JP7066174B2 - Motion analysis device - Google Patents

Description

The present disclosure relates to a motion analysis device for analyzing the motion of an animal.

In recent years, techniques for analyzing the movement of muscles of humans and animals have been developed. The movement is analyzed using a muscle model that expresses the mechanical function of human muscles on a computer. Regarding the muscle motion analysis technique, Japanese Patent Application Laid-Open No. 2017-037553 (Patent Document 1) discloses a muscle model of a human body adopting a spring mass damper model method.

In order to realize more complicated motion analysis, it is necessary to consider the interference between parts in the body. Japanese Patent Application Laid-Open No. 2006-302098 (Patent Document 2) relates to a technique for avoiding interference between three-dimensional objects. Patent Document 2 discloses "a three-dimensional object control device capable of calculating interference between objects arranged in a virtual space by a simple calculation". The three-dimensional object control device disclosed in Patent Document 2 is not premised on avoiding interference between parts in the body.

Japanese Unexamined Patent Publication No. 2017-037553 Japanese Unexamined Patent Publication No. 2006-302098

Each part of the body is in constant contact with other parts, and it is necessary to always consider the interference of each part in order to realize more complicated motion analysis. When the interference determination process and the interference avoidance process are performed on each part of the body, a huge amount of time is required for the process. Therefore, it is not realistic to always perform these processes. Therefore, it is necessary to realize the avoidance of interference of each part in the body by a method different from the conventional method.

The present disclosure has been made to solve the above-mentioned problems, and the purpose in a certain aspect is a motion analysis capable of realizing an interference avoidance operation of each part in the body by a method different from the conventional method. To provide the device.

According to a certain aspect, the motion analysis device for analyzing the motion of an animal includes a storage unit for storing a plurality of three-dimensional models representing each part of the body of the animal. Each of the plurality of three-dimensional models is represented by a plurality of mass points. Each of the above-mentioned mass points is associated with a coordinate value in three dimensions and a mass, and is related to the other by a mechanical relationship depending on the distance from other associated mass points in the three-dimensional model. It is combined with the mass point of. The plurality of three-dimensional models include a first to third part model representing the first to third parts in the body of the animal. A part or all of the plurality of mass points constituting the third site model may be any of the plurality of mass points constituting the first site model and the plurality of mass points constituting the second site model. It is associated with the mass point. A calculation unit for calculating the coordinate values for each of the plurality of mass points of the first to third part models is provided. For each of the plurality of mass points of the first to third part models, the arithmetic unit receives the mass point from the other mass point according to the distance between the mass point and the other mass point associated with the mass point. The force is calculated, and for each of the plurality of mass points of the first to third part models, the mass point is based on a motion equation that depends on the mass associated with the mass point and the force calculated for the mass point. The coordinate values of are calculated sequentially.

Preferably, the third site model is composed of a plurality of mass points constituting the first surface of the third site and a plurality of mass points constituting the second surface of the third site. There is.

Preferably, some or all of the plurality of mass points constituting the first surface are associated with at least one mass point constituting the first site model. Some or all of the plurality of mass points constituting the second surface are associated with at least one mass point constituting the second site model.

Preferably, the third site model represents the skin of the animal. A part or all of the plurality of mass points constituting the first surface is any one of the plurality of mass points constituting the first site model and the plurality of mass points constituting the second site model. Is associated with the mass point of.

Preferably, the first site model represents the first muscle in the animal's body. The second site model represents a second muscle in the body of the animal.

In a certain aspect, it is possible to realize the avoidance of interference of each part in the body by a method different from the conventional method.

The above and other objects, features, aspects and advantages of the invention will become apparent from the following detailed description of the invention as understood in connection with the accompanying drawings.

It is a figure which shows the example of the muscle model schematically. It is a figure which shows the mass point information 112 about each mass point of a muscle model. It is a block diagram which shows an example of the functional structure of a motion analysis apparatus. It is a conceptual diagram which shows the arrangement of a thread in a GPU (Graphics Processing Unit) at the time of coordinate calculation of each mass point. It is a figure for demonstrating the interference avoidance processing between muscle models. It is a figure which shows an example of the data structure of the join information. It is a figure for demonstrating the interference avoidance processing between a muscle model and a skin model. It is a figure which shows the musculoskeletal model schematically. It is a block diagram which shows the main hardware composition of a motion analysis apparatus. It is a figure which shows an example of the simulation result of this proposed method.

Hereinafter, embodiments according to the present invention will be described with reference to the drawings. In the following description, the same parts and components are designated by the same reference numerals. Their names and functions are the same. Therefore, the detailed description of these will not be repeated. Moreover, the modification described below may be selectively combined as appropriate.

[Coordinate system of 1 muscle model]
A coordinate system for representing the muscle model 50 according to the embodiment will be described with reference to FIG. FIG. 1 is a diagram schematically showing an example of a muscle model 50.

The muscle model 50 is represented by a plurality of mass points. A mass point is an element that represents each part that constitutes a muscle. Each mass point is associated with a three-dimensional coordinate value and mass.

