Control device for vibrator of haptic / force presentation device, and haptic / force presentation device

Abstract

A control device (40) comprises a CPU (41) as an execution device, and a storage device (44). The storage device (44) stores vibration mode data (VMD) that indicates a plurality of vibration modes for rehabilitation. The storage device (44) stores model data (MD) that defines a learning model that outputs a mode variable indicating the type of haptic vibration mode that a haptic presentation device (10) outputs in response to the input of a body variable indicating body information of a user. The model data (MD) is data learned by machine learning. In an acquisition process, the CPU (41) acquires a plurality of body variables of the user. In a mode selection process, the CPU (41) selects a specific mode on the basis of the mode variable corresponding to the input of a plurality of body variables acquired in a mode acquisition process. In a drive process, the CPU (41) drives a vibrator (20) in the specific mode.

Description

[Technical Field]

[0001] The present invention relates to a control device for a vibrator of a haptic sense presentation device, and the haptic sense presentation device. [Background technology]

[0002] The haptic presentation device described in Patent Document 1 includes a housing, a vibrator, and a control device. The vibrator is located inside the housing. The housing is used by being held, for example, by a user's hand. The control device controls the vibration pattern of the vibrator to a specific pattern. In this way, the haptic presentation device presents various haptic sensations to a user who is touching the housing. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2005-190465 Summary of the Invention [Problem to be solved by the invention]

[0004] One possible situation in which the haptic presentation device described in Patent Document 1 would be used is when the haptic presentation device is used as an assistive tool during rehabilitation to improve the user's motor function. However, in such a case, it is unclear what kind of haptic sensation should be presented to the user in order to enhance the effect of improving the motor function. [Means for solving the problem]

[0005] In order to solve the above problem, one aspect of the present disclosure is a control device for a vibrator of a haptic presentation device, comprising a storage device and an execution device, and the control target is a vibrator of a haptic presentation device, the storage device stores vibration mode data indicating multiple vibration modes for rehabilitation, and model data that defines a learning model that takes as input physical variables indicating the user's physical state and outputs mode variables indicating the type of vibration mode of the haptic sensation output by the haptic presentation device, the model data being learned data through machine learning, and the execution device is a control device for a vibrator of a haptic presentation device that performs an acquisition process to acquire the multiple physical variables of the user, a mode selection process to select a specific mode from the multiple vibration modes based on the mode variable that takes as input the multiple physical variables acquired in the acquisition process, and a drive process to drive the vibrator in the specific mode selected in the mode selection process.

[0006] In order to solve the above problem, one aspect of the present disclosure is a haptic presentation device that includes a vibrating body, and a control device having a memory device and an execution device and controlling the vibrating body, wherein the memory device stores vibration mode data indicating multiple vibration modes for rehabilitation and model data that defines a learning model that takes as input physical variables indicating a user's physical state and outputs mode variables indicating the type of haptic vibration mode, the model data being learned data through machine learning, and the execution device executes an acquisition process that acquires the multiple physical variables of the user, a mode selection process that selects a specific mode from the multiple vibration modes based on the mode variable that takes as input the multiple physical variables acquired in the acquisition process, and a drive process that drives the vibrating body in the specific mode selected in the mode selection process.

[0007] According to the above configurations, it is possible to select a vibration mode that provides an appropriate haptic sensation in accordance with the user's physical condition. That is, it is possible to provide an appropriate haptic sensation in accordance with the user's physical condition. [Effects of the Invention]

[0008] Appropriate haptic sensations can be presented according to the user's physical condition. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a schematic diagram showing a haptic feedback device. [Figure 2] FIG. 2 is a schematic diagram showing the learning model. [Figure 3] FIG. 3 is a flowchart showing a series of processes performed by the control program. [Figure 4] FIG. 4 is a schematic diagram showing a learning system. [Figure 5] FIG. 5 is a flowchart showing a series of processes performed by the learning program. DETAILED DESCRIPTION OF THE INVENTION

[0010] (One embodiment) An embodiment of a control device for a haptic sense presentation device will be described below with reference to the drawings. First, the haptic sense presentation device will be described.

[0011] <Tactile sensation presentation device> As shown in FIG. 1, the haptic sense presentation device 10 includes a vibrator 20, an input / output device 30, and a control device 40.

