ELECTRODERMATIC DEVICE
Patent Information
- Application Number
- DE602021021534
- Authority / Receiving Office
- DE · DE
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2020-09-23
- Filing Date
- 2021-09-23
- Publication Date
- 2024-11-06
- Estimated Expiration
- 2041-09-23
AI Technical Summary
Existing methods for measuring galvanic skin response and electrodermal activity rely on user-initiated learning, leading to subjective and unreliable results due to dependency on user behavior and exposure to varying events, and require complex signal processing to filter noise from physiological signals.
A wearable device with stamped stainless steel electrodes and integrated sensors for heart rate and temperature, utilizing techniques like Active Noise Cancellation and Empirical Mode Decomposition to process signals independently of user movements, providing digital data for emotional state analysis through S arousal and valence values.
The solution enables reliable, objective measurement of emotional states by filtering noise and reducing subjectivity, providing accurate and reproducible emotional state assessments without requiring user intervention.
Description
Field of invention
[0001] The present invention relates to the field of measuring galvanic skin response and electrocutaneous activity, and in particular to portable devices and methods for measuring the electrodermal activity of a user's skin.
[0002] Electrodermal activity is biological electrical activity recorded on the surface of the skin and reflecting the activity of the sweat glands and the autonomic nervous system and therefore, among other things, the individual's perception and involuntary behavior rather than that of a response he wishes to give.
[0003] The scientific study of the galvanic skin reflex began in the early 1900s. Psychologist C.G. Jung was among the first to use his recording in psychoanalysis and he recounted this experience in his work "Studies in Word Analysis" published in 1906. Wilhelm Reich also studied the electrodermal response in experiments conducted at the Institute of Psychology, University of Oslo, in 1935 and 1936, with the aim of confirming the existence of a bioelectrical charge behind his concept of "vegetative currents of pleasure."
[0004] In the 1960s and until the end of the 1970s, the psychogalvanic response was used in all fields of psychology for a wide variety of research. Psychogalvanic phenomena are linked to the activity of the sympathetic nervous system, and are manifested by the variation of the electrical resistance of the skin, following an emotion.
[0005] Since then, much work has focused on the use of GSR (Galvanic Skin Response) or RED (electrodermal resistance) signals representing the variation in the electrical resistance of the skin caused by the activity of the sweat glands following sensory excitation, called psychogalvanic reflex, for various applications such as lie detectors, emotion analysis or the control of video games by new human-machine interfaces. State of the art
[0006] Known in the state of the art is patent FR3063425 proposing a system for determining a user's emotion comprising a sensor of the user's electrodermal activity, a first circuit adapted to: a / detect strong emotional moments of the user based on an analysis of information from the sensor (electrodermal activity, b / in the event of detection of a strong emotional moment in step / a / , ask the user to classify said strong emotional moment.
[0007] This solution requires user-initiated learning, which requires the user to identify a "strong emotional moment" and then classify it. It is obvious that this solution leads to erratic or even fanciful behavior, and is in no way reproducible or reliable because it depends on the user's subjectivity, their ability to successfully complete this learning, and the nature of the events to which the user is exposed during this learning phase.
[0008] Patent application WO2017202626 relates to a sensor for measuring skin conductance. An amplifier is used to convert the skin conductance into an analog output voltage, which is then converted to the digital domain, so that both the tonic skin conductance increase and the phasic skin conductance response are obtained in the digital domain. The amplifier has a non-linear logarithmic gain, with a decreasing gain for increasing skin conductance values. The sensor can detect both the increase in tonic and phasic signals over a wide skin conductance range. It allows optimal use of the analog-to-digital converter, so that a lower resolution and therefore a cheaper converter can be used.
[0009] US Patent Application 2013 / 183646 describes an apparatus for interoperably utilizing multiple biosensor data attached to the user's finger. These sensors may be electrodermal response (EDR) sensors, or photoplethysmograph (PPG) signals, or temperature, or three-axis acceleration, and combinations thereof. The biometric card records and processes the sensor signals and communicates them to a mobile device that interoperably utilizes multiple sensor information to determine aspects of the user's emotional state.