As shown in FIG. 1, each mass point constituting the muscle model 50 is represented by a three-dimensional grid coordinate system (i, j, k). The "k" in the coordinate system represents the number of the cross section orthogonal to the longitudinal direction of the muscle (that is, the direction from the beginning to the end of the muscle). The number of each mass point included in each cross section of the muscle model 50 is managed by the two-dimensional lattice coordinates (i, j).

The number of quality points in the _i direction is Ni. The number of quality points in the _j direction is Nj. The number of quality points in the k direction is N _k . That is, one line is composed of elongated polygons formed by arranging (N _k -1) hexahedron polygons in series in parallel (N _i -1) * (N _j -1). The serial number n of the (i, j, k) th mass point is represented by "n = i + (N _i * j) + (N _i * N _j * k)". For example, when _{Ni = 5 and N j =} 7, the serial number of the mass point at (i, j, k) = (0, 2, 1) is 45 (= 0 + (5 * 2) + (5 * 7). * 1)). By adjusting the values of N _i , N _j , and N _k , the muscle model 50 can be represented by various shapes such as elongated muscles and flat muscles.

This shape expression can be used not only for muscles but also for ligaments connecting bones and aponeurosis that bundles multihead muscles. As a result, the shape of each part of the human body (for example, muscle, tendon, ligament, aponeurosis, etc.) is centrally managed. Further, by such a shape representation, the shapes of all human body parts can be calculated by the kernel function of the same GPU as described later.

[2 Muscle model motion analysis]
(2.1 Overview)
For the muscle model 50 (see FIG. 1), a spring mass damper model in which adjacent mass points are mechanically connected is adopted. The feature of the spring mass damper model is that the equation of motion of each mass point is expressed by the same formula, and the integral calculation of one time step (Δt) of the simulation of each mass point can be performed independently of other mass points. It is a point. As a result, the integral calculation of a large number of mass points in one time step can be processed in parallel. As a method of time integration, for example, Verlet's method is used.

(2.2 Equation of motion)
A method of motion analysis of the muscle model 50 (see FIG. 1) by the motion analysis device 100 will be described with reference to FIGS. 2 and 3. FIG. 2 is a diagram showing mass point information 112 for each mass point of the muscle model 50. FIG. 3 is a block diagram showing an example of the functional configuration of the motion analysis device 100.

As shown in the following equation (1), the equation of motion of the nth mass point is adjacent to the passive force "f ^Passive " received from the adjacent mass point and the active muscle contraction force "f ^Active ". It is determined by the external force "f ^Col " received from the muscle or bone.

“M _n ” in the formula (1) represents the mass of the nth mass point. The mass " _mn " of each mass point is defined in, for example, the mass point information 112 of FIG. As an example, the mass " _mn " of each mass point is obtained in advance from the human body model that defines the structure of the human body. “A _n ” in the equation (1) represents the acceleration vector of the nth mass point.

Equation (1) is calculated by, for example, the arithmetic unit 150 included in the motion analysis device 100. The arithmetic unit 150 corresponds to a CPU (Central Processing Unit) 102 or GPU 104, which will be described later. The calculation unit 150 sequentially calculates the coordinate values for each mass point based on the fact that the mass points constituting the surface of the muscle receive an external force among the plurality of mass points constituting the muscle model 50. The calculation unit 150 depends on the mass associated with the mass point, the passive force "f ^Passive " for the mass point, and the external force "f ^Col " for the mass point receiving the external force "f ^Col ". The coordinate values of the mass points are sequentially calculated by the equation of motion. For a mass point that has not received an external force "f ^Col ", the calculation unit 150 uses an equation of motion that depends on the mass associated with the mass point and the force "f ^Passive " calculated for the mass point to obtain the coordinates of the mass point. Calculate the values sequentially. That is, the arithmetic unit 150 solves the equation of motion of the equation (1) after setting the external force “f ^Col ” to zero for the mass point that has not received the external force. When calculating the coordinate values of each mass point in more detail, an active muscle contraction force "f ^Active " is added to the equation of motion as shown in the equation (1).

As shown in FIG. 3, the calculation unit 150 includes a first calculation unit 151, a second calculation unit 152, a third calculation unit 153, and a position calculation unit 160 as a functional configuration. The first calculation unit 151 calculates the passive force "f ^Passive " of the equation (1). The second calculation unit 152 calculates the muscle contraction force “f ^Active ” represented by the equation (1). The third calculation unit 153 calculates the external force "f ^Col " represented by the equation (1). The position calculation unit 160 calculates the coordinate values of each mass point of the muscle model 50 by solving the equation of motion of the above equation (1).

(A) First calculation unit 151
Hereinafter, a method of calculating the force "f ^Passive " received from adjacent mass points will be described. The force "f ^Passive " is calculated by the first calculation unit 151. The first calculation unit 151 calculates "f ^Passive " by regarding the mechanical coupling relationship between the target mass point and the adjacent mass point as a spring. That is, each mass point of the muscle model 50 is connected to the adjacent mass point by a mechanical relationship depending on the distance from the adjacent mass point, and the first calculation unit 151 is between the target mass point and the adjacent mass point. The force that the target mass point receives from the adjacent mass point is calculated according to the distance.