[0012] Although not shown, the vibrating bodies 20 are housed inside the housing of the haptic presentation device 10. The vibrating bodies 20 also include voice coil motors, weights corresponding to each voice coil motor, and a cubic case that houses them. The weights vibrate due to a force generated when a current flows through the coil of the voice coil motor. When the weights vibrate, the case vibrates due to the vibration of the weights. Therefore, by controlling the current flowing through the coil of the voice coil motor, the vibrating bodies 20 vibrate in a direction along an axis perpendicular to the surface of the case. More specifically, the vibrating bodies 20 are as described in, for example, JP 2005-190465 A.

[0013] The input / output device 30 is a device for the user to input physical variables BV. The input / output device 30 is configured, for example, with a touch display. Therefore, the input / output device 30 can also provide information to the user through images. The input / output device 30 is operated by the user. A plurality of physical variables BV indicating physical information are input to the input / output device 30 by the user's operation.

[0014] In detail, two or more of the plurality of physical variables BV are variables indicating the presence or absence of each of the predetermined physical disabilities. For example, if there are n types of predetermined physical disabilities, the n physical variables BV are variables indicating the presence or absence of each of the physical disabilities. If a person has a first type of physical disability, the physical variable BV is input as 1. Similarly, if a person has a second type of physical disability, the physical variable BV is input as 1. On the other hand, if a person does not have the corresponding type of physical disability, the physical variable BV is input as 0.

[0015] Two or more of the physical variables BV are variables that indicate the degree of each predetermined physical disability. For example, if there are n predetermined types of physical disabilities, the n physical variables BV are variables that indicate the degree of each physical disability. For example, the greater the degree of the corresponding type of physical disability, the closer the physical variable BV is to 1, and the smaller the degree of the corresponding type of physical disability, the closer the physical variable BV is to 0. In this embodiment, if a person does not have the corresponding type of physical disability, the physical variable BV is 0.

[0016] One of the plurality of physical variables BV is a variable indicating the user's age. Another of the plurality of physical variables BV is a variable indicating a part of the body that is the target of rehabilitation. Another of the plurality of physical variables BV is a variable indicating the user's height. Another of the plurality of physical variables BV is a variable indicating the user's weight.

[0017] The control device 40 controls the vibrating body 20. The control device 40 controls the vibration pattern of the vibrating body 20 to a specific vibration pattern according to the haptic sense to be presented. The vibration pattern is, for example, a vibration pattern represented by a nonlinear waveform. In this way, the haptic sense presentation device 10 presents a haptic sense or a kinematic sense to the user. The haptic sense includes an illusionary tactile sense. The illusionary tactile sense includes an illusionary tactile sense and an illusionary kinematic sense. The illusionary tactile sense is an illusion in which the user's brain feels as if it is touching an uneven surface when the vibration of the vibrating body 20 is presented to the user. The illusionary kinematic sense is an illusion in which the user's brain feels as if it is being given a force when the vibration of the vibrating body 20 is presented to the user.

[0018] The control device 40 includes a CPU 41 as an execution device, a peripheral circuit 42, a ROM 43, a storage device 44, and a bus 45. The bus 45 connects the CPU 41, the peripheral circuit 42, the ROM 43, and the storage device 44 so that they can communicate with each other. The peripheral circuit 42 includes a circuit that generates a clock signal that regulates internal operations, a power supply circuit, a reset circuit, etc. The ROM 43 pre-stores various programs that the CPU 41 uses to execute various controls. In particular, the ROM 43 stores a control program P1 that performs control based on a learning model LM, which will be described later. The CPU 41 controls the vibrating body 20 by executing the various programs stored in the ROM 43.

[0019] The storage device 44 stores vibration mode data VMD indicating a plurality of vibration modes VM for rehabilitation. The vibration modes VM are modes for presenting haptic sensations according to the type of rehabilitation and its combination.

[0020] The storage device 44 also stores data indicating vibration patterns of the vibrating body 20 for realizing each vibration mode VM. The data stored in the storage device 44 determines the haptic sense presented by the haptic sense presentation device 10.

[0021] The storage device 44 stores model data MD that defines the learning model LM. As shown in Fig. 2, the learning model LM is configured, for example, by a neural network NN and a softmax function SF that normalizes the output of the neural network NN.

