KR20160065505A - Method for Generating and controlling Locomotion, Locomotion Controller - Google Patents

Abstract

Disclosed are a method for generating and controlling a walking motion on a virtual character or a robot using an inverse kinematics solver, and a walking controller. The method for generating a walking motion comprises the following steps: receiving capture data on the virtual character or the robot; generating angle information of a leg joint on a target position of an end effector; and generating rotation information of a pelvis on the target position by the angle information when the end effector does not reach the target position.

Description

[0001] The present invention relates to a method for generating and controlling a walking motion,

The present invention relates to a walking motion generating and controlling method and a walking controller, and more particularly, to a walking maneuver generation and control method for a virtual character or a robot using an inverse kinematic solver and a walking controller.

In computer animation, it takes a lot of money to create the natural motion of the virtual character or to create the natural motion of the robot. For example, an operation is photographed using a manual animation of an experienced animator or a motion capture device, and a gait operation is generated through correction. Or an inverse kinematics solver to generate a gait motion.

The inverse kinematic solver is a tool for calculating the angles of the joints so that the end-effector, such as a hand or foot, is precisely positioned at the desired location. A numerical optimization-based inverse kinematic solver is used to find a solution to the total degree of freedom of the character and provides a variety of functions, but is relatively slow compared to analytical inverse kinematic solvers. In the case of an inverse kinematic solver that uses a database, it can be used to learn the natural posture from the database and create a new posture over a large area.

As described above, Korean Patent Publication No. 2010-0133232 discloses a technique for generating a gait using an inverse kinematic solver and a related prior art document.

Conventionally, there is a disadvantage in that the walking motion is not adaptively generated to the terrain, and the walking motion is unnatural such as kneeling in a staircase or a sloping area, which is not flat. And a method of adjusting the leg length to solve an unnatural walking motion, etc., are not suitable for the robot simulation area.

It can be used both in the bipedal robot simulation area where the walking motion needs to be corrected in real time according to the external environment or in the game field using the virtual character to provide the realistic graphic effect, and generates a natural, There is a need for research on how to do this.

The present invention is to provide a gait controller for generating and controlling a gait motion for a virtual character or robot using an inverse kinematic solver.

According to an aspect of the present invention, there is provided a method for generating a walking motion for a virtual character or robot using an inverse kinematics solver, the method comprising: Receiving an input; Generating angle information of a leg joint with respect to a target point of the end effector; And generating rotation information of the pelvis with respect to the target point when the end effector fails to reach the target point, based on the angle information.

According to another aspect of the present invention, there is provided a method for generating a gait operation for a virtual character or a robot using an inverse kinematic solver, the method comprising: receiving capture data for the virtual character or the robot; Generating angular information of an arm joint with respect to a target point of the end effector; And generating rotation information of the upper body with respect to the target point when the end effector can not reach the target point by the angle information.

According to another aspect of the present invention, there is provided a method for controlling a virtual character or a robot using an inverse kinematics solver, the method comprising: Receiving data and terrain information; Generating end effector of the captured data and locus information of the pelvis according to the topographic information; Generating angle information of a leg joint with respect to a target point of the end effector based on the locus information; And generating rotation information of the pelvis with respect to the target point based on the locus information.

According to another aspect of the present invention, there is provided a walking controller for a virtual character or robot using an inverse kinematics solver, the controller comprising: An information receiving unit for receiving the terrain information; A trajectory information generating unit for generating trajectory information of an end effector and a pelvis of the captured data according to the terrain information; And an inverse kinematic analysis unit for generating angle information of a leg joint with respect to a target point of the end effector and rotation information of the pelvis based on the locus information.

According to the present invention, it is possible to generate a natural walking motion by using the inverse kinematics to not only generate the angle information of the leg joint necessary for the walking operation but also consider the rotation of the pelvis.

According to the present invention, the trajectory information of the end effector and the pelvis is generated according to the terrain, and the leg joint angle and the pelvis rotation angle are calculated based on the trajectory information, thereby generating and controlling the walking operation adaptable to the terrain .

1 is a view for explaining a walking controller according to an embodiment of the present invention.
2 is a view for explaining a method of generating end effector trajectory information according to an embodiment of the present invention.
3 and 4 are views for explaining a method of generating pelvis locus information according to another embodiment of the present invention.
5 is a diagram for explaining a walking motion generating method according to an embodiment of the present invention.
6 and 7 are views for explaining a walking operation according to the present invention.
8 is a view for explaining a walking operation control method according to an embodiment of the present invention.

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that the invention is not intended to be limited to the particular embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention. Like reference numerals are used for like elements in describing each drawing.

