JP2010088628A - Device and method for estimating finger joint position - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a device for estimating finger joint position, estimating easily and accurately estimating the position of a finger joint, and a method for estimating finger joint position using the finger joint position estimating device. <P>SOLUTION: The device for estimating finger joint position includes: measuring terminals disposed on a finger bone at the distal end side and on a finger bone at the root of a finger with a joint of the finger in-between in the longitudinal direction of the finger, and a measuring portion estimating the position of the joint by acquiring and analyzing motion data from the measuring terminals. The accuracy in estimating the position of the joint by the finger joint position estimating device is improved by selecting and analyzing specific data from the motion data obtained from the measuring terminals at the measuring position, and the device for estimating finger joint position is used in the method for estimating finger joint position. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a finger joint position estimation device that can easily and accurately estimate the joint position of a finger, and a finger joint position estimation method that uses the finger joint position estimation device.

  Conventionally, a technique for quantitatively capturing the movement of a human finger has been considered. According to such a technique, it is possible to measure and reproduce techniques such as a craftsman, pianist, pitcher, and surgeon who are involved in traditional performing arts, which are difficult to grasp only from images. In addition, it is considered that the movement of fingers of patients with arthritis or patients undergoing rehabilitation after a fracture can be quantitatively evaluated. Furthermore, it is expected that an artificial joint suitable for the interphalangeal joint mechanism can be produced by clarifying the joint fluctuation when the finger bends / extends.

  As a technique for quantitatively capturing the movement of a human finger, for example, Patent Document 1 discloses an apparatus that records and reproduces a delicate movement of a finger using information obtained from a plurality of sensors attached to the finger. Techniques related to this are disclosed. However, this apparatus only measures changes in the position and posture of the sensor attached to the finger, not the actual finger movement, and the reproducibility of the finger movement is insufficient. In order to accurately reproduce the movement of a finger based on information obtained from a plurality of sensors attached to the finger, it is necessary to know in advance the outline of the finger and the length of the phalange.

The outer shape of the finger can be measured accurately and easily using a laser scanner or the like. On the other hand, for measuring the length of the phalange, a method using computer tomography or the like is conceivable, but this method requires a lot of labor. Therefore, as a method for easily knowing the length of the phalange, a method of estimating the joint position of the finger and measuring the length of the phalange from the estimated joint position and the outer shape of the finger can be considered. As a technique for simply estimating the position of a joint, for example, Non-Patent Document 1 discloses a technique related to a method for estimating the position of a joint of a human limb.
JP 2007-236602 A James O'Brien and three others, “Automatic Joint Parameter Estimate from Magnetic Motion Capture Data”, Graphics Interface (Canada), May 2000, p. 53-60

  However, in the conventional technique disclosed in Non-Patent Document 1, the error between the estimated joint position and the actual joint position is large, and when this technique is applied to a finger joint, The estimation accuracy of was insufficient.

  Therefore, an object of the present invention is to provide a finger joint position estimation device that can easily and accurately estimate a finger joint position, and a finger joint position estimation method using the finger joint position estimation device.

  In order to solve the above-described problem, the first aspect of the present invention provides a measurement terminal disposed on the phalange on the distal end side of the finger and the phalange on the base side with the joint of the finger interposed therebetween, and the measurement Provided is a finger joint position estimation device including a measurement part that can estimate the position of a joint by acquiring and analyzing motion data from a terminal. By using such a finger joint position estimation device, the finger joint position can be easily estimated.

  In the first aspect of the present invention and the present invention described below (hereinafter sometimes simply referred to as “the present invention”), “motion data” means information related to the position and orientation of the measurement terminal. The “measurement terminal” is not particularly limited as long as it is a sensor that can be attached to a finger and can measure the operation data by some method. However, as will be described later, the measurement terminal used in the present invention. Is preferably a triaxial orthogonal coil. Furthermore, the “measurement site” refers to the position of the joint by acquiring and analyzing motion data from two or more measurement terminals installed on both sides of the joint whose position is to be estimated in the finger length direction. There is no particular limitation as long as it is possible.

  In the finger joint position estimation device according to the first aspect of the present invention, the measurement terminal is preferably a three-axis orthogonal coil.

  In the present invention, the “three-axis orthogonal coil” means a coil composed of three-axis isotropic coils that are orthogonal to each other and have a common center. When the measurement terminal is composed of a three-axis orthogonal coil, the position and orientation of the measurement terminal can be easily measured.

  In the finger joint position estimation apparatus according to the first aspect of the present invention, the measurement terminal may be a group of three or more optical sensors. When an optical sensor is used as a measurement terminal, the position and orientation of the measurement terminal can be measured by regarding at least three groups as one measurement terminal.