One or more mass points are adjacent to each mass point. In order to perform a stable simulation, it is necessary to devise a mechanical connection relationship between mass points (that is, how to connect springs). This is because the muscle model 50 may not be able to maintain its shape only by arranging the springs in the i, j, and k directions (see FIG. 1). Therefore, it is preferable that diagonal springs are arranged in all the cross sections in the i, j, and k directions. As a result, the motion analysis device 100 can stably perform shape simulation on the muscle model 50. For example, the passive force "f ^Passive " is expressed by the following equation (2).

The superscript "e" in equation (2) represents the eastern direction (see FIG. 1). The superscript "w" represents the west direction. The superscript "s" represents the south direction. The superscript "n" represents the north direction. The superscript "f" represents the forward direction (downward on the paper in FIG. 1). The superscript "b" represents the backward direction (the direction on the paper in FIG. 1). For example, the force "f ^ne " in equation (2) represents the force received from the mass point (spring) in the northeast direction. If there are no adjacent mass points, the force received from that direction is zero. For example, it is assumed that the mass point in the cross section of k = 0, which is the starting end of the muscle, receives no force from the posterior direction (that is, the bone side). Actually, the mass point at the beginning keeps a certain position with respect to the bone by receiving a force from the bone, but the method of implementing such a constraint condition will be described later.

As shown in the following equation (3), the force "f _nm " received by the nth mass point from the adjacent mth mass point is the sum of the elastic force "f _nm ^S " and the viscous force "f _nm ^D ". It is represented by. The force "f _nm " corresponds to each term on the right side of the equation (2).

The elastic force "f _nm ^S " in the formula (3) is expressed by the following formula (4).

“K _S ” in the equation (4) is a spring constant. "X _nm " means "x _m -x _n ". "X _m " represents the coordinates of the mth mass point. “X _n ” represents the coordinates of the nth mass point. The coordinates of each mass point are defined in, for example, the mass point information 112 in FIG. "| X _nm |" represents the distance between the nth mass point and the mth mass point. "L _nm " represents the distance (that is, the natural length) between the nth mass point and the mth mass point when no force is applied. The natural length between each mass point is predetermined, for example, in the mass point information 112 of FIG.

The viscous force "f _nm ^D " of the formula (3) is expressed by the following formula (5).

“ _Kd ” in the equation (5) represents a viscous drag coefficient. “V _n ” represents the velocity of the nth mass point. "V _m " represents the velocity of the mth mass point. The velocity of each mass point is calculated, for example, by dividing the travel distance of the mass point from the previous position to the current position by the travel time.

(B) Second calculation unit 152
The calculation method of the muscle contraction force "f ^Active " represented by the above formula (1) will be described. The muscle contraction force "f ^Active " is a force generated by the active activity of the muscle. The muscle contraction force "f ^Active " is calculated by the second calculation unit 152. The muscle contraction force "f ^Active " works only in the k direction, and its magnitude is the relationship between muscle activity, tension and length "F _l (| x _nm |)", and tension and velocity. It is determined by "F _v (| v _nm |)" and the maximum isometric contractile muscle strength "F _max ". More specifically, the muscle contraction force "f ^Active " is expressed by the following equation (6).

(C) Third calculation unit 153
As described above, the third calculation unit 153 calculates the external force "f ^Col " applied to the muscle model 50. The external force "f ^Col " is generated as, for example, a repulsive force or an interference force caused by the muscle model 50 colliding with another object. The object that exerts an external force on the muscle model 50 includes, for example, at least one of the other muscle model and the bone model. The external force "f ^Col " may be calculated by a calculation method described later, or may be given by an operator of the motion analysis device 100 at an arbitrary value.

The external force "f ^Col " is applied to the mass points constituting the surface of the muscle model 50. That is, the external force "f ^Col " is not applied to the mass points constituting the inside of the muscle model 50.

(D) Position calculation unit 160
The position calculation unit 160 calculates the acceleration of each mass point by solving the equation of motion shown in the above equation (1) for each mass point. The position calculation unit 160 calculates the coordinates of each mass point by integrating the acceleration of each mass point twice in time after reflecting the initial position of each mass point and the initial velocity of each mass point (typically zero). do. As a method of time integration, for example, Verlet's method is used.

The position calculation unit 160 rewrites the current position in the mass point information 112 in FIG. 2 with the new coordinate value of each mass point based on the calculation of the new coordinate value of each mass point, and the position before rewriting is the mass point information. Store in the history of 112.

(2.3 Implementation method on GPU)
(A) Outline As shown in the above equations (1) to (6), the new coordinate value of each mass point is calculated from the coordinate value of each current mass point and the coordinate value of each previous mass point. That is, if the current coordinate value of each mass point and the previous coordinate value of each mass point are known, the calculation of the new coordinate value of each mass point is not affected by the calculation result of the new coordinate value of the other mass points. Focusing on this point, the motion analysis device 100 performs the calculation of the equation of motion (see the above equation (1)) for each mass point of the muscle model 50 in parallel for a part or all of them. As a result, the motion analysis device 100 can significantly reduce the processing time required to analyze the motion of the muscle model 50.