The present invention proposes a gait generation and control method for a virtual character or robot using an inverse kinematic solver, and a gait controller.

The present invention can generate a natural walking motion by using the inverse kinematics to not only generate the angle information of the leg joint necessary for the walking operation but also consider the rotation of the pelvis together. That is, according to the present invention, when the end effector does not reach the target point according to the angle information of the leg joint, the end effector can reach the target point by rotating the pelvis.

According to the present invention, the end effector and the pelvis trajectory information is generated according to the terrain, and the leg joint angle and the pelvis rotation angle are calculated according to the above-described method based on the trajectory information, can do.

A method and a control method for generating a walking motion according to the present invention can be performed in an electronic device including a processor, for example, a PC, a notebook, a server, a robot and a mobile device. Further, the description of the hardware aspects referred to in this specification can be easily interpreted from a process viewpoint, and the description of the process viewpoint can be easily interpreted from a hardware point of view.

Hereinafter, embodiments according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a view for explaining a walking controller according to an embodiment of the present invention.

The walking controller 100 according to the present invention controls a walking operation for a virtual character or a robot using an inverse kinematics solver. The walking controller 100 may be included in a walking motion simulator of the robot or may provide information to a walking motion simulator of the robot. Alternatively, the walking controller 100 can directly generate or control the virtual character to walk.

The walking controller 100 according to the present invention includes an information receiving unit 110, a locus information generating unit 120, and an inverse kinematics analyzing unit 130, as shown in FIG.

The information receiving unit 110 receives the capture data and the terrain information for the virtual character or the robot. The capture data may be provided, for example, via motion capture for a walking motion for virtual character generation.

The trajectory information generating unit 120 generates the end effector of the captured data and the trajectory information of the pelvis using the captured data of the walking operation as the reference motion. The locus information generating unit 120 can generate locus information according to a walking condition input from a user, for example, a walking speed and a walking direction. The end effector can also be a foot or an ankle.

The locus information generating unit 120 may include a pendulum trajectory generator 121, a stride computing unit 123, and a terrain adaptation unit 125. [

The inverse pendulum trajectory generator 121 generates an inverse pendulum trajectory by using an invert pendulum on a cart (IPC) model to prevent the captured data from falling due to user input. Since the center of gravity is changed when the virtual character changes direction or fast step according to the user's input, there is a difference from the actual capture data. Therefore, it is necessary to modify the captured data to match the generated inverse pendulum trajectory. The reverse pendulum trajectory can be generated by changing the position of the cart so as to receive the desired speed and rotation speed from the initial pendulum state, balance the poles so that they do not fall, and move at a desired speed .

The stride calculator 123 can obtain the position of the footprint of the captured data from the inverse pendulum trajectory and calculate the stride length using an operation planner, thereby generating an ent effector and a locus of the pelvis in an environment such as a flat ground .

The terrain adaptation unit 125 corrects the entropy effect of the captured data and the trajectory of the pelvis using the terrain data. The terrain adaptation unit 125 may generate the trajectory information of the end effect using the convex hull for the terrain. The end effect and the method of generating the locus information of the pelvis are described in detail in Figs.

The inverse kinematic analysis unit 130 generates angular information of a leg joint with respect to a target point of the end effector based on the locus information and generates rotation information of the pelvis to a target point of the pelvis. The inverse kinematic analysis unit 130 may include a pelvis locus correcting unit 131 and a walking information generating unit 133 according to the embodiment.

The pelvis locus corrector 131 may perform Gaussian filtering on the locus information of the pelvis generated by the locus information generation unit 120 to correct the pelvis locus. The walking information generating unit 133 can generate the angle information and the rotation information according to the walking motion generating method according to the present invention, which is described in detail in FIG.

The pelvis locus correcting unit 131 and the walking information generating unit 133 exchange information with each other and update the information. The pelvis locus corrector 131 updates the locus information of the pelvis using the displacement of the pelvis according to the rotation information, and the gait information generator 133 updates the angle information and the rotation information based on the updated locus information .

The walking controller 100 according to the present invention can generate and control the walking operation using the information generated by the walking information generator 133. [ Depending on the embodiment, each of the gait generation and control may be performed in a separate device.

According to the present invention, it is possible to generate and control a natural walking motion adaptable to the terrain, using an example operation such as the capture data for the walking motion.

2 is a view for explaining a method of generating end effector trajectory information according to an embodiment of the present invention. In FIG. 2, the generation of foot trace information according to the step topography is described as an embodiment.