  In the finger joint position estimation device according to the first aspect of the present invention, the measurement part has data acquisition means for acquiring motion data from the measurement terminal, and an axis in a direction parallel to the finger width direction on the phalanges on both sides of the joint. Each local coordinate system is set, and the movement of the local coordinate system set for the other phalange is converted to Axis-angle with respect to the local coordinate system set for one of the phalanges on both sides of the joint using motion data. The conversion data is created, the conversion data is classified into multiple sections for the axis component in the direction parallel to the finger width direction, and the operation based on the conversion data included in the section with the largest number of conversion data It is preferable to include an analysis unit that estimates the position of the joint using the data. By using such a measurement site, when moving a finger, the position of the joint is estimated using only motion data when the directions of the axes (joint axes) serving as the centers of movement (rotation) of the fingers are substantially the same. Therefore, the joint position of the finger can be estimated easily and accurately.

  In the present invention, the “finger width direction” means a direction orthogonal to the direction from the palm of the hand to the back of the hand and the direction from the base of the finger to the tip.

  In the finger joint position estimation apparatus according to the first aspect of the present invention, when the conversion unit classifies the converted data into a plurality of sections for the axial component in the direction parallel to the finger width direction, the axial component in the direction parallel to the finger width direction is classified. It is preferable to divide the value into 100 equal parts.

  In the second aspect of the present invention, the measurement terminals are arranged on the phalange on the distal end side and the phalange on the base side of the finger with the joint of the finger interposed therebetween, and the operation data is obtained from the measurement terminal by the measurement site. Provided is a finger joint position estimation method for estimating a joint position by acquiring and analyzing the joint position. By using such a finger joint position estimation method, the joint position of the finger can be easily estimated.

  In the finger joint position estimation method according to the second aspect of the present invention, the measurement terminal is preferably a three-axis orthogonal coil. When the measurement terminal is composed of a three-axis orthogonal coil, the position and orientation of the measurement terminal can be easily measured.

  In the finger joint position estimation method of the second aspect of the present invention, the measurement terminal can be a group of three or more optical sensors. When an optical sensor is used as a measurement terminal, the position and orientation of the measurement terminal can be measured by regarding at least three groups as one measurement terminal.

  In the finger joint position estimation method according to the second aspect of the present invention, at the measurement site, the operation data is obtained from the measurement terminal, and the finger bones on both sides of the joint have axes in a direction parallel to the finger width direction. Each local coordinate system is set, and the movement of the local coordinate system set for the other phalange is converted to Axis-angle with respect to the local coordinate system set for one of the phalanges on both sides of the joint using motion data. The conversion data is created, the conversion data is classified into multiple sections for the axis component in the direction parallel to the finger width direction, and the operation based on the conversion data included in the section with the largest number of conversion data It is preferable to perform a process including an analysis step of estimating a joint position using data. By using such a measurement site, when moving a finger, the position of the joint is estimated using only motion data when the directions of the axes (joint axes) serving as the centers of movement (rotation) of the fingers are substantially the same. Therefore, the joint position of the finger can be estimated easily and accurately.

  In the finger joint position estimation method according to the second aspect of the present invention, when the conversion data is classified into a plurality of sections with respect to the axis component values in the direction parallel to the finger width direction in the analysis step, the direction of the finger parallel to the finger width direction is determined. It is preferable to divide the axial component equally.

  According to the first aspect of the present invention, there is provided a finger joint position estimation device capable of easily and accurately estimating a finger joint position.

  According to the second aspect of the present invention, there is provided a finger joint position estimation method capable of easily and accurately estimating a finger joint position.

  The structure of the finger joint is not uniaxial, and the joint axis when moving the finger is not constant. When a finger flexes / extends at the joint, one bone is pulled by the tendon, and the other bone moves while sliding on the bone end joint surface, so that the joint axis fluctuates. It is considered that the fluctuation of the joint axis greatly affects the estimation accuracy of the joint position. The present invention proposes a finger joint position estimation device and a finger joint position estimation method using the finger joint position estimation device, in which the estimation accuracy of the finger joint position is improved in consideration of the fluctuation of the joint axis. .

  The finger joint position estimation device of the present invention is a measurement terminal disposed on the finger bone on the tip side of the finger and the finger bone on the base side across the finger joint (joint whose position is to be estimated), and A measurement part is provided that can estimate the position of the joint by acquiring and analyzing motion data from the measurement terminal.

  The measurement terminal that can be used in the present invention is not particularly limited as long as it can emit information on its own position and orientation. However, from the viewpoint of facilitating measurement and downsizing of the measurement terminal, it is preferable to use a three-axis orthogonal coil as the measurement terminal. In the following description of the present invention, the description is limited to the case where a triaxial orthogonal coil is used as a measurement terminal.

