JP7072853B2 - Exercise training systems, control methods, and programs - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Act January 17-19, 2018 Published at the "Industrial-Academia Collaboration Robot Forum" exhibition at the "2nd Robotex Robot Development and Utilization Exhibition" January 2018 Published in an oral presentation using materials at the "Industry-Academia Collaboration Robot Forum" at the "2nd Robotex Robot Development and Utilization Exhibition" on the 17th June 13-15, 2018 "Robot Industry Matching Fair Kitakyushu" Published at the exhibition at "2018"

The present disclosure relates to exercise training systems, control methods, and programs that support and analyze exercise for trainers.

Conventionally, a rehabilitation system for rehabilitating a trainee is known (see Patent Document 1). This rehabilitation system is equipped with an actuator device including a limb manipulator for moving the upper limbs. The limb manipulator measures limb movement and provides kinesthetic feedback to the trainee. The rehabilitation system also measures various measurements of brain function, such as SCP (slow cortical potential), MAC (exercise-related brain potential associated with voluntary movements), murhythms, and physiological measurements, and controllers and software. The module inputs measurements to process and monitor central (brain) activity.

Japanese Patent Publication No. 2008-510560

The rehabilitation system of Patent Document 1 repeats the control and measurement of a single motion a plurality of times, and repeats the control and measurement of a more complicated motion a plurality of times. However, by repeating the same exercise, the same stimulus is given to the brain, and the same exercise becomes accustomed. In this case, the stimulation to the brain becomes constant, and it may be difficult to improve the motor ability.

This disclosure is made in view of the above-mentioned conventional situation, and exercises that can be expected to improve athletic ability by giving different stimuli to the brain by variously moving any part of the trainer's body. Provides training systems, control methods, and programs.

One aspect of the present disclosure is an exercise training system that supports and analyzes exercises for a trainer, the control device that controls the movement of a slider on which a part of the trainer's human body is placed, and the above-mentioned. The control device comprises an analyzer that acquires brain function information regarding the brain function of the trainer during the training period of the trainer using the slider and analyzes the brain function information, and the control device relates to the slider. In the first period of the training period, the periodic movement is instructed, and in the second period of the training period, unlike the periodic movement, the time position, the movement amplitude, and the movement of the slider are different. Instructed to perform aperiodic movement in which at least one of the cycles is random, the analyzer extracts a plurality of feature quantities in the brain function information, and the brain function is based on the plurality of feature quantities. The score of the information is calculated, and based on the score, the brain function information is used as the first brain function information regarding the brain function of the trainer in the first period, or the trainer in the second period. It is an exercise training system that is classified into the second brain function information regarding the brain function of the brain.

One aspect of the present disclosure is a control method of an exercise training system that supports and analyzes exercise for a trainer, wherein a part of the trainer's human body is mounted on the control device of the exercise training system. The step of controlling the movement of the slider and the analyzer of the exercise training system acquire the brain function information regarding the brain function of the trainer during the training period of the trainer using the slider, and obtain the brain function information. The step of having the analysis and the step of controlling the movement of the slider by the control device instructs the slider to perform periodic movement during the first period of the training period. In the second period of the training period, unlike the periodic movement, a step of instructing the slider to perform an aperiodic movement in which at least one of the time position, the movement amplitude, and the movement cycle is random is performed. The step of analyzing the brain function information by the analyzer includes a step of extracting a plurality of feature quantities in the brain function information and a step of calculating a score of the brain function information based on the plurality of feature quantities. And, based on the score, the brain function information is used as the first brain function information regarding the trainer's brain function in the first period, or the brain function of the trainer in the second period. It is a control method including a step of classifying into 2 brain function information.

One aspect of the present disclosure is a program for causing a computer to execute a control method of an exercise training system that supports and analyzes exercise for a trainee, and the control method is a part of the human body of the trainee. A step of controlling the movement of the slider on which the slider is placed, a step of acquiring brain function information regarding the brain function of the trainer during the training period of the trainer using the slider, and a step of analyzing the brain function information. The step of controlling the movement of the slider is to instruct the slider to perform periodic movement during the first period of the training period and during the second period of the training period. Unlike the periodic movement, the brain function information is analyzed, including a step instructing the slider to perform an aperiodic movement in which at least one of the time position, the movement amplitude, and the movement cycle is random . The steps include a step of extracting a plurality of feature amounts in the brain function information, a step of calculating a score of the brain function information based on the plurality of feature amounts, and a step of calculating the brain function information based on the score. Includes a step of classifying into a first brain function information relating to the trainer's brain function in the first period or a second brain function information relating to the trainer's brain function in the second period. It is a program.

According to the present disclosure, various movements of any part of the trainer's body can be expected to give different stimuli to the brain and improve athletic ability.

Block diagram showing a configuration example of the exercise training system of the embodiment Perspective view showing an external example of an exercise training device A diagram showing an example of the command value of the position of the slider of the right foot and the detected value of the position of the slider of the right foot in chronological order. A diagram showing an example of the command value of the position of the slider of the left foot and the detected value of the position of the slider of the left foot in chronological order. A diagram showing an example of the load value applied to the right foot slider and the right hand action value for the right foot slider in chronological order. A diagram showing an example of the load value applied to the left foot slider and the right hand action value for the left foot slider in chronological order. The figure which shows the display example by a display device Flow chart showing an operation example of exercise training according to the exercise schedule by the motion control device Diagram to illustrate bilateral control Diagram to illustrate haptic control External view showing an example of a trainee wearing a brain measuring device on his head The figure which shows an example of light transmission using a light transmitting probe and a light receiving probe. The figure which shows an example of the light transmission position by a light transmission probe and the light reception position by a light receiving probe in the trainee's head. A diagram showing an example of a brain function measurement channel of the head in a trainee The figure which shows an example of the command value of the position of the slider of the right foot at the time of a simple repetitive motion. The figure which shows an example of the command value of the position of the slider of the left foot at the time of a simple repetitive motion. The figure which shows an example of the command value of the position of the slider 20a of the right foot at the time of a non-simple repetitive operation. The figure which shows an example of the command value of the position of the slider 20b of the left foot at the time of a non-simple repetitive operation. A diagram showing changes in oxygenated hemoglobin concentration in chronological order when simple repetitive movements are performed by an exercise training device. A diagram showing changes in oxygenated hemoglobin concentration in chronological order when non-simple repetitive movements are performed by an exercise training device. A flowchart showing an operation example when analyzing NIRS data by a brain analyzer. The figure which shows an example of the value of the 1st principal component and the 2nd principal component after the dimension reduction of the feature quantity for each operation of the exercise training apparatus in the 1st analysis example. A diagram showing an example of the relationship between the values of the first and second principal components and the classification score in the first analysis example. The figure which shows an example of the classification result of NIRS data in the 1st analysis example. The figure which shows an example of the classification result of the NIRS data of a trainee tr1. The figure which shows an example of the classification result of the NIRS data of a trainee tr2. The figure which shows an example of the classification result of the NIRS data of a trainee tr3. The figure which shows an example of the classification result of the NIRS data of a trainee tr4. The figure which shows an example of the classification result of the NIRS data of a trainee tr5. The figure which shows an example of the limitation of the motion range used for the analysis of the exercise training apparatus when the simple repetitive motion is performed by the exercise training apparatus. The figure which shows an example of the limitation of the motion range used for the analysis of the exercise training apparatus when the non-simple repetitive motion is performed by the exercise training apparatus. The figure which shows an example of the value of the 1st principal component and the 2nd principal component after the dimension reduction of the feature quantity for each operation of the exercise training apparatus in the 3rd analysis example. A diagram showing an example of the relationship between the values of the first and second principal components and the classification score in the third analysis example. The figure which shows an example of the classification result of NIRS data in the 3rd analysis example. The figure which shows the setting example of the selected brain function measurement channel The figure which shows an example of the value of the 1st principal component and the 2nd principal component after the dimension reduction of the feature quantity for each operation of the exercise training apparatus in the 4th analysis example. A diagram showing an example of the relationship between the values of the first and second principal components and the classification score in the fourth analysis example. The figure which shows an example of the classification result of NIRS data in the 4th analysis example. The figure which shows an example of the value of the 1st principal component and the 2nd principal component after the dimension reduction of the feature quantity for each operation of the exercise training apparatus in the 5th analysis example. A diagram showing an example of the relationship between the values of the first and second principal components and the classification score in the fifth analysis example. The figure which shows an example of the classification result of NIRS data in the 5th analysis example.

Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.

(First Embodiment)
FIG. 1 is a block diagram showing a configuration example of the exercise training system 100. The exercise training system 100 includes a motion control device 10, a servo amplifier 15 (15a to 15d), a slider 20 (20a to 20d), a motor 22 (motors 22a to 22d), a motion control generation device 25, a display device 30, and a brain analysis device. 40, a brain measuring device 45, and an acceleration sensor 50 (50a, 50b) are provided. The exercise training system 100 is used for exercise training of trainee tr, and is expected to be used for exercise training of healthy elderly people, exercise training of healthy people in a training gym, rehabilitation of patients, and the like. ..

Each device (for example, motion control device 10, motion control generation device 25, display device 30, brain analysis device 40, brain measurement device 45, acceleration sensor 50) included in the exercise training system 100 is configured to be included in the information processing device. May have a part.

For example, each device may include a processor, a memory, a communication device, an input device, and a display device. The processor may include a CPU (Central Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), and the like. The memory may include, for example, a primary storage device, a secondary storage device, and a tertiary storage device. The primary storage device may include a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The secondary storage device may include an HDD (Hard Disk Drive), an SSD (Solid State Drive), and the like. The tertiary storage device may include an optical disk, a flash memory, an SD card, and the like. The communication device may communicate via wireless or wired. The input device may include, for example, various buttons, keys, keyboards, touch panels, microphones, and the like. The display device may include an organic EL (Electro Luminescence) device, a liquid crystal device, and the like.

The motion control device 10 may have a position control function of the slider 20, a speed control function of the slider 20, a disturbance observation function, a bilateral control function, and a haptics control function. The motion control device 10 controls the movement of the slider 20 according to, for example, an arbitrary exercise schedule. In this case, the operation control device 10 calculates the torque command value for controlling the movement of the slider 20, and the torque according to the torque command value is passed through the servo amplifier 15 and the motor 22 that provides the driving force to the slider 20. The command may be transmitted to the slider 20. The exercise schedule may be stored in memory. A plurality of exercise schedules may be prepared according to the content of exercise training, the level of the trainer tr, and the like. Further, the operation control device 10 may include a DA converter, an AD converter, a counter, and the like.

The servo amplifier 15 controls the operation of the slider 20 via the motor 22 in response to the torque command acquired from the operation control device 10. The servo amplifier 15 includes a servo amplifier 15a for the right foot slider 20a, a servo amplifier 15b for the left foot slider 20b, a servo amplifier 15c for the right hand slider 20c, and a servo amplifier 15d for the left hand slider 20d. include.

The servo amplifier 15 may acquire a detected value of the angle of the motor 22 (for example, the rotation angle of the motor 22) from the motor 22 and control the motor 22 (for example, torque control). The angle of the motor 22 may be detected by an angle detector (encoder), and the angle detector may notify the servo amplifier 15 of the angle of the motor 22. The slider 20 may be moved by the rotation of the motor 22, and the position of the slider 20 may be controlled according to the angle control of the motor 22.

