JP5640537B2 - Driver's fatigue estimation device and fatigue estimation method - Google Patents

Description

  The present invention relates to a fatigue estimation technique for estimating a fatigue state of a driver who drives a vehicle.

  Patent Literature 1 describes a steering operation reaction force adjusting device that reduces an operator's fatigue by adjusting an operation reaction force. This device detects the driver's posture (driver state) by calculating the shoulder joint position and elbow joint position of the driver relative to the steering device from the seat slide position, seat back angle, seat cushion angle, and steering operation angle. To do. From the detected posture of the driver, the muscle force required for the driver to hold the current upper limb, that is, the load of each muscle of the current upper limb is obtained. Then, the reaction force of the steering device is adjusted so that the operational feeling (load) felt by the driver is constant.

  The muscular strength (load of each muscle) is detected based on a musculoskeletal model using the dimensions of the human skeletal specimen (such as the angle of the elbow joint and the distance from the shoulder joint to the elbow joint).

JP-A-8-301136

  In general, the musculoskeletal model has a redundant problem that the muscular strength cannot be uniquely determined because the number of muscles that control the movement is large with respect to the degree of freedom of the skeletal model. For this reason, in general, the muscle strength is estimated by optimization under preset boundary conditions, but there is no guarantee that the muscle strength is physiologically appropriate.

  For example, in the above prior art, an operation reaction force at the time of driving operation is input to the musculoskeletal model. However, since the force canceled inside the driver is not reflected in the operation reaction force, the fatigue state cannot be accurately estimated. There is a fear. Specifically, even if the operation reaction force is 1N, the muscle strength difference between the antagonizing muscles is 1N, and each muscle is working, and the antagonizing muscles (for example, triceps and biceps) Since the muscular strength borne by one of the muscles is unknown, the estimation accuracy of the fatigue state is deteriorated.

  The present invention focuses on the above points, and has an object to make it possible to more accurately estimate the driver's fatigue state.

  In order to solve the above-described problem, the present invention provides the entire musculoskeletal portion estimated separately when estimating the fatigue state of the driver's musculoskeletal portion used when operating the operator based on the musculoskeletal model. The accuracy of the estimated fatigue state is improved by utilizing the hardness state.

  According to the present invention, by estimating the hardness state of the target musculoskeletal part, it is possible to judge the validity of the estimated value estimated by the musculoskeletal model, or to appropriately narrow down when estimating the estimated value by the musculoskeletal model. Is possible. As a result, the accuracy of the estimated value is improved, so that the driver's fatigue state can be estimated more accurately.

It is a conceptual diagram of the fatigue estimation apparatus which concerns on embodiment based on this invention. It is a figure which shows the example of a musculoskeletal model. It is explanatory drawing of the driver | operator's fatigue estimation apparatus which concerns on embodiment based on this invention. It is explanatory drawing of the driver | operator's fatigue estimation apparatus which concerns on 1st Embodiment based on this invention. It is explanatory drawing of the driver | operator's fatigue estimation apparatus which concerns on 1st Embodiment based on this invention. 1 is a system diagram of a driver fatigue estimation device according to a first embodiment based on the present invention. FIG. It is a figure explaining disturbance torque A. It is a figure explaining the disturbance torque B. FIG. It is a block diagram of impedance measurement. It is a figure which shows the frequency characteristic of an impedance. It is a conceptual diagram which calculates | requires the model of a musculoskeletal model. It is a figure explaining a muscle model. It is a figure which shows the relationship between length change and force. It is a figure which shows the relationship between contraction speed and force. It is a figure which shows the time-dependent change of the stimulation of a contraction element, and activation. It is a figure which shows the relationship between the force and length of a contraction element. It is a figure which shows the relationship between the force of a parallel elastic element, and length. It is a figure which shows numerical formula. It is a figure which shows the model of the viscoelasticity of an arm. It is an image figure of calculation of viscosity, elasticity, and inertia using a musculoskeletal model. It is a conceptual diagram of a musculoskeletal model. It is a figure which shows the procedure of the determination (selection) of a muscular strength estimated value. It is a figure which shows the relationship of the holding | grip with respect to a steering wheel. It is a figure which shows the relationship between the position of a palm, and impedance. It is a figure explaining the relationship between the position of a palm, elasticity, etc. FIG. It is a figure explaining the process of the fatigue estimation apparatus which concerns on 1st Embodiment based on this invention. It is a figure which shows the relationship between muscular strength and duration. It is a figure explaining the process of fatigue determination. It is a figure explaining the operation | movement etc. which concern on 1st Embodiment based on this invention. It is a figure explaining the measurement of an impedance. It is a figure explaining the process of the fatigue estimation apparatus which concerns on 2nd Embodiment based on this invention. It is a figure which shows the model of an arm. It is a figure explaining the boundary conditions based on impedance measurement. It is a figure which shows the relationship of the main muscle and antagonistic muscle in the estimation result of a musculoskeletal model.

“First Embodiment”
Next, embodiments of the present invention will be described with reference to the drawings.
(Concept of fatigue estimation device)
FIG. 1 is a conceptual diagram of this embodiment.
Here, the operator 100 is for giving a driving instruction to the vehicle when operated by the driver.

The driver's fatigue estimation device of this embodiment includes disturbance torque applying means 101, hardness estimation means 102, operation reaction force acquisition means 103, posture acquisition means 104, estimated value calculation means 105, and fatigue estimation means.
The disturbance torque applying means 101 applies a desired reaction force to the operation element 100. The disturbance torque applying means 101 is not always necessary. When a required disturbance is input to the operation element 100, the disturbance can be used as a reaction force.

  The hardness estimation means 102 estimates the hardness state of the driver's musculoskeletal part used when operating the operator 100. Specifically, the hardness estimation unit 102 of the present embodiment acquires, as the hardness state, the characteristic of the resistance (suppressed) state by the muscles of the entire musculoskeletal part when the reaction force is input to the operation element 100. . The above characteristic is a frequency characteristic in the present embodiment. The hardness estimation unit 102 outputs information on the estimated hardness state to the fatigue estimation unit.

  The hardness estimation means 102 of this embodiment includes disturbance torque detection means 102A and resistance characteristic detection means 102B. The disturbance torque detecting means 102A detects the disturbance torque input to the operation element 100. The resistance characteristic detection unit 102B detects an impedance, which is a resistance characteristic of the musculoskeletal portion, with respect to the disturbance torque detected by the disturbance torque detection unit 102A as the hardness state.