  Hereinafter, the present invention will be specifically described with reference to the drawings.

  FIG. 1 is a diagram schematically showing measurement terminals and measurement parts used in the present invention.

  The measurement terminal 1 (hereinafter referred to as “receiver 1”) is composed of a three-axis orthogonal coil, and the measurement part 2 is a transmitter 3, a drive circuit 4, a detection circuit 5 (data acquisition means) composed of a three-axis orthogonal coil, And it is comprised by the computer and the software 6 for analysis (analysis means). Since the sizes of the transmitter 3 and the receiver 1 are kept very small compared to the distance between the transmitter 3 and the receiver 1, the transmitter 3 and the receiver 1 can be regarded as one point.

<Data acquisition means>
A means for acquiring operation data from the measurement terminal will be described below.

  When the coils in the axial directions constituting the transmitter 3 are sequentially excited by a signal from the drive circuit 4, an alternating magnetic field (pulse magnetic field) having a predetermined frequency is generated in the space around the transmitter 3. When the receiver 1 receives this pulse magnetic field, an electromotive force is generated in the receiver 1 by electromagnetic induction. As will be described later, the electromotive force generated in the receiver 1 varies depending on the relative position and posture of the receiver 1 and the transmitter 3. The electromotive force generated in the receiver 1 is detected by the detection circuit 5.

  Therefore, the electromotive force generated in the receiver 1 is the operation data of the measurement terminal, and the detection circuit 5 is the data acquisition means.

  In the description of FIG. 1 and the above-described data acquisition means, for simplicity, one form of the receiver 1 is illustrated. However, when estimating the joint position of the finger, the receiver 1 is provided on each side of the joint in the longitudinal direction of the finger. And the operation data obtained from the two receivers 1 are used.

<Analysis means>
A means for estimating the position of the joint using the operation data obtained from the receiver 1 installed on both sides of the joint whose position is to be estimated will be described below.

(Position and posture of receiver 1)
First, the principle by which the position and orientation of the receiver 1 can be grasped quantitatively will be described.

  The pulse magnetic field emitted from the transmitter 3 is inversely proportional to the cube of the distance from the transmitter 3. Therefore, the distance between the receiver 1 and the transmitter 3 can be obtained by digitizing the magnitude of the electromotive force generated in the receiver 1 detected by the detection circuit 5 via the computer and the analysis software 6. In addition, an electromotive force proportional to the relative posture of the receiver 1 with respect to the transmitter 3 is generated in each coil constituting the receiver 1. Therefore, by analyzing the information from each coil constituting the receiver 1 by the computer and the analysis software 6, the relative attitude of the receiver 1 with respect to the transmitter 3 can be obtained. In this manner, the position and orientation of the receiver 1 can be quantitatively captured using the operation data of the receiver 1 acquired by the detection circuit 5.

(Joint position estimation)
Next, the process of estimating the joint position from the operation data of the receiver 1 will be described using the model shown in FIG. 2 for simplicity.

  In the model 20 shown in FIG. 2, two rigid bodies (upper segment 22 and lower segment 23) are regarded as phalanges, and one bar (joint 21) is regarded as a joint. The upper segment 22 and the lower segment 23 (hereinafter, the upper segment 22 and the lower segment 23 may be collectively referred to as “segments 22, 23”) are connected by the joint 21, and the joint 21 is moved as a joint axis. Can do. On the upper segment 22 and the lower segment 23, the receivers 1 described above are respectively installed. The receivers 1 and 1 are installed at an angle to avoid collision during operation of the upper segment 22 and / or the lower segment 23. In the following description of the present invention, the receiver 1 installed on the upper segment 22 and the receiver 1 installed on the lower segment 23 are distinguished from each other and may be referred to as an upper receiver 1a and a lower receiver 1b, respectively.

  By obtaining the operation data of the upper receiver 1a and the lower receiver 1b by the above-described method, a local coordinate system (hereinafter, the local coordinate system is referred to as “LC”) with the centers 11a and 11b of the upper receiver 1a and the lower receiver 1b as origins. Can be set, and the change of those LCs can also be captured.

  The position vectors 12a and 12b from the origins 11a and 11b to the joint 21 do not change even when the upper segment 22 and / or the lower segment 23 are operating. Therefore, it is considered that the estimation process of the coordinates (joint position) of the joint 21 can be treated as an optimization problem described below.

Consider a case where a point P indicated by an arbitrary position vector V _UP (x _UP , y _UP , z _UP ) is set from the LC of the higher-order receiver 1a. Even when the upper segment 22 and / or the lower segment 23 are operating, the magnitude of V _UP does not change. At this time, the magnitude | V _LP (i) | of the position vector V _LP (x _LP , y _LP , z _LP ) from the LC of the lower receiver 1b to the point P in the i-th frame in operation is It is expressed as

While the upper segment 22 and / or the lower segment 23 are in operation, the above | V _LP (i) | varies. That is, the variance s ² of | V _LP (i) | in all frames of the operation data is expressed by the following equation 2.