The servo amplifier 15 may receive a torque command value from the operation control device 10, generate a drive current for driving the motor 22, and send the drive current to the motor 22.

The slider 20 operates according to the control by the servo amplifier 15. Each slider 20 is operated by a driving force supplied from the motor 22. The slider 20 may reciprocate each part of the trainee tr's limbs (right foot, left foot, right hand, left hand) along a predetermined direction. The slider 20 includes a right foot slider 20a on which the right foot is placed, a left foot slider 20b on which the left foot is placed, a right hand slider 20c on which the right hand is placed, and a left hand slider 20d on which the left hand is placed. And, including. The placement of the hand on the hand slider 20 may include gripping the hand slider 20 by hand.

The slider 20 may be used for walking training. Therefore, during the period of periodic movement, the right foot slider 20a and the left foot slider 20b may move in opposite directions, and the right hand slider 20c and the left hand slider 20d may move in opposite directions. Further, during the period of periodic movement, the slider 20a of the right foot and the slider 20c of the right hand may move in opposite directions, and the slider 20b of the left foot and the slider 20d of the left hand may move in opposite directions.

The motor 22 is connected to the slider 20 via a coupling. The motor 22 receives a driving current from the servo amplifier 15, rotates the motor 22 to generate a driving force, and supplies the driving force to the slider 20. The slider 20 moves according to the driving force from the motor 22.

The motor 22 is provided corresponding to each slider 20. That is, the motor 22a supplies a driving force for moving the slider 20a. The motor 22b supplies a driving force for moving the slider 20b. The motor 22c supplies a driving force for moving the slider 20c. The motor 22d supplies a driving force for moving the slider 20d.

Therefore, the slider 20 may be independently controllable with multiple axes (for example, four axes of both hands and feet). That is, the exercise training device 5 may be capable of performing ascending / descending exercise training using a 4-axis actuator.

The motion control generator 25 generates, for example, an motion control program executed by the motion control device 10. The motion control program may include, for example, exercise schedule information for operating each slider 20. The motion control generator 25 may generate a motion control program according to the analysis result by the brain analyzer 40. The exercise according to the exercise schedule may include a period of performing periodic movements and a period of performing aperiodic movements. The aperiodic operation may include the time position of the slider 20 and the movement amplitude and movement cycle of the slider 20 being random. The periodic operation may include a periodic operation other than a simple reciprocating operation.

The motion control generator 25 generates control software (motion control program) by using, for example, MATLAB or Simulink, which is mathematical formula calculation software provided by MatheWorks. The motion control device 10 may download the generated motion control program, execute the motion control program, and give various instructions in the motion training. The motion control device 10 and the motion control generation device 25 may be integrally formed.

The display device 30 displays various data and various information. The display device 30 may display data or information used when generating an operation control program by the operation control generation device 25, for example. Further, the display device 30 may display information for assisting the trainer tr who performs the exercise training in the exercise training executed according to the motion control program.

The brain analyzer 40 acquires brain function information from the brain measuring device 45. The brain analyzer 40 analyzes the acquired brain function information and sends the analysis result to the motion control generation device 25.

The brain measuring device 45 measures the oxygen state on the surface of the brain using Near-Infrared Spectroscopy (NIRS). This measurement result may be color-mapped in real time as the activity state of the brain. The brain measuring device 45 sends the measurement result of the oxygen state on the brain surface to the brain analyzer 40.

The acceleration sensor 50 detects acceleration in three axial directions (for example, two directions perpendicular to each other in the horizontal direction and three directions perpendicular to the horizontal direction). A plurality of acceleration sensors 50 may exist, and include, for example, acceleration sensors 50a and 50b. The measurement result of the acceleration sensor 50 is sent to the motion control device 10. The accelerometer 50 may be provided in the exercise training device 5 (see FIG. 2). The acceleration sensor 50 may be attached to the trainer tr, and the acceleration may be detected in accordance with the exercise training of the trainer tr. The acceleration sensor 50 may detect acceleration in three axial directions, and the acceleration in each direction may be combined. Further, the acceleration sensor 50 may be used to give a stimulus from the outside to generate a torque command in response to the external stimulus. In this case, the trainer tr or the manager of the exercise training may vibrate the acceleration sensor 50, so that the motion control device 10 may manually give the slider 20 a motion different from the normal exercise schedule.

The accelerometer 50 may be gripped by the trainer tr's hand or may be attached to a part of the trainer tr's body. Vibration torque may be applied by the vibration of the acceleration sensor 50. Two accelerometers may be provided corresponding to each of the left foot and right foot sliders 20. The acceleration sensor 50a may be an acceleration sensor for the right hand, and the acceleration sensor 50b may be an acceleration sensor for the left hand.

Next, an example of the measurement method by the brain measuring device 45 will be described.

Humans take in information such as sight, hearing, touch, smell, and taste from sensory organs such as eyes and ears, convert them into electrical signals, and transmit them to the brain. Then, about 100 billion neurons in the brain determine the next action by transmitting and processing the information to each other via synapses. In this case, oxygenated hemoglobin (oxyHb) supplies oxygen via capillaries. The brain measuring device 45 measures the activity state of the brain in real time by capturing the enzyme state on the surface of the brain using NIRS, and analyzes the functional localization of the brain.

Hemoglobin, a blood component, scatters light. The degree of light absorption / scattering changes depending on whether or not oxygen is added to hemoglobin. The brain measuring device 45 measures the degree of absorption and scattering of light by hemoglobin, and measures the change in the concentration of oxygenated hemoglobin. In the optical measurement of living tissue, for example, near-infrared light having a wavelength of 700 to 900 nm is used. In the wavelength band of visible light (400 to 700 nm), hemoglobin and other biological constituents are largely absorbed. Further, in a wavelength band having a wavelength longer than that of near-infrared light, water absorption becomes large and light cannot travel in the living body. On the other hand, in the near-infrared wavelength band, light easily passes through the living body.

Absorption of light in the wavelength band of near-infrared light is mainly caused by oxygenated hemoglobin (oxyHb) and deoxygenated hemoglobin (deoxyHb). OxyHb and deoxyHb have different absorption spectra. Therefore, if the molar extinction coefficient of oxyHb and deoxyHb is known, the brain measuring device 45 can calculate the concentration change of oxyHb and the concentration change of deoxyHb by measuring the change in absorbance at two or more wavelengths.

Near-infrared light is relatively transparent to living organisms, but it is difficult to transmit it to the human head. Therefore, the brain measuring device 45 irradiates the brain with near-infrared light from the head surface using an optical fiber, and absorbs and scatters the light in the cerebral cortex, for example, an optical fiber on the head surface about 30 mm away. Condensate with. The irradiated light reaches a depth of about 20 mm from the head surface and is absorbed by hemoglobin in the cerebral cortex. Since the living body is a strong scatterer, the near-infrared rays introduced by the optical fiber are scattered by various parts of the tissue of the head, and a part of the scattered light reaches the optical fiber of the light receiving portion. The received light is guided to a photomultiplier tube and converted into an electric signal. The obtained electrical signals and data are also referred to as NIRS signals and NIRS data.

FIG. 2 is a perspective view showing an external example of the exercise training device 5. Here, walking training is illustrated as exercise training.

The exercise training device 5 has a guide member 26, a slider mechanism 27 (27a, 27b) for the foot attached to the guide member 26, and a gate shape arranged around the trainer tr for exercise training (for example, walking training). The frame fr and the hand slider mechanism 28 (28a, 28b) attached to the frame fr are included. The frame fr may have a shape other than the gate shape.

The slider mechanism 27 (27a, 27b) includes a foot slider 20 (20a, 20b), a motor 22 (22a, 22b), and a ball screw 24 (24a, 24b). The slider mechanism 28 (28c, 28d) includes a hand slider 20 (20c, 20d), a motor 22 (22c, 22d), and a ball screw (not shown). Here, the details of the slide method for the foot are illustrated, but the details of the slide method for the hand may be the same.

In the slider mechanisms 27a and 27b, the motors 22a and 22b rotate the ball screws 24a and 24b in the forward and reverse directions, respectively, so that the sliders 20a and 20b move diagonally forward and upward and diagonally backward and downward. Therefore, the left and right sliders 20a and 20b can be moved independently.

The angle of the guide member 26 with respect to the horizontal direction may be fixed or adjustable. For example, the exercise training device 5 enables walking training of sliding foot walking by moving the slider 20 in the front-back direction, and 0 moving diagonally forward and upward and diagonally backward and up and down enables walking training of exercise. You may do it.

The sliders 20a and 20b may move in opposite directions in synchronization with or without synchronization by driving the motor 22 for the sliders 20a and 20b. The sliders 20a and 20b may move in the same direction in synchronization or without synchronization by driving the motors 22 for the sliders 20a and 20b. By placing the foot on the sliders 20a and 20b, the trainee tr can experience the movement of the foot during walking in various walking methods.

The foot sliders 20a and 20b can be reciprocated in the forward diagonally upward direction and the backward diagonally downward direction. That is, the trainee tr can move up and down by the sliders 20a and 20b. The hand sliders 20c and 20 can reciprocate in the front-back direction. That is, the trainee tr can receive the support for maintaining the balance during the ascending / descending movement by the sliders 20a and 20b.

The right foot slider 20a and the right hand slider 20c may be moved so as to be in the same position in the horizontal direction. That is, the sliders 20a and 20c may be at the same position in each reciprocating orbit.

The foot sliders 20a and 20b are slidable in the direction of the arrow α shown in FIG. The hand sliders 20c and 20 are slidable in the direction of the arrow β shown in FIG. As a result, the trainee tr puts his / her foot on the sliders 20a and 20b and puts his / her hand on the sliders 20c and 20d, so that the foot moves with the sliders 20a and 20b and the hand moves with the sliders 20c and 20d. Therefore, you can experience the movement of the limbs during walking in a pseudo manner.

The right foot slider 20a and the right hand slider 20c and the left foot slider 20b and the left hand slider 20d may alternately and alternately perform reciprocating movements in the forward diagonally upward direction and the backward diagonally downward direction. This round-trip movement is a predictable movement for the trainee tr. The right foot slider 20a and the right hand slider 20c and the left foot slider 20b and the left hand slider 20d do not synchronize and move back and forth diagonally forward and diagonally upward and diagonally backward and downward with an unsteady and indefinite rhythm. You may go. This round-trip movement is unpredictable for the trainee tr.

The motor 22 of each slider 20 can move independently according to the control of each servo amplifier 15 without depending on the movement of other sliders. Therefore, the movement of the slider 20a of the right foot and the movement of the slider 20b of the left foot may be separate. For example, the cycle of movement of the slider 20a of the right foot and the cycle of movement of the slider 20b of the left foot may be different, and the range of movement of the slider 20a of the right foot (that is, the stride) and the range of movement of the slider 20b of the left foot may be different. It may be different.

Next, the details of the operation control device 10 will be described.

The motion control device 10 performs processes related to exercise training of the trainee tr (for example, data or information acquisition process, recording process, arithmetic process, display process, voice output process). The motion control device 10 may servo-control the motor 22 of each slider 20 to position the slider 20 at a target foot position. Multiple training courses for performing exercise training may be prepared. The results of walking training may be evaluated according to different evaluation criteria for each training course.