The operation reaction force acquisition unit 103 acquires an operation reaction force received when the driver operates the operation element 100.
The posture acquisition means 104 acquires information on the skeletal posture of the driver's musculoskeletal portion that is mainly used when operating the operator 100. The posture acquisition unit 104 includes, for example, one or more cameras 104A (imaging unit) attached to the vehicle, and an image processing unit 104B that performs image processing on an image captured by the camera 104A. When the operation element 100 is a steering wheel, the musculoskeletal part from the shoulder to the hand is mainly targeted. Then, the camera 104A images the musculoskeletal portion targeted by the driver. The image processing unit 104B performs image processing on an image captured by the camera, thereby obtaining posture information for specifying a posture such as an angle and a position of each joint constituting the target musculoskeletal portion, and a distance between the joints. Calculate. The image processing unit 104B outputs the calculated posture information to the estimated value calculation unit 105.

  The estimated value calculation means 105 uses the posture information calculated by the posture acquisition means 104 and the operation reaction force acquired by the operation reaction force acquisition means 103 as input values, and the muscle strength of the target musculoskeletal part from the driving equation based on the musculoskeletal model. Find an estimate for the value.

  As shown in FIG. 2, the musculoskeletal model includes a skeletal model in which a plurality of skeletons composed of rigid bodies are connected by a link structure, and a muscle / tendon / ligament system wire model (simply a muscle model) stretched on the skeleton of the skeleton model. Called). The muscle model is a model of a muscle as an actuator that exerts contractile force. And the estimated value regarding the muscular strength value of each muscle can be calculated | required by calculating the driving | operation equation expressing the said musculoskeletal model. The estimated value related to the muscle strength value may be a value related to the muscle strength value, muscle viscosity, elasticity, or inertia. However, since there is redundancy, it is necessary to set some evaluation function or constraint condition as a boundary condition.

  A database for musculoskeletal models is also provided. The database stores, for example, data used in equations of motion such as the characteristics of each muscle of the target musculoskeletal portion (standard maximum muscle strength value, etc.) and the muscle load ratio according to the posture.

  The estimated value calculation means 105 specifies the joint position of the skeletal model of the target musculoskeletal part based on the posture information calculated by the posture acquisition means 104, for example, and the joint torque of each joint part with respect to the operation reaction force In addition, from the above posture information, estimate the range of muscle strength that can be generated from the length of each muscle, the muscle strength ratio that is the burden ratio of each muscle, and calculate the inverse dynamics based on the equation of motion. Calculate the estimated value. At this time, the calculation has a certain degree of redundancy and indefinite elements. In the present embodiment, a plurality of reasonable estimated values are calculated without performing optimization.

  The fatigue estimation unit 106 estimates the driver's fatigue state based on the hardness state estimated by the hardness estimation unit 102 and the estimated value obtained by the estimated value calculation unit 105. The fatigue estimation unit 106 according to the present embodiment includes, for a plurality of estimated values estimated from the estimated value calculating unit 105, characteristics of the plurality of estimated values and a hardness state characteristic of the hardness estimating unit 102 (this embodiment). In the embodiment, an estimated value of the characteristic most similar to the hardness state characteristic is selected. And the fatigue estimation means 106 of this embodiment estimates a driver | operator's fatigue state from the selected estimated value.

(Specific configuration)
Next, the present embodiment will be described by exemplifying a specific configuration.

  In the following embodiments, the steering wheel 7 will be described as an example of the operation element 100. The target operating element 100 is not limited to the steering wheel 7 but may be a brake pedal, a shift lever, or the like. In short, it is applicable to any operator 100 that is operated when driving and that generates an operation reaction force on the driver.

In the present embodiment, the body part from the shoulder part to the hand is the target musculoskeletal part.
The index of fatigue may be muscle strength itself, or muscle viscosity, elasticity, or inertia. In addition, muscle viscosity, elasticity, and inertia can be used as indicators of fatigue.

FIG. 3 is an overall view for explaining the fatigue estimation device of the present embodiment mounted on a vehicle.
Reference numeral 1 denotes a steering gear mechanism. The steering gear mechanism 1 is a device that converts the rotation of the steering into movement in the vehicle width direction (lateral direction) by a rack / pinion mechanism or the like.

Reference numerals 2 and 3 are steered wheels. The tires of the steered wheels 2 and 3 may be of any type (flatness, tire diameter, radial / studless, etc.) as long as they are attached to a general sedan type.
Reference numeral 4 denotes a reaction force device motor angle sensor. The reaction device motor angle sensor 4 is a motor angle sensor installed between the steering column shaft and the steering reaction force actuator 5, and detects and outputs the current angle of the rotating steering wheel.

  Reference numeral 5 denotes a steering reaction force actuator that constitutes a disturbance applying means. The steering reaction force actuator 5 is a motor installed on the steering column. The steering reaction force actuator 5 is a device that can be used to amplify and assist the driver's force required to rotate the steering, or to reduce unnecessary disturbance that enters from the tire side. The steering reaction force actuator 5 can generate a desired reaction force by inputting a control current according to a command from the controller 6.

Reference numeral 6 denotes a steering reaction force controller (steering reaction force control means). The steering reaction force device controller 6 outputs a control current corresponding to a target steering reaction force and steering angle. The controller for the steering reaction force device is a control device for driving the actuator 5.
Reference numeral 7 denotes a steering wheel (steering means) as the operator 100 operated by the driver. The steering wheel 7 is connected to the steering gear mechanism 1 via the steering.

Reference numeral 8 denotes a torque sensor with a built-in steering link that constitutes the operation reaction force acquisition means 103. The torque sensor 8 is a sensor for measuring the force generated when the driver operates the steering wheel 7.
Reference numeral 9 denotes a steering information transmission harness. The steering information transmission harness 9 includes a cable for transmitting a steering force, a steering angle, and a disturbance torque signal from the steering reaction force actuator 5 to the muscle force calculation / instruction device use determination device 10.

Reference numeral 10 denotes a muscular strength calculation / instruction device use determination device. The muscular strength calculation / instruction device use determination device 10 is an arithmetic device for estimating the muscular strength of the upper limb body (target musculoskeletal portion) of the driver.
Reference numeral 11 denotes a driver camera. The driver camera 11 is an imaging device for detecting the posture of the driver. Examples of the camera include a CCD camera and a CMOS camera.

Reference numeral 12 denotes a tire lateral force detection sensor. The tire lateral force detection sensor 12 is a sensor that measures the lateral force acting on the tire.
Reference numeral 13 denotes a pointing device. When the instruction device 13 detects that the driver is in a fatigued state or a state that is likely to lead to fatigue in the future, for example, the instruction device 13 prompts the driver to take a break or use a lane keeping assist device. It is a pointing device.