(However, | V _LP | _m : average of | V _LP | in all frames)

In the above formula 2, if s ² is 0, the point P overlaps with the position of the joint 21, and thus it can be said that the V _UP coordinate at this time is the position of the joint 21. That is, the position of the joint 21 can be estimated by obtaining V _LP that minimizes s ² .

  However, when the joint position estimation process is handled as such an optimization problem, the deviation between the actual position of the joint 21 and the estimated position is large. Therefore, in the present invention, it was considered to improve the estimation accuracy by adding a restriction to change the region estimated as the position of the joint 21 from three dimensions to two dimensions.

  Hereinafter, a method for estimating the joint position when the estimation region is constrained to two dimensions will be described.

The expression of the plane in the space is expressed as the following equation type.
ax + by + cz + d = 0 (Formula 3)

In order to consider a three-dimensional space, there is always one z for any value of x and y. In order to obtain the plane equation, as shown in FIG. 3, if the position coordinates of three points (P ₁ , P ₂ , P ₃ ) on the surface are known, the coefficients a, b, and c are obtained by simultaneous linear equations. Can do. Note that (a, b, c) is a normal vector and represents a vector perpendicular to the plane. Further, when the plane in the space is represented by the parameter type by the parameters t and s, the following is obtained.

However,

A plane in space can be defined in parameter form by two different vectors on the plane. That is, the point P ₁ represents the coordinates of a certain point on the plane, and v ₂₁ (v _21x , v _21y , v _21z ), v ₃₁ (v _31x , v _31y , v _31z ) represent two vectors on the plane. . Note that n represents a normal vector.

From equation 4, the normal vector (a, b, c) can be derived from the coordinates at which the plane intersects the x-axis, y-axis, and z-axis. For example, the coordinates intersecting the x axis can be calculated by obtaining s and t satisfying y = 0 and z = 0. Accordingly, the reciprocal of the calculated x coordinate becomes a of the x component in the normal vector. The same calculation can be performed for b and c. The normal vector can be calculated in the same manner by taking the outer product of v ₂₁ and v ₃₁ as follows.

The normal vector obtained by the outer product has the same direction as the normal vector (a, b, c) of the plane formula, but a different size. Therefore, by substituting the point P ₁ is the starting point of the normal vector to the expression of the surface, it is necessary to adjust the magnitude of the normal vectors Motoma' outer product.

  When the estimation area is constrained to a plane, the following optimization problem is solved.

  “Create an arbitrary plane in a three-dimensional space and find an explanatory variable that minimizes an objective function within that plane.”

  The conditional expressions are shown below.

here,

The estimated position of the joint position can be obtained by solving the above optimization problem by, for example, the simplex method (other methods such as the least square method are available). The above condition is a condition in the case of setting the point P indicated by LC upper receiver 1a at an arbitrary position vector V _UP, estimates the joint position. The same conditional expression applies when the point P is set from the LC of the lower receiver 1b and the joint position is estimated.

  The initial value setting method is shown below.

As shown in FIG. 4, a perpendicular is drawn from the point V arranged at an arbitrary position to the plane constituted by P ₁ , P ₂ and P ₃ . Let the point P be the intersection of the perpendicular and the plane. From equation 4, the components (a, b, c) of the normal vector are obtained. Substituting P ₁ into Equation 3 to obtain the coefficient d in Equation 3 which is a plane equation is as follows.

However,

  Further, the distance l from the point V to the point P is as follows.

  Using the above results, the vector VP can be expressed as follows.

  When Expression 8 is expanded, the point P can be expressed as follows.

  The position of the point P calculated by Equation 9 is the initial value of the simplex method.

After defining the initial value on the plane, when substituting the point P into Equation 3 and calculating the parametric variables s and t, case classification is required. The conditions are as follows.
1. 1. When the constructed plane is parallel to the yz plane → Solve simultaneous equations of y and z components 2. When the constructed plane is parallel to the xz plane → Solve simultaneous equations of x and z components When the constructed plane is parallel to the xy plane → Solve simultaneous equations of x and y components

In the simplex method, a polyhedron is determined by the number of objective function variables (in the case of n variables, a polyhedron is created with n + 1 vertices), the objective function value at the vertex of this polyhedron (simplex) is evaluated, A method for finding the optimal solution by repeating the simplex vertex rearrangement, reflection, expansion, reduction, and contraction for the point with the worst function value (the objective function value that is the maximum value for minimization) It is. The procedure of the simplex method is shown below.
1. Rearrange vertices (reorder values at n + 1 vertices)
2. Reflection
3. Expand
4). Contraction
Outside contract
Inside contract
5). Shrink
6). Determine the minimum value by changing the vertex in this way and repeating the process.