[0010] We also know the article “RUI GUO ET AL: "Pervasive and unobtrusive emotion sensing for human mental health", PERVASIVE COMPUTING TECHNOLOGIES FOR HEALTHCARE (PERVASIVEHEALTH), 2013 7TH INTERNATIONAL CONFERENCE ON, IEEE, May 5, 2013 (2013-05-05), pages 436-439, XP032439656, DOi: 10.4108 / ICST.PERVASIVEHEAL TH.2013.252133, ISBN: 978-1 -4 799-0296-5.
[0011] Another article JORN BAKKER ET AL: "What's Your Current Stress Level? Detection of Stress Patterns from GSR Sensor Data", DATA MINING WORKSHOPS (ICDMW), 2011 IEEE 11TH INTERNATIONAL CONFERENCE ON, IEEE, December 11, 2011 (2011-12-11), pages 573-580, XP032100119, DOi: 10.1109 / ICDMW.2011.178, ISBN: 978-1-4673-0005-6 concerns stress detection phenomena.
[0012] Also known is patent KR20160085577 describing another solution for determining a psychological state according to an embodiment of the present invention comprises: a bio-signal acquisition unit for acquiring a first bio-signal and a second bio-signal of a user generated by a first stimulus; and a psychological state determination unit for extracting a first feature and a second feature by analyzing the first bio-signal and the second bio-signal and determining the psychological state of the user on the basis of the first feature and the second feature. The first bio-signal and the second bio-signal are bio-signals indicating changes in autonomic nerves measured by different bio-sensors. For example, the first bio-signal is a PPG signal measured by a pulse wave sensor, and the second bio-signal is a GSR signal measured by a skin conductivity sensor. Solution provided by the invention
[0013] In order to overcome the drawbacks of the solutions of the prior art, the present invention relates, in its most general sense, to a system in accordance with claim 1.
[0014] Advantageously, said electrodes consist of pads formed by a stamped and tinned steel sheet.
[0015] Optionally, at least one of said electrodes has a housing capable of receiving an additional sensor.
[0016] According to one variant, said electrodermal equipment further comprises a temperature sensor.
[0017] Advantageously, the temperature sensor is positioned on another dedicated steel electrode or probe. The thickness of this dedicated probe is specific in order to avoid too much inertia in the signal and to lose important micro-variations in the monitoring of emotional reactions.
[0018] The electrodermal device further comprises a heart rate sensor, preferably a heart rate sensor delivering physiological signals by photoplethysmography (PPG). The heart rate sensor is positioned under a glass (PPG) and is not connected to the metal electrode.
[0019] Techniques applicable to the PPG signal include: Techniques such as SVD and ICA, which use only a single PPG channel and therefore provide relatively "poor" information but can be tested very quickly. Fourier Series Analysis with automatic determination of the number of coefficients to retain (as for ICA) is a possible technique but is quite complex due to the noise induced by movements which can, in certain transient circumstances, be more powerful than the PPG signal. Periodic Moving Average Filter processing can be effective but will induce a lot of latency in the signal. Wavelet Denoising processing is an effective alternative to other methods. The "Active Noise Cancellation" technique based on adaptive filtering is the method which seems to me to be the most appropriate for our multimodal approach and can be coupled with Empirical Mode Decomposition to further improve denoising.
[0020] The invention also relates to a system comprising an electrodermal sensor according to at least one of the preceding variants and a computer delivering at least one digital data item representative of an emotional state, characterized in that said computer executes a program controlling: The processing of digital conductance data provided by said processor over a sliding time window, the result of which provides an S arousal value.