[0022] The neural network NN has an input layer IL, a hidden layer ML, and an output layer OL. The input layer IL has a plurality of nodes. The number of nodes is equal to the number of variables of the plurality of physical variables BV indicating the physical state of the user. The neural network NN also has an activation function in the hidden layer ML. The activation function is, for example, a hyperbolic tangent. The output layer OL outputs the matching probability of each vibration mode VM.

[0023] In addition, the softmax function SF is a function for making the sum of the matching probabilities of each vibration mode VM output to the output layer OL equal to 1. Therefore, the matching probabilities of each vibration mode VM output to the output layer OL are normalized by the softmax function SF and output from the learning model LM.

[0024] In the learning model LM configured in this way, multidimensional input variables are input to the input layer IL, and the sum of values ​​multiplied by weights according to each transmission path is input to the activation function. The output value of the activation function is then input to the next layer. By repeating this calculation, the matching probability of each vibration mode VM is output from the output layer OL. The matching probability of each vibration mode VM is then normalized by the softmax function SF and finally output. Note that in Figure 2, the transmission paths connecting nodes in adjacent layers are omitted.

[0025] The model data MD is data indicating the learning model LM, and therefore includes data relating to weights according to each transmission path that have been updated through learning.

[0026] <Control based on learning models> 1, the CPU 41 executes a control program P1 stored in the ROM 43 to perform a series of processes for controlling the vibrator 20. As a result, the CPU 41 executes an acquisition process for acquiring a plurality of body variables BV of the user, a mode selection process for selecting a specific mode SM that is a specific mode from a plurality of vibration modes VM, and a drive process for driving the vibrator 20 in the specific mode SM.

[0027] When the power of the haptic sense presentation device 10 is turned on, the CPU 41 executes the control program P1 stored in the ROM 43. For example, when the power of the haptic sense presentation device 10 is turned off, the power of the haptic sense presentation device 10 is turned on by operating the input / output device 30. In other words, the control program P1 is a program that causes the CPU 41 to execute an acquisition process, a mode selection process, and a drive process.

[0028] 3, when the CPU 41 starts the control program P1, it first executes the process of step S11. In step S11, the CPU 41 performs a request process. Specifically, in the request process, the CPU 41 outputs image data to the input / output device 30 so that the user can select a physical variable BV indicating the requested physical information of the user. That is, the input / output device 30 displays an icon indicating an input field for each physical variable BV on the touch display.

[0029] More specifically, in the request process, the CPU 41 outputs the name of each of a plurality of predetermined physical disabilities and a corresponding option of "yes" or "no" for each of the physical disabilities. The result of the user's input here is treated as a physical variable BV indicating whether the user has a physical disability.

[0030] Furthermore, in the request processing, the CPU 41 outputs an input field for information indicating the degree of disability corresponding to the name of each physical disability, and options for information to be entered into the input field. Examples of options include three levels: strong, medium, and weak. If "none" is selected from the options indicating the presence or absence of a physical disability, the CPU 41 does not output the input field for information indicating the degree of disability or the options. The user's input result here is treated as a physical variable BV indicating the user's degree of physical disability. Furthermore, if the physical disability is "none," the physical variable BV indicating the user's degree of physical disability is considered to have been entered as 0.

[0031] Furthermore, in the request processing, the CPU 41 outputs an input field for inputting the user's age and options to be input into the input field. Examples of the options are integers from 10 to 100. The user's input result here is treated as a physical variable BV indicating the user's age. The CPU 41 then outputs an input field for selecting a body part to be rehabilitation and options to be input into the input field. Examples of the options are body parts such as the right arm, left arm, right foot, and left leg. The user's input result here is treated as a physical variable BV indicating the body part to be rehabilitation.

[0032] The above selection fields may be displayed simultaneously on one screen of the input / output device 30, or each time an option is selected, the next option may be displayed. After that, the CPU 41 advances the process to step S12.

[0033] In step S12, the CPU 41 determines whether or not the physical variables BV have been input. Specifically, the CPU 41 determines whether or not any option has been selected for all items in the request processing of step S11. If an option has been selected for all items, the CPU 41 determines that the physical variables BV have been input from the input / output device 30. On the other hand, if an option has not been selected for any item, the CPU 41 determines that the input of the physical variables BV from the input / output device 30 has not been completed.

[0034] If the user's input from the input / output device 30 has not been completed (S12: NO), the CPU 41 returns the process to step S11. On the other hand, if the user's input from the input / output device 30 has been completed (S12: YES), the CPU 41 advances the process to step S13.