  A method for numerically solving the above optimization problem is not particularly limited, and a known method may be used.

In this way, the joint position estimated by setting the point P indicated by LC upper receiver 1a in any vector V _UP, hereinafter referred to as a P _U. Set the point indicated by the arbitrary vector from LC lower receiver 1b made it similarly to estimate the joint position, joint positions estimated in this case, hereinafter referred to as a P _L.

(Joint axis fluctuation)
If the uniaxial joint (joint 21) is provided as in the model 20 shown in FIG. 2, the orientation of the joint axis is constant during the operation of the upper segment 22 and / or the lower segment 23. When estimating the joint position, it is considered that the optimization problem described above should be solved without considering the fluctuation of the joint axis. On the other hand, in the case of a finger joint, the joint axis fluctuates during the movement of the finger. Therefore, it is considered that the fluctuation of the joint axis needs to be taken into account in order to estimate the joint position of the finger. In the present invention, the variation of the joint axis of the finger is quantitatively captured using Axis-angle.

  With reference to FIG. 5, a method for quantitatively grasping joint axis fluctuations using Axis-angle will be described. 5, components having the same configuration as in FIG. 2 are denoted by the same reference numerals, and description thereof will be omitted as appropriate.

  As shown in FIG. 5A, LCs are set for the upper segment 22 and the lower segment 23. Since the upper segment 22 is a rigid body, the LC origin of the upper segment 22 may be set at any position within the upper segment 22. The same applies to the lower segment 23. Using this property, let us consider a state in which the LCs of the segments 22 and 23 are translated to the position of the joint 21 as shown in FIG. In the case of the model 20, since the joint 21 is a uniaxial joint, the LC y-axis of the segments 22 and 23 and the direction (longitudinal direction) of the joint 21 coincide even when the upper segment 22 and / or the lower segment 23 are operating. To do.

  During the operation of the upper segment 22 and / or the lower segment 23, the LC of the upper segment 22 and the LC of the lower segment 23 have the same y-axis, so the x axis of the LC of the upper segment 22 and the x of the LC of the lower segment 23 It can be said that the angle θ between the axes is the bending angle of the joint. Further, in the model 20, when the variation of the LC of the lower segment 23 with respect to the LC of the upper segment 22 is expressed by Axis-angle, the direction of the Axis-angle vector substantially coincides with the direction of the joint 21.

  On the other hand, the actual joint is not a perfect uniaxial joint like the model 20. Therefore, LC is set for the phalange existing through the joint whose position is to be estimated, similarly to the upper segment 22 and the lower segment 23 of the model 20, and the LC set for the other phalange with respect to the LC set for one phalange. When the change of Axis is represented by Axis-angle, it is considered that the orientation of the Axis-angle vector varies.

  For a finger and a model having a uniaxial joint that simulates a finger (hereinafter referred to as “mock”), the movement of the joint axis during operation is converted into Axis-angle, and the motion data for the y component of Axis-angle FIG. 11 shows a graph in which these are classified. 11A is a graph in which the y component (x axis) of Axis-angle is 0 to 1, and FIG. 11B is a graph in which the y component (x axis) of Axis-angle is 0.9 to 1. It is. The measurement conditions will be described in detail in the examples.

  As can be seen from FIG. 11, in the case of mock, the orientation of the Axis-angle vector does not vary and is almost parallel to the y-axis, whereas in the case of a finger, the orientation of the Axis-angle vector varies. It is done.

  In this way, it is possible to identify the movement when the joint axis is substantially constant during the movement of the finger by quantitatively capturing the fluctuation of the joint axis of the finger using Axis-angle. Then, it is considered that the joint position estimation accuracy can be improved by estimating the joint position using only the motion data when the joint axis is substantially constant.

  Hereinafter, a method of expressing the variation of the LC of the lower segment 23 with respect to the LC of the upper segment 22 by Axis-angle using the operation data obtained from the receivers 1 and 1 will be specifically described.

First, the correspondence between the LC of the transmitter (not shown) fixed in an arbitrary location and the LC of the segments 22 and 23 is clarified, and the correspondence between the LC of the transmitter and the LC of the segments 22 and 23 is clarified (hereinafter, the correspondence between the LC of the transmitter and the LC of the segments 22 and 23). In this case, the stationary state is referred to as “standard state”), and data is acquired for several seconds. The attitude data of the upper receiver 1a and the lower receiver 1b are averaged and converted into xyz type rotation matrices R _UR (0) and R _LR (0), respectively.

