WO2020222392A1 - Method for robot safety evaluation on basis of collision force bigdata that enables real-time robot collision risk monitoring using graphic information - Google Patents

Abstract

According to the present invention, a method for robot safety evaluation on the basis of collision force bigdata that enables real-time robot collision risk monitoring using graphic information may comprise the steps of: preparing an artificial neural network trained with data on collision force and collision factors; acquiring a three-dimensional image or a three-dimensional model of a test robot; inputting profile information to configure a moving time and a moving path of the test robot; extracting, in real time, the collision force from the artificial neural network prepared in advance; determining whether the magnitude of the extracted collision force falls within preconfigured maximum collision force, so as to evaluate safety of the test robot; and converting a safety evaluation result of the robot into a numerical value, converting the numerical value into graphic information in a simulation program, and providing the graphic information as visual information.

Description

Robot safety evaluation method based on collision physics big data that enables real-time robot collision risk monitoring using graphic information

The present invention relates to a method for evaluating the safety of a robot based on collision physical force big data that enables real-time robot collision risk monitoring using graphic information. More specifically, the collision physical force according to the collision factor extracted from the robot is pre-data mapped or through a neural network. It is possible to learn, and extract the learned artificial neural network in real time, to grasp the safety of the robot in real time according to the movement of the test robot, and to display the safety of the robot to the user using graphic information such as color. It relates to a robot safety evaluation method based on collision physics big data that can monitor the risk level.

Recently, due to the realization of high performance of robots, various efforts have been made to maximize driving convenience and to secure safety by preventing collisions with workers during operation of the robot.

Since such a robot can be installed in the same work space as a human, accidents due to collisions occur frequently during work. Therefore, what is essential to the robot is 1) precision of motion and 2) safety of motion. In the case of motion precision, which is the first requirement, the situation has entered a certain trajectory through the development of motor precision control technology, but in the case of motion safety, the second requirement, the technical completion is very high compared to the precision of motion. It is inadequate.

In particular, in recent years, as the term safety has emerged as a key topic in robotic systems, various studies are being conducted to increase the safety of robots.

The currently used way to ensure safety is to lower the speed of the robot below a certain speed. However, this method does not control the speed of the robot by evaluating the appropriateness of the speed of the robot, but to secure stability by significantly lowering the speed of the robot, so safety can be increased, but there is a problem that the efficiency is not good due to a significant decrease in productivity. .

In order to solve this problem, a fence was built on the robot and the robot was operated normally only when the person is not in the fence, thereby increasing productivity, but safety in case the robot and the person are in the same space, such as cooperative robots and service robots. The improvement solution has not yet been established. Therefore, it is necessary to prepare a solution to prevent injuries by robots while humans and robots work in the same space.

In addition, the safety evaluation method of most of the robots currently developed is to install and evaluate a separate device for obtaining the collision pressure, collision force, movement speed, etc. in the actual test robot, so there is a problem that the evaluation cost increases.

In addition, when evaluating the safety, since the test robot repeats stopping and operating more than necessary, work efficiency is reduced, and there is a problem that an excessive force is applied to the test robot.

In order to solve this problem, a method of mathematically predicting the collision force exerted on the body according to the movement of the test robot was implemented, but the pressure (ISO/TS 15066 robot collision safety evaluation), which is a physical index directly related to body injury. Factor) is not predicted, so there is a problem in that the accuracy of safety evaluation is low because a constant pressure is calculated regardless of the shape of the impact site, or there is a limitation in predicting injury due to a collision because the pressure is not calculated.

Most of the methods for predicting the pressure due to collision are to perform collision simulation using the finite element method, but it takes a long time to analyze and it is impossible to grasp the safety of the robot in real time according to the movement of the test robot.

An object of the present invention is to calculate the collision pressure and collision force applied to the operator according to the movement speed and movement path of each part in consideration of the shape of the test robot, and determine whether the calculated value falls within the International Standardization Organization (ISO) standard. It is to provide a method for evaluating the safety of a robot with improved accuracy of safety evaluation.

In addition, the subject of the present invention is to calculate the collision pressure and collision force applied to the operator according to the movement speed and movement path of each part in consideration of the shape of the test robot, and the calculated value conforms to the standards of the International Organization for Standardization (ISO). It is to provide a method of improving the safety of a robot that controls the speed and posture of a test robot to be satisfied.