(B) Thread management method The motion analysis device 100 realizes, for example, parallel processing of the equations of motion of each mass point on the GPU. The GPU has a plurality of cores (for example, hundreds of cores). The GPU divides the program for realizing the motion analysis of the muscle model 50 into a large number of threads, and assigns threads corresponding to each core to process the motion analysis of the muscle model 50 in parallel.

A method of managing threads by the GPU will be described with reference to FIG. FIG. 4 is a conceptual diagram showing the arrangement of threads on the GPU at the time of coordinate calculation of each mass point.

As shown in FIG. 4, threads are managed in two hierarchical structures, "grid" and "block", within the GPU. A plurality of blocks are arranged in two dimensions in each grid. As shown in FIG. 4 (A), one line is assigned to one block. Preferably, the blocks in the first column are assigned to the muscles of the left half of the body, and the blocks in the second column are assigned to the muscles of the right half of the body.

A plurality of threads are arranged in each block in two dimensions. As shown in FIG. 4B, the motion analyzer 100 allocates one muscle fiber to one thread. Since most human muscle shapes can be represented within the range _{Ni = 7, N j =} 16, N _k = 33, the maximum number of threads in the i direction is set to 7, and the maximum number of threads in the j direction. Is set to 16.

(C) Effective use of shared memory In order to speed up GPU parallel processing, it is necessary to effectively use memory with a high access speed. The global memory in the DRAM (Dynamic Random Access Memory) can be accessed from all blocks, but has the drawback of slow access speed. On the other hand, the shared memory in the SM (Streaming Multiprocessor) can be accessed only from within the same block, but has the advantage of high access speed. Since the information in the same muscle model (for example, position, natural length, etc.) that is often referred to when calculating the equation of motion of the above equation (1) is placed in the shared memory, the processing is speeded up. The motion analysis device 100 is designed so that one muscle is assigned to one block.

However, since the capacity of the shared memory is as small as several tens of KB (for example, three types of 16KB, 32KB, and 48KB), it is difficult to allocate all the information of one source to the shared memory. For example, when the muscle size is _{Ni = 7, N j =} 16, N _k = 33, and each mass point is a float type (4 bytes) 24 variables (18 variables: natural length L, If you have 3 variables: the current coordinate value and 3 variables: the past coordinate value), the total size is about 346.5KB (= 354816B = 7 * 16 * 33 * 24 * 4B), which is the capacity of the shared memory. Will exceed.

To solve this, here's how to save shared memory. The information required to calculate the equation of motion of a mass point is sufficient if the information contained in the adjacent cross section is available. That is, when calculating the equation of motion of the mass point included in the kth cross section, it is sufficient to have the information of the three k-1th, kth, and k + 1th cross sections. As far as the three cross sections are concerned, the data capacity allocated to the shared memory can be saved to about 31.5 kB (= 32256B = 7 * 16 * 3 * 24 * 4B). In order to realize such a saving, as shown in FIG. 4B, the motion analysis device 100 allocates one muscle fiber to one thread, and the coordinate value of each mass point of each cross section is k =. Calculate in order from 0 to N _k . More specifically, the motion analysis device 100 arranges the information of the mass points included in the three cross sections of k = 0, k = 1, and k = 2 in the shared memory at the time of calculating the mass points in the cross section of k = 1. Then, the coordinate value of each mass point in the cross section of k = 1 is calculated. When all the threads have completed the calculation, the motion analysis device 100 rewrites the shared memory with the information of the mass points included in the cross sections of k = 1, k = 2, and k = 3, and in the next cross section of k = 2. Proceed to the mass calculation process.

[Interference avoidance processing of 3 muscle models]
(3.1) Outline The interference calculation for the muscle model 50 will be described below. The motion analysis device 100 according to the present embodiment also applies the above-mentioned motion analysis method based on the spring mass damper model to the interference avoidance process between the muscle model 50 and other parts. As described above, the characteristics of the spring mass damper model are that the equation of motion of each mass point is expressed by the same formula, and the integral calculation of one time step (Δt) of the simulation of each mass point is independent of the other mass points. It is a point that can be done. As a result, the integral calculation of a large number of mass points in one time step is processed in parallel, so that the interference avoidance processing can be speeded up as well.

Hereinafter, the interference avoidance process for the muscle model 50 will be described, but the interference avoidance process described in the present specification can also be applied to a three-dimensional model representing a part other than the muscle.

(3.2) Interference avoidance processing between muscle models An example of interference avoidance processing will be described with reference to FIG. FIG. 5 is a diagram for explaining interference avoidance processing between muscle models.

FIG. 5 shows a muscle model 50A (first part model) representing the first muscle (first part) of the human body and a muscle model 50B (second part) representing the second muscle (second part) of the human body. 2 site model) is shown. FIG. 5 shows an example in which the muscle model 50A and the muscle model 50B are separated from each other for convenience of explanation. In fact, the muscle models 50A and 50B are in contact with each other or close to each other. ing.

In order to avoid the interference of the muscle models 50A and 50B, the buffer object 60 is arranged between the muscle models 50A and 50B. The buffer object 60 may be manually placed by the user or may be automatically placed between sites that may interfere with each other.