The memory of the motion control device 10 may hold various setting information related to exercise training. The setting information includes, for example, information regarding the movement of each slider 20 in the exercise schedule for exercise training (for example, the position of each slider 20 at the time from the start of exercise training, the slide cycle of each slider 20, and the slide width of each slider 20. , Etc.) may be included.

The memory of the motion control device 10 may record and store data obtained by walking training (for example, a torque command value, a value of the angle sensor 23 of the motor 22, and a torque detection value corresponding to the value of the angle sensor). The motion control device 10 may determine the torque command value at the next timing based on the torque command value of each slider 20 and the sensor value of the angle sensor 23 of each slider 20. Thereby, the exercise training system 100 may estimate how to apply the weight of the trainee tr and the walking balance based on the displacement between the slider 20a of the right foot and the slider 20b of the left foot. For example, the motion control device 10 may be adjusted as the subsequent torque command value so as to make it easier to balance the walking of the trainee tr.

The operation control device 10 acquires the command value of the position of the slider 20 and the detected value of the position of the slider 20. The command value of the position of the slider 20 is determined according to the exercise schedule, and is the value of the position where the sliders 20a and 20b should be arranged at a predetermined time. The detected value of the position of the slider 20 is a value based on the detected value of the angle sensor 23, and indicates the actual positions of the sliders 20a and 20. For example, the motion control device 10 may calculate the detected value of the position of the slider 20 based on the rotation angle of the motor 22 and the moving distance of the slider 20 for one round of the ball screw 24.

FIG. 3A is a diagram showing the command value of the position of the slider 20a of the right foot and the detected value of the position of the slider 20a of the right foot in time series, and is also referred to as FIG. Fg1. FIG. 3 is a diagram showing the command value of the position of the slider 20b of the left foot and the detected value of the position of the slider 20b of the left foot in time series, and is also referred to as FIG. Fg2. In FIGS. 3A and 3B, since the command value and the detection value at the positions of the sliders 20 (20a, 20b) match, the waveforms overlap and become one.

The operation control device 10 may acquire the value of the load (force) applied to the sliders 20 (20a, 20b). The load applied to the sliders 20a and 20b may be a load based on the rotation angle of the motor 22 detected by the angle sensor 23 (23a, 23b) of the motor 22 (22a, 22b). The load may be expressed in terms of force or may be converted into torque. Here, as an example, it is shown as a torque detection value. The torque detection value may reflect not only the torque command value but also the load applied to the slider 20 due to insufficient weight transfer of the trainee tr.

The motion control device 10 may acquire the value of the action of the right hand with respect to the slider 20. The action here means that the accelerometer 50 is artificially vibrated, for example. The value of the detected action may be, for example, a torque command value corresponding to the vibration of the acceleration sensor 50. That is, the action value is the magnitude of moving (for example, vibrating) the left-handed or right-handed accelerometer 50 (50a, 50b) converted into torque, and follows, for example, a normal exercise schedule. It is added to the torque command value.

FIG. 4A is a diagram showing the value of the load (force) applied to the slider 20a of the right foot and the value of the action of the right hand with respect to the slider 20a of the right foot in chronological order, and is also referred to as FIG. Fg3. FIG. 4B is a diagram showing the value of the load (force, weight) applied to the slider 20b of the left foot and the value of the action of the left hand with respect to the slider 20b of the left foot in chronological order, and is also referred to as FIG. Fg4.

FIG. 5 is a diagram showing a display example of the display device 30.

The display device 30 displays FIGS. fg1 to fg4. The display device 30 displays the balance meter fg5. The value of the balance meter fg5 may be the difference between the load applied to the slider 20a of the right foot and the load applied to the slider 20b of the left foot. The value of the balance meter fg5 may be the difference between the position of the slider 20a of the right foot and the position of the slider 20b of the left foot. This difference may be estimated by a disturbance observer.

The display device 30 displays the balance graph fg6. The balance graph fg6 is a graph in which the values of the balance meter fg5 are arranged in chronological order.

The motion control device 10 calculates a balance score based on the balance of the loads applied to the left and right sliders 20a and 20b. The motion control device 10 sets the penalty value as an initial value (for example, a value) when the difference between the load applied to the slider 20a of the right foot and the load applied to the slider 20b of the left foot is not 0, that is, when the left and right balance is lost. The balanced score may be calculated by subtracting from 100). The motion control device 10 may periodically (for example, constantly) calculate and update the balance score during exercise training. The display device 30 may display the balanced score fg7.

The motion control device 10 calculates the auxiliary force applied to at least one of the sliders 20a and 20b based on the balanced score. Auxiliary force is applied to the slider 20a or the slider 20b in order to equalize the load applied to the left and right sliders 20a and 20b. Therefore, the motor 22 for driving the sliders 20a and 20b may be given a torque command value for supplying an auxiliary force together with a torque command value supplied according to the exercise schedule. The operation control device 10 may calculate a parameter related to the torque command value for supplying the auxiliary force as a balance auxiliary value (value “15” in FIG. 5). The balance auxiliary value corresponds to the parameter _KB related to the bilateral control described later. The display device 30 may display the balance auxiliary value fg8.

Next, the operation of the exercise training system 100 will be described.

FIG. 6 is a flowchart showing an operation example of exercise training according to an exercise schedule.

The operation control device 10 instructs to display the operation command information related to the operation of each slider 20 (S11). In this case, the motion control device 10 may transmit a display instruction to the display device 30 via the motion control generation device 25. That is, the motion control device 10 may teach the trainee tr an image of the ascending / descending motion through the display by the display device 30. The operation command information may include, for example, a command value of the positions of the left and right sliders 20 and a command value of torque.

The motion control device 10 instructs the motor 22 via the servo amplifier 15 to alternately move both feet up and down according to the exercise schedule (S12). The ascending / descending motion of both feet alternately here is also referred to as a normal ascending / descending motion, and is an motion in which the amplitude and frequency at which the slider 20 moves are constant, that is, the motion of the slider 20 is periodic. In this case, the operation control device 10 normally generates a torque command value for raising and lowering, and transmits the torque command value to the motor 22 via the servo amplifier 15.

The motion control device 10 instructs the motor 22 via the servo amplifier 15 to alternately move both feet up and down according to the exercise schedule (S13). The alternating ascending / descending motion of both feet here is an motion in which the amplitude of movement of the slider 20 (20a, 20b) changes irregularly (for example, suddenly), that is, an motion in which the movement of the slider 20 becomes aperiodic. In this case, the operation control device 10 generates a torque command value whose amplitude changes irregularly and transmits it to the motor 22 via the servo amplifier 15. The aperiodic movement is an unexpected movement for the trainee tr.

The motion control device 10 instructs the motor 22 via the servo amplifier 15 to alternately move both feet up and down according to the exercise schedule (S14). The alternating ascending / descending operation of both feet here is an operation in which the frequency at which the slider 20 (20a, 20b) moves changes irregularly (for example, suddenly), that is, an operation in which the frequency of the slider 20 becomes aperiodic. In this case, the operation control device 10 generates a torque command value whose frequency changes irregularly and transmits it to the motor 22 via the servo amplifier 15.

The motion control device 10 instructs the motor 22 via the servo amplifier 15 to move both feet up and down at the same time according to the exercise schedule (S15). When both feet are raised and lowered at the same time, the positions of the sliders 20a and 20b in the left and right slider mechanisms 27a and 27b are the same. The simultaneous ascending / descending operation of both feet here may be an operation in which the amplitude and frequency of movement of the slider 20 (20a, 20b) are constant, that is, an operation in which the movement of the slider 20 is periodic. The simultaneous raising and lowering operation of both feet may include an operation in which the amplitude and frequency of movement of the slider 20 (20a, 20b) are indefinite.

The motion control device 10 instructs the motor 22 via the servo amplifier 15 to move one leg up and down according to the exercise schedule (S16). The raising and lowering motion of one foot may be the right foot or the left foot. Further, the running motion of one foot may be performed on the right foot and the left foot in order.

The operation control device 10 instructs the display device 30 to display operation command information (for example, position command value, torque command value) and operation results of the right foot and left foot (for example, position detection value, torque detection value). (S17). In this case, the motion control device 10 may transmit a display instruction to the display device 30 via the motion control generation device 25.

As described above, the exercise training system 100 can monitor and display the motion command in the motion of the exercise training according to the exercise schedule shown in FIG. 6, and can visually or audibly teach the trainee tr for the rehabilitation training. This operation command may be command information regarding the movement of the slider 20, and may be, for example, information indicating a position command value or a torque command value.

Further, in S12, the exercise training system 100 can cause the trainee tr to perform the expected ascending / descending motion alternately with both feet in accordance with this motion command. In S13, the exercise training system 100 allows the trainee tr to perform an ascending / descending motion in which the motion amplitude of one of the legs suddenly changes during the ascending / descending motion of both feet. In S14, the exercise training system 100 allows the trainee tr to perform an ascending / descending motion in which the operating frequency of one of the legs suddenly changes during the ascending / descending motion of both feet. In S15, the exercise training system 100 allows the trainee tr to perform an ascending / descending motion that suddenly changes from an alternating ascending / descending motion of both feet to a simultaneous ascending / descending motion. In S16, the exercise training system 100 allows the trainee tr to perform an elevating motion that suddenly changes from elevating both feet to elevating one foot.

Therefore, it is expected that similar stimulation will be given to the brain by performing the same walking training for a certain period of time (for example, alternating ascending / descending motions of both feet having the same amplitude and frequency, simultaneous ascending / descending motions of both feet, and ascending / descending motions of one leg). The exercise training system 100 can stimulate the brain through the movements of the legs subjected to different exercise training by incorporating motions having different amplitudes and frequencies irregularly. This can be confirmed by referring to the brain function information measured by the brain measuring device 45 and analyzed by the brain analyzing device 40.

Next, the details of the bilateral control by the operation control device 10 will be described. FIG. 7 is a diagram for explaining bilateral control. FIG. 7 describes a function related to bilateral control.

The operation control device 10 has each function of the disturbance observer 11 (11a, 11b), the PID control unit 12 (12a, 12b), and the bilateral control unit 13. In FIG. 7, the right component has an “a” at the end of the code, and the left component has a “b” at the end of the code.

The PID control unit 12 acquires position reference information from the memory of the operation control device 10. The position reference information may include a command value or a torque command value v1 of the position of the slider 20 determined according to the exercise schedule, as shown in FIGS. 3A and 3B, for example. The PID control unit 12 may acquire (for example, calculate) the torque command value v2 derived by the bilateral control unit 13. The torque command value v2 from the bilateral control unit 13 may be a value in consideration of the load balance on the left and right sliders.

The PID control unit 12 adds, for example, torque command values v1 and v2 based on the torque command value v1 from the position reference information and the torque command value v2 from the bilateral control unit 13, and provides them to the motors 22a and 22b. The torque command value v3 to be calculated is calculated.

The torque command value of the slider for the right foot is given an a at the end of the code, and the torque command value of the slider for the left foot is given a b at the end of the code. That is, in the drawing, the torque command values v1a, v1b, v2a, v2b, v3a, v3b are described.