[0035] In step S13, the CPU 41 performs an acquisition process. In the acquisition process, the CPU 41 acquires the body variables BV. Specifically, the CPU 41 acquires each body variable BV corresponding to the option input to the input / output device 30. After that, the CPU 41 proceeds to step S14.

[0036] In step S14, the CPU 41 performs a mode selection process. In the mode selection process, the CPU 41 inputs the body variables BV acquired in the acquisition process as input variables to the learning model LM. Then, the CPU 41 selects a specific mode SM from among the vibration modes VM based on the mode variables MV output from the learning model LM. Specifically, the CPU 41 acquires the compatibility probability of each vibration mode VM output by the learning model LM as the mode variable MV, which is an output variable. Then, the CPU 41 selects the vibration mode VM corresponding to the compatibility probability with the highest value among the compatibility probabilities for each vibration mode VM as the specific mode SM. After that, the CPU 41 proceeds to step S15.

[0037] In step S15, the CPU 41 performs a drive process. In the drive process, the CPU 41 drives the vibrator 20 in the specific mode SM selected in the mode selection process. Note that, while the haptic sense presentation device 10 is driven in the specific mode SM in this manner, the user performs rehabilitation while using the haptic sense presentation device 10 as an assistive tool. That is, the user performs exercise operations associated with rehabilitation while receiving haptic sensations from the haptic sense presentation device 10. Then, the CPU 41 performs the drive process for a certain period of time, and then terminates the series of processes.

[0038] <Learning System> <Learning model learning method> Next, a learning method for the learning model LM will be described. First, a learning system 60 that generates model data MD that defines the learning model LM will be described.

[0039] 4, the learning system 60 includes the above-described haptic sense presentation device 10, a measuring device 70, and a setting device 80. In the learning system 60, each device is connected to each other so as to be able to communicate with each other.

[0040] The measuring device 70 is a device that measures activity parameters that indicate the activity state of a user who has received a haptic sensation from the haptic sensation presentation device 10. For example, the measuring device 70 measures the activity state of the user's brain. Specifically, the measuring device 70 is a brain measuring instrument that uses near-infrared spectroscopy.

[0041] The measuring device 70 measures activation parameters at each location of the brain. For example, the measuring device 70 measures the blood flow rate at each location of the brain as the activation parameter. The greater the blood flow rate, the higher the degree of activation of the brain. The measuring device 70 measures activation parameters at locations in the primary motor cortex of the brain that correspond to each part of the body. The measuring device 70 then transmits the activation parameters at each location of the brain to the setting device 80.

[0042] The setting device 80 is a device for updating model data MD indicating the learning model LM. The setting device 80 controls the vibrator 20 of the haptic presentation device 10 via the control device 40. The setting device 80 includes a CPU 81, a peripheral circuit 82, a ROM 83, a storage device 84, and a bus 85. The bus 85 connects the CPU 81, the peripheral circuit 82, the ROM 83, and the storage device 84 so that they can communicate with each other. The peripheral circuit 82 includes a circuit for generating a clock signal that defines internal operations, a power supply circuit, a reset circuit, etc. The ROM 83 pre-stores various programs that the CPU 81 uses to execute various controls. In particular, the ROM 83 stores a learning program P2 for learning the learning model LM. The CPU 81 learns the learning model LM by executing the learning program P2 stored in the ROM 83. The storage device 84 stores model data MD that defines the learning model LM.

[0043] The CPU 81 executes a series of processes for training the learning model LM by executing the learning program P2 stored in the ROM 83. As a result, the CPU 81 executes a test driving process, a parameter acquisition process, and a correct label determination process (to be described later) as a training data generation process. The CPU 81 also executes a model calculation process and a model update process as a model data update process.

[0044] When the haptic presentation device 10 is powered on while the haptic presentation device 10, the measuring device 70, and the setting device 80 are connected, the CPU 81 executes the learning program P2 stored in the ROM 83. In other words, the learning program P2 is a program that causes the CPU 81 to execute a training data generation process and a model data update process.

[0045] 5, when the learning program P2 is started, the CPU 81 first executes the processes of steps S21 to S23. The processes of steps S21 to S23 are the same as steps S11 to S13 in the control program P1 described above. After step S23, the CPU 81 proceeds to the process of step S24.