The buffer object 60 can be likened to a part of the human body. In this example, the buffer object 60 is likened to the fascia of the human body.

The buffer object 60 is defined by the mass point information 112 described with reference to FIG. 2 in the same manner as the muscle models 50A and 50B. That is, the buffer object 60 is represented by a plurality of mass points, and each mass point is associated with a mass, a natural length of a spring, and the like. The natural length of the spring is the distance between each mass point in the initial posture.

For example, "Nj" representing the number of quality points in the _j direction of the buffer object 60 is set to 2. That is, the buffer object 60 has a plurality of mass points constituting the first surface (that is, mass points corresponding to j = 0) and a plurality of mass points constituting the second surface (that is, mass points corresponding to j = 1). ) And. Each mass point constituting the buffer object 60 is spring-loaded with adjacent mass points in the buffer object 60.

Each mass point constituting the buffer object 60 is connected not only to the mass point in the buffer object 60 but also to each mass point of the muscle models 50A and 50B in contact with the buffer object 60 by a virtual spring. That is, some or all the mass points constituting the buffer object 60 include a plurality of mass points constituting the muscle model 50A (first site model) and a plurality of mass points constituting the muscle model 50B (second site model). It is associated with one of the mass points of.

In the example of FIG. 5, each mass point representing the first surface of the buffer object 60 (that is, the mass point corresponding to j = 0) is connected to the mass point of the muscle model 50A. The mass points constituting the first surface of the buffer object 60 can be regarded as the fascia of the muscle model 50A. On the other hand, each mass point representing the second surface of the buffer object 60 (that is, the mass point corresponding to j = 1) is connected to the mass point of the muscle model 50B. The mass point representing the second surface of the buffer object 60 can be regarded as the fascia of the muscle model 50B.

The mass points in the muscle models 50A and 50B coupled to the buffer object 60 are automatically registered by, for example, the motion analysis device 100. More specifically, the motion analysis device 100 calculates the distance from all the mass points in the muscle model 50A for each mass point (that is, the mass point corresponding to j = 0) constituting the first surface in the buffer object 60. do. Then, the motion analysis device 100 combines each mass point constituting the first surface in the buffer object 60 with one or more mass points having the closest distance among the mass points constituting the muscle model 50A. Similarly, the motion analysis device 100 calculates the distance from all the mass points in the muscle model 50B for each mass point (that is, the mass point corresponding to j = 1) constituting the second surface in the buffer object 60. Then, the motion analysis device 100 combines each mass point constituting the second surface in the buffer object 60 with one or more mass points having the closest distance among the mass points constituting the muscle model 50B. The number of mass points of the muscle models 50A and 50B coupled to each mass point of the buffer object 60 is, for example, about two points.

The connection result between each mass point of the muscle models 50A and 50B and each mass point of the buffer object 60 is stored as the connection information 113 shown in FIG. FIG. 6 is a diagram showing an example of the data structure of the coupling information 113.

As shown in FIG. 6, the mass points of the muscle model 50A are associated with each mass point of the buffer object 60 in the coupling information 113. As an example, the mass point of the mass point number “n1” of the buffer object 60 is associated with the mass point of the mass point number “n10” of the muscle model 50A.

Although the connection information 113 between the muscle model 50A and the buffer object 60 is shown in FIG. 6, in reality, other muscle models (for example, muscle model 50B) are also combined with the other muscle models. There is coupling information 113 with the buffer object 60.

The motion analysis device 100 performs motion analysis on each mass point constituting the muscle models 50A and 50B and the buffer object 60 by the calculation method described in the above-mentioned "motion analysis of the muscle model". That is, the motion analysis device 100 calculates the force received from the other mass points of the coupling destination for each mass point constituting the muscle models 50A and 50B and the buffer object 60, and determines the force and the mass associated with each mass point. The coordinate values of each mass point are sequentially calculated by the dependent equation of motion (that is, the above equation (1)).

More specifically, in the motion analysis device 100, among the mass points constituting the muscle models 50A and 50B, for each mass point bonded to the buffer object 60, each mass point of the connection destination in the muscle models 50A and 50B. The position of each mass point in the muscle models 50A and 50B is calculated according to both the force received from and the force received from each mass point of the coupling destination in the buffer object 60. On the other hand, the motion analysis device 100 receives the force received from each of the mass points of the connection destinations in the muscle models 50A and 50B for each mass point not bound to the buffer object 60 among the mass points constituting the muscle models 50A and 50B. The position of each mass point in the muscle models 50A and 50B is calculated according to the above.

Further, the motion analysis device 100 receives the force received from each mass point of the coupling destination in the buffer object 60 and the force received from each mass point of the coupling destination in the muscle models 50A and 50B for each mass point constituting the buffer object 60. The position of each mass point in the buffer object 60 is calculated according to both of the above.

As a result, the motion analysis device 100 can not only calculate the force exchange between each mass point in the muscle models 50A and 50B, but also exchange the force between the muscle model 50A and the muscle model 50B via the buffer object 60. It can be calculated and interference between the muscle model 50A and the muscle model 50B can be avoided.