The rotation matrix of the segments 22 and 23 in the standard state is a unit matrix, and the rotation matrix R _US (i) of the upper segment 22 and the rotation matrix R _LS (i) of the lower segment 23 in the i-th frame in operation are And from Equation 11.

R _LS (i) = R _LR (i) · R _LRS (Formula 10)
(However, R _LRS = R _LR (i) ⁻¹ · R _LS (0))

R _US (i) = R _UR (i) · R _URS (Formula 11)
(However, R _URS = R _UR (i) ⁻¹ · R _LS (0))

The posture change R _ULS (i) of the lower segment 23 with respect to the upper segment 22 is obtained from the following equation 12.

R _ULS (i) = R _US (i) ⁻¹ · R _LS (i) (Formula 12)

R _ULS (i) in all frames is calculated and converted to Axis-angle. Since there are two Axis-angle solutions for a certain posture, the y component of the calculated vector is set to be positive.

  Similarly, in the case of the finger, the change of the joint axis can be expressed using Axis-angle, and the data converted into Axis-angle is classified into a plurality of sections with respect to the y component. By classifying the data converted to Axis-angle with respect to the y component, it is possible to specify when the direction of the joint axis is substantially constant during the movement of the finger. Then, using the motion data that is the basis of the conversion data included in the section with the largest number of data, the joint position estimation accuracy is improved by solving the above optimization problem and estimating the joint position. Can do.

  Hereinafter, the present invention will be described in more detail by way of examples. However, the present invention is not limited to the examples.

  The components of the finger joint position estimation apparatus of the present invention used for the measurement described below will be described.

<Measurement terminal>
A 3-axis orthogonal coil was used as a measurement terminal (receiver). Since the existing triaxial orthogonal coil has a large coil part and a thick cable, a receiver in which the coil part is miniaturized and the cable is thinned by Akita University and POLHEMUS. The coil portion of the manufactured receiver has a teardrop shape, and the size is 9.6 mm long × 9.6 mm wide × 9.6 mm high, and can be sufficiently mounted on the fingertip. The thickness of the cable is about 2 mm in diameter and does not substantially affect the movement of the finger. Furthermore, the receiver was molded with free plastic in order to facilitate attachment to fingers.

<Measurement site>
As the measurement site, a transmitter, a drive circuit, a detection circuit, and a computer equipped with a computer and analysis software were used in the same manner as illustrated in FIG.

(Transmitter)
The transmitter (TX-1) manufactured by Polhemus was used as the transmitter. The size of TX-1 was length 22.9 mm × width 27.9 mm × height 15.2 mm. The measurement range of TX-1 is on a semicircle with a radius of about 200 mm. Moreover, although the thickness of the cable of TX-1 is about 4.2 mm in diameter, since it fixes to a wrist, it does not interfere with finger operation. In order to make it easy to attach to fingers, Velcro was attached to the back of TX-1.

(Drive circuit, detection circuit, computer and analysis software)
Liberty manufactured by Polhemus was used as the drive circuit, detection circuit, computer and analysis software. Liberty has a sampling rate of 240 Hz per receiver and can measure operation data at high speed. In addition, the sampling rate can be changed to 120 Hz. Furthermore, the Liberty can be controlled and the measurement data can be stored by connecting the Liberty to the computer via USB. For the control of Liberty and the measurement and storage of operation data, PiMgr provided by Polhemus was used.

  The validity of the finger joint position estimation method of the present invention was examined using the above-described finger joint position estimation apparatus of the present invention.

<Examination of validity of finger joint position estimation method of the present invention by mock>
Prior to estimating the finger joint position by the finger joint position estimation method of the present invention, the joint position of the mock was estimated, and the validity of the finger joint position estimation method of the present invention was verified by examining the estimation results.

  FIG. 6 schematically shows the configuration of the produced mock 40. The mock 40 was made of wood that did not disturb the magnetic field. Two joints (joint shafts) 41a and 41b were made of a cylinder having a diameter of 4 mm, and the three segments 42, 43 and 44 were connected by the joints 41a and 41b. The names of the two joints 41a and 41b are DIP joint 41b and PIP joint 41a, respectively, and the names of the two segments 43 and 44 connected to the DIP joint 41b are upper segment 43 and lower segment 44, respectively.

  As shown in FIG. 7, the lower receiver 1b and the upper receiver 1a connected to the Liberty are arranged on the lower segment 44 and the upper segment 43, and the transmitter (TX-1) connected to the Liberty on the remaining segment 42 3 was placed. The lower receiver 1b and the upper receiver 1a are installed at an angle in order to avoid collision during the operation of the upper segment 43 and / or the lower segment 44. At the time of measuring the movement, the segment 42 on which the transmitter 3 was placed was firmly fixed to a wooden frame as a base. The measurement environment was set so that there was no object that disturbed the magnetic field in the measurement range.