[0046] In step S24, the CPU 81 starts a test drive process. In the test drive process, the CPU 81 drives the vibrator 20 in a plurality of vibration modes VM in a predetermined order for a predetermined period of time. In addition, in the test drive process, the CPU 81 starts measuring the activity state of the user's brain using the measurement device 70. Note that, during the execution of this test drive process, it is preferable that the user is performing rehabilitation while using the haptic presentation device 10 as an assistive device. Thereafter, the CPU 81 advances the process to step S25.

[0047] In step S25, the CPU 81 executes a parameter acquisition process. In the parameter acquisition process, the CPU 81 acquires each activation parameter when the vibrating body 20 is driven in each vibration mode VM measured by the measurement device 70. Then, Parameter acquisition process After the above steps are completed, the CPU 81 advances the process to step S26.

[0048] In step S26, the CPU 81 executes a correct label determination process. In the correct label determination process, the CPU 81 determines a correct mode CM to be selected as a specific mode SM from among a plurality of vibration modes VM based on the activation parameters in each vibration mode VM in the parameter acquisition process. Then, the CPU 81 acquires the correct mode CM as a correct label.

[0049] Specifically, first, the CPU 81 outputs the activation parameters of each part of the brain during the period when the vibrator 20 is driven in the first vibration mode VM. At this time, the CPU 81 may output the activation parameter values ​​themselves, or may output the activation parameter values ​​for each part of the brain in the form of an image visually representing them. Then, the CPU 81 outputs the activation parameters of each part of the brain for the second and all subsequent vibration modes VM.

[0050] A medical professional such as a doctor or physical therapist refers to the activation parameters for each vibration mode VM output as described above and the user's behavior when receiving the haptic sensation in each vibration mode VM. The medical professional comprehensively evaluates these and identifies the vibration mode VM that is deemed to be most effective as the correct mode CM. The medical professional then inputs the correct mode CM via the input / output device 30 of the haptic sensation presentation device 10. In this way, the CPU 81 determines the correct mode CM as the correct label for the multiple body variables BV acquired in the acquisition process, and generates a set of training data. Then, the process proceeds to step S27.

[0051] In step S27, the CPU 81 determines whether the number of training data is equal to or greater than a predetermined number. If the number of training data is less than the predetermined number (S27: NO), the CPU 81 repeats the processes of steps S21 to S26 for a different user or a different rehabilitation target area. On the other hand, if the number of training data is equal to or greater than the predetermined number (S27: YES), the CPU 81 proceeds to step S28.

[0052] In step S28, the CPU 81 executes a model calculation process. In the model calculation process, for each training data, the CPU 81 inputs the data of the plurality of body variables BV acquired in the acquisition process from the training data as input data to the input layer IL of the learning model LM, and calculates the matching probability of each vibration mode VM. Then, the CPU 81 proceeds to step S29.

[0053] In step S29, the CPU 81 executes a model update process. In the model update process, the CPU 81 adjusts the weights in the neural network NN so that the rate at which the matching probability of each vibration mode VM calculated in the model calculation process does not match the correct label is reduced.

[0054] Specifically, the CPU 81 calculates the total matching probability of each vibration mode VM as "1" for one set of training data. Next, the CPU 81 compares the vibration mode VM with the highest matching probability with the vibration mode VM indicated by the correct label. Next, if they match, the CPU 81 determines that the result of the model calculation process for the training data matches the correct label. The CPU 81 repeats this process for multiple sets of training data. Then, the CPU 81 calculates the proportion of the number of times it is determined that there is a match among the total number of sets of training data as the matching probability. After that, the CPU 81 proceeds to step S30.

[0055] In step S30, the CPU 81 determines whether the matching probability of each vibration mode VM calculated in the model calculation process matches the correct label is equal to or greater than a predetermined specified probability. If the matching probability is less than the specified probability (S30: NO), the CPU 81 repeats the processes of steps S28 and S29. On the other hand, if the matching probability is equal to or greater than the specified probability (S30: YES), the CPU 81 determines that learning is complete. Then, the CPU 81 updates the model data MD that defines the learning model LM as learned data. After that, the CPU 81 ends the series of processes.

[0056] <Operation of the embodiment> According to the above embodiment, a user, for example a patient, inputs a plurality of body variables BV indicating the user's own physical state when using the haptic presentation device 10. The haptic presentation device 10 selects a specific mode SM from a plurality of vibration modes VM based on the output of the learning model LM when the input body variables BV are input to the learning model LM.