(3.3) Interference avoidance processing between the muscle model and the skin model Next, another example of the interference avoidance processing will be described with reference to FIG. 7. FIG. 7 is a diagram for explaining the interference avoidance process between the muscle model and the skin model.

FIG. 7 shows a muscle model 50A (first part model) representing the first muscle (first part) of the human body and a muscle model 50B (second part) representing the second muscle (second part) of the human body. 2 site model) and the buffer object 60 are shown. FIG. 7 shows an example in which the muscle models 50A, 50B and the buffer object 60 are separated from each other for convenience of explanation. In fact, the muscle models 50A, 50B and the buffer object 60 are separated from each other. They are in contact with each other or close to each other.

Basically, the buffer object 60 shown in FIG. 7 is the same as the buffer object 60 shown in FIG. In the buffer object 60 shown in FIG. 5, the mass points of the muscle model 50A are connected to a plurality of mass points constituting the first surface of the buffer object 60 (that is, mass points corresponding to j = 0), and the muscle model The mass points of 50B were combined with a plurality of mass points (that is, mass points corresponding to j = 1) constituting the second surface of the buffer object 60. On the other hand, in the buffer object 60 shown in FIG. 7, the mass points of the muscle model 50A and the mass points of the muscle model 50B are both mass points corresponding to the first surface of the buffer object 60 (that is, j = 0). ) Is combined with the mass points that make up. The buffer object 60 arranged in this way can be regarded as the skin (third part) of the human body.

The mass points in the muscle models 50A and 50B coupled to the buffer object 60 are automatically registered by, for example, the motion analysis device 100. More specifically, the motion analysis device 100 has all the mass points in the muscle model 50A and the muscle model 50B for each mass point (that is, the mass point corresponding to j = 0) constituting the first surface in the buffer object 60. Calculate the distance to. Then, the motion analysis device 100 combines each mass point constituting the first surface in the buffer object 60 with one or more mass points having the closest distance among the mass points constituting the muscle models 50A and 50B. The number of mass points of the muscle models 50A and 50B coupled to each mass point of the buffer object 60 is, for example, about two points.

The motion analysis device 100 includes not only each mass point constituting the first surface in the buffer object 60, but also each mass point constituting the second surface in the buffer object 60 (that is, a mass point corresponding to j = 1). ) May also be combined with any mass point in the muscle models 50A and 50B. In this case, all the mass points in the buffer object 60 are combined with any of the mass points in the muscle models 50A and 50B. More specifically, the motion analysis device 100 calculates the distances of all the mass points in the buffer object 60 from the mass points in the muscle models 50A and 50B. Then, the motion analysis device 100 combines all the mass points in the buffer object 60 with one or more mass points having the closest distance among the mass points constituting the muscle models 50A and 50B.

The motion analysis device 100 performs motion analysis on each mass point constituting the muscle models 50A and 50B and the buffer object 60 by the calculation method described in the above-mentioned "motion analysis of the muscle model". That is, the motion analysis device 100 calculates the force received from the other mass points of the coupling destination for each mass point constituting the muscle models 50A and 50B and the buffer object 60, and determines the force and the mass associated with each mass point. The coordinate values of each mass point are sequentially calculated by the dependent equation of motion (that is, the above equation (1)).

More specifically, the motion analysis device 100 receives each mass point bonded to the buffer object 60 from each mass point of the coupling destination in the buffer object 60 among the mass points constituting the muscle models 50A and 50B. The position of each mass point in the muscle models 50A and 50B is calculated according to both the force and the force received from each mass point of the connection destination in the muscle models 50A and 50B. On the other hand, the motion analysis device 100 receives the force received from each of the mass points of the connection destinations in the muscle models 50A and 50B for each mass point not bound to the buffer object 60 among the mass points constituting the muscle models 50A and 50B. The position of each mass point in the muscle models 50A and 50B is calculated according to the above.

Further, the motion analysis device 100 receives the force received from each mass point of the coupling destination in the buffer object 60 and the force received from each mass point of the coupling destination in the muscle models 50A and 50B for each mass point constituting the buffer object 60. The position of each mass point in the buffer object 60 is calculated according to both of the above.

As a result, the motion analysis device 100 can not only calculate the force exchange between each mass point in the muscle models 50A and 50B, but also calculate the force exchange between the muscle models 50A and 50B and the buffer object 60. It is possible to avoid interference between the muscle models 50A and 50B and the buffer object 60.

[4 Integration of muscle model and bone model]
A model whose shape changes, such as a muscle model, can be combined with a rigid body model whose shape does not change. A musculoskeletal model in which a rigid body model and a musculoskeletal model are combined will be described with reference to FIG. In the following, the musculoskeletal model will be described by taking the bone model, which is an example of the rigid body model, as an example, but the rigid body model is not limited to the bone model. FIG. 8 is a diagram schematically showing a musculoskeletal model.

In order to handle the musculoskeletal model, the motion analysis device 100 needs to handle not only the bone model but also the equation of motion of the muscle model (see the above equation (1)) at the same time. In order to handle two types of equations of motion at the same time, the motion analysis device 100 is subject to the constraint that the mass points (hereinafter, also referred to as “end points”) contained in the ends of the muscles are immobile with respect to the bone. Calculate both equations of motion. The end point refers to a mass point included in the start end (that is, the cross section of k = 0) and the end point (that is, the cross section of k = N _k ) of the streak.