The terrain adaptation unit 125 corrects the initial trajectory curve a of the foot corresponding to the footstep (next footstep) generated by the trajectory information generation unit 120 according to the convex hull b, And generates curve d. That is, the terrain adaptation unit 125 can generate an optimized trajectory curve d by sampling the height of the terrain according to the initial trajectory information for the example operation, and simplifying it to the convex hull b. The terrain adaptation unit 125 may generate the trajectory information according to the terrain by mixing the initial trajectory curve a and the optimized trajectory curve d.

The piece-wise linear curve obtained through the convex hull (b) is not smooth enough to be used immediately, so optimization proceeds with constraint point (c). The constraint point (c) is the upper support vertex of the convex hull (b), connecting the lines located at the top of the convex hull from the previous plant position to the next plant position.

As a result, the terrain adaptation unit 125 can generate an optimized trajectory curve d using Equation (1). Equation (1) can be interpreted by a finite difference method and a least squares method, and a curve in which S is minimized can be an optimized trajectory curve (d). At this time, the above-described constraint point (c) becomes a constraint condition.

Here, t represents the frame constituting the walking operation, and x is the height value of the curve.

3 and 4 are views for explaining a method of generating pelvis locus information according to another embodiment of the present invention. In Fig. 3, locus information generation of the pelvis along the step topography is described as an embodiment.

The terrain adaptation unit 125 mixes the segmented linear curve a connecting the footprint and the initial trajectory curve b of the pelvis generated by the trajectory information generation unit 120 according to the walking motion of the captured data, And generates the locus information (c) of the pelvis.

Since the pelvis moves unnaturally when the walking operation is performed by the pelvis locus information (c), or the knee joint can move unnaturally between the pelvis locus information (c) and the foot locus information, the pelvis locus corrector 131 May perform Gaussian filtering to smoothly process, i.e. smooth, the locus information c of the generated pelvis.

As shown in FIG. 4, the knee joint is straightened before the smoothing (a), but after the smoothing (b), the knee joint is naturally curved and a more natural walking motion can be generated.

5 is a diagram for explaining a walking motion generating method according to an embodiment of the present invention.

The walk information generating unit 133 receives the capture data for the virtual character or the robot, that is, receives the input (S501). Then, the angle information of the leg joint with respect to the target point T1 of the first end effector of the captured data is generated (S503). Here, the leg joint includes a hip joint, a knee joint, and an ankle joint, and the first end effector may be a foot or an ankle.

When the first end effector can not reach the target point T1 due to the angle information, that is, the distance between the position of the first end effector and the target point T1 is equal to or larger than the threshold value , The rotation information of the pelvis with respect to the target point T1 is generated (S505). Here, the rotation information is rotation angle information of the pelvis centered around the root joint, which is a joint connecting the vertebrae and the pelvis.

In the case of conventional inverse kinematics, when angle information of the leg joint is generated and the target point is not reached, an unnatural walking motion in which the knee bounces is generated or the leg length is adjusted to reach the target point. The present invention can provide a more natural gait operation by allowing the first end effector to reach the target point by rotating the pelvis similarly to a human gait operation. Further, the present invention can provide a more natural gait operation through the pelvis movement or the upper body movement, which will be described later.

값을 최소화하는 r을 회전 정보로 이용할 수 있다. 컨쥬게이트 그레디언트 솔버(conjugate gradient solver)를 통해 [수학식 2]를 반복연산하여 회전 벡터 r을 산출할 수 있다.More specifically, the walking information generating unit 133 generates the rotation information of the pelvis using the distance between the first end effector and the hip joint based on the distance between the hip joint and the target point T1 of the first end effector and the angle information In Equation 2,

R which minimizes the value can be used as rotation information. The rotation vector r can be calculated by repeating [Equation 2] through the conjugate gradient solver.

은 고관절과 타겟 지점(T1) 사이의 거리를 나타내며,

은 제1엔드 이펙터와 고관절 사이의 거리를 나타낸다. a, b는 스칼라 가중치이며,

는 제1엔드 이펙터에 대한 가중치로서, 0에서 1사이 값이다.

는 k>0 일 때

이며 그렇지 않으면 0인 연산자이다.here,

Represents the distance between the hip joint and the target point T1,

Represents the distance between the first end effector and the hip joint. a and b are scalar weights,

Is a weight for the first end effector, and is a value between 0 and 1.

When k> 0

Otherwise 0.

와

의 차이 값은 제1엔드 이펙터와 타겟 지점(T1) 사이의 거리를 나타낸다.

가

보다 크거나 같은 경우는 제1엔드 이펙터가 타겟 지점(T1)을 초과하거나 타겟 지점(T1)에 도달한 경우로서,

의 값은 0이기 때문에, 골반의 회전 정보는 생성되지 않는다.