[0021] It further comprises a step of processing said second series of heart rate signals consisting of bandpass filtering of the frequencies between 0.04 and 0.26 Hz and detection of peaks and inter-peak RR time measurement, over a sliding time window, the result of which provides a value S valence. Detailed description of a non-limiting example of the invention
[0022] [ Fig. 1 ] there [ Fig. 1] represents a perspective view of an electrode according to the invention [ Fig.2 ] there [ Fig.2 ] represents a partial view of a bracelet equipped with two sensors according to the invention. Hardware architecture
[0023] The invention relates to a device for acquiring physiological signals via two (or more) electrodes in contact with the epidermis of a user, and for exploiting different types of signals for estimating a reliable emotional state independent of artifacts such as the user's movements. The spacing between the electrodes is large and determined, so that the amount of skin crossed by the signal is sufficient to have a usable signal. Typically a spacing of 2 to 20 millimeters is suitable.
[0024] The invention uses two metal electrodes, the [ Fig. 1 ] is an example of a perspective view.
[0025] An electrode is formed by stamping a stainless steel sheet, for example INOX 316L with a thickness of 0.5 millimeters to form an electrical pad (10) with a height of 2 millimeters, and 11 millimeters long, 8 millimeters wide, of rectangular section with rounded corners (1 to 4) with a radius of curvature of 1.5 millimeters, extended by an edge (15) of rectangular section with a width of 1.5 millimeters, with rounded corners (11 to 13) with a radius of curvature of 3 millimeters. The radius of curvature of the edges is 0.3 millimeters. The edge is pierced with holes (20) with a section of 0.8 millimeters for fixing on a bracelet or a support plate. Notches (5, 6) with a semi-circular section and a radius of 1.1 millimeters allow the passage of electrical cables.The outer surface of the electrode is in contact with the epidermis and allows, with a second electrode placed nearby, the measurement of the electrical conductivity observed on the surface of the user's skin and providing a signal representative of electrodermal activity (EDA).
[0026] The electrode (10) may optionally integrate a plurality of sensors intended to obtain measurements of physiological signals associated with the user's emotions, for example: a sensor capable of measuring the user's heart rate via a light formed in the metal electrode, and a window closing said light; a sensor capable of measuring the surface temperature of the skin fixed to the bottom by a thermal glue; a sensor consisting of a three-axis or multi-axis accelerometer such as a 9-axis inertial module capable of allowing the measurement of movements on a limb of the user, for example a piezoelectric accelerometer or a three-axis MEMS accelerometer and a gyroscope (wrist rotation).
[0027] The signals provided by the accelerometer are representative of the user's movements and can be used to filter physiological signals in order to eliminate transient signals during sudden movements, which can alter electrical conduction or the positioning of the electrode on the epidermis. This noisy signal filtering functionality is particularly important when physiological signals are analyzed over a long time window, of several seconds, because taking into account abnormal signals would seriously distort the evaluations produced by these temporal analyses. Bracelet detail
[0028] There [ Fig.2] represents an example of mounting two electrodes (10) of 8x5 millimeters and a thickness of 0.5 millimeters, arranged transversely, with a spacing of 10 millimeters, on a bracelet (30). The bracelet has a clasp (31) and a housing (32) for receiving the electronic circuits. The assembly can be overmolded to ensure insensitivity to humidity and dust.
[0029] The bottom of this area has a temperature sensor (33) and an optical sensor (34) comprising a photodiode and LEDs to capture signals related to blood circulation, for example heart rate and oxygenation rate.
[0030] The electrodes are spaced and polarized by an electronic circuit implementing an ADC converter circuit LTC2487 ANALOG DEVICES (trade name) with two differential inputs and ensuring 16-bit sampling as well as a temperature compensation circuit. It is also possible to use the ADC of the Nordic nRF52 microcontroller which allows to have two signal modalities with the same components. This sensor makes it possible to measure the resistance of the skin via two divider bridges at the input of an analog-to-digital converter (ADC) in differential from a common point and a measurement of the voltage induced by the variation of the resistance of the human body, typically between 500 and 300 Ohm. Sensors fitted to the bracelet
[0031] The bracelet (30) can be equipped with several sensors: a. A galvanic sensor using the two electrodes (10) providing a “glavanic skin response” signal sampled at 8 Hz, used by the computer to characterize the Arousal score and the level of vigilance b. A 3-axis accelerometer sampled at 64 Hz, used by the computer to characterize steps and falls c. A blood sensor (34) sampled at 50 Hz, used by the computer to characterize the valence score, biometric recognition, heart rate and stress level d. A temperature sensor (33) sampled at 1 Hz for measuring body temperature.