FIG. 4 is a diagram illustrating the configuration of the muscle strength calculation / instruction device usage determination device. FIG. 5 is a conceptual diagram thereof.
Reference numeral 14 denotes a disturbance torque signal detector that constitutes the disturbance torque detector 102A. The disturbance torque signal detection unit 14 is based on an actually measured torque signal for a certain past time, and a self-aligning torque, a road surface disturbance, and a force signal acting on the tire due to a crosswind disturbance (tire generation torque signal), that is, a tire generation torque signal A predetermined disturbance torque that can be used to measure the driver's impedance is detected from the history of the measured torque. The vehicle signal of the tire generation torque signal is acquired from the tire generation torque signal detection unit 15.

The tire generation torque signal used for measurement and the detection conditions of the measured torque are as follows.
That is, the detection condition is that the coherence with the actually measured torque is 0.9 or more and the phase of the self-aligning torque is earlier at a frequency of 1 Hz to 100 Hz.

  When this detection condition is not satisfied, the disturbance torque applying unit 101 uses a predetermined disturbance torque signal prepared in advance. That is, the disturbance torque applying means 101 transmits a predetermined disturbance torque to the steering wheel 7 using the steering reaction force actuator 5.

Reference numeral 15 denotes a tire generated torque signal detector. The tire generation torque signal detection unit 15 estimates the self-aligning torque. The estimation method estimates the self-aligning torque generated in the front wheels that are steered wheels from the product of the tire lateral force and the caster trail.
Reference numeral 16 denotes an attitude detection unit constituting the attitude acquisition unit 104. The posture detection unit 16 detects the joint position and joint angle of the body (mainly the arm) by performing image processing on the image captured by the driver camera 11.

  Reference numeral 17 denotes a muscle-specific viscoelasticity estimation unit that constitutes the estimated value calculation means 105. The muscle-specific viscoelasticity estimation unit 17 refers to a musculoskeletal model registered in a computer in the vehicle or a musculoskeletal model registered in a server installed outside the vehicle via a network, and determines a predetermined arm. Estimate the viscosity, elasticity, and inertia of a single muscle that occurs in a region. The viscoelasticity estimation unit 105 receives the driver's joint position, joint angle, steering angle, and measured torque as input signals, and outputs viscosity, elasticity, and inertia as calculation results. Here, there are a plurality of solutions for individual muscle viscosity, elasticity, and inertia. For example, when there is a joint composed of two muscles, when the rotational moment applied to the joint is distributed to the two muscles by inverse mechanical calculation, there are multiple combinations of allocation.

Reference numeral 18 denotes a hardness estimation unit 102, which is an impedance measurement unit for the entire muscle in the target musculoskeletal unit. The impedance measuring unit 18 measures the impedance of the entire muscle. Here, there are several methods for impedance measurement. The impedance measuring unit 18 of the present embodiment uses a technique for obtaining the arm frequency characteristics using disturbance torque. That is, the impedance measuring unit 18 measures the measured torque and the steering angle while applying the disturbance torque to the steering wheel using the steering angle, the measured torque, and the disturbance torque as input signals. Alternatively, the measured torque and the steering angle are measured while executing the processing of the tire generation torque signal detection unit 15. Based on the disturbance torque, measured torque, and steering angle signals obtained at the same time, that is, in a corresponding relationship, the impedance is calculated. Here, a value obtained by dividing the cross spectrum of the disturbance torque Ta and the actually measured torque Tb by the cross spectrum of the disturbance torque and the steering angle is defined as impedance. That is, the impedance is calculated by the following formula.

Impedance calculation formula:
Hnms = −STaTb / STAx [Nm * Nm / Nm * deg]

  Reference numeral 19 denotes a transfer function calculation unit constituting the estimated value calculation means 105. The transfer function calculation unit 19 is a transfer function in which transfer functions of a plurality of muscles constituting the arm are integrated into one, and the transfer function and solutions of viscosity, elasticity, and inertia obtained by the individual viscoelasticity estimation unit are obtained. Based on this, the arm transfer function is calculated. Here, there are as many transfer functions as calculated above as the number of solutions obtained by the individual viscoelasticity estimation unit. This transfer function constitutes the estimated value.

  Reference numeral 20 denotes a frequency characteristic comparison unit constituting the fatigue estimation means 106. The frequency characteristic comparison unit 20 compares the gain characteristic and phase characteristic of the impedance measured by the impedance measurement unit with respect to the frequency characteristic (gain characteristic and phase characteristic) of each transfer function obtained by the transfer function calculation unit 19. Although there are as many comparison results as the number of solutions (estimated values) obtained by the viscoelasticity estimation unit 17, the solution closest to the impedance gain characteristic and phase characteristic obtained by the impedance measurement unit 18 is determined as the optimum solution. The comparison may be performed using at least the gain characteristic.

  Reference numeral 21 denotes a muscle strength calculation unit for each muscle constituting the fatigue estimation means 106. The muscular strength calculator 21 calculates the muscular strength from the viscosity, elasticity, inertia, and individual muscle states (muscle length, muscle contraction speed, etc.) determined by the frequency characteristic comparator. Predetermined values are used for the muscle length and the contraction speed.

  Reference numeral 22 denotes a palm state correction unit 22. The palm state correction unit 22 corrects the muscle strength estimation value of the muscle strength calculation unit, the impedance of the impedance measurement unit, or the frequency characteristic of the frequency characteristic comparison unit 20 according to the position and orientation of the arm with respect to the steering wheel 7. When the arm position is on the lower side of the steering wheel, the palm state correcting unit 22 has an effect that the impedance of the impedance measuring unit becomes large even if no force is applied to the arm because the arm is supported by the trunk. Correct the direction to decrease.

  Reference numeral 23 denotes a determination unit using the pointing device that constitutes the fatigue estimation means 106. The determination unit 23 using the pointing device detects a fatigue state in a single muscle or a plurality of muscles or a state likely to lead to fatigue in the future.

FIG. 6 is a system configuration diagram for explaining the fatigue estimation device.
The fatigue estimation device functionally includes an impedance acquisition process A, a muscle force estimated value acquisition process B using a musculoskeletal model, and a fatigue index value determination process C.

  The impedance acquisition process A corresponds to the hardness estimation unit 102 and corresponds to the process in the disturbance torque signal detection unit 14, the tire generation torque signal detection unit 15, and the impedance measurement unit 18.

The impedance acquisition process A is performed as follows.
First, frequency characteristics (gain characteristics, phase characteristics) indicating the characteristics of the hardness state (impedance) of the body at the musculoskeletal position targeted by the driver are obtained.

  Specifically, the measured torque and the steering angle are measured while applying disturbance torque to the steering wheel 7. Alternatively, the measured torque and the steering angle are measured while executing the processing of the tire generation torque signal detection unit 15. Next, the impedance is calculated based on the disturbance torque, measured torque, and steering angle signals obtained at the same time. In the calculation, the impedance is a value obtained by dividing the cross spectrum of the disturbance torque and the actually measured torque by the cross spectrum of the disturbance torque and the steering angle.

