Mental assistance system and mental assistance method

Abstract

This disclosure provides a mental assistance system and mental assistance method that can bring the target emotion closer to the desired emotion. [Solution] The mental assist system 10 according to this disclosure comprises an emotion estimation unit 11 that estimates the emotion of the subject from the subject's brainwaves, a parameter generation unit 12 that generates parameters based on a target emotion that represents the target emotion of the subject, and an image determination unit 13 that determines an image to bring the subject's emotion closer to the target emotion based on the parameters.

Description

[Technical Field]

[0001] The present disclosure relates to a mental assistance system and a mental assistance method. [Background Art]

[0002] Mental assistance techniques for caring mental health have been developed. For example, Non-Patent Document 1 discloses an image set in which a subject experiences positive or negative emotions by presenting images. Non-Patent Document 2 discloses an approach based on a learning algorithm for solving an infinite-time linear quadratic tracker (LQT) for an unknown discrete-time system. [Prior Art Documents] [Non-Patent Documents]

[0003] [Non-Patent Document 1] Benedek Kurdi, Shayn Lozano & Mahzarin R. Banaji,“Introducing the Open Affective Standardized Image Set (OASIS)”,Behavior Research Methods, February 23, 2016, Volume 49, pages 457-470, (2017) [Non-Patent Document 2] Bahare Kiumarsi, Frank L. Lewis, Hamidreza Modares, Ali Karimpour, Mohammad-Bagher Naghibi-Sistani“Reinforcement Q-learning for optimal tracking control of linear discrete-time systems with unknown dynamics”,Automatica, April 2014, Volume 50, pages 1167-1175 [Summary of Invention] [Problems that the invention aims to solve]

[0004] Non-patent document 2 designs the input using error feedback, which results in significant noise in the input waveform for weak signals such as electroencephalograms. Therefore, controlling the target emotion to approximate the desired emotion was difficult.

[0005] This disclosure is made in view of these circumstances and provides a mental assistance system and mental assistance method that can bring the target emotion closer to the desired emotion. [Means for solving the problem]

[0006] The mental assistance system related to this disclosure is An emotion estimation unit that estimates the emotion of the subject from the subject's brainwaves, A parameter generation unit generates parameters based on a target emotion that represents the target emotion of the aforementioned subject, The system includes an image determination unit that determines an image to bring the target emotion closer to the desired emotion based on the parameters generated by the parameter generation unit.

[0007] The mental assistance system described herein uses parameters based on a target emotion to determine an image that brings the subject's emotion closer to the target emotion. This configuration allows the subject's emotions to be brought closer to the target emotion.

[0008] The parameter generation unit may generate a first parameter based on the target emotion, and a second parameter based on the target emotion and the target emotion. Even with such a configuration, the target emotion can be brought closer to the target emotion.

[0009] The parameter generation unit may generate the second parameter in such a way as to minimize the deviation between the target emotion and the desired emotion. This configuration makes it possible to more accurately bring the target emotion closer to the desired emotion.

[0010] The parameter generation unit may generate the second parameter using a learning model that takes the target emotion and the target emotion as inputs and outputs the second parameter. With such a configuration, the target emotion can be brought closer to the target emotion with greater accuracy.

[0011] The mental assistance method related to this disclosure is The subject's emotions are estimated from the subject's brainwaves. Based on the target emotion, which represents the target emotion of the aforementioned subject, parameters are generated. Based on the aforementioned parameters, an image is determined to bring the target emotion closer to the desired emotion. The computer performs the process.

[0012] The mental assistance method disclosed herein uses parameters based on a target emotion to determine an image that brings the subject's emotion closer to the target emotion. By using this configuration, the subject's emotions can be brought closer to the target emotion. [Effects of the Invention]

[0013] This disclosure provides a mental assistance system and mental assistance method that can bring the target emotion closer to the desired emotion. [Brief explanation of the drawing]

[0014] [Figure 1] This is a block diagram showing an example of the configuration of a mental assistance system. [Figure 2] This is a diagram showing a flowchart of a mental assistance system. [Figure 3] This diagram shows the relationship between arousal and valence values ​​in electroencephalography (EEG) to indicate emotion. [Figure 4] This figure shows the change in valence value over time. [Figure 5] This figure shows the change in the arousal value over time. [Modes for carrying out the invention]

[0015] Embodiments of this disclosure will now be described with reference to the drawings. In each drawing, the same or corresponding elements are denoted by the same reference numeral, and redundant explanations are omitted where necessary for clarity. In addition, some reference numerals have been omitted to avoid cluttering the drawings.