The disturbance observer 11 may acquire the result of PID control from the PID control unit 12. The result of PID control may include a torque command value v3 to the motor 22 that drives the slider 20 for the right foot or the left foot. Further, the disturbance observer 11 may detect the actual load applied to the slider 20 for the right foot or the left foot. In this case, the disturbance observer 11 may acquire a torque detection value converted into torque. The disturbance observer 11 may acquire, for example, the angle of the motor 22 detected by the angle sensor 23, and calculate a torque value (torque detection value) or a load corresponding to the angle.

That is, the disturbance observer 11b may acquire the result (for example, the torque command value v3) of the PID control unit 12a for the right foot. The disturbance observer 11a may detect the actual load on the slider 20a for the right foot. Further, the disturbance observer 11b may acquire the result (for example, torque command value v3) of the PID control unit 12b for the left foot. The disturbance observer 11b may detect the actual load on the slider 20b for the left foot.

The disturbance observer 11a calculates the difference h_Dis1 between the result of the PID control unit 12a for the right foot (for example, torque command value v3) and the actual load (for example, torque detection value). The disturbance observer 11b calculates the difference h_Dis2 between the result of the PID control unit 12b for the left foot (for example, torque command value v3) and the actual load (for example, torque detection value).

The bilateral control unit 13 acquires the differences h_Dis1 and h_Dis2 from the disturbance observers 11a and 11b. The bilateral control unit 13 determines the torque command value v2 based on the difference h_Dis1 and _{h_Dis2} and the parameter KB used for PID control (for example, the value “15” in FIG. 5). The value of the parameter _KB may be input, for example, by the administrator who manages the exercise of the trainee tr via an operation unit (not shown), or a predetermined value is set and is held in the memory. May be good. The parameter KB indicates a magnification with respect to the difference between the difference _{h_Dis1} and the difference h_Dis2.

That is, the value of ( _{h_Dis1} -h_Dis2) × KB may be fed back to at least one of the left and right PID control units 12a and 12b as the torque command value v2. In this case, for example, the torque command value v2 is applied to the PID control unit 12 (for example, the right PID control unit 12a) corresponding to the slider 20 (for example, for the right) with the smaller load detected by the left and right sliders 20 (for example, for the left). Is sent. The PID control unit 12 generates a torque command value v3 based on the fed-back torque command value v2.

That is, for example, the motion control device 10 maintains the left-right balance by estimating the load fluctuation due to the unreasonable movement by the disturbance observer 11 and equalizing the load applied to the left and right axes (left and right sliders 20a and 20b). It has a bilateral control function.

In this way, the exercise training system 100 transfers the operating resistance of the range of motion that can be smoothly moved and the range of motion that is difficult to move smoothly in the body of the trainee tr to the load disturbance that returns to the motor 22. It can be estimated as a force. That is, when the trainee tr has insufficient weight transfer between the left and right sliders 20a and 20b, the load based on the body weight appears as the load disturbance force. The exercise training system 100 can estimate the load applied to the left and right feet by a disturbance observer or the like, and control (bilateral control) to maintain the left-right balance so that the loads applied to the left and right feet are equal. In this case, the exercise training system 100 can prevent the trainer tr from becoming an unreasonable exercise by limiting the movement of the slider 20 or the like. Therefore, the exercise training system 100 can reduce the possibility that the trainee tr loses the left-right balance and falls during exercise training (for example, during rehabilitation).

Next, the details of the haptics control by the motion control device 10 will be described. FIG. 8 is a diagram for explaining haptic control. FIG. 8 describes a function related to haptic control.

The operation control device 10 has the functions of the PID control unit 12 (12a, 12b) and the haptics control unit 14. In FIG. 8, the right component has an “a” at the end of the code, and the left component has a “b” at the end of the code.

The haptics control unit 14 acquires the acceleration detected by the acceleration sensor 50 (50a, 50b). The haptics control unit 14 derives (for example, calculates) a torque command value v4 based on the acquired acceleration and sends it to the PID control unit 12. In this case, the torque command value v4a based on the acceleration sensor 50a may be sent to the right PID control unit 12a, and the torque command value v4b based on the acceleration sensor 50b may be sent to the right PID control unit 12b. The relationship between the magnitude of the acceleration and the magnitude of the torque command value v2 may be predetermined and may be held in a memory and referred to.

The PID control unit 12 acquires position reference information from the memory of the operation control device 10. The position reference information may include a command value or a torque command value v1 of the position of the slider 20 determined according to the exercise schedule, as shown in FIGS. 3A and 3B, for example. The PID control unit 12 may acquire (for example, calculate) the torque command value v4 derived by the haptics control unit 14. The torque command value v2 from the haptics control unit 14 may be a value in consideration of the vibration given to the acceleration sensor 50.

The PID control unit 12 is provided to the motors 22a and 22b by adding, for example, torque command values v1 and v4 based on the torque command value v1 from the position reference information and the torque command value v4 from the haptics control unit 14. The torque command value v5 is calculated.

The torque command value of the slider 20a for the right foot is suffixed with a, and the torque command value of the slider 20b for the left foot is suffixed with b. That is, in the drawing, the torque command values v1a, v1b, v4a, v4b, v5a, v5b are described.

Therefore, the torque command value v1 based on the position reference information and the torque command value v4 based on the acceleration are input to the PID control unit 12. The operation control device 10 performs PID control based on the torque command value v1 based on the position reference information and the torque command value v4 based on the acceleration, calculates the torque command value v5 to the motor 22, and sends it to the motor 22. .. The motor 22 moves the slider 20 by driving according to the torque command value v5. Therefore, the trainee tr can experience, for example, the vibration artificially applied to the acceleration sensor 50 as the vibration of the slider 20.

That is, the motion control device 10 has a haptics control function that sensuously transmits an external signal measured by the acceleration sensor 50 or the like to the drive shaft that drives the motor 22 as vibration torque.

As described above, the exercise training system 100 can be estimated as a load disturbance force (for example, vibration torque due to vibration) introduced from a viewpoint different from PID control based on position reference information, based on the acceleration acquired from the acceleration sensor 50. The exercise training system 100 can perform control (haptics control) for sensuously transmitting this load disturbance force.

As described above, the exercise training system 100 can suppress the unreasonable exercise by the trainer tr by implementing the bilateral control and the haptics control, improve the balance maintenance performance, and suppress the fall of the trainer tr. Therefore, the exercise training system 100 can improve the safety during exercise training.

Further, the exercise training system 100 can realize a balance maintaining function, an external stimulating function, and a load estimation function by performing bilateral control and haptic control. As a result, the trainer tr can avoid unreasonable rehabilitation training.

Next, the details of the brain measuring device 45 and the brain analysis device 40 will be described.

FIG. 9 is an external view showing an example of a trainee tr in which the mounting portion 46 of the brain measuring device 45 is mounted on the head. FIG. 10 is a diagram showing an example of light transmission using the light transmitting probe pr1 and the light receiving probe pr2.

The brain measuring device 45 includes a mounting unit 46, a light transmitting probe pr1, a light receiving probe pr2, and a processing unit 47. The mounting portion 46 is mounted on the head of the trainee tr. The light transmission probe pr1 is arranged in the mounting portion 46 and transmits (irradiates) light toward the head of the trainee tr. The light receiving probe pr2 receives light from the head of the trainee tr. One or more of the light transmitting probe pr1 and the light receiving probe pr2 are provided. The light transmitted from the light transmission probe pr1 may be near-infrared light. The light received by the light receiving probe pr2 may be near infrared light. The light transmitting probe pr1 and the light receiving probe pr2 may be connected by wire (for example, an optical fiber).

The processing unit 47 performs various processes by executing a program by the processor. For example, the processing unit 47 may control the light transmission timing by the light transmission probe pr1 and the light reception timing by the light reception probe pr2. The light transmission by the light transmission probe pr1 and the light reception by the light reception probe pr2 may be performed periodically (for example, always) during the exercise training period. The processing unit 47 acquires the NIRS signal as a measurement result based on the light received by the light receiving probe pr2 using the communication function. The processing unit 47 may transmit to another device (for example, the brain analysis device 40) via wired or wireless. The NIRS signal may include information on the concentration and concentration change of oxygenated hemoglobin (oxyHb) (see, for example, FIGS. 13E and 13F).

FIG. 11 is a diagram showing an example of the light transmission position Tp by the light transmission probe pr1 and the light reception position Rp by the light reception probe pr12 on the head of the trainee tr. The “p” of the light transmission position Tp and the light reception position Rp is an integer of 1 or more. That is, it is indicated by the light transmission positions T1, T2, ..., And the light reception positions R1, R2, .... In FIG. 11, there are 16 light transmitting positions Tp and 16 light receiving positions Rp. That is, there are 16 transmission channels and 16 reception channels.

FIG. 12 is a diagram showing an example of the brain function measurement channel CHq of the head in the trainee tr. “Q” of the brain function measurement channel CHq is an integer of 1 or more. That is, it is indicated by brain function measurement channels CH1, CH2, .... In FIG. 12, there are 52 brain function measurement channels CHq. The brain function measurement channel Chq is a channel for measuring brain function at a position between an arbitrary light transmission position Tp and a light reception position Rp located in the vicinity thereof.

Therefore, the brain measuring device 45 utilizes the light transmitting signal transmitted by the light transmitting probe pr1 and the light receiving signal received by the light receiving probe pr2, and a plurality of (for example, 52 or a part thereof) brain function measuring channels Chq. The cerebral blood flow oxygen concentration (for example, the concentration of oxygenated hemoglobin) may be measured (measured) during exercise training (for example, during ascending / descending motion). Information on cerebral blood flow enzyme concentration is an example of brain function information.

The number of light transmitting channels and light receiving channels is 16, and the number of brain function measuring channels CHq is 52. That is, the light corresponding to the light from one light transmission channel (for example, T6) may be acquired by a plurality of light receiving channels (for example, R2, R6, R7, R10). Further, for example, when the light is received by the light receiving channel R6 in response to the light transmitted by the light transmitting channel T6, the information of the brain function measuring channel CH18 is obtained.

Next, a specific example of data based on the NIRS signal will be described.

The exercise training device 5 operates in response to an operation command from the operation control device 10. This action includes a simple repetition action (repetition) and a non-simple repetitive action (illusion). The simple repetitive motion is an example of a periodic motion, and the non-simple repetitive motion is an example of an aperiodic motion different from the periodic motion.

In the simple iterative operation, the sliders 20a to 20d move in a simple iterative manner and move periodically. FIG. 13A is a diagram showing an example of a command value of the position of the slider 20a of the right foot during a simple repetitive operation. FIG. 13B is a diagram showing an example of a command value of the position of the slider 20b of the left foot during a simple repetitive operation. That is, FIGS. 13A and 13B are examples of operation commands at the time of simple repetitive operation.

In the non-simple iterative operation, the sliders 20a to 20d move in a non-simple iterative manner and move aperiodically. FIG. 13C is a diagram showing an example of a command value of the position of the slider 20a of the right foot during the non-simple repetitive operation. FIG. 13D is a diagram showing an example of a command value of the position of the slider 20b of the left foot during the non-simple repetitive operation. That is, FIGS. 13A and 13B are examples of operation commands during non-simple repetitive operation.

