JP5228155B2 - Omnidirectional power assist device and control method for omnidirectional power assist device - Google Patents

Description

The present invention relates to a device for improving the operability of a wheel drive device using power assist and a control method therefor.

Power assist is a technique for reducing the burden on the operator by generating a driving force based on the force to be operated when the operator tries to move the object.

In Japan, the proportion of people aged 65 and over per population is rapidly increasing, an aging society is progressing, and the development of wheelchairs that reduce the burden on caregivers is required. Under such circumstances, a device using power assist for driving a wheel is widely used in the field of care such as a wheelchair because it can move with a heavy object or a person by assisting human power.

Patent Document 1 is a cart that includes a power assist device that calculates a driving force based on a force operated by an operator and can move in all directions.

Patent Document 2 is a wheelchair that uses power assist technology that operates by driving an assisting push handle provided on the vehicle body and a motor, and the height of the steering bar handle can be adjusted to improve operability. Is.

Patent Document 3 discloses a power assist that can comfortably perform work using a power assist device by changing a virtual mass, a virtual viscosity coefficient, and a virtual spring constant that the worker feels according to the preference of the worker. Device.
JP 2001-233219 A JP 2002-153517 A JP-A-2005-154047

However, in Patent Document 1, Patent Document 2, and Patent Document 3, when an operator applies an operating force using the power assist technology, there is an individual difference in the recognition of the center of gravity position of the device, and the force is compared with the human sensor However, due to the fact that it cannot be applied accurately as intended, there is a problem that the device operates in a direction not intended by the operator.

An object of the present invention is to provide a user-friendly power assist device by inferring the direction of operation in accordance with a habit when an operator of a wheel drive device using power assist operates.

In addition, when inferring the operation direction of the wheel drive device using power assist as described above, it is also an object to provide a device having an easy-to-understand interface for an operator to communicate his intention to the inference unit. .

In order to solve the above-mentioned problems, the present inventor has intensively studied, and as a result, has completed the following invention.

The first invention is a wheel drive device that uses power assist technology, an interface unit for transmitting an operation direction of a device desired by an operator to the device, and an operation direction transmitted from the interface unit as a teacher signal And a direction reasoning unit for inferring the operation direction of the device.

A second invention is the device according to the first invention, wherein the direction inference unit infers the operation direction of the device using a fuzzy neural network.

According to a third aspect of the present invention, the interface unit converts information input from the image display unit for performing information input and device trajectory information display and information input from the image display unit into direction information, and transmits information to the direction inference unit. The apparatus according to the first aspect or the second aspect, comprising a conversion unit.

According to a fourth aspect of the present invention, there is provided a method for controlling a wheel drive device using a power assist technique, wherein the operation direction of the device is inferred using the operation direction of the power assist device desired by the user as a teacher signal. .

A fifth aspect of the invention is the method according to the fourth aspect of the invention, wherein the motion direction inference is performed using a fuzzy neural network.

According to a sixth invention, in the above invention, the operation direction of the device desired by the user is input from the image display device, the input information is given as a teacher signal, and the locus of actual operation of the device is displayed. It is a characteristic method.

According to the above-described apparatus and method of the present invention, the operation direction desired by the operator is given as a teacher signal for direction inference through the interface unit , and based on this, the direction that the operator wants to truly operate is learned. It is possible to prevent the device from moving in an operation direction unintended by the operator due to personal traps.

Next, embodiments of the present invention will be described, but the technical scope of the present invention is not limited to these embodiments, and can be implemented in various forms without changing the gist of the invention. .

As shown in FIG. 1, the embodiment of the present invention is a power assist controller that detects an operation force of an operator and converts it into an operation speed, and transmits an operation direction of the operator to a direction inference unit. The touch panel that is the interface unit, the direction inference unit that infers the direction of the operation speed input from the power assist controller using the fuzzy neural network based on the teacher signal from the touch panel, and the vibration when the device operates It consists of a vibration control part for the purpose, and an actual wheel part.

Detailed description will be given below. The power assist controller shown in FIG. 1 converts the operating force applied by the operator as shown in Equation 1 into a speed. Here, fx indicates an operating force in the x-axis direction, fy indicates an operating force in the y-axis direction, and mz indicates a moment for operating around the z-axis.

At this time, it is desirable to use a speed control method that generates a speed proportional to the force as the power assist method. This is because a speed corresponding to the applied force can be obtained even if the passenger's load changes, and the operator can operate with the same operating force. Further, when such a speed control method is used, the speed command value is zero if no force is applied even on a slope, so that there is an advantage that the wheel is locked and does not slide down due to gravity.

When converting the operating force to the speed of the device, as is clear from Newton's second law, the operating force corresponds to acceleration, and it is natural that the pushing force such as vibration due to walking and hand shake is vibrational. It can be said. Since the device performs speed control, the power assist controller requires an integral element. In the case of a controller with only a simple integration, when the operator releases his hand while applying the operating force, the speed command value at that moment is held and the operation continues. For this reason, a term proportional to acceleration, that is, speed is given a viscous characteristic so that when the assistant releases the hand, it stops naturally.