[0057] <Effects of the embodiment> (1) According to the above embodiment, the haptic presentation device 10 selects a specific mode SM from a plurality of vibration modes VM based on the mode variable MV output by the learning model LM in the mode selection process. Therefore, the haptic presentation device 10 can drive the vibrator 20 in the vibration mode VM that presents an appropriate haptic sensation in accordance with the user's body variable BV. In other words, by inputting the body variable BV that the user is aware of, the user can perform rehabilitation while receiving an appropriate haptic sensation from the haptic sensations that the haptic presentation device 10 can present.

[0058] In particular, in the above embodiment, the number of combinations of multiple body variables BV can be extremely large. Therefore, it would be extremely time-consuming to assign a vibration mode VM to each combination pattern based on a predetermined rule. Furthermore, the correspondence between the magnitude of the body variables BV and the optimal vibration mode VM is not necessarily simple, and it may be difficult to find a clear regularity between the two. In this regard, according to the above embodiment, the specific mode SM is selected using a learning model LM defined by model data MD learned by machine learning. Therefore, even if it is not possible to cover all combinations, a vibration mode VM that exhibits a haptic sensation that is expected to be appropriate based on the experience of a medical professional can be selected as the specific mode SM.

[0059] (2) According to the above embodiment, two or more of the physical variables BV are variables that indicate whether the user has a physical disability. Therefore, an appropriate haptic sensation can be presented depending on whether the user has a physical disability.

[0060] (3) According to the above embodiment, two or more of the physical variables BV are variables that indicate the degree of the user's physical disability. Therefore, it is possible to present an appropriate haptic sensation in accordance with the degree of the user's physical disability.

[0061] (4) According to the learning method of the learning model LM of the above embodiment, the correct label is selected by a medical professional based on the activation state of the user's body, specifically, the activation state of each part of the brain, when the vibrator 20 is driven in a plurality of vibration modes VM. Therefore, the experience of a medical professional, which is difficult to express in mathematical terms and in a map, can be reflected in the learning model LM. Therefore, even if a medical professional does not directly select a vibration mode VM each time, the user can use the haptic presentation device 10 with the vibration mode VM that is expected to be most effective.

[0062] (5) When a haptic sensation is presented to a user, even if the user's body, for example, arms or legs, does not appear to be moving, the brain's primary motor cortex may be activated in areas corresponding to each part of the body. According to the learning method of the learning model LM of the above embodiment, activation parameters for each part of the user's brain are measured when selecting a correct label for a medical professional. Therefore, even if the user's body does not appear to be moving, by referring to the activation parameters, a correct label that is effective in terms of brain activation can be obtained.

[0063] (Other embodiments) The above embodiment can be modified as follows: The above embodiment and the following modifications can be combined and implemented within the scope of technical compatibility.

[0064] In the above embodiment, the configuration of the vibrating body 20 is not limited to the configuration of the above embodiment. For example, the vibrating body 20 may be one that uses vibration by a motor, or one that has a piezoelectric element.

[0065] The number of vibrating bodies 20 may be plural. In this case, the vibration mode VM may include information on whether or not the plural vibrating bodies 20 vibrate and the order in which they vibrate. This allows for a greater variety of vibration modes VM.

[0066] In the above embodiment, the control device 40 is not limited to a device equipped with a CPU and ROM and executing software processing. For example, it may be a dedicated hardware circuit (such as an ASIC) that performs hardware processing on at least part of the software processing in the above embodiment. It may also be performed by That is, the control device 40 may have any of the following configurations (a) to (c): (a) A processing device that executes all of the above processes according to a program, and a program storage device such as a ROM that stores the program; (b) A processing device and program storage device that executes part of the above processes according to a program, and a dedicated hardware circuit that executes the remaining processes; (c) A dedicated hardware circuit that executes all of the above processes. Here, there may be multiple software execution devices that include a processing device and program storage device, and multiple dedicated hardware circuits. The same applies to the setting device 80.

[0067] Only one of the multiple physical variables BV may be a variable indicating whether or not the user has a physical disability. Also, for example, only one of the multiple physical variables BV may be a variable indicating the degree of the user's physical disability. For example, when presenting haptic feedback to a user with a specific physical disability, as long as the physical variable BV for one predetermined type of physical disability can be obtained in this way, variables for other types of physical disability are not necessary.