FIG. 13E is a diagram showing changes in oxygenated hemoglobin concentration in chronological order when a simple repetitive motion is performed by the exercise training device 5. That is, FIG. 13E shows an example of the measured value of the NIRS signal during the simple repetitive operation.

FIG. 13F is a diagram showing changes in oxygenated hemoglobin concentration in time series when a non-simple repetitive motion is performed by the exercise training device 5. That is, FIG. 13F shows an example of the measured value of the NIRS signal during the non-simple repetitive operation.

In FIGS. 13E and 13F, among the plurality of brain function measurement channels CHq, the values of the NIRS signals that measure the brain function of the brain function measurement channel CHq corresponding to the position of the head that controls the brain function near the right foot motor cortex. It is shown.

The brain analyzer 40 has information on operation commands (for example, command values of the position of the slider 20 and command values of torque) and information on operation results (for example, detection values of the position of the slider 20 and detection values of torque) from the exercise training device 5. May be obtained. Therefore, the brain analyzer 40 can analyze the brain function at the time of exercise training by observing the operation of the exercise training device 5 and the NIRS signal from the brain measuring device 45 in synchronization with the timing.

The brain analyzer 40 can acquire actually measured values of a plurality of (for example, a large number) NIRS signals from the brain measuring device 45. The brain analyzer 40 analyzes the relationship between the motion of the exercise training device 5 (simple repetitive motion and non-simple repetitive motion) and the measured values of the plurality of NIRS signals acquired during this motion.

The NIRS signal is measured at various measurement points (light transmission position Tp, light reception position Rp, brain function measurement channel CHq) in various trainer trs. Further, the NIRS signal indicates, for example, the oxygenated hemoglobin concentration, but since this signal varies from person to person, the measured values of the NIRS signals for a plurality of trainee trs who have performed the same exercise training may be different values. Therefore, the brain analyzer 40 looks at the relationship between the movements of the exercise training device 5 (simple repetitive movements and non-simple repetitive movements) and the actually measured values of the plurality of NIRS signals acquired during this movement. It can be difficult to derive.

The brain analyzer 40 uses a machine learning (for example, deep learning) technique to perform the motion of the exercise training device 5 (simple repetitive motion and non-simple repetitive motion) and the measured values of a plurality of NIRS signals acquired during this motion. You may analyze the relationship between and. As a result, the brain analyzer 40 can perform the motion of the exercise training device 5 (simple repetitive motion and non-simple repetitive motion) and a plurality of NIRS signals acquired during this motion even when it is difficult to derive the relationship at first glance. The relationship with the measured value of can be derived.

In machine learning, the brain analyzer 40 may extract feature points in measured values of NIRS signals according to, for example, CNN (Convolutional Neural Network). The brain analyzer 40 may diagnose the brain function of the trainee tr during exercise training based on sensor signal analysis (for example, analysis of actually measured values of NIRS signals) and machine learning.

Next, using the following first analysis example to fifth analysis example, the operation of the exercise training device 5 (simple repetitive operation and non-simple repetitive operation) and the measured values of a plurality of NIRS signals acquired during this operation are used. Consider the relationship between.

(Example of first analysis)
In the first analysis example, the following exercise training data can be obtained. The duration of one exercise training is 91 seconds, and 1784 samples are obtained during this period. The number of brain function measurement channels CHq in which brain function is measured is 52. Exercise training is carried out three times in a row. The motion commands for exercise training include two types of commands: simple repetitive motion commands and non-simple repetitive motion commands. There are five trainee trs. The NIRS data is not distinguished for each trainer tr.

FIG. 14 is a flowchart showing an operation example at the time of analysis of NIRS data by the brain analyzer 40. The operation example of FIG. 14 may be the same in the first analysis example to the fifth analysis example.

First, the brain analyzer 40 acquires a NIRS signal, and acquires a plurality of NIRS data which are data based on the NIRS signal (S21). Here, for example, 1784 samples × 52 channels × 3 times continuous × 2 types of operation commands × 5 people, NIRS data for minutes are input. The NIRS data may be an actually measured value of the NIRS signal.

The brain analyzer 40 extracts the feature amount of the input NIRS data (S22). The brain analyzer 40 may extract features using an autoregressive model. The autoregressive model is defined by the following equation (1).

なお、Ｘｎ：ＮＩＲＳデータ、ａ_０：定数項、ａｉ：モデルのパラメータ、ε_ｔ：ホワイトノイズ、である。Ｘｎは、例えば、時系列に並ぶ９１秒間に含まれる１７８４個のサンプル値（例えば図１３Ｅ、図１３Ｆ参照）を数式化したものである。脳分析装置４０は、Ｎ＝１０として、ＮＩＲＳデータを１０次元でモデル化し、重み（ａ_０，ａｉ）を同定（推定）してよい。式（１）で定義される自己回帰モデルは、５２（脳機能計測チャネル数）×２（動作指令の数）＝１０４個、生成されてよい。

Xn: NIRS data, a ₀ : constant term, ai: model parameter, ε _t : white noise. Xn is, for example, a mathematical expression of 1784 sample values (see, for example, FIGS. 13E and 13F) included in 91 seconds arranged in a time series. The brain analyzer 40 may model NIRS data in 10 dimensions and identify (estimate) weights (a ₀ , ai) with N = 10. The autoregressive model defined by the equation (1) may be generated in 52 (number of brain function measurement channels) × 2 (number of operation commands) = 104.

The brain analyzer 40 may apply each NIRS data to the equation (1) of the autoregressive model and perform dimension reduction of the feature amount by AutoEncoder (S23). The brain analyzer 40 may be dimensionally reduced by the AutoEncoder, for example, so that the 10-dimensional model becomes a two-dimensional model. The two-dimensional model may include a first principal component and a second principal component. By using the brain analyzer 40 as a two-dimensional model, the amount of information of the data to be handled can be reduced and it can be easily visualized. In addition, learning accuracy tends to improve by performing machine learning using data with reduced dimensions. The first principal component and the second principal component may correspond to the component having the largest influence and the component having the second largest influence as a result of the principal component analysis. The process of S23 may be omitted.

FIG. 15 is a diagram showing an example of the values of the first principal component and the second principal component after the dimension reduction of the feature amount for each operation of the exercise training device 5 in the first analysis example. The first principal component and the second principal component are the components having the first and second largest influence on the entire NIRS data among the plurality of components in the feature amount of the NIRS data.

The brain analyzer 40 classifies the feature quantities (for example, the first principal component and the second principal component) of each dimension-reduced NIRS data according to the motion (simple repetitive motion and non-simple repetitive motion) of the exercise training device 5 (simple repetitive motion and non-simple repetitive motion). S24).

The brain analyzer 40 may generate a classification model by a one-class SVM (Support Vector Machine) and perform the above classification according to the classification model. One-class SVM is a method of unsupervised learning. In the One-class SVM, the brain analyzer 40 sets one of the NIRS data during the simple iterative operation and the NIRS data during the non-simple iterative operation as a normal value, and sets the NIRS data and the non-simple repeat operation during the simple iterative operation as normal values. The above classification may be performed by using any one of the NIRS data during operation as an abnormal value.

Here, it is mainly exemplified that the NIRS data at the time of simple repetitive operation is set as a normal value and the data of any one of the NIRS data at the time of non-simple repetitive operation is set as an abnormal value. If the classification is not ideal, the NIRS data during the simple repetitive operation may be classified as an abnormal value, and the NIRS data during the non-simple repetitive operation may be classified as a normal value.

The brain analyzer 40 maps each NIRS data by the values of the first principal component and the second principal component according to the one-class SVM, and calculates the classification score in the case of the values of the first principal component and the second principal component. .. The classification score is also simply referred to as a score.

FIG. 16 is a diagram showing an example of the relationship between the values of the first principal component and the second principal component and the classification score in the first analysis example. For example, data other than the simple repetitive operation (that is, the NIRS data of the non-simple repetitive operation) is determined based on the five NIRS data at the time of the simple repetitive operation. The larger the classification score, the more the data indicates an abnormal value, that is, the NIRS data of non-simple repetitive motion. Therefore, the brain analyzer 40 has a normal value (NIRS data of simple repetitive motion) or an abnormal value (NIRS data of non-simple repetitive motion) depending on whether the classification score is equal to or less than the threshold value th. May be determined. The threshold value th may be a fixed value or a variable value. The brain analyzer 40 may set the threshold value th to an arbitrary value. Thereby, for example, the manager who manages the training of the trainer tr can set the threshold value th to a value that can classify the NIRS data into a desired state via the operation unit of the brain analyzer 40 or the like.

なお、式（２）中の「γρ（ｆ（ｘｉ））」は、損失係数であるが、損失係数は式（３）で表されてよい。

Specifically, the brain analyzer 40 may calculate the classification score based on, for example, the Surf function. The classification score (Score) may be calculated according to the formula (2), for example.

The "γρ (f (xi))" in the equation (2) is a loss coefficient, but the loss coefficient may be expressed by the equation (3).

Using the formula for calculating the classification score in Eq. (2), classify the NIRS data during non-simple repetitive operation so that the cluster of unexpected data (NIRS data during non-simple repetitive operation) becomes a cohesive range. The value of ρ in Eqs. (2) and (3) may be adjusted so that the score falls within a predetermined range. This ρ may correspond to the above threshold value th. Therefore, ρ may be adjusted so that the value of the classification score calculated by the equation (2) is large and the value of the loss function shown by the equation (3) is small.

The accuracy of ρ adjustment can be affected by the accuracy of classification by classification score. For example, in FIG. 16, the brain analyzer 40 determines that an abnormal value (NIRS data at the time of non-simple repetitive operation) is obtained when the classification score (Score) is a value of 5 or more by setting the threshold value th = 5 of the classification score. However, when the classification score (Score) is less than the value 5, it may be determined as a normal value (NIRS data at the time of simple repetitive operation). Even in this case, the normal value may be included in the range where the classification score is 5 or more, or the abnormal value may be included in the range where the classification score is less than 5. In other words, the classification may be incorrect.

Further, the brain analyzer 40 may accumulate the results of classification according to the One-class SVM, generate a classification model (classification algorithm), and perform machine learning. As a result, the brain analyzer 40 can improve the classification accuracy of the feature amount of the NIRS data for each operation of the exercise training device 5 performed at the timing after the machine learning.

FIG. 17 is a diagram showing an example of the classification result of NIRS data in the first analysis example. FIG. 17 shows the classification results of 5 NIRS data in the simple repetitive operation and 260 NIRS data in the non-simple repetitive operation. Here, five sample data may be randomly extracted from the NIRS data during the simple iterative operation, and a classification model may be generated based on the extracted sample data. According to the generated classification model, the brain analyzer 40 includes NIRS data as a result of simple repetitive motion (repetition) as data without stimulation, and NIRS data as a result of non-simple repetitive motion (illusion) as data with stimulation. It is classified into. The method of this classification is the same in the first analysis example to the fifth analysis example.