Here, the impedance calculation methods are classified into the following 1) to 3).
In the following 1) and 2), after obtaining the viscosity, elasticity, and inertia, the impedance is determined by substituting it into the second-order transfer function and obtaining its frequency characteristics, so only the method for obtaining the viscosity, elasticity, and inertia is described. . Further, 3) is a method for directly obtaining frequency characteristics without obtaining viscosity, elasticity, and inertia.

(Impedance calculation method)
1) Calculation method using simultaneous equations Impedances are calculated by solving simultaneous equations having viscosity, elasticity, and inertia as variables from measured torque measured at predetermined time intervals set in advance.
2) Method of calculating impedance by individually measuring viscosity, elasticity, and inertia Here, elasticity is measured by applying stepwise disturbance torque to the steering wheel 7. The viscosity is measured by rotating the steering wheel 7 at a constant speed. Inertia is determined using the weight and volume of the arm as reference values.

3) Method of calculating impedance by measuring measured torque and steering angle while applying disturbance torque to steering wheel 7. 3-1) Applying disturbance torque to operator operated by driver.
3-2) Measure the actual torque generated in the steering wheel 7 and the shift amount of the steering angle.
From the comparison of the frequency characteristics of 3-1) and 3-2), the impedance information of the driver's body is obtained.

  Of the above impedance calculation methods, 3) can measure impedance with the highest accuracy. Based on this, in this embodiment, the impedance is measured (calculated) by the method of 3). This impedance calculation method uses disturbance torque transmitted to the steering wheel 7.

Here, the following may be adopted as the target disturbance torque.
A) Predetermined predetermined signal: pseudo M series, swept waveform, white noise, etc. convolved with a sine wave in a frequency band of 0 to 100 Hz shifted in phase (see FIG. 7)
B) Disturbance torque obtained from travel data of the vehicle As the disturbance torque B, for example, time series data of the force transmitted from the tire to the steering or the force transmitted from the steering to the tire is recorded in the memory. When the waveform of the time series data has a predetermined shape and a predetermined spectrum set in advance, it is used as a disturbance torque (see FIG. 8).

FIG. 9 is a schematic diagram when disturbance torque is applied by the actuator 5 connected to the steering wheel 7 in a state where the driver is holding the steering wheel 7 as the operation unit.
In this schematic diagram
Ta: Disturbance torque
Tb: Actual torque
Tc: Difference between disturbance torque and measured torque (= Ta-Tb)
x: Steering angle.

  At this time, the driver's impedance can be obtained from “−STaTb / STAx (a value obtained by dividing the cross spectrum of the disturbance torque and the actually measured torque by the cross spectrum of the disturbance torque and the steering angle)”. FIG. 10 shows an example of the impedance measurement result. From this lower left figure, it is possible to estimate the state of force entering the arm, that is, the state of hardness of the musculoskeletal part. For example, it can be confirmed that the impedance increases in a frequency band of 1 Hz or less when the force for grasping the steering wheel increases.

Here, it supplements about how to give the white noise as the above-mentioned disturbance torque A. FIG.
The influence of viscosity and elasticity on impedance is limited to the low frequency side, but the influence of inertia affects relatively high frequencies.
For this reason, when measuring the impedance of an occupant with high accuracy, a disturbance torque as a white nozzle is generated with the following specifications.

・ Frequency band: From low frequency of 0 Hz to high frequency of 100 Hz ・ Types of waves: sine wave, triangular wave, square wave ・ Torque: The level at which the steering wheel 7 starts moving is the lower limit value, and the limit that can be overridden by the passenger is the upper limit value. Within this range, it is estimated that the occupant impedance can be measured.
-About how to rotate the steering wheel 7, the method of rotating differs depending on whether the phase of different wave signals is shifted or not during the generation of disturbance torque.

  When shifting the phase, the phase is randomly set between 0 and 2 pi. When the phase is shifted, the measurement tends to be performed for a long time. Therefore, it is preferable to perform the measurement only when traveling straight and at a constant steering.

On the other hand, when the phase is not shifted, the signals are superimposed with the same phase. When the phase is not shifted, the measurement can be performed for a short time, and the traveling scene at the time of measurement becomes gentler than when the phase is shifted.
Here, when the impedance of the occupant is simply measured, it may be carried out as follows.

  That is, the impedance of the entire frequency band of the occupant is measured using only a value equal to or higher than a predetermined frequency. Since the position of the arm changes only continuously, it is possible to measure the impedance in the entire frequency band using only signals having a predetermined frequency or higher. For example, an occupant mostly performs a steering operation at a resonance frequency or lower. The resonance frequency often appears in the vicinity of 1 Hz. Therefore, the impedance of 0 to 100 Hz can be measured using only a signal of 1 Hz or more. The change in reaction force of 1 Hz or more is less noticeable by the measurement because it is difficult for the passenger to notice. Also, if only high frequency components are used, the measurement time can be shortened.

The disturbance torque may be applied by a step response or a sweep waveform.
Further, the estimated muscle force value acquisition process B using the musculoskeletal model is performed by the posture detection unit 16 constituting the posture acquisition unit 104, the viscoelasticity estimation unit 17 and the transfer function calculation unit 19 constituting the estimated value calculation unit 105. It corresponds to processing.

The muscular strength estimated value acquisition process B is performed as follows.
First, an operation reaction force, a driver's joint angle, and a joint position at the time of driving operation are acquired and input to a musculoskeletal model to obtain a driver's individual viscosity, elasticity, and inertia. That is, as shown in FIG. 11A, the driving scenery is photographed by a camera installed in the vehicle interior. Next, as shown in FIG. 11B, the joint position and joint angle of the driver are detected from the driving scenery. Then, as shown in FIG. 11C, the viscosity, elasticity, and inertia of each muscle are estimated from the joint position and joint angle of the driver and the operation amount of the vehicle using the musculoskeletal model.

  Here, calculation of muscle strength of each muscle using a musculoskeletal model and calculation examples of viscosity, elasticity, and inertia of muscle strength of each muscle will be described. The estimation of the viscosity, elasticity, and inertia of each muscle is not limited to this, and a known method may be adopted.

  As a muscle model, a model shown in FIG. 12 is assumed. This muscle model is for calculating the muscle strength of a muscle alone. The contraction element CC corresponds to viscosity, and the total elastic element corresponds to elasticity. The inertia is determined by which muscle is modeled.