Satisfying the above, the mass is m, the viscous friction coefficient is c, the X-axis direction force fx [N] to the X-axis direction velocity vx ^P [m / s], the Y-axis direction force fy [N] to Y The power assist controller for obtaining the angular velocity ω ^P [rad / s] around the Z axis from the axial velocity vx ^P [m / s] and the moment mz [Nm] around the Z axis is as follows in the Laplace transformed s region: It is expressed in

Here, Kα represents the gain in the α-axis direction, and Tα represents the time constant of the controller in the α-axis direction.

The touch panel shown in FIG. 1 is composed of an image display device for information input and trajectory output, and a device for transmitting input information to the direction inference unit, and an intention of the operator to operate as a teacher signal to the direction inference unit. It is an interface unit for transmission. Further, at this time, it is desirable that a function capable of displaying a locus of actual operation of the apparatus is provided. The interface unit is not limited to this, as long as it is a device capable of communicating other intentions such as a combination of a keyboard and a display. However, the touch panel is considered to be the most appropriate communication device for the following reasons.
(1) It is possible to give intention data visually.
(2) An interface unit can be constructed only by creating a GUI (Graphical User Interface) that promotes input of intention .
(3) It is possible to display the locus of actual operation of the device.

On the touch panel, designation of angle, right rotation, and left rotation in translation movement can be instructed.
For example, if you want to set the translation data acquisition mode and give the angle you want to tune, touch the part on the touch panel screen to graphically display the size of the angle you want to operate. Make it possible to confirm the given angle.

In addition, when instructing the left / right rotation, a graphical display is provided so that it can be confirmed in which direction the rotation should be performed, and data can be easily acquired.

The direction inference unit shown in FIG. 1 is for inferring a direction that the operator really wants to operate, and preferably using a fuzzy neural network. The reason why this direction reasoning unit is necessary is as follows.

Let us consider a case where the device is translated (left true lateral translation) using power assist without performing direction inference. At this time, when actually performing the power assist of the apparatus, a person tries to apply a force in a direction in which the center of gravity of the apparatus is desired to be moved. The recognition of the position of the center of gravity differs depending on the person, and is found by searching while moving the device. Therefore, when the force is applied, the position of the center of gravity is determined by an individual's sense at the beginning, and the force is applied. Therefore, if the position of the center of gravity of the device and the center axis of the sensor are misaligned, And counterclockwise rotation moment), which leads to deterioration of operability. Here, the right hand operating force is f _R [N], the left hand operating force is f _L [N], the distance in the X-axis direction from the sensor center to the hand grip is L ₁ [m], and the sensor center to the hand grip is If the distance in the Y-axis direction is L ₂ [m] and the distance in the X-axis direction from the device center of gravity to the handgrip is Lomw (f _R and f _L have the same size), it acts on the device center of gravity. The force Fomw is as follows.

Here, fxomw is a force [N] in the X-axis direction acting on the center of gravity of the apparatus, fymw is a force in the Y-axis direction acting on the center of gravity of the OMW [N], and mzomw is a moment about the Z-axis acting on the center of gravity of the OMW [Nm]. It can be seen that the force acts only in the Y-axis direction. The force Fsensor acting on the sensor center is as follows.

Here, fx is a force [N] in the X-axis direction acting on the sensor center, fy is a force [N] in the Y-axis direction acting on the sensor center, and mz is a moment around the Z-axis acting on the sensor center [Nm]. It can be seen that operability deteriorates because mz, which is contrary to human intentions, is input to the sensor.

Also, in the case of right rotation, a force is applied to the handle to rotate the center of gravity of the apparatus to the right, so that the left lateral force and the right rotation moment are input to the sensor, resulting in deterioration of operability.

From the above, it is necessary to make an inference using the direction inference so that the power assist operation intended by the human being is achieved.

The direction reasoning unit is preferably described by fuzzy reasoning. This means that empirical knowledge can be treated as a linguistic expression like if-then format, and knowledge can be arranged in parallel.
Because there are advantages such as being able to draw conclusions from facts that do not necessarily match the antecedent part of the rule, being able to express ambiguous expressions, and not necessarily being theoretically exact It is.

The fuzzy reasoning suitable for the present invention is described by the if-then rule shown in FIG. Incidentally, in FIG. 2, _{f R} is the right hand operating force [N], _{f L} is left-handed force [N], fx is the X-axis direction of the operation force [N], fy is the Y-axis direction of the operation force [N], mz is the moment around the Z axis [Nm], vx ^P is the speed command value in the X axis direction generated by the power assist controller [m / s], and vy ^P is the speed command in the Y axis direction generated by the power assist controller. The value [m / s], ω ^P is the angular velocity command value around the Z axis generated by the power assist controller [rad / s], and vy ^F is the velocity command value in the Y axis direction generated by direction inference [m / s. ], Ω ^P represent angular velocity command values [rad / s] around the Z-axis generated by direction inference.