[0032] The bracelet includes a calculator for pre-processing data which is then transmitted in the form of digital packets via a radio frequency protocol by a radio module and an antenna. Processing of sampled signals
[0033] Different types of processing can be considered to provide emotional insights.
[0034] Fourier Series Analysis is a common solution, but it involves a constraint related to the automatic determination of the number of coefficients to be retained. This constraint is quite complex, as the noise induced by the movements risks being more powerful than the PPG signal without being present continuously. However, the use of a denoising solution based on signal filtering based on the information provided by the accelerometer makes it possible to reduce this difficulty.
[0035] The sensor provides a signal representative of a passive or endosomatic parameter corresponding to the skin conductance level (SCL) or an active or exosomatic parameter corresponding to the skin conductance response (SCR). These parameters make it possible to determine electrodermal activity (EDA) originating from characteristics of the epidermal membrane, and from eccrine sweat activity under the control of the autonomic and central nervous systems.
[0036] There are two methods of recording.
[0037] The first method, called endosomatic, translates the potential differences generated by the skin membranes and leads to the measurement of the electrodermal potential. In this case, the sensor is a skin conductivity sensor associated with a current-voltage converter, for example a skin resistivity sensor, equipped with a pair of noble steel electrodes.
[0038] The second method, called exosomatic, translates the variations of a current applied to the skin, the characteristics of which can lead to the measurement of various electrodermal signals, including the measurement of skin conductance, the most commonly used in the literature. Each of the electrodermal signals is subdivided into a tonic component and a phasic component.
[0039] The first identifies slow variations in the electrodermal signal while the second corresponds to rapid variations in the signal commonly called electrodermal responses. From these phasic measurements, different measurement parameters can be extracted, such as the frequency, latency or amplitude of electrodermal responses. The origins and variability of the measurement parameters of electrodermal activity make this activity a measurement sensitive to changes in our environment and to different mental processes under the control of the central nervous system, such as emotion, motivation, attention and mental load.
[0040] Each sensor is associated with a pre-processing circuit optionally performing analog processing (pre-amplification, filtration, delivery of an excitation signal), digitization (sampling, optional digital filtering, storage in a buffer memory, etc.) to deliver digital signals to a computer used to determine the pair of values representative of the emotional state. Physiological signal processing
[0041] The signals provided by a pair of electrodes are sampled at a frequency of 64 Hz, or even 130 Hz, and filtered in amplitude and frequency to remove aberrant signals. These signals constitute environmental information that complements the signals associated with emotions, for example to provide a context for the mode of movement and / or fall delivered by the accelerometer.
[0042] The signals provided by the electrical conductivity sensor (10) are sampled at a frequency of 8 Hz and then processed for the calculation of the arousal score and the alertness level.
[0043] The signals provided by the user's heart rate sensor are sampled at a frequency of 100 Hz and used by the calculator to determine the valence score, as well as for biometric recognition and stress level estimation.
[0044] The skin temperature measurement sensor is sampled at a low frequency of around 1Hz and completes the information used to characterize the emotional state. Particular implementation example
[0045] An example of implementation consists of equipping the patient with a wireless connected bracelet equipped with three physiological sensors (a single sensor may be sufficient) measuring electrodermal conductance (noted GSR for galvanic skin response) at a rate of 8Hz, cardiac activity (noted PPG for photoplethysmography) at a rate of 50Hz, body temperature (noted SKT for skin temperature) at a rate of 1Hz and one or more accelerometric and gyroscopic sensors (noted ACC) at a rate of 50Hz are used to synchronously record the data and the corresponding time stamps.
[0046] The GSR and PPG data are transmitted to a mobile terminal that performs the calculations for real-time identification of the emotional state. The GSR and PPG data are stored in calculation buffers whose durations vary depending on the variables calculated.