As shown in FIG. 17, of the 5 NIRS data (actual data) during the simple iterative operation, 4 NIRS data are classified into the data during the simple iterative operation, and 1 NIRS data is the non-simple iterative operation. It was classified as time data. That is, the classification correct answer rate of NIRS data at the time of simple repetitive operation is 80%. Further, out of 260 NIRS data (actual data) at the time of non-simple repetitive operation, 155 NIRS data are classified into data at the time of simple repetitive operation, and 103 NIRS data are classified as data at the time of non-simple repetitive operation. It was classified. That is, the classification correct answer rate of NIRS data at the time of non-simple repetitive operation is 40%.

The brain analyzer 40 presents the classification results of the derived NIRS data. In this case, the brain analyzer 40 may display, for example, the information shown in FIG. 17 on the display device 30 as a classification result of NIRS data.

(Second analysis example)
In the second analysis example, the following exercise training data can be obtained. The duration of one exercise training is 91 seconds, and 1784 samples are obtained during this period. The number of brain function measurement channels CHq in which brain function is measured is 52. Exercise training is carried out three times in a row. The motion commands for exercise training include two types of commands, a command for simple repetitive motion and a command for non-simple repetitive motion. There are five trainee trs. The NIRS data is distinguished for each trainer tr, and the above relationship is analyzed for each individual trainer tr.

In the second analysis example, the NIRS data may be analyzed by the same method as in FIG. The NIRS data acquired in S12 is distinguished for each trainer tr, and for example, 1784 samples × 52 channels × 3 consecutive times × 2 types of operation commands, and minutes of NIRS data are input for each trainer. ..

18A to 18E are diagrams showing an example of the classification result of NIRS data for each trainer tr (trainers tr1 to tr5) in the second analysis example. FIG. 18A shows the classification result for the trainee tr1. FIG. 18B shows the classification result for the trainee tr2. FIG. 18C shows the classification results for the trainee tr3. FIG. 18D shows the classification results for the trainee tr4. FIG. 18E shows the classification results for the trainee tr5.

As for the trainee tr1, as shown in FIG. 18A, 4 NIRS data out of 5 NIRS data (actual data) at the time of simple iterative operation are classified into data at the time of simple iterative operation, and 1 NIRS. The data was categorized as non-simple iterative data. That is, the classification correct answer rate of NIRS data at the time of simple repetitive operation is 80%. Of the 52 NIRS data (actual data) during non-simple repetitive operation, 32 NIRS data are classified as data during simple repetitive operation, and 20 NIRS data are classified as data during non-simple repetitive operation. It was classified. That is, the classification correct answer rate of NIRS data at the time of non-simple repetitive operation is 38%.

As for the trainee tr2, as shown in FIG. 18B, 4 NIRS data out of 5 NIRS data (actual data) at the time of simple iterative operation are classified into data at the time of simple iterative operation, and 1 NIRS. The data was categorized as non-simple iterative data. That is, the classification correct answer rate of NIRS data at the time of simple repetitive operation is 80%. Of the 52 NIRS data (actual data) during non-simple repetitive operation, 29 NIRS data are classified as data during simple repetitive operation, and 23 NIRS data are classified as data during non-simple repetitive operation. It was classified. That is, the classification correct answer rate of NIRS data at the time of non-simple repetitive operation is 44%.

As for the trainee tr3, as shown in FIG. 18C, 4 NIRS data out of 5 NIRS data (actual data) at the time of simple iterative operation are classified into data at the time of simple iterative operation, and 1 NIRS. The data was categorized as non-simple iterative data. That is, the classification correct answer rate of NIRS data at the time of simple repetitive operation is 80%. Of the 52 NIRS data (actual data) during non-simple repetitive operation, 31 NIRS data are classified as data during simple repetitive operation, and 21 NIRS data are classified as data during non-simple repetitive operation. It was classified. That is, the classification correct answer rate of NIRS data at the time of non-simple repetitive operation is 40%.

As for the trainee tr4, as shown in FIG. 18D, 4 NIRS data out of 5 NIRS data (actual data) at the time of simple iterative operation are classified into data at the time of simple iterative operation, and 1 NIRS. The data was categorized as non-simple iterative data. That is, the classification correct answer rate of NIRS data at the time of simple repetitive operation is 80%. Of the 52 NIRS data (actual data) during non-simple repetitive operation, 31 NIRS data are classified as data during simple repetitive operation, and 21 NIRS data are classified as data during non-simple repetitive operation. It was classified. That is, the classification correct answer rate of NIRS data at the time of non-simple repetitive operation is 40%.

As for the trainee tr5, as shown in FIG. 18E, 4 NIRS data out of 5 NIRS data (actual data) at the time of simple iterative operation are classified into data at the time of simple iterative operation, and 1 NIRS. The data was categorized as non-simple iterative data. That is, the classification correct answer rate of NIRS data at the time of simple repetitive operation is 80%. Of the 52 NIRS data (actual data) during non-simple repetitive operation, 30 NIRS data are classified as data during simple repetitive operation, and 22 NIRS data are classified as data during non-simple repetitive operation. It was classified. That is, the classification correct answer rate of NIRS data at the time of non-simple repetitive operation is 42%.

As described above, in the second analysis example, the classification correct answer rate of the NIRS data at the time of non-simple repetitive operation is within the range of 38% to 44%, and it can be understood that there is almost no individual difference in the classification result. Further, when the data is not distinguished for each trainer tr in the first analysis example, the classification correct answer rate of the NIRS data at the time of non-simple repetitive operation is 40%, so that the data may be distinguished for each trainer tr. It can be understood that there is almost no difference in the classification results even without it.

(3rd analysis example)
In the third analysis example, the following exercise training data can be obtained. Limitation of the time range of the measured values of the NIRS signal used for analysis is taken into consideration. The duration of one exercise training is 60 seconds, and 1177 samples are obtained during this period. The number of brain function measurement channels CHq in which brain function is measured is 52. Exercise training is carried out three times in a row. The motion commands for exercise training include two types of commands: simple repetitive motion commands and non-simple repetitive motion commands. There are five trainee trs. The NIRS data is not distinguished for each trainer tr.

FIG. 19A is a diagram showing an example of limitation of the operation range used for the analysis of the exercise training device 5 when the simple repetitive motion is performed by the exercise training device 5. FIG. 19B is a diagram showing an example of limitation of the operation range used for the analysis of the exercise training device 5 when the non-simple repetitive motion is performed by the exercise training device 5.

In FIGS. 19A and 19B, the NIRS signal during exercise training associated with the movement between the time 20 seconds (s) and the time 80 seconds (s) (period P1) is used for the analysis. This time range corresponds to a range in which the movement cycle of the slider 20 and the movement distance (movement amplitude) in the movement cycle of the slider 20 are different between the simple repetitive motion (repetitive) and the non-simple repetitive motion (illusion). As described above, in the third analysis example, the limitation of the time range of the measured value of the NIRS signal used for the analysis is taken into consideration. The same applies to the fifth analysis example described later.

In the third analysis example, the NIRS data may be analyzed by the same method as in FIG. The NIRS data acquired in S12 is time-limited, and for example, 1177 samples × 52 channels × 3 consecutive times × 2 types of operation commands × 5 people, minutes of NIRS data are input.

FIG. 20 is a diagram showing an example of the values of the first principal component and the second principal component after the dimension reduction of the feature amount for each operation of the exercise training device 5 in the third analysis example. In the third analysis example, the NIRS data for five trainers tr5 is used, and the NIRS data is analyzed by limiting the time range in which the brain stimulation is considered to be affected.

FIG. 21 is a diagram showing an example of the relationship between the values of the first principal component and the second principal component and the classification score in the third analysis example. FIG. 22 is a diagram showing an example of the classification result of NIRS data in the third analysis example.

In the third analysis example, as shown in FIG. 22, four NIRS data out of the five NIRS data (actual data) at the time of the simple iterative operation are classified into the data at the time of the simple iterative operation, and one NIRS. The data was categorized as data during non-simple iterative operation. That is, the classification correct answer rate of NIRS data at the time of simple repetitive operation is 80%. Further, of the 260 NIRS data (actual data) during the non-simple repetitive operation, 146 NIRS data are classified into the data during the simple repetitive operation, and 112 NIRS data are classified into the data during the non-simple repetitive operation. It was classified. That is, the classification correct answer rate of NIRS data at the time of non-simple repetitive operation is 43%.

(4th analysis example)
In the fourth analysis example, the following exercise training data can be obtained. The duration of one exercise training is 91 seconds, and 1784 samples are obtained during this period. The number of brain function measurement channels CHq in which brain function is measured is 12, and the number of brain function measurement channels (the number of measurement points in the head) is limited. Exercise training is carried out three times in a row. The motion commands for exercise training include two types of commands, a command for simple repetitive motion and a command for non-simple repetitive motion. There are five trainee trs. The NIRS data is not distinguished for each trainer tr.

In the fourth analysis example, 12 brain function measurement channels that are presumed to be closely related to exercise are selected from a plurality of brain function measurement channels, and NIRS data measured by the selected brain function measurement channels are analyzed. To.

FIG. 23 is a diagram showing a setting example of the selected brain function measurement channel. FIG. 23 shows brain function measurement channels in the primary somatosensory cortex and the higher motor cortex. The primary somatosensory cortex includes the sensorimotor cortex (SMC). Higher motor areas include premotor cortex (PMC), supplementary motor area (SMA), and premotor area (preSMA). The brain function measurement channels of SMC are brain function measurement channels CH33 and CH34. The brain function measurement channels of PMC are brain function measurement channels CH10, CH17, CH25, CH12, CH20, and CH27. The brain function measurement channels of SMA are brain function measurement channels CH18 and CH19. The brain function measurement channels of preSMA are brain function measurement channels CH3, CH18, CH4, and CH19. The selected channel is an example, and other brain function measurement channels may be selected.

In the fourth analysis example, the NIRS data may be analyzed by the same method as in FIG. The NIRS data acquired in S12 is limited to measurement points (channels). For example, 1784 samples x 12 channels x 3 consecutive times x 2 types of operation commands x 5 people, and NIRS data for 5 people are input. To.

FIG. 24 is a diagram showing an example of the values of the first principal component and the second principal component after the dimension reduction of the feature amount for each operation of the exercise training device 5 in the fourth analysis example. FIG. 25 is a diagram showing an example of the relationship between the values of the first principal component and the second principal component and the classification score in the fourth analysis example. FIG. 26 is a diagram showing an example of the classification result of NIRS data in the fourth analysis example.

In the fourth analysis example, as shown in FIG. 26, four NIRS data out of the five NIRS data (actual data) at the time of the simple iterative operation are classified into the data at the time of the simple iterative operation, and one NIRS. The data was categorized as data during non-simple iterative operation. That is, the classification correct answer rate of NIRS data at the time of simple repetitive operation is 80%. Further, of the 60 NIRS data (actual data) at the time of non-simple repetitive operation, 8 NIRS data are classified into the data at the time of simple repetitive operation, and 52 NIRS data are classified into the data at the time of non-simple repetitive operation. It was classified. That is, the classification correct answer rate of NIRS data at the time of non-simple repetitive operation is 87%.

As described above, in the fourth analysis example, it can be understood that the selection of the measurement point (that is, the brain function measurement channel) has a great influence on the classification accuracy of the NIRS data. In addition, it can be said that it has succeeded in stimulating the brain part, which is presumed to be closely related to exercise, through exercise training. Moreover, it can be said that the stimulus of non-simple repetitive motion and the stimulus of simple repetitive motion are recognized separately.