[0068] The physical variables BV may be any variables that indicate the user's physical information. For example, the physical variables BV may not include a variable that indicates whether the user has a physical disability, and the physical variables BV may not include a variable that indicates the degree of the user's physical disability.

[0069] Furthermore, for example, the plurality of physical variables BV may not include a variable indicating a location to be subject to rehabilitation. In this case, for example, if the user has only one physical disability, the location to be subject to rehabilitation corresponding to that physical disability may be estimated. Furthermore, the plurality of physical variables BV may include information other than that exemplified in the above embodiment. For example, the plurality of physical variables BV may include a variable indicating the user's blood pressure.

[0070] The learning model LM is not limited to one learned by the learning method of the above embodiment. The model data MD may be data learned by machine learning. For example, the learning method of the learning model LM is not limited to supervised learning, and the learning model LM may be defined by model data MD learned by reinforcement learning. The control program P1 may be executed using a learning model LM learned by other machine learning methods.

[0071] In the learning method of the learning model LM, the method of determining the correct label is not limited to the example in the above embodiment. For example, the oscillation mode VM with the highest brain activation parameter may be determined as the correct label.

[0072] The configuration of the learning model LM is not limited to the example of the above embodiment. For example, although only one hidden layer ML of the neural network NN is illustrated in Fig. 2, the neural network NN may have multiple hidden layers ML.

[0073] The vibration mode VM may be a mode that presents only tactile sensations, or may be a mode that presents only force sensations. The request process may be modified as needed to suit the input / output device 30, without outputting options, and instead allowing the user to input characters.

[0074] In the process of step S12, if an input completion button is displayed on the input / output device 30, it may be determined that input has been completed when the input completion button is pressed. The measuring device 70 is not limited to a brain measuring device using near-infrared spectroscopy. For example, the measuring device 70 may be a measuring device using functional magnetic resonance imaging (functional MRI). Also, for example, the measuring device 70 may be a measuring device using electroencephalography (EEG).

[0075] The measuring device 70 does not have to be a device that measures the activity state of the user's brain. For example, it may be a device that measures the activity state of a specific part of the user's body other than the brain. Specifically, if the part of the user that is the target of rehabilitation is the right leg, the measuring device 70 may be a device that measures the activity state of the right leg.

[0076] The setting device 80 may control the vibrating body 20 without going through the control device 40. For example, it is assumed that the storage device 84 of the setting device 80 stores data indicating a plurality of vibration modes VM. In this case, in the test driving process, the CPU 81 may drive the vibrating body 20 using the data.

[0077] The technical ideas that can be understood from the above-described embodiment and modified examples are additionally described below. <Appendix 1> a storage device and an execution device, the vibrator of the haptic presentation device being a control target; the storage device stores a plurality of vibration modes for rehabilitation and model data defining a learning model that receives as input a physical variable indicating a physical state of a user and outputs a mode variable indicating a type of the vibration mode of the haptic sensation output by the haptic presentation device; The model data is data that has been trained by machine learning, The execution device an acquisition process for acquiring a plurality of the physical variables of the user; a mode selection process for selecting a specific mode, which is a specific mode, from the plurality of vibration modes based on the mode variable having the plurality of body variables acquired in the acquisition process as input; a driving process for driving the vibrator in the specific mode selected in the mode selection process; A control device for a vibrator in a haptic presentation device.

[0078] <Appendix 2> One of the plurality of physical variables is a variable indicating whether or not the user has a physical disability. A control device for a vibrator of the haptic presentation device described in <Appendix 1>.

[0079] <Appendix 3> One of the plurality of physical variables is a variable that indicates the degree of the physical disability of the user. A control device for a vibrator of the haptic presentation device described in <Appendix 2>.

[0080] <Appendix 4> Two or more of the plurality of physical variables are variables that respectively indicate the presence or absence of each of a plurality of predetermined physical disabilities of the user. A control device for a vibrator of a haptic sense presentation device according to any one of <Supplementary Note 1> to <Supplementary Note 3>.

[0081] <Appendix 5> Two or more of the plurality of physical variables include variables that respectively indicate the degree of each of the plurality of physical disabilities of the user. A control device for a vibrator of the haptic presentation device described in <Appendix 4>.

[0082] <Appendix 6> the haptic sense presentation device has a plurality of the vibrators, The vibration mode includes information on whether or not vibration occurs for the plurality of vibrators and the order in which vibration occurs. A control device for a vibrator of the haptic sense presentation device according to any one of <Supplementary Note 1> to <Supplementary Note 5>.