Then, for each line, the following calculation is performed for each time step Δt.
First, the muscle parameters of the contraction element and the serial element are obtained for the target muscle. Also get muscle length and stimulus input. The activation level and length of the contraction element is then obtained from the previous time step. Also obtained from the time step before the length and tension of the series elastic element.

  Next, the muscle model subroutine is called and the following processes 1 to 4 are performed to obtain the individual muscle viscosity, elasticity, and inertia.

1. Inline elastic element tension = f (inline elastic element length)
This has a relationship as shown in FIG.
Contraction element force = inline elastic element tension Activation level of contraction element = f (stimulation)
3. Contraction element speed = f (force of contraction element, activation level, length)
4). Contraction element length = ∫ contraction element speed dt

The above 2 to 4 are in the relationship shown in FIGS.
FIG. 18 shows a basic relational expression.
Next, in the transfer function that integrates the transfer functions of multiple muscles that make up the arm, the transfer function of the arm is calculated based on the transfer function and the solution of viscosity, elasticity, and inertia obtained for each muscle. To do. Subsequently, frequency characteristics (gain characteristics and phase characteristics) of the calculated arm transfer function are obtained.

Here, the viscoelasticity equation and transfer function of the arm will be described.
The arm joint is a secondary system and can be represented by a mass spring model as shown in FIG.
That is, this mass spring model has viscosity and elasticity as shown in FIG. The mass of the arm is inertia.

  When the occupant moves his / her arm at the handle angle θ with respect to the target angle α, it can be expressed by the following (Expression 1). It should be noted that it is not necessary to strictly follow (Equation 1) as long as it is possible to approximately obtain parameters corresponding to viscosity, inertia, and elasticity even in systems other than secondary systems.

The transfer function corresponding to (Equation 1) can be expressed by the following (Equation 2). The gain G ₀ , the damping ratio ζ, and the resonance frequency ω _n ² include occupant viscosity, elasticity, and inertia parameters.

  Here, the musculoskeletal model is a mathematical model of the musculoskeletal mechanical properties (arm weight, muscle strength, speed). In computer analysis, it is used to estimate physical quantities that are difficult to measure from measurable information. FIG. 20 shows an image diagram for calculating viscosity, elasticity, and inertia using a musculoskeletal model. In order to obtain the muscular strength, the force that acts on the musculoskeletal model (external force acting on the hand), the musculoskeletal geometric shape (length, angle), and the musculoskeletal motion information (position, velocity) are required as input information. . Further, it is necessary to define in the computer at least a part of the body part desired to know in advance as shown in FIG. 21 as a musculoskeletal model.

Further, the fatigue index value determination process C corresponds to the fatigue estimation means 106 and corresponds to the processes in the frequency characteristic comparison unit 20 and the muscle strength calculation unit 21.
The fatigue index value determination process C is performed as follows.

  By comparing the impedance obtained in the impedance acquisition process A with the frequency characteristics (multiple solutions) of the arm transfer function corresponding to each estimated value obtained in the muscle force estimated value acquisition process B, the viscosity and elasticity of each muscle are compared. The inertia is determined, and the fatigue state is estimated based on the result. The comparison is performed by the number of solutions obtained by the muscle strength estimation value acquisition process B. And the solution calculated | required by the muscular strength estimated value acquisition process B estimated that the characteristic (at least a gain characteristic is compared) is the most approximate is selected as an optimal solution. The optimal solution is, for example, a value that minimizes the difference between the two frequency singularities or maximizes the ratio (result of muscle strength estimation value acquisition process B / result of impedance acquisition process A).

Once the optimal solution is found, muscle strength is estimated based on the viscosity, elasticity, and inertia values. And a driver | operator's fatigue state is estimated from this estimated muscular strength.
A method for comparing the frequency characteristics is shown in FIG. That is, first, in step 1, the muscular strength estimated value acquisition process B obtains a plurality of solutions based on the musculoskeletal model. In step 2, the muscle strength estimation value acquisition process B obtains a secondary transfer function of the entire upper arm for each of the plurality of obtained solutions. In step 3, impedance acquisition processing A calculates a frequency characteristic indicating impedance. In step S4, the fatigue index value determination process C compares the frequency characteristics and selects an optimum solution.

Next, the process of the palm state correction unit 22 will be described.
When the position of the arm (mainly the position of the palm) is below the center of the steering wheel 7 (see FIG. 23), the degree to which the arm is supported by the trunk increases. For this reason, the impedance measured by the impedance acquisition process A tends to increase even if no force is applied to the arm.

In order to reduce the influence of the palm position on the impedance, as shown in FIG. 24, the palm state correction unit 22 obtains the impedance in such a direction that the gain characteristic becomes smaller as the palm position is lower. The impedance obtained in process A is corrected.
It is also possible to make a determination using inertia instead of the palm position. That is, as shown in FIG. 25, when inertia increases and elasticity changes, it can be considered that the position of the palm has changed, and the palm state correction unit 22 similarly reduces the impedance. Apply correction to.

  Here, the secondary transfer function of the upper arm can be expressed, for example, by the following equation.

  Depending on the situation (such as when the palm is below the steering wheel, the side is open and the arm is not supported by the trunk), a condition that the impedance is not corrected may be added.

Next, the processing of the fatigue estimation device will be described with reference to the flowchart of FIG.
First, in order to generate a disturbance torque necessary for impedance measurement, in step S1, an operation amount of the driver is measured. Then, the process proceeds to step S2.
In step S2, the operation amount is recorded in the memory and held for a predetermined time. Subsequently, in step S3, frequency analysis is performed on the operation amount for a predetermined time set in advance.

  Next, in step S4, it is determined whether or not it can be determined that the power and phase characteristics are sufficient for impedance measurement in a predetermined frequency band set in advance. If it is determined that determination is possible, the process proceeds to step S6. On the other hand, when it determines with it being inadequate, it returns to step S1 which is measurement of the operation amount.

In step S6, the driver's impedance (hardness state) is measured by the above-described method. Then, the process proceeds to step S14.
In parallel with the above processing, steps S7 to S13 are performed.
In step S7, the driving scenery of the driver is photographed using an imaging means such as a camera. Then, based on the captured image, the joint position (or joint length) and angle of the driver are calculated.

Next, in step S8, the posture and dimensions are determined in order to obtain muscle strength with the musculoskeletal model.
Next, in step S9, an external force value is measured in order to obtain muscle strength using a musculoskeletal model.
Next, in step S10, the viscoelasticity of each muscle is calculated using the external force value, posture, and dimensions as inputs and using the musculoskeletal model.
Here, the solution of viscoelasticity of one muscle is determined depending on the number of muscles and tendons defined in the musculoskeletal model, and is basically a plurality of solutions.