(5th analysis example)
In the fifth analysis example, the following exercise training data can be obtained. Limitation of the time range of the measured values of the NIRS signal used for analysis is taken into consideration. The duration of one exercise training is 60 seconds, and 1177 samples are obtained during this period. The number of brain function measurement channels CHq in which brain function is measured is 12, and the number of brain function measurement channels (the number of measurement points in the head) is limited. Exercise training is carried out three times in a row. The motion commands for exercise training include two types of commands: simple repetitive motion commands and non-simple repetitive motion commands. There are five trainee trs. That is, in the fifth analysis example, both the time limit of the third analysis example and the measurement point (channel) limitation of the fourth analysis example are taken into consideration. The NIRS data is not distinguished for each trainer tr.

In the fifth analysis example, the NIRS data may be analyzed by the same method as in FIG. The NIRS data acquired in S12 is limited in time and measurement points. For example, 1177 samples x 12 channels x 3 consecutive times x 2 types of operation commands x 5 people, and NIRS data for 5 people are input. To.

FIG. 27 is a diagram showing an example of the values of the first principal component and the second principal component after the dimension reduction of the feature amount for each operation of the exercise training device 5 in the fifth analysis example. FIG. 28 is a diagram showing an example of the relationship between the values of the first principal component and the second principal component and the classification score in the fifth analysis example. FIG. 29 is a diagram showing an example of the classification result of NIRS data in the fifth analysis example.

In the fifth analysis example, as shown in FIG. 29, four NIRS data out of the five NIRS data (actual data) at the time of the simple iterative operation are classified into the data at the time of the simple iterative operation, and one NIRS. The data was categorized as data during non-simple iterative operation. That is, the classification correct answer rate of NIRS data at the time of simple repetitive operation is 80%. Further, of the 60 NIRS data (actual data) during the non-simple repetitive operation, 5 NIRS data are classified into the data during the simple repetitive operation, and 55 NIRS data are classified into the data during the non-simple repetitive operation. It was classified. That is, the classification correct answer rate of NIRS data at the time of non-simple repetitive operation is 92%.

As described above, in the fifth analysis example, it can be understood that selecting the measurement point (that is, the brain function measurement channel) and limiting the time range have a greater influence on the classification accuracy of the NIRS data.

Therefore, when a classification model is generated using the NIRS data at the time of simple iterative operation as a sample and the NIRS data at the time of simple iterative operation and the NIRS data at the time of non-simple iterative operation are classified according to the classification model, the assumption in the first analysis example ( Based on the assumption that the trainer tr is not distinguished, the time range limitation is not considered, and the measurement point limitation is not considered), the following matters can be said.

That is, as shown in the second analysis example, in the classification of NIRS data according to the classification model, the classification result does not differ greatly for each trainer tr, and who the trainer tr is is determined by the classification result. It can be understood that it does not affect.

Further, as shown in the third analysis example, it is understood that in the classification of NIRS data according to the classification model, whether or not the time range in which the NIRS data is acquired is limited does not significantly affect the classification result. can.

Further, as shown in the fourth analysis example, in the classification of NIRS data according to the classification model, it can be understood that whether or not the measurement points from which the NIRS data is acquired is limited has a great influence on the classification result. ..

Further, as shown in the fifth analysis example, in the classification of NIRS data according to the classification model, whether or not the time range is limited together with the measurement point from which the NIRS data is acquired has a greater influence on the classification result. Can be understood.

As described above, the exercise training system 100 was able to confirm by analyzing each NIRS data that the exercise training different from the trainer tr's expectation could affect the brain activation (stimulation).

In addition, although it was exemplified that the feature amount is detected for the NIRS data due to the expected movement (during simple repetitive movement) and the NIRS data due to the unexpected movement (during non-simple repetitive movement) is classified, it is unexpected. The feature amount may be detected for the NIRS data due to the exercise of the above, and the NIRS data due to the expected exercise may be classified.

It is illustrated that five sample data are used when generating a classification model and machine learning, but the number of data is an example and may be another number. For example, the classification results may be accumulated with a large number of trainers tr and a large number of sample data, and the classification model may be machine-learned with a large number of samples.

As described above, the exercise training system 100 can apply the robot control technique such as the exercise training device 5 to the brain rehabilitation to analyze the brain function of the trainer tr during the exercise training. In addition, the cerebral blood flow oxygen concentration can be measured by the cerebral measuring device 45. Further, the brain measuring device 45 can classify, analyze, and evaluate the brain function of the trainee tr by the machine learning technique.

Further, the exercise training system 100 can recognize that the measured cerebral blood flow pattern is different between the simple repetitive motion and the non-simple repetitive motion. Therefore, the exercise training system 100 can give the trainer tr's brain a non-stationary brain stimulus by the non-simple repetitive motion as well as a steady brain stimulus by the simple repetitive motion. The trainee tr can receive the unsteady brain stimulus as an unfamiliar stimulus and recognize the difference in the brain stimulus. Therefore, even if it is difficult to move a part of the trainer tr's body (for example, legs and hands), the slider 20 of the exercise training device 5 is used to forcibly move the trainer tr's body to give brain stimulation. be able to. That is, even if it is difficult to move a part of the trainer tr's body (for example, legs, hands), the trainer tr can generate an illusion of exercise in the brain.

As described above, the exercise training system 100 can be expected to improve the exercise ability of the trainer tr by giving different stimuli to the brain by variously moving any part of the body of the trainer tr.

Although the embodiments have been described above with reference to the drawings, it goes without saying that the present disclosure is not limited to such examples. It is clear that a person skilled in the art can come up with various modifications or modifications within the scope of the claims, which naturally belong to the technical scope of the present disclosure. Understood.

The above embodiment can also be applied to exercise training of a trainer tr in outer space. When staying in outer space, the brain becomes accustomed to the stimulation of the brain in outer space, the sense of balance that the trainer tr should have on the ground is lost, and the trainer tr may be injured. The exercise training system 100 can also be applied to the exercise training of the trainer tr located in outer space, and can be used for the exercise training for returning to the ground.

In the above embodiment, the brain analyzer 40 analyzes and classifies the relationship between the motion of the exercise training device 5 (simple repetitive motion and non-simple repetitive motion) and the measured values of the plurality of NIRS signals acquired during this motion. Although this is illustrated, a value other than the actually measured value (for example, an analysis value) may be used. For example, the brain analyzer 40 acquires the actually measured value of the NIRS signal from the brain measuring device 45 as the measurement result, and generates various analysis values based on the actually measured value of the NIRS signal. The analysis value may include, for example, a histogram showing variation in the measured value of the NIRS signal, and an eigenvalue obtained by frequency analysis of the measured value of the NIRS signal. Eigenvalues may be expressed in oxygenated hemoglobin concentration / frequency. The brain analyzer 40 may analyze or classify the relationship between the motion of the exercise training device 5 (simple repetitive motion and non-simple repetitive motion) and the analysis value at the time of this motion.

In the above embodiment, the brain measuring device 45 measures the oxygenated hemoglobin concentration as an example of the cerebral blood flow information regarding the cerebral blood flow, but the cerebral function information regarding the brain function other than the cerebral blood flow information is provided. It may be measured. The brain measuring device 45 may, for example, measure the brain waves during exercise training of the trainee tr. Further, instead of the brain measuring device 45, a measuring device for measuring the biological signal of the trainee tr for measuring the brain function information may be provided.

For example, the measuring device may be an electrocardiograph or an electromyogram. The electrocardiograph measures the electrocardiogram during exercise and non-exercise of the trainee tr, and obtains electrocardiogram information. The electrocardiograph sends the electrocardiogram information to the brain analyzer 40. The brain analyzer 40 may analyze or classify the relationship between the motion of the exercise training device 5 (simple repetitive motion and non-simple repetitive motion) and the electrocardiogram information during this motion. The electromyogram measures the electromyogram of the trainee tr during exercise and non-exercise, and obtains electromyogram information. The electromyogram sends the electromyogram information to the brain analyzer 40. The brain analyzer 40 may analyze or classify the relationship between the motion of the exercise training device 5 (simple repetitive motion and non-simple repetitive motion) and the electromyographic information during this motion.

In the above embodiment, the acceleration sensor 50 is used to artificially apply vibration, and the motion control device 10 controls the movement of the slider 20 by the vibration torque. The motion control device 10 may control the movement of the slider 20 according to the signal. For example, the motion control device 10 may perform a torque command in synchronization with the heartbeat in conjunction with the electrocardiogram to control the movement of the slider 20.

[Effects of exercise training system, etc.]
As described above, the exercise training system 100 supports and analyzes the exercise for the trainee tr. The exercise training system 100 is a motion control device 10 (an example of a control device) that controls the movement of the slider 20 on which a part of the human body of the trainer tr is placed, and a training period of the trainer tr using the slider 20. It is provided with a brain analyzer 40 (an example of an analyzer) that acquires brain function information regarding the brain function of the trainee 54 and analyzes the brain function information. The motion control device 10 instructs the slider 20 to perform periodic movement during the simple repetitive motion period (an example of the first period) in the training period, and the non-simple repetitive motion period (second) in the training period. (Example of the period) is instructed to perform an aperiodic movement different from the periodic movement. The brain analyzer 40 uses the brain function information as the first brain function information regarding the trainer tr's brain function during the simple repetitive motion period, or the second brain function regarding the trainer tr's brain function during the non-simple repetitive motion period. Classify into information.

Thereby, the exercise training system 100 can classify the brain function information into brain function information in a periodic movement period or brain function information in an aperiodic movement period. In other words, it was possible to change the pattern of cerebral blood flow of the trainee by periodic exercise training and aperiodic exercise training. Therefore, it can be understood that the brain becomes accustomed to the periodic exercise training and the stimulation to the brain is slowed down, but the stimulation to the brain can be achieved by incorporating the aperiodic exercise training. In addition, the brain can be non-invasively stimulated from a part of the human body (feet, hands, etc.) trained for exercise. Therefore, it can stimulate the brain non-invasively and improve safety.

The slider 20 may be repeatedly movable along a predetermined direction. The periodic movement may be a periodic repetitive movement along the predetermined direction. The aperiodic movement may be an aperiodic repetitive movement along the predetermined direction.

As a result, in the exercise training system 100, the movement of the body due to the exercise of the trainer tr becomes a reciprocating motion, so that the range of movement of the body can be reduced and safety can be considered.

In the periodic movement, the movement cycle of the slider 20 and the movement distance in one movement cycle of the slider 20 may be invariant. In the aperiodic movement, at least one of the movement cycle of the slider 20 and the movement distance in one movement cycle of the slider 20 may change.

Thereby, the exercise training system 100 can give the trainee tr a stimulus to the brain due to the constant movement distance in the movement cycle of the slider 20 and one movement cycle of the slider 20. Further, the exercise training system 100 can give the trainee tr a stimulus to the brain due to the movement cycle of the slider 20 and the movement distance in one movement cycle of the slider 20 being indefinite.

The slider 20 includes a slider 20a (an example of the first slider n) of the right foot on which the right foot of the trainee tr (an example of the first foot) is placed and a left foot (an example of the second foot) of the trainer tr. It may include a left foot slider 20b (an example of a second slider) on which it is placed.

