JPH0584679A - Robot programming method - Google Patents

Abstract

PURPOSE:To provide a robot programming method based on dynamic analysis. CONSTITUTION:Knowledge of force, torque, etc., applied to a joint, obtained by analyzing actual human movement to analyze the actual movement of the human body obtained by an analytic result, is used to design a program relating to new action desired by a programmer. The program is divided into three steps of dynamics, restricting condition and reverse dynamics, and in the first step of dynamics, each element of a machine is discriminated to take out its movement in a shape of dynamic equation. In the second step of the restricting condition, the restricting condition of containing a mutual connection relation of the element and a range of respective movement is checked. In the third step of inverse dynamics, the movement and force checked by these restricting condition are calculated by using the dynamic equation to display a result of these movement and force or the like.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a robot programming method for inputting movements of humans or animals to be imitated in a robot or the like which is controlled by a computer and imitates movements of humans or animals. is there.

[Prior Art]

Recently, robots have been used in various fields in place of humans. In order to allow these robots to work on behalf of humans, an operation called teaching is required to teach the robots human movements in advance. For this teaching, there is an online programming method called teaching play back that teaches human movements directly to the robot, and an offline programming method that simulates human movements by a computer and programs robot movements. is there.

Of these, teaching playback is a teaching method in which a person manually moves a robot in accordance with a work procedure to memorize a motion, and is widely used in industrial robots because it is easy to operate. In this case, it is necessary to suspend the work being performed by the robot. On the other hand, the offline programming method is a method in which programming is performed separately from the robot, and programming can be performed even while the robot is performing work. This is an effective means for applying to a robot that is doing work that you do not want to do.

Usually, in this off-line programming method, the contents of a trajectory or motion planned by using an environment model are described in a robot language, the contents are confirmed by a motion simulation, and then fine adjustment is performed at a work site. There is. As a result, offline programming tasks are complex. This environment model includes geometric information such as the shape and dimensions of the work object and peripheral equipment, position and orientation, physical information such as material and weight, and technical information such as functions, applications, and usage. , The programmer plans work using this environment model.

In order for a robot to perform work on behalf of a human, it is necessary to analyze the human motion and design the motion based on the knowledge obtained as a result. Since the obtained incorrect knowledge is used, it is difficult to say that the obtained robot movement is rational. In addition, it is difficult to design a new motion because the dynamics used to analyze and design the motion are also based on kinematics using only position, velocity and acceleration.

Further, in order to design the movement of the robot using a computer, a method of responding in real time by an easy-to-use interactive form is required, but as described above, in the conventional off-line programming method, a motion simulation is performed. It requires trial and error of operation confirmation and fine adjustment at the work site, and it does not respond in real time in an interactive manner.

As a method for discussing the motion of an object, there is a method called dynamics for discussing the motion of an object in relation to a force in addition to the kinematics using position / velocity and acceleration. If this method is applied to a robot programming method, complicated movements can be programmed with a small number of operations. Further, according to the robot programming method using the dynamics, another great advantage, that is, it is possible to design a new motion that cannot be performed by the robot programming method using the kinematics.

However, in order to program the motion of the robot using dynamics, data such as moment of inertia, center of gravity, joint friction, elasticity of muscles / ligaments, etc., which are difficult to measure, are necessary, and there is no such data. And it becomes an irrational movement as well as when using kinematics. Therefore, it has been impossible to reproduce the complete movement of the human body by the conventional robot programming method using dynamics.

In this method, the calculation amount O (f (n)) is n ⁴ when n is the number of minimum units of movement in the human body.
Is a function O (n ⁴ ) and has a large amount of calculation, which is not practical because the calculation by the computer is expensive.
On the other hand, there is also a kinetic method using a calculation in which the amount of calculation is a function O (n) of n, but this method is valid only when there is no rotation about the axis. Therefore, this method cannot be used when there is rotation around the axis.

As described above, the conventional robot programming methods are carried out by trial and error, and do not respond in real time in an interactive manner. In view of this, the present invention proposes a new robot programming method that can utilize dynamics to program complex tasks.

[0011]

In the present invention, a new motion of the robot can be created by using the knowledge about the motion of the human obtained by dynamically analyzing the actual motion of the human. Propose a robot programming method. That is, an object of the present invention is to provide a method of interactively creating a robot program using a computer without trial and error or intuition of a programmer.

[0012]

In order to solve the above-mentioned problems, in the robot programming method of the present invention, firstly, the basic motion of a human being is analyzed, and the force and torque generated at each joint of the human body are analyzed. The data of the dynamic parameters including is input to the database as knowledge about basic movement.

Next, the programmer accesses the database and processes the obtained data, but the computer feeds back the result of the inverse dynamics to the programmer in the form of force in the form of motion with constrained conditions. Then, the motion is interactively designed by repeating this process until a satisfactory result is obtained.

The calculation amount of this motion analysis method is a function O of n.
(N) solves the problem of high computational cost. In addition, it is possible to create a natural robot movement in an interactive manner without trial and error or the intuition of a programmer.

[0015]

Embodiments of the present invention will be described below with reference to the drawings. A flow chart of the present invention is shown in FIG.
This flow chart is 1. Create a human body model 2. Input actual human movements 3. Analyze the input motion 4. Operation design 5. Apply kinetics 6. Apply constraints 7. Apply inverse dynamics 8. It consists of stages for displaying results.

In the first step, "creating a human body model", the human body is decomposed into parts which are the minimum units of motion, and the constraint conditions such as the characteristics peculiar to each part, the mutual relation, and the range of joint motion are considered. A human body model is created based on it, and it is input to the computer as a database.

In the second step "inputting actual human motion", the motion to be analyzed is input in frame units of video images or frame units of film images. In this case, shadows are simultaneously impaired from a plurality of directions. By using the image, the analysis of the next stage can be performed more concretely.