Next, in step S11, the viscoelasticity of the entire arm is calculated using a predetermined combination among a plurality of solutions obtained in step S10.
Next, in step S12, assuming that the driver's arm system is the same secondary system as the steering, a secondary transfer function is calculated from the calculated viscoelasticity of the entire arm.
Next, in step S13, a Bode diagram of the transfer function calculated in step S12 is obtained. Then, the process proceeds to step S14.

  In step S14, the impedance measurement result is compared with the Bode diagram. Subsequently, in step S15, it is determined whether the result of the predetermined combination is closest to the impedance measurement result. If the determination result is the closest, the process proceeds to step S17. On the other hand, if the determination result is not close, the process proceeds to step S11, and the calculations of steps S12 to S16 are performed again with another combination.

In step S17, the muscular strength value of one or more muscles is calculated from a combination of viscoelasticity at close values (optimum values).
Next, in step S18, the muscular strength value is decreased as the palm is positioned below the steering wheel 7. As described above, the correction process in step S18 may be performed after step S6 or before step S17.

Next, in step S19, if it is determined that there is a sign of fatigue based on the history of muscle strength estimation values of a single muscle or a plurality of muscle groups, the driver is notified. The notification is performed, for example, by voice or vibration applied to the seat, thereby providing driving assistance to the driver.
Here, the processing method of the driver's fatigue estimation in the determination unit 23 constituting the fatigue estimation means 106 based on the muscle strength estimation value that is an index of the fatigue level will be described.

In general, the duration of muscle strength (instantaneous value) has a relationship as shown in FIG. Therefore, when the muscular strength is large, the duration in which the muscle continues to exert force is short. Conversely, if the muscle strength is small, the duration that the muscle can continue to exert force is long. As shown in FIG. 27, the duration decreases exponentially as the muscular strength increases.
Based on this, the following methods 1 and 2 can be exemplified as fatigue estimation methods.

(Measure 1 for fatigue estimation)
Fatigue is determined from the frequency of instantaneous muscle strength. That is, based on the duration of the instantaneous value from the relationship of FIG. 27 (fatigue curve), the degree of fatigue state and the fatigue state to be notified to the driver are estimated.

  For the above estimation, recognition processing based on statistical knowledge may be performed. Examples of statistical processing include Bayesian estimation, fuzzy estimation, support vector machine, genetic algorithm, neural network, and the like.

(Measure 2 for fatigue estimation)
Depending on whether or not the instantaneous value of the muscular strength exceeds a preset threshold value, it is estimated how much the fatigue state is and whether the fatigue state should be notified to the driver. There may be one threshold value or a plurality of threshold values.

Here, when a muscle exerts a force of 100% or more, a phenomenon called a crane occurs. Therefore, the threshold is set based on the extreme state.
Next, the fatigue estimation process employing the above-described measure 1 will be described with reference to FIG.
First, as preprocessing, muscle strength calculation and a histogram of muscle strength and time are calculated.

Next, in the feature extraction, the area or size of the histogram is used as a feature amount.
Next, in post-processing, an average, median, and mode value of feature quantities are calculated, or a feature space is created based on a plurality of different feature quantities.
In the next determination, it is determined whether or not a preset numerical value or a value within a predetermined range of the feature space (a value exceeding a threshold value). If this determination is not satisfied, the process returns to the preprocess and the process is repeated. On the other hand, when this determination is satisfied, it is determined that the driver is required to be notified of fatigue, and the process ends.

(Operation other)
When estimating based on the musculoskeletal model from the reaction force (operation reaction force) acting on the steering wheel 7 that is the operation element 100 and the joint position that is posture information, there is a possibility that the muscle force cannot be estimated accurately.

  For example, the same operation reaction force “1 (N)” is added to the steering wheel 7 even if the muscular strength of each muscle is different between (a) and (b) of FIG. At this time, as shown in FIG. 29, the muscular strength generated in the muscle B should be different between the case where the muscular strength of the muscle A is 2N and 10N. However, such a force cancelled inside the arm cannot be obtained from the musculoskeletal model.

Normally, the muscle estimation value is calculated by applying a preset boundary state. However, as described above, the accuracy of the muscle estimation value may be poor.
On the other hand, in the present embodiment, in order to solve this problem, the force canceled inside the arm is estimated from the impedance measurement to obtain a more appropriate muscle estimation value.

  In the impedance measurement, as shown in FIG. 30, the impedance (hardness state) of the arm that antagonizes the movement of the steering wheel 7 to be rotated by the input disturbance torque is measured. Here, when the target musculoskeletal part (body) is in a hard state, even if an external force is input to the operation element 100, the operation element 100 does not move according to the input external force. Based on this, it is possible to measure the hardness state of the driver's body from the force input to the operating element 100 and the movement (position, actual generated force, etc.) of the operating element 100.

  Since the measured impedance is the state of the overall muscular strength of the body part (the musculoskeletal part to be dealt with) that resists disturbance torque, it reflects information on the internal force of the arm in the musculoskeletal part to be dealt with. It also shows the actual characteristics of actual muscle strength.

  Based on this, by selecting a muscle estimation value close to the measured impedance characteristic, that is, by combining with the musculoskeletal model, a more accurate muscle estimation value can be estimated in consideration of the internal force. As a result, the muscle strength can be accurately estimated including the internal force information, so that the fatigue estimation can be performed more accurately.

(Effect of this embodiment)
(1) The hardness estimation means 102 estimates the hardness state of the driver's musculoskeletal part used when operating the operator 100 for giving a driving instruction to the vehicle. The operation reaction force acquisition unit 103 acquires an operation reaction force when the driver operates the operation element 100. The posture acquisition means 104 acquires information on the skeletal posture of the musculoskeletal portion of the driver. The estimated value calculating means 105 obtains an estimated value related to the muscle force of the musculoskeletal part from the musculoskeletal model based on the operation reaction force and the skeletal posture information. The fatigue estimation unit 106 estimates the driver's fatigue state based on the hardness state estimated by the hardness estimation unit 102 and the estimated value obtained by the estimated value calculation unit 105.
The fatigue state can be estimated with higher accuracy by estimating the fatigue state in consideration of the force offset within the driver.

(2) The estimated value calculation means 105 obtains a plurality of estimated values based on the musculoskeletal model. The fatigue estimation unit 106 selects an estimated value closest to the characteristic of the hardness state estimated by the hardness estimation unit 102 from the plurality of estimated values obtained by the estimated value calculation unit 105, and the selected estimation Estimate the fatigue state based on the value.
The estimated value obtained from the musculoskeletal model inputs the operation reaction force at the time of the driving operation, but the force canceled inside the driver is not reflected on the operation reaction force. On the other hand, since the estimated value with higher accuracy can be selected by selecting the estimated value closest to the characteristics of the actual hardness state, the fatigue state can be estimated with higher accuracy.