The exchange of force between the bone model and the muscle model takes place at the endpoints. The forces exerted by the muscle model on the bone model are the force vectors "f ₁ " and "f ₂ " shown in FIG. 8 (A).

The exchange of force between the bone model and the muscle model also takes place at the point of interference between the bone model and the muscle model. At the point of interference, the repulsive forces "f ^Col " in opposite directions of equal magnitude are applied to the bone model and the muscle model, respectively. Summarizing the above, the equation of motion of the bone model incorporating the force received from the muscle model is expressed by the following equations (7) and (8), taking the humerus shown in FIG. 8 as an example.

Here, "m ₁ " represents the mass of the upper arm. "G" represents gravitational acceleration. "F _s " represents the inter-articular force of the shoulder joint. "F _e " represents the inter-articular force of the elbow joint. “Fi _{, j} ^End ” represents the force exerted on the humerus by the i-th muscle fiber in the j-th muscle. “A _g1 ” represents the acceleration of the position of the center of gravity of the upper arm. "L _s " represents a vector from the position of the center of gravity of the upper arm to the shoulder joint. "L _e " represents a vector from the position of the center of gravity of the upper arm to the elbow joint. "L _{i, j} ^End " represents a vector from the position of the center of gravity of the upper arm to the i-th end point of the j-th muscle. “L ^Col ” represents a vector from the position of the center of gravity of the upper arm to the point of interference. I ₁ represents the inertia tensor of the upper arm. “Ω ₁ ” with dots represents the angular acceleration vector of the upper arm. “Ω ₁ ” without dots represents the angular velocity vector of the upper arm.

Although the end points are represented by the equation of motion of the above equation (1) of the muscle model, they are mass points that should substantially follow the simulation results of the bone model because there is a constraint condition that they are immobile with respect to the bone. That is, for the end points, the motion analysis device 100 updates the position after one time step with reference to the simulation result of the bone model. On the other hand, for mass points other than the end points, the motion analysis device 100 updates the position after one time step using the simulation result of the muscle model as it is. In this way, the motion analysis device 100 can handle the bone model and the muscle model in an integrated manner by treating the end points and the other mass points separately.

[5 Hardware configuration of motion analysis device 100]
An example of the hardware configuration of the motion analysis device 100 will be described with reference to FIG. 9. FIG. 9 is a block diagram showing a main hardware configuration of the motion analysis device 100. As shown in FIG. 9, the motion analysis device 100 includes a ROM (Read Only Memory) 101, a CPU 102, a RAM (Random Access Memory) 103, a GPU 104, a network interface 105, a display interface 106, and a storage device. Including 110.

The ROM 101 stores an operating system, a control program executed by the motion analysis device 100, and the like. The CPU 102 controls the operation of the motion analysis device 100 by executing various programs such as an operating system and a motion analysis program 114. The RAM 103 functions as a working memory and temporarily stores various data necessary for executing a program.

The GPU 104 is an arithmetic unit for speeding up the motion analysis processing of the muscle models 50A and 50B and the buffer object 60. The GPU 104 describes the motion analysis of each mass point constituting the muscle model 50A, each mass point constituting the muscle model 50B, and each mass point constituting the buffer object 60 in the above-mentioned "2.3 Implementation method to GPU". Parallel processing is performed by the above method.

The display interface 106 is a terminal for connecting the motion analysis device 100 and the display 107. The display 107 includes, for example, a liquid crystal display, an organic EL (Electro Luminescence) display, or other display device. When the motion analysis program 114 is executed, the display 107 displays the motion analysis results of the muscle model and the bone model.

The storage device 110 is, for example, a storage medium such as a hard disk or an external storage device. As an example, the storage device 110 includes mass point information 112 (see FIG. 2) for expressing each part of the human body on a computer, and connection information 113 (see FIG. 6) that defines a connection relationship between each part. , A motion analysis program 114 for realizing various processes according to the present embodiment is stored.

The motion analysis program 114 may be provided as a part of any program, not as a stand-alone program. In this case, the process according to the present embodiment is realized in cooperation with an arbitrary program. Even a program that does not include such a part of the modules does not deviate from the purpose of the motion analysis device 100 according to the present embodiment. Further, some or all of the functions provided by the motion analysis program 114 according to the present embodiment may be realized by dedicated hardware. Further, the motion analysis device 100 and the server may cooperate with each other to realize the processing according to the present embodiment. Further, the motion analysis device 100 may be configured in the form of a so-called cloud service in which at least one server realizes the processing according to the present embodiment.

[7 Summary]
(7.1 Advantage)
(A) Reduction of calculation time By using the above-mentioned motion analysis method, the motion analysis device 100 can keep the distance between the muscle model 50A and the muscle model 50B constant through the buffer object 60. As a result, when the muscle model 50A and the muscle model 50B are in contact with each other in the initial posture, the contacted state is maintained.