[0047] In each buffer, signal processing is performed before extracting the different variables used for emotional state identification analysis: GSR Signal: Bandpass filtering (4th order butterworth) is applied to the signal with a bandwidth of 0.05 - 1Hz. PPG Signal: Bandpass filtering (4th order butterworth) is applied to the signal with a bandwidth of 0.5 - 5Hz.
[0048] The variables used for emotional state identification analysis are then extracted from the processed signals. The variable Arousal is obtained in a 20-second calculation buffer from the normalized spectral power of the GSR signal calculated over the 0.045 - 0.25Hz band by a Hilbert-Huang transformation. Example of treatment
[0049] The variable Mdiffis recorded in a 2-second calculation buffer from the average of the absolute value of the first derivative of the GSR signal.
[0050] The variable Valence is obtained in a 60-second calculation buffer by calculating the cardiac coherence ratio. For this, peak detection in the PPG signal is performed using a dedicated function to deduce the peak-to-peak time intervals noted RR intervals. Then, the heart rate noted BPM is calculated from the RR intervals.
[0051] From the signal BPM, the maximum peak of the power spectrum is identified on the 0.04 - 0.26Hz band (the frequency range within which coherence can be observed). The power of this peak noted Peak Power is then determined by calculating the integral over a 0.030Hz wide window, centered around the peak. The total power over the 0.0033 - 0.4Hz band of the BPM signal noted Total Poweris then calculated. The normalized valence level is obtained by the following calculation: [Math.1] Valence = Peak Power Total Power − Peak Power
[0052] Every second, the new GSR and PPG values recorded by the bracelet allow the calculation of new values Arousal, Mdiff And Valence.
[0053] Mdiff is stored in memory the last minute to allow dynamic calibration of the system for detecting punctual variations in the level of physiological excitation. A weighting coefficient is applied to this calibration data to make the contribution of the most recent values more important during calibration. It is then possible to calculate the dynamic thresholds allowing the variable to be classified respectively Mdiff. The calculation of thresholds can be detailed as follows: [Math.2] Seuil t = Max Mdif f x 1 ⋮ x n + Moyenne Mdif f x 1 ⋮ x n 2
[0054] With Threshold ( t ) the value of the dynamic threshold at the instant tAnd Mdif ƒ x 1 ⋮ x n the values of the variable Mdif f over the entire duration of the calibration period.
[0055] Every second, a new value of Threshold is obtained and compared to Mdif f. If Mdif f is higher than its threshold value, then an emotional reaction is detected. Other methods of processing
[0056] Processing using the “Periodic Moving Average Filter” solution can be effective but induces a lot of latency in the signal.
[0057] Wavelet Denoising is a relevant solution, as is Active Noise Cancellation based on adaptive filtering. This is the most appropriate method for a multimodal approach and can be coupled with Empirical Mode Decomposition to further improve noise reduction. Specific examples of processing
[0058] The aim of the invention is to automatically provide, without human intervention, digital S arousal and / or S valence signals representative of a person's emotional state.
[0059] Efficient recognition of emotions from human physiological activity can be achieved through the use of a simple emotional model. Indeed, emotions can be projected into a multidimensional space, the most common being the valence-arousal plane. The valence level represents the positivity and negativity of an emotion, while the arousal level describes the intensity of the emotion. These two emotional components are expressed at the physiological level.
[0060] During stress, the sympathetic nervous system predominates and leads to an increase in the level of physiological arousal. An acceleration of the heart rate or an acceleration of the interbeat interval (IBI) is characteristic of this state. Denoising is all the more important for this indicator from which Valence is derived. Light is more sensitive to movement (PPG).
[0061] At rest, on the contrary, the parasympathetic nervous system is activated, reflecting a decrease in the physiological state of arousal and heart rate. Furthermore, the alternation of accelerations and decelerations of the heart rate becomes regular and coherent (state of cardiac coherence) in states of well-being, calm, mastery (positive emotional valence) while in states of stress, anxiety, anger (negative emotional valence), the tachogram corresponding to the S arousal and S valence couple becomes irregular, its chaotic pattern and its magnitude will decrease.