[0016] (Embodiment 1) <Mental Assist System> A mental assistance system according to Embodiment 1 will now be described. Figure 1 is a block diagram showing an example of the configuration of the mental assistance system. Figure 2 is a flowchart showing the mental assistance system.

[0017] As shown in Figure 1, the mental assist system 10 comprises an emotion estimation unit 11, a parameter generation unit 12, and an image determination unit 13. As shown in Figure 1, the parameter generation unit 12 comprises a first parameter generation unit 21 and a second parameter generation unit 22. The following will describe the overview of the mental assist system 10, followed by each functional block.

[0018] <Overview> This section outlines the Mental Assist System 10. The Mental Assist System 10 uses electroencephalography (EEG) to estimate the user's emotions and determine an appropriate image. Then, by presenting the determined image to the target, the system helps to bring the target's emotions closer to the target emotion. For example, the Mental Assist System 10 can change a target's negative emotions into positive emotions. It should be noted that emotions are a type of feeling. Emotions, such as joy, anger, sadness, and happiness, are temporary and relatively strong, and are easily changed by external stimuli.

[0019] Referring to Figure 2, the operation overview of the mental assist system 10 will be explained. As shown in Figure 2, the emotion estimation unit 11 estimates the target's emotion (the target's current emotion) from the target's brainwaves (step ST1). Next, the mental assist system 10 receives input for the target emotion, which is the target emotion (step ST2). Then, the first parameter generation unit 21 generates the first parameter based on the target emotion (step ST3).

[0020] Next, the second parameter generation unit 22 generates a second parameter based on the target emotion (the current emotion of the target) and the target emotion (step ST4). Next, the image determination unit 13 determines an image based on the first parameter and the second parameter to bring the target emotion closer to the target emotion (step ST5). Next, the mental assist system 10 displays the determined image (step ST6). This brings the target emotion closer to the target emotion.

[0021] <Emotional Inference> The method for estimating emotions from brain waves in the emotion estimation unit 11 will be explained in detail. The emotion estimation unit 11 estimates the arousal value (level of arousal) from the brain waves. The arousal value (level of arousal) indicates the degree to which the person is awake or relaxed.

[0022] Specifically, the Fp1 electrode of the electroencephalograph (EEG) is placed on the left frontal lobe, and the Fp2 electrode is placed on the right frontal lobe. Then, using a bandpass filter, the instantaneous amplitudes of alpha and beta waves are determined from the EEG data from the Fp1 and Fp2 electrodes. Fp1 |, |β Fp1 |, |α Fp2 |, |β Fp2 Extract the | symbol. This will give you the state variable x of the arousal value of the electroencephalogram. a This is calculated using formula (1). Note that x a The signal is processed in real time using a high-pass filter so that the average amplitude becomes 0 μV.

[0023]

number

[0024] In addition, the emotion estimation unit 11 estimates a valence value (pleasure degree) from an electroencephalogram. The valence value (pleasure degree) indicates the degree of pleasure and displeasure. Similarly to the arousal value (arousal degree), the instantaneous amplitudes of α waves and β waves |α Fp1 |, |β Fp1 |, |α Fp2 |, |β Fp2 | are extracted. Thereby, the state quantity x of the valence value of the electroencephalogram v is calculated using formula (2). Note that x v is processed in real time using a high-pass filter so that the average value of the amplitude is 0 μV.

[0025]

Math

[0026] Next, the emotion estimation unit 11 estimates an emotion using the calculated state quantity x of the arousal value of the electroencephalogram a and the state quantity x of the valence value of the electroencephalogram v . FIG. 3 is a diagram showing emotions derived from the relationship between the arousal value and the valence value of an electroencephalogram. In FIG. 3, four basic emotions: joy, anger, sorrow and pleasure, are shown as an example of emotions. FIG. 3 is created based on sampled data, for example, and threshold values for the four basic emotions are set.

[0027] For example, when the calculated state quantity x of the arousal value of the electroencephalogram a and the state quantity x of the valence value of the electroencephalogram v become a plot P1, the emotion estimation unit 11 estimates that the target emotion of the subject is joy. Further, for example, when the calculated state quantity x of the arousal value of the electroencephalogram a and the state quantity x of the valence value of the electroencephalogram v become a plot P2, the emotion estimation unit 11 estimates that the target emotion of the subject is pleasure.