After preparing the measurement environment as described above, the lower segment 44 of the mock 40 was moved by hand for about 15 to 30 seconds like the movement of fingers flexing / extending. The measured operation data was recorded using PiMgr provided by Polhemus. The above process was repeated five times to obtain five motion data. And the position (P _L and P _U ) of the DIP joint 41b was estimated by solving the optimization problem described above.

<Examination of validity of finger joint position estimation method of the present invention using fingers>
Two receivers were connected to the Liberty, one of the two receivers was placed on the index finger phalanx as a lower receiver, and the other was placed on the middle phalanx of the index finger to be an upper receiver. The receiver placed on each phalange was placed at the center of the phalange so as not to interfere with contact between adjacent receivers and articulation. The measurement environment was set so that there was no object that disturbed the magnetic field in the measurement range. A transmitter (TX-1) connected to the Liberty was placed on the wrist.

After preparing the measurement environment as described above, bending / extending of fingers was repeated for about 15 seconds to 30 seconds. The measured operation data was recorded using PiMgr provided by Polhemus. The above process was repeated five times to obtain five motion data. And the position (P _L and P _U ) of the joint (DIP joint) between the distal phalanx and the middle phalanx was estimated by solving the optimization problem described above.

<Estimation result>
The validity of the joint position estimation result was examined by calculating the distance between the P _L and P _U. As estimated joint position is closer to the actual joint position, the distance of the P _L and P _U are considered to be small.

Figure 8 shows a comparison of the distance of the _{P L} and _{P U} in mock 40 and fingers. As it can be seen from FIG. 8, while the distance between _{P L} and _{P U} in the case of estimating the position of the DIP joint 41b is about 2mm mock 40, and _{P L} in the case of estimating the position of the index finger of the DIP joints distance distance P _U is is about 8 mm, not able to estimate the joint positions in a reasonable position. This was thought to be due to the difference in the joint structure between the mock 40 and the fingers. The segments 43 and 44 of the mock 40 are connected by a single shaft 41b and have a center of rotation. On the other hand, there is no center of rotation in the finger joint, and the lower phalange is pulled by the tendon in the flexion / extension movement, and the joint axis changes because the upper phalange end joint surface moves while sliding. As a result, it was speculated that the fluctuation of the joint axis had a great influence on the accuracy of joint position estimation.

  In the present invention, as described above, in order to estimate the joint position in consideration of the fluctuation of the joint axis, the fluctuation of the joint axis is quantitatively captured using Axis-angle.

  In the case of the mock 40, as shown in FIG. 7, when LCs are set in the lower segment 44 and the upper segment 43, the y-axis directions of these LCs coincide with the DIP joint 41b. As with the mock 40, if the finger joint is a perfect uniaxial joint, the Axis-angle vector should match the y-axis even when the phalange is moving. Therefore, the posture data of the mock 40 and the finger are converted into Axis-angle, and the change of the axis-angle vector in the mock 40 and the finger is examined, thereby calculating the variation of the joint axis at the joint of the mock 40 and the finger. As described above, the motion data of the mock 40 and the fingers were measured 5 times each. Here, the five motion data measured were combined and handled as one data.

  The Axis-angle of the mock 40 and finger DIP joint calculated by the above-described method was calculated. The calculated Axis-angle vector is plotted in the three-dimensional space as shown in FIG. There are two Axis-angle solutions for a certain posture. For this reason, the y component of the calculated vector is set to be positive. FIG. 10 shows the relationship between the Axis-angle vector component and the rotation component (angle). Since the Axis-angle vector is a unit vector, it can be seen that it greatly depends on the y component.

  From the above results, FIG. 11A shows a graph obtained by classifying the operation data with respect to the y component value. It can be seen that in both cases of mock and fingers, the y component is present in the range of 0.9 to 1 in an amount of 90% or more. FIG. 11B shows a graph obtained by further classifying the motion data distribution in the section where the y component is 0.9 to 1 into 10 equal parts. Mock considers that 70% or more of the data exists in the interval from 0.99 to 1, and the joint axis does not vary and is maintained substantially parallel to the y-axis of the coordinate system. On the other hand, when attention is paid to the data distribution of fingers, there is a variation in data in the section of 0.96 to 1. The fact that the value of the y component varies during the movement of the finger indicates that the joint axis is fluctuating, and this is considered to be a cause of deteriorating the joint position estimation result. Therefore, an attempt was made to estimate the joint position using the motion data in the section 0.97 to 0.98 having the largest number of data in the section 0.96 to 1 in which data variation occurred.