[0062] By extracting the level of coherence of the heart rate from the PPG signal, it becomes possible to obtain a robust indicator of the level of emotional valence and to calculate dynamic thresholds beyond which this valence level changes significantly.
[0063] Once the valence level has been estimated, it is then possible to verify the level of arousal by controlling the level of physiological activation in the spectral domain from the GSR signal to deduce in real time the emotional state of the individual and communicate it to the multimedia system with which the latter interacts.
[0064] The emotional state can be characterized according to the following table: [Tables 1] S valence + + EDGY ALERT S arousal NERVOUS EXITE UPSET SAD HAPPY DREPRESSIVE THRILLED BORED SERENE - RELAXED
[0065] During stress, the sympathetic nervous system predominates and leads to an increase in the level of physiological arousal. An acceleration of the heart rate is characteristic of this state. At rest, on the contrary, the parasympathetic nervous system is activated, reflecting a decrease in the state of physiological arousal and heart rate. Moreover, the alternation of accelerations and decelerations of the heart rate becomes regular and coherent (state of cardiac coherence) in states of well-being, calm, mastery (positive emotional valence) while in states of stress, anxiety, anger (negative emotional valence), the tachogram becomes irregular, its pattern chaotic and its magnitude will decrease.
[0066] By extracting the level of coherence of the heart rate from the PPG signal, it becomes possible to obtain a robust indicator of the level of emotional valence and to calculate dynamic thresholds beyond which this valence level changes significantly.
[0067] Once the valence level has been estimated, it is then possible to verify the level of arousal by controlling the level of physiological activation in the spectral domain from the GSR signal to deduce in real time the emotional state of the individual and communicate it to the multimedia system with which the latter interacts. Learning the characterization criteria
[0068] To construct a characterization model, the invention proposes a variant implementing a preparatory step of supervised learning.
[0069] This solution consists of proposing to a panel of users equipped with a physiological data acquisition device mentioned above, experimental plans formed by a succession of emotionally labeled video sequences each associated with a digital descriptor ID (t), and of recording the pairs of S arousal and S valence signals and their evolution over time, for each of the members of the panel.
[0070] These structured data (S arousal and S valence (t); ID (t)) for each of the panel members are then injected into a neural network, to develop a characterization model.
[0071] Participants will be equipped with a connected bracelet conforming to the invention equipped with three physiological sensors measuring cardiac activity (denoted PPG for photoplethysmography), body temperature (denoted SKT for skin temperature) and electrodermal conductance (denoted GSR). The bracelets communicate with a portable acquisition computer allowing synchronous recording of the data and the corresponding timestamp with an acquisition frequency of 50Hz, 1Hz and 4Hz for the PPG, the SKT and the GSR, respectively for the connected bracelet or connected watch and with an acquisition frequency of 64Hz, 4Hz and 4Hz for the PPG, the SKT and the GSR.
[0072] An HTC Vive virtual reality system will be used to display the stimuli selected to induce an emotional reaction and provide additional immersion (unprecedented in emotional stimulation protocol) Experimental design
[0073] For each participant, data will be recorded over a single twenty-minute session. The experimental design is: Sn (participants) * V6 (six emotional videos).