[0028] In addition, the emotion estimation unit 11 uses formula (3) to obtain the state quantity of the electroencephalogram x=[xv x a ] T From the coordinates in Figure 3, the coordinates of the emotion of the target object are y=[y v y a ] T Specifically, the emotion estimation unit 11 uses the following formula (3) to estimate the coordinates of the target emotion y = [y v y a ] T We estimate this.

[0029]

number

[0030] <Parameter generation> The parameter generation method of the parameter generation unit 12 will be described in detail. The parameter generation unit 12 generates parameters to bring the target emotion closer to the target emotion, based at least on the target emotion. More specifically, the first parameter generation unit 21 generates the first parameter u ff,k The second parameter generation unit 22 generates the second parameter u fb,k Generates.

[0031] Referring to equation (4), the first parameter u ff,k and the second parameter u fb,k This will be explained. The first parameter u ff,k is, r k This is a Ford-forward term that takes (parameters derived from the target emotion) as direct input.

[0032] Second parameter u fb,k is, r k (Parameters derived from target emotion) and y k This is a feedback term to minimize the deviation of (parameters obtained from the estimated emotion of the target). That is, the second parameter generation unit 22 generates y kと r k To minimize the deviation from, the second parameter u fb,kThis generates the target emotion. By using this configuration, it is possible to more accurately bring the target emotion closer to the desired emotion. Note that in the following, simply, r k and y k This is how it is written.

[0033] Here, the image determination unit 13 receives a control input u obtained using equation (4). k Based on this, an image is determined to bring the target emotion closer to the desired emotion.

[0034]

number

[0035] In other words, the parameter generation unit 12 receives the control input u k To obtain r k The first parameter u is a Ford-Forward term that takes the input directly. ff,k and, r k and y k The second parameter u is a Ford-forward term based on the above. fb,k And, it generates.

[0036] Furthermore, the second parameter generation unit 22 uses a learning model that takes the target emotion and the target emotion as inputs and outputs a second parameter, thereby generating the second parameter u fb,k The second parameter generation unit 22 may generate the second parameter u by utilizing the learned LQR (Linear Quadratic Regulator). fb,k The second parameter generation unit 22 may generate r k and y k Using a learning model that takes and as input and outputs a second parameter, the second parameter u fb,k It is also possible to generate a second parameter. In this way, by using a learning model to generate a second parameter, the target emotion can be brought closer to the desired emotion more accurately.

[0037] <Second parameter> Second parameter u fb,kWe will explain this in detail. The second parameter u fb,k Using equations (5) to (8), the matrix H∈R 4×4 The value function Q(e) is the following equation, consisting of the following: k ,u e,k The theoretical input u that minimizes ) ^ e,k This is determined by (see equation (8)). Note that in equation (8), u ^ e,k is ∂Q(e k ,u ^ e,k ) / ∂u ^ e,k It satisfies = 0.

[0038]

number

[0039]

number

[0040]

number

[0041]

number

[0042] In equation (8), by constructing matrix H from the data, the optimal gain K can be calculated without requiring a mathematical model of the system. Here, the learning model is updated so that matrix H has the optimal value.

[0043] Specifically, from the reward r in equation (9), we derive equation (10) of the Bellman equation using the value function Q.

[0044]

number

[0045]

number

[0046] In updating matrix H, the update is repeated using equation (11), which is created from equations (5) and (10). i →H i+1 By doing so, we can approach the optimal value.

[0047]

number

[0048] Furthermore, in updating matrix H, the least squares solution H of equation (11) for the 4(N+1) data points (equation (12)) acquired in real time is obtained. i+1 We seek.

[0049]

number

[0050] From equation (8), the relationship in equation (13) is obtained, and the next gain K i+1 This can be determined.

[0051]

number

[0052] From equations (7) and (8), the relationship in equation (14) holds. From equation (4), the second parameter u fb,k This is determined by formula (15).

[0053]

number

[0054]

number

[0055] <Emotional changes (simulation)> Next, with reference to Figures 4 and 5, the emotional state of a subject using the mental assist system according to Embodiment 1 and the emotional state of a subject using the mental assist system according to the comparative example will be described. Figure 4 shows the change in valence value over time. Figure 5 shows the change in arousal value over time.