가

보다 작은 경우에는 제1엔드 이펙터가 타겟 지점(T1)에 도달하지 못한 경우로서,

의 값이 0이 아니기 때문에,

값을 최소화하는 r을 회전 정보가 생성될 수 있다.In other words,

Wow

Represents the distance between the first end effector and the target point T1.

end

Is greater than or equal to the target end point T1, the first end effector exceeds the target point T1 or reaches the target point T1,

The rotation information of the pelvis is not generated.

end

When the first end effector does not reach the target point T1,

Is not 0,

Rotation information that minimizes the value r can be generated.

In the meantime, when the first end effector does not reach the target point T1 due to the rotation information of the pelvis, the walking motion generating method according to the present invention may include generating the movement information of the pelvis with respect to the target point T1 (S507 ). The walking information generating unit 123 can generate the movement information using the distance between the hinge and the target point T1 and the distance between the first end effector and the hip joint based on the angle information and the rotation information. Here, the movement information may be the displacement information of the pelvis relative to the root joint position, which is the reference point, or the position information of the pelvis.

값을 최소화하는 t를 이동 정보로 이용할 수 있다. 컨쥬게이트 그레디언트 솔버를 통해 [수학식 3]을 반복 연산하여 3차원 공간에서의 변위 벡터 t를 산출할 수 있다.More specifically, the walking information generating unit 133 can generate the pelvis movement information using Equation (3). In Equation (3)

T, which minimizes the value, can be used as movement information. The displacement vector t in the three-dimensional space can be calculated by repeating [Equation 3] through the conjugate gradient solver.

는 고관절과 타겟 지점(T1) 사이의 거리를 나타내며,

는 다리 각도 정보 및 골반 회전 정보에 의한 제1엔드 이펙터와 고관절 사이의 거리를 나타낸다. c 및 d는 스칼라 가중치이다.here,

Represents the distance between the hip joint and the target point T1,

Represents the distance between the first end effector and the hip joint by the leg angle information and the pelvis rotation information. c and d are scalar weights.

의 값이 0이 아닐 때,

를 최소화하는 t 값이 산출될 수 있다.As shown in Equation (2)

Is not 0,

Lt; / RTI > can be computed.

Meanwhile, the walking motion generating method according to the present invention can be applied to the arm joints. That is, the walking information generating unit 133 may generate the angle information of the arm joint with respect to the target point T2 of the second end effector. Here, the arm joint includes the shoulder joint, the elbow joint and the wrist joint, and the second end effector may be a hand or a wrist.

When the second end effector does not reach the target point T2 due to the angle information of the arm joint, that is, the distance between the position of the second end effector and the target point T2 is equal to or greater than the threshold value, The unit 133 can generate rotation information of the upper body with respect to the target point T2.

At this time, the first end effector corresponds to the second end effector, the hip joint corresponds to the shoulder joint, and the target point T1 corresponds to the target point T2. The rotation information of the upper body may be rotational angle information about the spinal joint corresponding to the position of the body part, and may be generated in the same manner as the method of generating the pelvis rotation information.

On the other hand, the above-described step can be repeatedly performed until the end effector reaches the target point, and the result of the previous step is used in the next step. For example, when the rotational angle of the pelvis is generated in step S503, the angle information of the leg joint generated in step S501 is reflected.

As a result, according to the present invention, it is possible to generate a more natural gait through rotation and movement of the pelvis and rotation of the upper body.

6 and 7 are views for explaining a walking operation according to the present invention.

In Fig. 6, the end effector is an ankle, and the target point is a blue hole 600. Fig. 6 (a) shows that the right ankle does not touch the blue hole 600 as a walking operation by the analytical inverse kinematic solver. On the other hand, in FIG. 6B showing the walking operation according to the present invention, it can be seen that the right leg reaches the blue hole 600 as the pelvis rotates and moves.

FIG. 7 is a diagram showing a distance between a left foot and a target point when the walking motion generating method according to the present invention is repeatedly performed. FIG. In Fig. 7, the horizontal axis represents the image frame, and the vertical axis represents the left-hand error, i.e., the distance between the left foot and the target point.

As shown in FIG. 7, it can be seen that the maximum error is 0.15 m or more at the time of one execution, but the error decreases as the repetition is repeated, and almost no error occurs at the sixth execution.

8 is a view for explaining a walking operation control method according to an embodiment of the present invention. In Fig. 8, a walking motion control method performed in the walking controller described in Fig. 1 will be described as an embodiment.