[0074] Each video corresponds to an emotional extreme: Video 1: Rest (40s) - Sadness-type emotional induction phase (30s) - After-effect (30s) Video 2: Rest (40s) - Joy-type emotional induction phase (30s) - After-effect (30s) Video 3: Rest (40s) - Disgust-type emotional induction phase (30s) - After-effect (30s) Video 4: Rest (40s) - Fear-type emotional induction phase (30s) - After-effect (30s) Video 5: Rest (40s) - Neutral-type emotional induction phase (30s) - After-effect (30s) Video 6: Rest (40s) - Relaxation-type emotional induction phase (30s) - After-effect (30s)
[0075] The resting phase will constitute a reference period to initialize the calculation of physiological variables. For each participant, the order of presentation of the videos will be random to avoid any order effect. In addition, to enrich the dataset, two videos will be available for the emotions of fear and joy. For each participant, the choice of the video used for each of these two emotions will be random. Data acquisition procedure
[0076] Each participant is first equipped with one or more connected bracelets and a virtual reality system allowing them to isolate themselves from external stimuli and optimize their attentional focus. The experimenter then checks the quality of the physiological signals. Each participant will be generally instructed to view six 30-second videos. During the 40 seconds before the video and the 30 seconds after the video, the instruction given will be to remain calm and still. Once all the videos have been viewed, the experimenter helps the participant remove the virtual reality headset and bracelet and then conducts a debriefing to check that everything went well. Data analysis
[0077] For each participant, the recorded physiological data will be preprocessed as follows:
[0078] For the PPG signal, signal jumps will be corrected using a dedicated function. A bandpass filtering (4th order butterworth) will then be applied to the signal with a bandwidth of 0.5 - 5Hz and then the signal will be normalized using a Hilbert transform and smoothed using a 16-second Gaussian window. Regarding the SKT signal, a lowpass filtering (4th order butterworth) will be applied to the signal with a cutoff frequency at 0.05Hz. Finally, a bandpass filtering (4th order butterworth) will be applied to the GSR signal with a bandwidth of 0.05 - 3Hz. All these variables will constitute the input data for the emotional classification algorithms. Graphical representation of treatment results
[0079] Every second, the variables obtained are represented by the detected emotional state display system (noted overlay) as follows:
[0080] The diameter of the circle corresponds to the normalized value of the value S Arousal. The larger the diameter, the higher the level of arousal.
[0081] The color of the circle corresponds to the normalized value of the value S Valencia. When the color tends towards green, the valence level is higher. When the color tends towards red, the valence level is lower.
[0082] When an emotional reaction is detected from the evolution parameter Mdif f, the The circle outline animates. The heart rate value updates in the center of the circle.
[0083] The emotional state is communicated to the multimedia system with which the individual interacts ([ Fig.2 ]). It is important to note that the dock / mobile provides for updating the bracelet on the one hand and the methods for calculating the values and detection levels on the other hand via an internet connection. Applications
[0084] The method implemented by a system according to the invention makes it possible to deliver control signals for controlling equipment such as a robot, in particular an empathic robot, or for controlling functional parameters of electronic equipment such as the sound level, light level, rhythm, etc.
[0085] These signals also make it possible to control the adaptation of the speed of an individual / collective transport vehicle and the management of security agents, control, pilots, drivers but also the reduction of the number of information in delicate piloting situations (fighter plane or F1).
Claims
1. A system (30) comprising a computer, a heart rate sensor (34) and an electrodermal sensor comprising at least two electrodes (10), each of which is configured to be in contact with a dermal region, and an electronic circuit having an analog-to-digital converter and a processor that is configured to execute computer-executable instructions in order to bias said electrodes (10), to measure the current flowing between pairs of electrodes, to apply digital preprocessing to the measured electrical signals and to determine the conductance of the dermal layer, said sensor further comprising an accelerometer, delivering electrical signals to said processor, said preprocessing taking into account the level of said signals that are delivered by said accelerometer, characterized in that said computer executes a program controlling: - the processing of the digital conductance data provided by said processor over a first sliding time window, the result of which provides a value Sarousal representative of the intensity of an emotion - and processing of a series of heart rate signals consisting of bandpass filtering of the frequencies between 0.04 and 0.26 Hz and detecting peaks and measuring the time between the peaks RR, over a second sliding time window, the result of which provides a value Svalence representative of the positivity and negativity of the emotion.
2. The system according to claim 1, characterized in that said electrodes (10) are formed by studs formed by a stamped steel sheet.
3. The system according to claim 1, characterized in that it further comprises a temperature sensor (33).
4. The system according to claim 1, characterized in that said heart rate sensor (34) delivers physiological signals by photoplethysmography (PPG).