When performing direction inference using fuzzy theory, it is important how to construct a membership function that indicates the degree of membership in a fuzzy set. Since the membership function is influenced by the characteristics of the operator's individual, it is necessary to tune the membership function to be optimal for each individual. The inventors of the present invention have devised a direction inference method using fuzzy theory as follows. First, the antecedent membership function in fuzzy inference is expressed as follows.

By simply changing the four parameters a1, c1, a2, and c2 in Equation 5, the shapes of positive, zero, and negative functions can be represented.

The membership function of the consequent part of the fuzzy reasoning part is as follows. Tables relating to this membership function are shown in FIGS.

The four parameters of the antecedent part and the parameters of A, B, and C of the consequent part are learned using the neural network shown in FIG. As a result, the fuzzy inference of the optimal operation direction reflecting the operator's intention was realized. The neural network is a back-propagation type, and the output unit is performed by the least square method.

The vibration damping unit shown in FIG. 1 is composed of an HSA controller, which is designed based on the hybrid shaping method devised by the present inventors. By providing the vibration control unit, it is possible to suppress unpleasant vibrations that occur when the apparatus operates and to realize a power assist apparatus that is more user-friendly. This is particularly important in the nursing field.

The algorithm used in the vibration control unit is the same as that of Patent Document 4 and Non-Patent Document 1 which are the prior studies of the present inventors, but this algorithm is applied to the power assist device for the first time. Is an attempt. This algorithm is
(1) Controller selection (2) Formulation of design specifications (3) Formulation of optimization problem (4) Calculation of controller.
JP 2002-351551 A “Vibration control by hybrid shaping considering time and frequency specifications”, Kenichi Yano, Naoki Oguro, Kazuhiko Terashima, Society of Instrument and Control Engineers Vol 37, No 5, 2001

Hereinafter, each step of the algorithm in the vibration control unit will be described. In the (1) controller selection step, a controller is selected based on characteristics to be added, such as proportional gain, integral gain, differential gain, low-pass characteristics, high-pass characteristics, notch characteristics, and band-pass characteristics.

In the (2) design specification formulation step, the design specification in the time domain and the frequency domain is expressed by a penalty function.

In the step (3) formulation of the optimization problem, a constrained optimization problem to which a penalty term is added is formulated.

In the (4) controller calculation step, the obtained optimization problem is solved using the simplex method, and the controller parameters are determined.

The wheel unit shown in FIG. 1 includes a wheel, a speed detection device (such as a six-axis force sensor), and a steering handle.

EXAMPLES Although this invention is demonstrated still in detail based on an Example, this invention is not limited only to these Examples.

When applying the present invention to an omnidirectional wheelchair and performing direction inference using fuzzy reasoning, the parameters of the membership function of a specific person are learned by a neural network, and the result of operation characteristics when used by another person is obtained. FIG. 8 shows the result of the operation characteristics when learning the parameters of membership function for each individual using a neural network when performing direction inference using fuzzy reasoning for the case of operating at 45 degrees diagonally. Show.

From the results, it can be seen that FIG. 8 in which the fuzzy parameters are tuned for each individual, which is the characteristic of the present invention, has better operational characteristics. Also, it can be executed in about 5 minutes to learn several conditions such as left / right translation, straight travel, rotation, etc. After that, even if the operator drives with his / her own scissors with the built controller, the operator's intention It is practical because it moves through.

It is certain that the demand for power assist devices will expand in the future in the field of nursing care such as wheelchairs. What is required of a device used in the field of care is to be kind to the care recipient and reduce the burden on the caregiver. In that sense, the present invention allows the device to be moved in the direction in which it is desired to operate, taking into account the personal characteristics of the operator who is the caregiver, and further provides vibration during operation so as not to be uncomfortable for the care recipient. It is expected to spread in the field of nursing care from the point of doing suppression.

The conceptual diagram of this invention is shown. If-then rule when direction inference is performed using fuzzy inference Membership function of velocity in x direction when direction inference is performed using fuzzy inference Membership function of velocity in y direction when direction inference is performed using fuzzy inference Membership function of rotation direction when direction inference is performed using fuzzy inference Block diagram of fuzzy neural network Operation results when parameters optimized for a specific person are used for other persons Operation results when using parameters tuned for each individual

None

Claims (3)

In the wheel drive device using power assist, an interface unit including an image display unit for performing information input and device trajectory information display for transmitting an operation direction of the device desired by an operator to the device; An apparatus comprising: a direction inference unit that infers an operation direction of the apparatus using a motion direction transmitted based on an operator's intention input from an interface unit as a teacher signal. The apparatus according to claim 1, wherein the direction inference unit infers an operation direction of the apparatus using a fuzzy neural network. The method according to claim 1, wherein the interface unit displays a trajectory in which the apparatus actually operates.

Images