The walking controller 100 receives the capture data and the terrain information for the virtual character or robot (S801). The end effector and the pelvic trajectory information of the captured data according to the terrain are generated using the capture data for the walking operation (S803). At this time, the walking controller 100 can apply the convex hull to the terrain to generate the trajectory information of the end effector. The end effector may be a foot or an ankle. Gaussian filtering can be performed on the locus information of the pelvis.

Based on the sign information generated in step S803, the walking controller 100 generates information necessary for the walking operation according to the walking operation generating method described in Fig. More specifically, the walking controller 100 generates angular information of a leg joint with respect to a target point of the end effector (S805), and generates rotation information of the pelvis with respect to the target point (S807). At this time, the walking controller 100 can generate rotation information of the pelvis if the end effector can not reach the target point, based on the angle information of the leg joint.

On the other hand, using the corrected displacement of the pelvis according to the rotation information, the locus information of the pelvis can be updated. Then, the walking controller 100 generates information necessary for the walking operation based on the updated sign information.

The above-described technical features may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be those specially designed and constructed for the embodiments or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks, and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, the present invention has been described with reference to particular embodiments, such as specific elements, and specific embodiments and drawings. However, it should be understood that the present invention is not limited to the above- And various modifications and changes may be made thereto by those skilled in the art to which the present invention pertains. Accordingly, the spirit of the present invention should not be construed as being limited to the embodiments described, and all of the equivalents or equivalents of the claims, as well as the following claims, belong to the scope of the present invention .

Claims (15)

A method for generating a gait operation for a virtual character or robot using an inverse kinematics solver,
Receiving capture data for the virtual character or robot;
Generating angle information of a leg joint with respect to a target point of the end effector; And
Generating rotation information of the pelvis with respect to the target point when the end effector fails to reach the target point by the angle information;
And generating a walking motion.
The method according to claim 1,
The leg joint
Including the hip, knee and ankle joints
How to Generate Walking Motion.
3. The method of claim 2,
The step of generating rotation information of the pelvis
The rotation information is generated using the distance between the hinge and the target point and the distance between the end effector and the hip joint by the angle information
How to Generate Walking Motion.
The method according to claim 1,
The rotation information
The rotation angle information of the pelvis centered at the root joint
How to Generate Walking Motion.
3. The method of claim 2,
Generating movement information of the pelvis with respect to the target point when the end effector does not reach the target point due to rotation information of the pelvis;
Further comprising the steps of:
6. The method of claim 5,
The movement information
The displacement information of the pelvis with respect to the position of the root joint as the reference point
How to Generate Walking Motion.
6. The method of claim 5,
The step of generating movement information of the pelvis
The movement information is generated using the distance between the hinge and the target point and the distance between the end effector and the hip joint by the angle information and the rotation information
How to Generate Walking Motion.
The method according to claim 1,
The end effector
Foot or ankle
How to Generate Walking Motion.
A method for generating a walking motion for a virtual character or robot using an inverse kinematic solver,
Receiving capture data for the virtual character or robot;
Generating angular information of an arm joint with respect to a target point of the end effector; And
Generating rotation information of the upper body with respect to the target point when the end effector fails to reach the target point by the angle information;
And generating a walking motion.
A method for controlling a walking motion of a virtual character or robot using an inverse kinematics solver,
Receiving captured data and terrain information for the virtual character or robot;
Generating end effector of the captured data and locus information of the pelvis according to the topographic information;
Generating angle information of a leg joint with respect to a target point of the end effector based on the locus information; And
Generating rotation information of the pelvis with respect to the target point based on the locus information
Wherein the walking operation is performed by the driver.
11. The method of claim 10,
The step of generating rotation information of the pelvis
And when the end effector does not reach the target point, the rotation information of the pelvis is generated by the angle information
A method for controlling walking motion.
11. The method of claim 10,
The step of generating the locus information
And applying a convex hull to the terrain information to generate locus information of the end effector
A method for controlling walking motion.
11. The method of claim 10,
Updating the locus information of the pelvis using the displacement of the pelvis according to the rotation information
Further comprising the steps of:
11. The method of claim 10,
Performing Gaussian filtering on the locus information of the pelvis
Further comprising the steps of:
1. A walking controller for a virtual character or robot using an inverse kinematics solver,
An information receiver for receiving captured data and terrain information for the virtual character or robot;
A trajectory information generating unit for generating trajectory information of an end effector and a pelvis of the captured data according to the terrain information; And
Based on the locus information, an inverse kinematic analysis unit for generating angle information of a leg joint with respect to a target point of the end effector and rotation information of the pelvis,
And a controller.

Images