(3) The disturbance torque detecting means 102A detects the disturbance torque input to the operation element 100. The resistance characteristic detection unit 102B detects the resistance characteristic of the musculoskeletal portion with respect to the disturbance torque detected by the disturbance torque detection unit 102A as the hardness state.
Thereby, the hardness state can be detected.

(4) The hardness estimation unit 102 includes disturbance torque applying unit 101 that applies disturbance torque to the operation element 100.
This makes it possible to stably estimate the hardness state even in a traveling state in which no disturbance is input to the operating element 100 such as traveling on a flat road surface.

(5) The disturbance torque is torque transmitted from the road surface to the steering wheel 7.
As a result, the hardness state can be estimated without the disturbance torque applying means 101. For example, it can be estimated while traveling on a rough road.
(6) The hardness state estimated by the hardness estimation unit 102 is corrected by the gripping state of the driver's steering hole.
By reflecting the relationship between the relative position and orientation of the operator 100 and the body, it is possible to estimate fatigue more accurately.

(Modification)
(1) In the above embodiment, the case where the steering wheel 7 is used as the operation element 100 has been described. The operation element 100 to be applied may be, for example, a shift lever, an accelerator pedal, or a brake pedal. However, it is preferable to select the operation element 100 with a large operation amount and operation time by the driver.

“Second Embodiment”
Next, a second embodiment will be described with reference to the drawings. In addition, about the apparatus similar to the said embodiment, the same code | symbol is attached | subjected and demonstrated.

The basic configuration of this embodiment is the same as that of the first embodiment.
However, in the first embodiment, among the plurality of estimated values (solutions) obtained using the musculoskeletal model, an estimated value having the characteristic closest to the impedance is set as the optimum value, and the optimum value is used as an index of fatigue level. The fatigue state was estimated as follows.

  On the other hand, in the second embodiment, the boundary condition of the musculoskeletal model is determined based on the impedance, an estimated value is obtained, and the optimum estimated value as an index of the degree of fatigue is estimated from the estimated value.

  That is, in the first embodiment, a combination (solution) of a plurality of estimated muscle strength values is obtained from the musculoskeletal model by setting no boundary condition or setting the boundary condition loosely. Then, impedance measurement results and characteristics are compared for all combinations of muscle strength estimation values obtained from the musculoskeletal model. Therefore, the number of calculations for this comparison tends to increase.

  In contrast, in the second embodiment, boundary conditions based on the impedance measurement result are set in advance in the process of calculating the muscular strength using the musculoskeletal model. This makes it possible to reduce the number of calculations compared to the first embodiment.

Here, in order to simplify the description of the present embodiment, the operation will be described with the elbow joint in the extended state, and the muscles will focus on the biceps and triceps muscles attached to the elbow joint. However, in the following description, the word biceps or triceps is not used, the biceps muscle is mainly referred to as the leading muscle, and the triceps is the leading muscle. An antagonistic muscle that works by antagonizing muscles.
The process of the fatigue estimation apparatus of this embodiment will be described with reference to FIG.

First, in step S100, the impedance (hardness state) of the entire target musculoskeletal portion (the entire arm) is measured.
Next, in step S110, based on the measured impedance measurement result of the entire arm, the level of co-contraction of the main muscle and antagonist muscle (muscle distribution condition) is calculated.
That is, the measured impedance is regarded as indicating the state of the force generated in the entire arm. In other words, the level of the muscular strength of the entire arm can be obtained (= determining the level of co-contraction) using the impedance as a value corresponding to the sum of the muscular strengths of the main muscle and antagonist muscle.
This impedance measurement result is used as a boundary condition in the musculoskeletal model.

Next, in step S120, an estimated muscle strength value is calculated based on the musculoskeletal model.
Next, in step S130, the co-contraction balance is calculated from the muscle strength estimation value calculated in step S120.
Specifically, the balance of the muscle strength of the main muscle and the antagonistic muscle used in the operation of the operating element 100 is determined from the operating reaction force with respect to the operating element 100 (= the balance of co-contraction is determined).

  The estimated muscle strength value obtained from the musculoskeletal model indicates the difference in muscle strength between the main muscle and the antagonist muscle, that is, the co-contraction balance.

Next, in step S140, an estimated muscle strength value that satisfies both the muscle strength level (distribution condition) of the entire arm obtained in step S110 and the co-contraction balance obtained in step S130 is determined (selected).
Next, in step S150, fatigue estimation is performed by the method described in the first embodiment, based on the muscle force estimation value determined in step S140.

(Action and others)
(1) Setting of boundary condition based on impedance measurement result The entire arm can be shown by a model as shown in FIG. 32. From the viscosity B, elasticity K, inertia I, and displacement (angle or position) x, the following equation is obtained. Based on the above, the muscle strength (or torque) F of the entire arm can be calculated.

        F = Ix ″ + Bx ′ + Kx

  Here, if the maximum value of the muscle strength of the entire arm is M, there is a relationship of F ≦ M, and F can be expressed simply as the sum of the main muscle strength and antagonistic muscle strength. When the difference between the main muscle strength and the antagonistic muscle strength is, for example, “1 (N)”, the muscle strength of each muscle is as follows. This is the distribution condition.

(Lead muscle strength, antagonist muscle strength) = (F, 0)
(F-1, 1)
(F-2, 2)
...
(0, F)
That is, it can be expressed by a relationship (distribution condition) as shown in FIG.
Note that the more accurate F is resolved with reference to the musculoskeletal model and muscle information, the better the accuracy of this boundary condition.

(2) About muscle strength estimation by the musculoskeletal model When the maximum muscle strength M is set, for example, when the difference in muscle strength is “1 (N)”, the muscle strength acting on the main muscle and the muscle strength acting on the antagonist muscle are as follows: There is such a relationship. That is, such a muscle force estimated value can be estimated from the musculoskeletal model. The difference in muscle strength corresponds to the operation reaction force.

(Lead muscle strength, antagonist muscle strength) = (1, 0)
(2,1)
(3, 2)
...
(M, M-1)
That is, it can be expressed by the relationship as shown in FIG.

(2) Determination of muscle strength estimation value Therefore, the intersection of the boundary condition (co-contraction level) based on the impedance measurement result and the balance of the muscle strength of the main muscle and antagonist muscle (the solid and broken lines in FIG. 34) From the intersection point), it is possible to determine the estimated strength values of the two muscles.

Even if the number of muscles increases, if a plurality of muscle groups corresponding to the main muscles and a plurality of muscle groups corresponding to antagonistic muscles are defined, the muscle strength can be estimated by the same method.
Here, the information on the balance between the muscle strength of the main muscle and the antagonist muscle can be used as a distribution coefficient (distribution condition) when the rotational moment around the joint obtained from the operation force of the operation element 100 is distributed to the muscle force. It can also be used as a starting point for solving muscle instability problems.