In the third step "Analysis of the input motion", the motion input in the second stage is calculated using inverse dynamics, and the center of gravity of each part, the force and torque acting on each joint, The center of gravity, the force acting on the center of gravity, and the torque are analyzed, and the data of the analysis result is input to the database.

Next, a case where a robot program having a new movement is created using the above analysis result will be described. In the fourth stage, "Motion Design," the programmer first selects a basic motion from the database.
The data is represented by a control graph as shown in FIG. 2, for example. This control graph is a control graph created based on the database of the movement of the left elbow of the human body. In this control graph, the horizontal axis represents time, and the vertical axis represents forces generated at each joint of the body, x, y, The three axes of z are shown.
As a matter of course, the two forces generated in the same joint have the same magnitude and opposite directions. Also, complex movements are represented by several control graphs.

A robot program is created on the basis of this control graph, but physical variables including enlargement / reduction of axes of the control graph are changed in order to correspond to the size and material of the object to be handled, the operating range of the robot, and the like. .

In the fifth stage, "Applying Dynamics,"
The movement of each element of the robot is calculated based on the force and dynamic equations specified by the programmer. In that case,
Originally, each element is connected to each other, but in order to reduce the amount of calculation, each element is cut off from other elements, and the constraints on the mutual connection relationship and movement range of each element are also temporarily ignored. To be done.

In order to calculate the movement of each element, in the present invention, Newton's equation is used to obtain the linear acceleration, and Euler's equation is used to obtain the angular acceleration. When the linear acceleration and the angular acceleration are obtained, These are integrated to obtain the velocity, and further integrated to obtain the position.

In the sixth step "Applying constraint conditions", two physical constraint conditions of the mutual connection relation of each element and the range of motion are checked for the calculation result of the motion of each element.
This processing starts from the basic element, and the position and arrangement direction of each of the lower elements are sequentially checked. Here, two checks are made as to whether or not the lower element is always connected to the upper element and whether or not the operation of each element exceeds a predetermined range.

As a result, when the lower element is not connected to the upper element, the lower element is moved in parallel so that the lower element is connected to the upper element, and the movement of each element is constant. If the value is out of the range, rotate it to adjust the movement of the element within the range.

In the seventh step "Applying inverse dynamics", the force that occurs at the connection point of each element is calculated using the Lagrangian equation expressing the relationship between force and motion.

When programming a new motion, if a satisfactory result is not obtained, the steps 5 to 7 are repeated to interactively design the new motion.

In the "display result" of the eighth step, a new movement in the middle of programming or after programming is displayed on the screen.

In the present invention, since a simple linear regression algorithm is used as shown in FIG. 1, the amount of calculation for performing inverse dynamics is O (n) which is a function of n. Further, according to the present invention, the movement of the human body can be designed interactively using a computer without trial and error or the intuition of the designer.

[0029]

As described above, the dynamic robot programming method of the present invention comprises two processes: analysis of the basic movement of an actual human and creation of a new dynamic robot program. Then, analysis of basic human movements involves creating a human body model, inputting actual movements,
Proceed through the three stages of analysis of the input motion, and the creation of the dynamics robot program is 3 of dynamics, constraint conditions, and inverse dynamics.
In the dynamics stage, the machine is divided into independent elements, and the movements of the individual elements are calculated using Newton's equation and Euler's equation, apart from the movements of other elements. In the constraint stage, the mutual connection relation of elements and the range of movement are checked. In the inverse dynamics stage, the force that produces the new motion modified by the constraints is calculated. In that case, the total calculation amount is 0 (n).

[Brief description of drawings]

FIG. 1 is a flowchart of a robot programming method according to the present invention.

FIG. 2 is a control graph of an example of a force acting on a joint.

Claims (2)

[Claims] 1. A robot programming method for programming a motion based on human motion in a robot, the method breaking down a human body into parts which are the minimum units of motion, and a characteristic peculiar to each part, Applying the inverse dynamics to the steps of creating a human body model based on mutual relationships and constraints such as the range of joint movements and inputting it into the database, inputting actual human movements, and inputting movements The step of calculating and calculating the center of gravity of each part / force / torque acting on each joint, the center of gravity / force / torque acting on the entire center of gravity, and the step of selecting a basic operation from the database and changing its physical variables Then, each part of the body is cut off from the other part, and the actual motion is replaced by a new motion based only on the specified force and dynamic equations, ignoring the mutual connection relation and the constraint condition regarding the range of motion of the joint. Calculation step, a step of checking and correcting the mutual constraint relationship and the physical constraint condition of the range of joint movement in the calculation result, and the movement of the robot is displayed on the screen using the calculated result. A robot programming method comprising the steps of: 2. A robot programming method for programming a robot to perform movements based on human movements, the method decomposing a human body into parts which are the minimum units of movement, and the characteristics peculiar to each part, Applying the inverse dynamics to the steps of creating a human body model based on mutual relationships and constraints such as the range of joint movements and inputting it into the database, inputting actual human movements, and inputting movements The step of calculating and calculating the center of gravity of each part / force / torque acting on each joint, the center of gravity / force / torque acting on the entire center of gravity, and the step of selecting a basic operation from the database and changing its physical variables Then, each part of the body is cut off from the other part, and the actual motion is replaced by a new motion based only on the specified force and dynamic equations, ignoring the mutual connection relation and the constraint condition regarding the range of motion of the joint. The step of calculating the force, the step of correcting the mutual connection relation and the physical constraint condition of the range of motion of the joint for the calculation result, and the step of calculating the force-motion relationship by applying the inverse dynamics. And a step of synthesizing the motion calculated by the dynamics and the force and the center of gravity calculated by the inverse dynamics and displaying the combined result on the screen.