[0056] The upper part of Figure 4 and the upper part of Figure 5 show the emotional state of a subject using the mental assistance system of the comparative example. The mental assistance system of the comparative example is r k This is a simulation of what happens when (parameters derived from the target emotion) are directly input.

[0057] The lower sections of Figures 4 and 5 show the emotional state of a subject using the mental assistance system according to Embodiment 1. The mental assistance system according to Embodiment 1 is a simulation that reproduces the method shown in Figure 2.

[0058] In Figures 4 and 5, the parameter y is obtained from the subject's emotion after image presentation. v , y a and parameters r obtained from the target emotion v , r a This is shown. Below, simply, y v , y a、 r v , r a This is how it is written.

[0059] In the upper part of Figure 4 (Mental Assist System Related to Comparative Example), r v and y v A phase difference can be observed. In contrast, in the lower part of Figure 4 (Mental Assist System according to Embodiment 1), r v and y vThe phase difference is reduced. In other words, from the valence value in Figure 4, it can be seen that by using the mental assist system according to Embodiment 1, the target emotion is approaching the target emotion.

[0060] In the upper part of Figure 5 (Mental Assist System Related to Comparative Example), r a and y a A difference in amplitude can be observed. In contrast, in the lower part of Figure 5 (Mental Assist System according to Embodiment 1), r a and y a The amplitude difference is reduced. In other words, from the arousal value in Figure 5, it can be seen that by using the mental assist system according to Embodiment 1, the target emotion is approaching the target emotion.

[0061] Thus, as shown in the simulation results in Figures 4 and 5, the mental assist system according to Embodiment 1 can bring the target emotion closer to the desired emotion. For example, the mental assist system according to Embodiment 1 can change the target's negative emotion into a positive emotion.

[0062] In the mental assistance system 10 described above, the target is typically a person, but is not limited to this; it may also be an animal, such as a dog or a cat. This allows the animal's emotions to be brought closer to the target emotions. It can also be used to provide mental care for animals.

[0063] Furthermore, some or all of the processing in the mental assist system according to Embodiment 1 described above can be implemented as a computer program. Such a program can be stored using various types of non-temporary computer-readable media and supplied to a computer. Non-temporary computer-readable media include various types of tangible recording media. Examples of non-temporary computer-readable media include magnetic recording media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memories (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (Random Access Memory)). The program may also be supplied to the computer by various types of temporary computer-readable media. Examples of temporary computer-readable media include electrical signals, optical signals, and electromagnetic waves. Temporary computer-readable media can be supplied to the computer via wired communication channels such as electric wires and optical fibers, or via wireless communication channels.

[0064] It should be noted that the present invention is not limited to the embodiments described above, and can be modified as appropriate without departing from the spirit of the invention. In other words, the above description has been omitted and simplified as appropriate for the sake of clarity, and those skilled in the art can easily change, add, and modify each element of the embodiments within the scope of the present invention. [Explanation of symbols]

[0065] 10 Mental Assist System 11. Emotion Estimation Unit 12 Parameter generation unit 21 First Parameter Generation Unit 22 Second Parameter Generation Unit 13 Image Determination Unit

Claims

1. An emotion estimation unit that estimates the emotion of the subject from the subject's brainwaves, A parameter generation unit generates parameters based on a target emotion that represents the target emotion of the aforementioned subject, An image determination unit determines an image that brings the target emotion closer to the desired emotion based on the parameters generated by the parameter generation unit, Equipped with, Mental assistance system.

2. The parameter generation unit, Based on the aforementioned target emotion, a first parameter is generated. Based on the aforementioned target emotion and the aforementioned target emotion, a second parameter is generated. The mental assistance system according to claim 1.

3. The parameter generation unit generates the second parameter in such a way as to minimize the difference between the parameter obtained from the target emotion and the parameter obtained from the objective emotion. The mental assistance system according to claim 2.

4. The parameter generation unit, The second parameter is generated using a learning model that takes the aforementioned target emotion and the aforementioned target emotion as inputs and outputs the second parameter. The mental assistance system according to claim 2.

5. The subject's emotions are estimated from the subject's brainwaves. Based on the target emotion, which represents the target emotion of the aforementioned subject, parameters are generated. Based on the aforementioned parameters, an image is determined to bring the target emotion closer to the desired emotion. The computer performs the process. Mental support methods.

Images