the movements mentioned in the present list, or any other sensory modality able to be replaced by vibrations, such as the warmth or configuration of a myoelectric prosthesis (11) of the subject (10).

4. Method for controlling according to any one of claims 1 to 3, wherein the device (4) for measuring myoelectric activity signals comprises a plurality of electrodes (6), the method comprising selecting two electrodes (6) from the plurality of electrodes (6) enabling a maximum acquisition of the myoelectric activity signals of each of the two antagonistic muscles.

5. Method for controlling according to one of claims 1 to 4, wherein determining the minimum and maximum myoelectric activities from the envelope of the signal comprises: setting the raw calibration myoelectric activity signal to a zero mean value, determining the absolute value of the raw calibration myoelectric activity signal set to zero mean value, filtering the signal in absolute value to obtain the envelope therefrom, and from the envelope of the filtered signal, determining the minimum value and the maximum value of the envelope of the filtered signal.

6. System for controlling a limb of a virtual avatar (24) by the myoelectric activities of a limb (8) of a subject (10), the system comprising: - a device (4) for measuring myoelectric activity signals resulting from contractions of first and second antagonistic muscles of the limb (8) of the subject (10), suitable for acquiring first and second raw calibration myoelectric activity signals resulting from contractions of variable intensities of first and second antagonistic muscles of the limb (8) of the subject (10) respectively, during a given period, and first and second raw myoelectric activity signals resulting from contractions of the two antagonistic muscles of the limb (8) of the subject (10), during a given period, - a processing device (14) for processing the myoelectric activities comprising at least one display device (23) configured for displaying the virtual avatar (24), the processing device (14) being configured for implementing the method for controlling a limb of a virtual avatar (24) by the myoelectric activities of a limb (8) of a subject (10) according to any of the preceding claims.

7. System according to claim 6, wherein the device (4) for measuring comprises a plurality of electrodes (6) and is in the form of a bracelet.

8. System according to claim 6 or 7, furthermore comprising a sensory feedback device (16).

9. System according to claim 8, wherein the sensory feedback device (16) comprises a plurality of vibrating elements (18).

10. System according to claim 8 or 9, wherein the processing device (14) is configured for activating the sensory feedback device (16) selectively and discretely according to predetermined ranges of values of a sensory parameter.

11. System according to any one of claims 6 to 10, wherein the device (4) for measuring signals of myoelectric activity comprises a remote communication device (12) configured for remotely communicating with the processing device (14).