  In addition, if muscle strength is measured from the operation state (operation reaction force, joint position, joint angle) and the musculoskeletal model, and the correspondence between the operation state and the impedance measurement result is made into a database, the muscle strength is measured each time using the musculoskeletal model. Without calculation, muscle strength can be estimated from only the operation state and the impedance measurement result.

(Effect of this embodiment)
(1) Based on the hardness state estimated by the hardness estimating means 102, a muscle force distribution condition for obtaining an estimated value based on the musculoskeletal model is obtained, and the estimated value calculating means 105 is determined based on the obtained distribution condition. The optimum value of the estimated value to be estimated is determined, and the fatigue state is estimated based on the optimum value.

  Thereby, since it is not necessary to compare all the estimated values with the characteristics of the hardness state, the fatigue state can be calculated in a shorter time.

4 Reaction force device motor angle sensor 5 Steering reaction force actuator (disturbance torque applying means)
6 Steering reaction force controller 7 Steering wheel (operator)
8 Torque sensor (Operation reaction force acquisition means)
DESCRIPTION OF SYMBOLS 10 Instruction apparatus utilization determination apparatus 11 Driver camera (attitude acquisition means)
12 tire lateral force detection sensor 13 indicating device 14 disturbance torque signal detection unit (disturbance torque detection means)
15 Tire generation torque signal detection unit 16 Posture detection unit (posture acquisition means)
17 Viscoelasticity estimation unit (estimated value calculation means)
18 Impedance measurement unit (hardness estimation means)
19 Transfer function calculation unit (estimated value calculation means)
20 Frequency characteristic comparison part (fatigue estimation means)
21 Strength calculation part (fatigue estimation means)
22 palm state correction unit 23 determination unit (fatigue estimation means)
DESCRIPTION OF SYMBOLS 100 Manipulator 101 Disturbance torque provision means 102 Estimation means 102A Disturbance torque detection means 102B Resistance characteristic detection means 103 Operation reaction force acquisition means 104 Posture acquisition means 104A Camera 104B Image processing part 105 Estimate value calculation means 106 Fatigue estimation means A Impedance acquisition process B Muscle strength estimated value acquisition process C Fatigue degree index value determination process

Claims (8)

Hardness estimation means for estimating the hardness state of the driver's musculoskeletal part used when operating an operator for giving driving instructions to the vehicle;
An operation reaction force acquisition means for acquiring an operation reaction force when the driver operates the operation element;
Posture acquisition means for acquiring information on the skeletal posture of the musculoskeletal part of the driver;
Based on the operation reaction force and the skeletal posture information, from the musculoskeletal model, an estimated value calculating means for obtaining an estimated value related to the muscular strength of the musculoskeletal portion;
Fatigue estimating means for estimating the driver's fatigue state based on the hardness state estimated by the hardness estimating means and the estimated value obtained by the estimated value calculating means ,
The estimated value calculating means obtains a plurality of estimated values based on the musculoskeletal model,
The fatigue estimation means selects an estimated value closest to the characteristic of the hardness state estimated by the hardness estimation means from the plurality of estimated values obtained by the estimated value calculation means, and based on the selected estimated value A driver fatigue estimation device characterized by estimating a fatigue state . Hardness estimation means for estimating the hardness state of the driver's musculoskeletal part used when operating an operator for giving driving instructions to the vehicle;
An operation reaction force acquisition means for acquiring an operation reaction force when the driver operates the operation element;
Posture acquisition means for acquiring information on the skeletal posture of the musculoskeletal part of the driver;
Based on the operation reaction force and the skeletal posture information, from the musculoskeletal model, an estimated value calculating means for obtaining an estimated value related to the muscular strength of the musculoskeletal portion;
Fatigue estimating means for estimating the driver's fatigue state based on the hardness state estimated by the hardness estimating means and the estimated value obtained by the estimated value calculating means ,
Based on the hardness state estimated by the hardness estimation means, a muscle force distribution condition for obtaining an estimated value based on the musculoskeletal model is obtained, and an estimated value estimated by the estimated value calculation means based on the obtained distribution condition A driver's fatigue estimation device characterized by determining an optimal value of the vehicle and estimating a fatigue state based on the optimal value . The hardness estimation means is
Disturbance torque detection means for detecting disturbance torque input to the operation element;
Driving resistance characteristics of the musculoskeletal unit to the disturbance torque according to claim 1 or claim 2, characterized in that and a resistance characteristic detecting means for detecting a state the hardness which the disturbance torque detection means for detecting Fatigue estimation device. 4. The driver fatigue estimating apparatus according to claim 3 , wherein the hardness estimating means includes disturbance torque applying means for applying disturbance torque to the operator. The operation element is a steering wheel,
The driver's fatigue estimation device according to claim 3 or 4 , wherein the disturbance torque is a torque transmitted from a road surface to a steering wheel. The operation element is a steering wheel,
The hardness state in which the hardness estimating means for estimating, driver fatigue estimation device driver described in any one of claims 1 to 5, characterized in that the correction by the gripping state of the steering Hall . Estimating the hardness state of the driver's musculoskeletal part used when operating the operator to give driving instructions to the vehicle,
Based on the operation reaction force when the driver operates the operation element and the skeletal posture of the driver's musculoskeletal part, from the musculoskeletal model, an estimated value related to the muscular strength of the musculoskeletal part is obtained,
Based on the hardness state and the estimated value, the driver's fatigue state is estimated ,
As an estimated value related to the muscle strength of the musculoskeletal part, a plurality of estimated values are obtained based on the musculoskeletal model,
The driver's fatigue state is estimated by selecting an estimated value closest to the estimated hardness state characteristic from the plurality of estimated values obtained above and estimating the fatigue state based on the selected estimated value. A method for estimating driver fatigue. Estimating the hardness state of the driver's musculoskeletal part used when operating the operator to give driving instructions to the vehicle,
Based on the operation reaction force when the driver operates the operation element and the skeletal posture of the driver's musculoskeletal part, from the musculoskeletal model, an estimated value related to the muscular strength of the musculoskeletal part is obtained,
Based on the hardness state and the estimated value, the driver's fatigue state is estimated ,
The estimation of the driver's fatigue state is based on the estimated hardness state, obtaining a muscle force distribution condition when obtaining an estimated value based on the musculoskeletal model, and the estimated value estimated based on the obtained distribution condition. A driver's fatigue estimation method characterized by determining an optimal value of the vehicle and estimating a fatigue state based on the optimal value .

Images