The distance of _{P L} and _{P U} y component of Axis-angle is estimated using the operation data in the interval of 0.97 to 0.98 is shown in FIG. 12. To a distance of _{P L} and _{P U} in the case (A) with all the operation data that is 7.09Mm, distance _{P L} and _{P U} in the case estimated using the classification data (B) is The accuracy of joint position estimation was improved to 4.21 mm.

  From the above results, it is considered that when the motion data of fingers is classified by the y component of Axis-angle and the joint position is estimated using the data of the section with the largest data distribution, the joint position estimation accuracy is improved.

  While the present invention has been described in connection with embodiments that are presently the most practical and preferred, the present invention is not limited to the embodiments disclosed herein. Rather, the finger joint position estimation device and the joint position estimation method can be changed as appropriate without departing from the spirit or concept of the invention that can be read from the claims and the entire specification. Should be understood as being included in the scope.

It is a figure which shows roughly the structure of the finger joint position estimation apparatus of this invention. It is a figure for demonstrating the principle by which a joint position is estimated. It is a figure which shows three points which comprise a plane. It is a figure for demonstrating the initial position setting method. It is a figure for demonstrating the estimation method of the joint position in consideration of the fluctuation | variation of a joint axis. It is a figure which shows the structure of a mock schematically. It is a figure which shows roughly a mode that the receiver and the transmitter were arrange | positioned at the mock. It shows a comparison of the distance of the P _L and P _U in mock and fingers. It is a figure which shows the vector distribution of Axis-angle. It is a figure which shows the relationship between an Axis-angle vector distribution and an angle. It is a figure which shows distribution of operation data. In the case of using a specific operation data selected with the case of using all the operation data, it shows a comparison of the distance of the P _L and P _U in hand.

Explanation of symbols

1 Receiver (measurement terminal)
2 Measurement site 3 Transmitter 4 Drive circuit 5 Detection circuit 6 Computer and analysis software 21 Joint 22 Upper segment 23 Lower segment

Claims (10)

A measurement terminal disposed on the phalange on the distal end side and the phalange on the base side of the finger joint across the joint of the finger, and by obtaining and analyzing operation data from the measurement terminal, A finger joint position estimation device comprising: a measurement site capable of estimating a position. The finger joint position estimation apparatus according to claim 1, wherein the measurement terminal is a three-axis orthogonal coil. The finger joint position estimation apparatus according to claim 1, wherein the measurement terminal is a group of three or more optical sensors. The measurement site is
Data acquisition means for acquiring operation data from the measurement terminal;
A local coordinate system having an axis parallel to the width direction of the finger is set on the phalanges on both sides of the joint, and the phalanges of one of the phalanges on both sides of the joint are set using the motion data. A conversion data obtained by converting a variation of the local coordinate system set in the other phalange to an Axis-angle with respect to the local coordinate system set in the axis is generated, and the converted data is converted into an axial component in a direction parallel to the finger width direction. Analyzing means for estimating the joint position using motion data based on the conversion data included in the section with the largest number of conversion data,
The finger joint position estimation apparatus according to any one of claims 1 to 3, further comprising: In the analysis means, when the conversion data is classified into a plurality of sections for the axial component in the direction parallel to the finger width direction, the value of the axial component in the direction parallel to the finger width direction is divided into 100 parts. Item 5. The finger joint position estimation apparatus according to Item 4. By placing measurement terminals on the phalange on the distal side and the phalange on the base side of the finger across the joint of the finger, and obtaining and analyzing operation data from the measurement terminal by the measurement site, A finger joint position estimation method for estimating a position of the joint. The finger joint position estimation method according to claim 6, wherein the measurement terminal is a three-axis orthogonal coil. The finger joint position estimation method according to claim 6, wherein the measurement terminal is a group of three or more optical sensors. In the measurement site,
Acquiring operation data from the measurement terminal, a data acquisition step,
A local coordinate system having an axis parallel to the width direction of the finger is set on the phalanges on both sides of the joint, and the phalanges of one of the phalanges on both sides of the joint are set using the motion data. A conversion data obtained by converting a variation of the local coordinate system set in the other phalange to an Axis-angle with respect to the local coordinate system set in the axis is generated, and the converted data is converted into an axial component in a direction parallel to the finger width direction. An analysis step of classifying into a plurality of sections and estimating the position of the joint using motion data based on the conversion data included in the section with the largest number of conversion data;
The finger joint position estimation method according to any one of claims 6 to 8, comprising: In the analysis step, when the conversion data is classified into a plurality of sections with respect to an axial component in a direction parallel to the finger width direction, the axial component value in a direction parallel to the finger width direction is divided into 100 parts. Item 10. The finger joint position estimation method according to Item 9.