In the conventional method, when the interference determination is performed, the three-dimensional object is often treated as a set of triangles (that is, a polygon object). In this case, it is necessary to select one triangle from each object and determine the interference between the two triangles. For elaborate objects with a large number of triangles, this can be very time consuming. In addition, when an interference determination is made, interference is avoided by applying a force (that is, a penalty force) in the opposite direction to both objects according to the interference depth.

The proposed method can eliminate many of the disadvantages of the conventional method. The first merit of the proposed method is that it is not necessary to determine the interference between three-dimensional objects, and the calculation time can be significantly shortened. Since the force vector generated between the mass points of the buffer object 60 and the mass points of the muscle models 50A and 50B is determined only by the positional relationship between the mass points, it is not necessary to use the concept of a polygon object.

The second merit of the proposed method is that it is not necessary to use a penalty force that tends to cause a sudden change in force. In the method of giving a penalty force only when interference is determined, a large force may be suddenly generated at the time of determination, and the simulation is not stable. On the other hand, in the motion analysis method according to the present embodiment, since a non-zero force always acts between the mass points of the buffer object 60 and the mass points of the muscle models 50A and 50B, sudden changes in force are unlikely to occur, and the simulation Is stable.

(B) Improvement of simulation performance Conventionally, the simulation was often unstable near the humeral large nodule ridge where many muscles such as the pectoralis major muscle, deltoid muscle, and biceps brachii muscle overlap. In particular, when the abduction angle of the shoulder joint was increased to about 90 degrees, the simulation sometimes diverged. On the other hand, when the proposed method was used, it did not diverge, and it became possible to continue the simulation stably even if the abduction angle of the shoulder joint exceeded 90 degrees. FIG. 10 is a diagram showing an example of the simulation result of the proposed method. As shown in FIG. 10, in the proposed method, the simulation continues stably even when the abduction angle of the shoulder joint exceeds 90 degrees.

(7.2 Usage example)
The proposed method is useful when reproducing a situation in which a plurality of deformable objects are always in contact with each other, such as a human body CG (Computer Graphics) model. In particular, in the case of the human musculoskeletal model, each muscle needs to be modeled individually, so that the number of objects and the number of interferences between objects become enormous. Therefore, in the interference avoidance method by the penalty force method, the simulation becomes unstable everywhere, and the proposed method that can solve this is extremely epoch-making.

In addition, the proposed method can be applied not only to muscles but also to other parts such as internal organs and blood vessels. Like muscle, internal organs and blood vessels are loosely connected to other tissues via connective tissue. Therefore, continuous control by arranging the buffer object 60 is more effective than discrete control by interference determination.

It should be considered that the embodiments disclosed this time are exemplary in all respects and not restrictive. The scope of the present invention is shown by the scope of claims rather than the above description, and is intended to include all modifications within the meaning and scope equivalent to the scope of claims.

50, 50A, 50B muscle model, 60 buffer object, 100 motion analyzer, 101 ROM, 102 CPU, 103 RAM, 104 GPU, 105 network interface, 106 display interface, 107 display, 110 storage device, 112 point information, 113 coupling Information, 114 motion analysis program, 150 calculation unit, 151 first calculation unit, 152 second calculation unit, 153 third calculation unit, 160 position calculation unit.

Claims (5)

It is a motion analysis device for analyzing the motion of animals.
A storage unit for storing a plurality of three-dimensional models representing each part of the animal's body is provided.
Each of the plurality of three-dimensional models is represented by a plurality of mass points.
Each of the plurality of mass points is associated with a coordinate value in three dimensions and a mass, and is said to be the other due to a mechanical relationship depending on the distance from other associated mass points in the three-dimensional model. Combined with the mass point of
The plurality of three-dimensional models include a first to third site model representing the first to third sites in the animal's body.
A part or all of the plurality of mass points constituting each of the one or more surfaces of the third site model constitutes the plurality of mass points constituting the first site model and the second site model. It is associated with one of the masses with multiple masses and
A calculation unit for calculating the coordinate values for each of the plurality of mass points of the first to third part models is provided.
The arithmetic unit
For each of the plurality of mass points of the first to third site models, the force that the mass point receives from the other mass points is calculated according to the distance between the mass point and the other mass points associated with the mass point.
For each of the plurality of mass points of the first to third part models, the coordinate values of the mass points are sequentially set by an equation of motion depending on the mass associated with the mass point and the force calculated for the mass point. A motion analyzer that calculates. The third site model is composed of a plurality of mass points constituting the first surface of the third site and a plurality of mass points constituting the second surface of the third site. Item 1. The motion analysis apparatus according to Item 1. Some or all of the plurality of mass points constituting the first surface are associated with at least one mass point constituting the first site model.
The motion analysis apparatus according to claim 2, wherein some or all of the plurality of mass points constituting the second surface are associated with at least one mass point constituting the second site model. The third site model represents the skin of the animal.
A part or all of the plurality of mass points constituting the first surface is any one of the plurality of mass points constituting the first site model and the plurality of mass points constituting the second site model. The motion analysis apparatus according to claim 2, which is associated with the mass point of the above. The first site model represents the first muscle in the animal's body.
The motion analysis device according to any one of claims 1 to 4, wherein the second site model represents a second muscle in the body of the animal.

Images