[0083] <Appendix 7> a control device having a vibrator and a storage device and an execution device, and controlling the vibrator; The storage device stores vibration mode data indicating a plurality of vibration modes for rehabilitation, and model data defining a learning model that receives as input a physical variable indicating a physical state of a user and outputs a mode variable indicating a type of the vibration mode of the haptic sense, The model data is data that has been trained by machine learning, The execution device an acquisition process for acquiring a plurality of the physical variables of the user; a mode selection process for selecting a specific mode, which is a specific mode, from the plurality of vibration modes based on the mode variable having the plurality of body variables acquired in the acquisition process as input; a driving process for driving the vibrator in the specific mode selected in the mode selection process; Tactile sensation presentation device. [Explanation of symbols]

[0084] 10...Tactile sensation presentation device 20...Vibration body 30... Input / output device 40...Control device 41...CPU 42...Peripheral circuit 43...ROM 44…Storage device 45...Bus 60...Learning System 70...Measuring device 80...Setting device BV: Body Variables CM...correct mode IL…Input layer LM…Learning Model MD...Model data ML…middle layer MV...Mode variable NN...Neural network OL…Output layer P1...Control program P2: Learning Program SF...Softmax function SM...Specific mode VM…Vibration mode VMD...Vibration mode data

Claims

1. a storage device and an execution device, the vibrator of the haptic presentation device being a control target; The storage device stores vibration mode data indicating a plurality of vibration modes for rehabilitation, and model data defining a learning model that receives as input a physical variable indicating a physical state of a user and outputs, for each vibration mode, a mode variable indicating a compatibility probability of the vibration mode of the haptic sense output by the haptic sense presentation device, and The model data is data that has been trained by machine learning, The execution device an acquisition process for acquiring a plurality of the physical variables of the user; a mode selection process for selecting a specific mode, which is a specific mode among the plurality of vibration modes, based on the mode variable having the plurality of body variables acquired in the acquisition process as input; a driving process for driving the vibrator in the specific mode selected in the mode selection process, In the mode selection process, inputting the plurality of body variables acquired in the acquisition process into the learning model to output the mode variables for each of the vibration modes; Identifying the mode variable that shows the highest matching probability based on the output mode variables; The vibration mode corresponding to the identified mode variable is selected as the identified mode. A control device for a vibrator in a haptic presentation device.

2. One of the plurality of physical variables is a variable indicating whether or not the user has a physical disability. A control device for a vibrator of the haptic feedback device according to claim 1.

3. One of the plurality of physical variables is a variable indicating the degree of the physical disability of the user. A control device for a vibrator of a haptic feedback device according to claim 2.

4. Two or more of the plurality of physical variables are variables that respectively indicate the presence or absence of each of a plurality of predetermined physical disabilities of the user. A control device for a vibrator of the haptic feedback device according to claim 1.

5. Two or more of the plurality of physical variables include variables that respectively indicate the degree of each of the plurality of physical disabilities of the user. A control device for a vibrator of a haptic feedback device according to claim 4.

6. the haptic sense presentation device has a plurality of the vibrators, The vibration mode includes information on whether or not vibration occurs for the plurality of vibrators and the order in which vibration occurs. A control device for a vibrator of the haptic feedback device according to claim 1.

7. a control device having a vibrator and a storage device and an execution device, and controlling the vibrator; The storage device stores vibration mode data indicating a plurality of vibration modes for rehabilitation, and model data defining a learning model that receives as input a physical variable indicating a physical state of a user and outputs a mode variable indicating a compatibility probability of the vibration mode of the haptic sense for each of the vibration modes, The model data is data that has been trained by machine learning, The execution device an acquisition process for acquiring a plurality of the physical variables of the user; a mode selection process for selecting a specific mode, which is a specific mode among the plurality of vibration modes, based on the mode variable having the plurality of body variables acquired in the acquisition process as input; a driving process for driving the vibrator in the specific mode selected in the mode selection process, In the mode selection process, inputting the plurality of body variables acquired in the acquisition process into the learning model to output the mode variables for each of the vibration modes; Identifying the mode variable that shows the highest matching probability based on the output mode variables; The vibration mode corresponding to the identified mode variable is selected as the identified mode. Tactile sensation presentation device.

Images