As a result, the exercise training system 100 can perform walking training using the first slider and the second slider. In addition, by training the exercise of the part of the human body that supports the weight of the trainee 54, it can be expected that the amount of stimulation of the brain will increase and it will be easy to obtain brain function information.

The motion control device 10 may acquire a first load value applied to the slider 20a of the right foot and a second load value applied to the slider 20b of the left foot. The motion control device 10 applies a driving force to the slider 20a of the right foot based on the first load value and the second load value so that the difference between the first load value and the second load value becomes small. The torque command value to the supplied motor 22a (an example of the first motor) is determined, and the torque command value to the motor 22b (an example of the second motor) to supply the driving force to the slider 20b of the left foot is determined. good.

As a result, the exercise training system 100 can maintain the balance of the slider 20 of the left foot. Therefore, the trainee tr can reduce the possibility of falling due to the imbalance between the left and right even when the exercise ability of either the left or right foot is insufficient and it is difficult to sufficiently move the weight.

The exercise training system 100 may further include an acceleration sensor 50 that detects acceleration. The motion control device 10 may acquire the acceleration detected by the acceleration sensor 50 and determine the torque command value to the motor 22 that supplies the driving force to the slider 20 based on the acceleration.

As a result, when acceleration is applied to the acceleration sensor 50, for example, an external force is applied so that the acceleration sensor 50 vibrates, so that the exercise training system 100 supplies a torque command based on the detection of the acceleration sensor 50 to the motor 22. And the slider 20 can be moved. Therefore, the exercise training system 100 can give a disturbance stimulus to a part of the human body on the slider 20 by using the acceleration sensor 50, and can give a stimulus to the brain through the part of the human body.

The brain analyzer 40 may extract a plurality of feature quantities in the brain function information. The brain analyzer 40 may calculate a score of brain function information based on a plurality of feature quantities. The brain analyzer 40 may classify brain function information based on the score.

Thereby, the exercise training system 100 can extract the feature amount of the brain function information by using, for example, principal component analysis, and classify the brain function information according to the feature amount.

The brain analyzer 40 may machine-learn a classification model for classifying brain function information using a plurality of feature quantities. The brain analyzer 40 may classify the plurality of brain function information according to the machine-learned classification model based on the score.

As a result, the exercise training system 100 can improve the classification accuracy by the classification model by, for example, accumulating the feature amounts of the brain function information of a large number of trainers and causing the classification model to be machine-learned. Therefore, the exercise training system 100 can improve the classification accuracy of the brain function information by following the machine-learned classification model.

Further, the exercise training system 100 may further include a brain measuring device 45 (an example of the measuring device) for measuring brain function information. The brain measuring device 45 is a first position at an arbitrary position on the head of the trainer tr. The light may be transmitted. The brain measuring device 45 may receive a second light corresponding to the first light from an arbitrary position on the head of the trainee tr. The brain measuring device 45 may receive light. The cerebral blood flow information regarding the blood flow in the brain of the trainee tr may be measured based on the amount of light of the second light. The brain measuring device 45 transmits the measured cerebral blood flow information to the brain analyzer 40. It's okay.

Thereby, the exercise training system 100 can acquire the cerebral blood flow information indicated by oxygenated hemoglobin (oxyHb) measured by, for example, near-infrared spectroscopy.

The brain measuring device 45 is included in the simple repetitive motion period, and is the first in the period P1 (an example of the third period) in which the movement cycle of the slider 20 and the movement distance in one movement cycle of the slider 20 are constant. Light may be transmitted and a second light may be received. The brain measuring device 45 may transmit the first light and receive the second light during the period P1 in the non-simple repetitive operation period.

As a result, the brain measuring device 45 operates in synchronization with the exercise training controlled by the motion control device 10, for example, so that the difference between the periodic movement and the aperiodic movement of the slider 20 appears best. Brain function information can be measured during the period P1. The brain analyzer 40 can improve the classification accuracy by acquiring and classifying the brain function information.

The brain measuring device 45 may send the first light to the positions of the primary somatosensory area and the higher motor area on the head of the trainee tr. The brain measuring device 45 may receive a second light corresponding to the first light from the positions of the primary somatosensory area and the higher motor area on the head of the trainee 54.

As a result, for example, the brain measuring device 45 selects a measurement point highly related to exercise and measures the brain function information, so that it can be expected that the brain function information changes in response to the exercise training.

The exercise training system 100 may further include a display device 30 that displays information regarding instructions for movement to the slider 20 by the motion control device 10.

As a result, in the exercise training system 100, the trainee tr can predict the movement of the slider 20, and the safety can be improved. Specifically, the trainee tr can confirm the position command, the torque command value, and the like of the slider 20 that change periodically or aperiodically by checking the information displayed on the display device 30. Further, when the display device 30 displays the balance display function or the like, the trainee tr can perform exercise training as if playing a game, and can enjoy rehabilitation. Further, the information regarding the instruction to move to the slider 20 may be notified to the trainee tr by a presentation device (for example, a speaker or vibration) other than the display device 30.

The present disclosure is useful for exercise training systems, control methods, programs, etc. that can be expected to improve athletic ability by giving different stimuli to the brain by variously moving any part of the trainer's body.

100 Exercise training system 5 Exercise training device 10 Motion control device 11, 11a, 11b Disturbance observer 12, 12a, 12b PID control unit 13 Bilateral control unit 14 Haptic control unit 15, 15a, 15c, 15c, 15d Servo amplifier 20, 20a , 20b, 20c, 20d Slider 22, 22a, 22b, 22c, 22d Motor 24, 24a, 24b Ball screw 25 Motion control generator 26 Guide member 27, 27a, 27b Slider mechanism 28, 28a, 28b Slider mechanism 30 Display device 40 Brain analyzer 45 Brain measuring device 46 Mounting unit 47 Processing unit 50, 50a, 50b Acceleration sensor pr1 Light transmission probe pr2 Light receiving probe

Claims (13)

An exercise training system that supports and analyzes exercise for trainees.
A control device that controls the movement of the slider on which a part of the trainee's human body is placed, and
An analyzer that acquires brain function information regarding the brain function of the trainer during the training period of the trainer using the slider and analyzes the brain function information.
Equipped with
The control device is
The slider is instructed to perform a periodic movement during the first period of the training period, and during the second period of the training period, unlike the periodic movement, the time of the slider. Instructed to perform aperiodic movement in which at least one of position, movement amplitude, and movement cycle is random .
The analyzer is
A plurality of features in the brain function information are extracted, and
Based on the plurality of features, the score of the brain function information is calculated.
Based on the score, the brain function information is used as the first brain function information regarding the trainer's brain function in the first period or the second brain function regarding the trainer's brain function in the second period. Classify into brain function information,
Exercise training system. The slider can be repeatedly moved along a predetermined direction.
The periodic movement is a periodic repetitive movement along the predetermined direction.
The aperiodic movement is an aperiodic repetitive movement along the predetermined direction.
The exercise training system according to claim 1. In the periodic movement, the movement cycle of the slider and the movement distance in one movement cycle of the slider are unchanged.
In the aperiodic movement, at least one of the movement cycle of the slider and the movement distance in one movement cycle of the slider changes.
The exercise training system according to claim 2. The slider is
A first slider on which the trainee's first foot is placed, and
Includes a second slider on which the trainee's second foot rests.
The exercise training system according to claim 2 or 3. The control device is
The first load value applied to the first slider and the second load value applied to the second slider are acquired.
Based on the first load value and the second load value, a driving force is supplied to the first slider so that the difference between the first load value and the second load value becomes small. The torque command value to the first motor is determined, and the torque command value to the second motor that supplies the driving force to the second slider is determined.
The exercise training system according to claim 4. Further equipped with an accelerometer that detects acceleration,
The control device is
The acceleration detected by the acceleration sensor is acquired, and the acceleration is acquired.
Based on the acceleration, the torque command value to the motor that supplies the driving force to the slider is determined.
The exercise training system according to any one of claims 1 to 5. The analyzer is
A classification model for classifying the brain function information is machine-learned using the plurality of features.
Based on the score, the plurality of brain function information is classified according to the machine-learned classification model.
The exercise training system of claim 1 . Further equipped with a measuring device for measuring the brain function information,
The measuring device is
A first light is sent to an arbitrary position on the trainee's head,
The second light corresponding to the first light is received from an arbitrary position on the trainee's head.
Based on the amount of the second light, the cerebral blood flow information regarding the blood flow in the trainer's brain is measured.
The measured cerebral blood flow information is transmitted to the analyzer.
The exercise training system according to any one of claims 1 to 7 . The measuring device is
The first light is transmitted in the third period within the first period included in the first period and in which the movement cycle of the slider and the movement distance in one movement cycle of the slider are constant. , Receiving the second light,
The first light is transmitted in the third period within the second period included in the second period and in which the movement cycle of the slider and the movement distance in one movement cycle of the slider are constant. , Receiving the second light,
The exercise training system according to claim 8 . The measuring device is
The first light is transmitted to the positions of the primary somatosensory cortex and the higher motor cortex on the trainee's head.
The second light corresponding to the first light is received from the positions of the primary somatosensory cortex and the higher motor cortex on the trainee's head.
The exercise training system according to claim 8 or 9 . A display device for displaying information regarding an instruction to move to the slider by the control device is further provided.
The exercise training system according to any one of claims 1 to 10 . It is a control method of an exercise training system that supports and analyzes exercise for trainees.
A step in which the control device of the exercise training system controls the movement of the slider on which a part of the trainer's human body is placed.
A step in which the analyzer of the exercise training system acquires brain function information regarding the brain function of the trainer during the training period of the trainer using the slider, and analyzes the brain function information.
Have,
The step in which the control device controls the movement of the slider is
The slider is instructed to perform a periodic movement during the first period of the training period, and during the second period of the training period, unlike the periodic movement, the time of the slider. Includes steps instructing an aperiodic movement in which at least one of the position, movement amplitude, and movement cycle is random .
The step in which the analyzer analyzes the brain function information is
The step of extracting a plurality of features in the brain function information and
A step of calculating a score of the brain function information based on the plurality of features, and
Based on the score, the brain function information is used as the first brain function information regarding the trainer's brain function in the first period or the second brain function regarding the trainer's brain function in the second period. Brain function information, including steps to classify,
Control method. It is a program for making a computer execute the control method of the exercise training system that supports and analyzes the exercise for the trainee.
The control method is
The step of controlling the movement of the slider on which a part of the trainee's human body is placed, and
A step of acquiring brain function information regarding the brain function of the trainer during the training period of the trainer using the slider and analyzing the brain function information.
Have,
The step of controlling the movement of the slider is
The slider is instructed to perform a periodic movement during the first period of the training period, and during the second period of the training period, unlike the periodic movement, the time of the slider. Includes steps instructing an aperiodic movement in which at least one of the position, movement amplitude, and movement cycle is random .
The step of analyzing the brain function information is
The step of extracting a plurality of features in the brain function information and
A step of calculating a score of the brain function information based on the plurality of features, and
Based on the score, the brain function information is used as the first brain function information regarding the trainer's brain function in the first period or the second brain function regarding the trainer's brain function in the second period. Brain function information, including steps to classify into,
program.

Images