In addition, the present invention calculates the collision pressure and collision force applied to the operator according to the movement speed and movement path of each part in consideration of the shape of the test robot, and stores the calculated values to collect the data to build a database. , It is to train it on an artificial neural network to predict the collision physics of the robot in motion through the trained artificial neural network in real time.

In addition, an object of the present invention is to provide a real-time collision safety monitoring method using a graphical representation of the collision physical force of the robot predicted in real time according to the movement of the test robot.

A collision physics big data-based robot safety evaluation method capable of monitoring a real-time robot collision risk using graphic information according to an embodiment of the present invention includes a collision physical force including at least one of a collision pressure and a collision force, and a collision affecting the collision force. Preparing an artificial neural network from which data on factors is learned; Obtaining a 3D image or a 3D model of the test robot including shape information of the actual robot; Setting a movement time and a movement path of the test robot by inputting profile information including movement time information and movement path information of the test robot; Extracting in real time a collision physical force including at least one of a collision pressure and a collision force applied to the object to be collided according to the collision factor of the test robot from the artificial neural network prepared in advance; Evaluating the safety of the test robot by determining whether the magnitude of the extracted collision physical force falls within a preset maximum collision physical force; And converting the result of the safety evaluation of the robot into a numerical value and converting it into graphic information in the simulation program and notifying the robot user as visual information.

In addition, in the step of preparing the artificial neural network in which data on the collision physical force including at least one of the collision pressure and the collision force and the collision factor affecting the collision is learned, a collision experiment between an actual robot and a collided object is performed, or In a method of performing finite element simulation or mathematical calculation, the collision factor including at least one of the shape of the collision site, the effective mass, the moving speed, the collision direction, and the physical properties of the human body, and the collision force according to the collision factor. It may be characterized by acquiring data.

In addition, in the preparing of the artificial neural network in which the collision physical force and data on the collision factors affecting the collision are learned, the steps of considering the collision factors and determining a combination of the collision factors; Obtaining a collision physical force according to a combination of the collision factors; And learning the acquired collision force data to the artificial neural network. It may include.

In addition, the step of evaluating the safety of the test robot by determining whether the magnitude of the extracted collision physical force falls within a preset maximum collision physical force may be performed by comparing the extracted collision physical force with the International Robot Collision Safety Standard (ISO). The safety of the robot can be evaluated.

In addition, the test robot may be a 3D image formed by inputting shape information of the robot into the simulation program or a 3D model formed through a 3D measurement sensor.

In addition, the simulation program may be a Computer Aided Engineering (CAE) program.

In addition, among the methods of obtaining data on the collision physical force according to the collision factor, the mathematical calculation method is to change the posture of the test robot, and according to the change of the posture, the avoidance is caused by the risk of injury of the test robot. It may include calculating the collision pressure and the collision force applied to the collider.

In addition, in the mathematical calculation method, the contact pressure applied to the object to be collided is calculated according to the area of each part of the test robot, and at least one injury to the test robot is caused through the calculated contact pressure value. It may further include the step of setting the risk area.

In addition, the mathematical calculation method may calculate a collision pressure and a collision force applied to the collided object for each predetermined time unit.

In addition, the step of converting the result of the safety evaluation of the robot into a numerical value and converting it into graphic information in a simulation program and notifying the robot user as visual information may include: selecting an injury-causing site that may be dangerous during a collision in the robot; Calculating collision information including at least one of force, pressure, collision energy, and deformation amount that will occur when a human collides with the injury-causing site during movement of the robot; Converting the calculated risk level of the collision information into color; And inputting a color of the injury-causing area drawing as the converted color value on a robot simulation that tracks the movement of an actual robot in real time.

In addition, in the step of displaying the color of the drawing of the corresponding part in the converted color in the robot simulation that tracks the movement of the real robot in real time, the color of the part where the collision information is not calculated in the drawing of the robot is the part where the collision information is calculated. The color value of can be interpolated and input.

According to the present invention, the magnitude of the collision pressure and the collision force applied to the object to be collided by the test robot can be obtained by simulating through a computer program such as a CAE (Computer Aided Engineering) program, or a calculation algorithm is applied to the test robot system. By applying it, collision pressure and collision force can be calculated in real time, so that the safety of the test robot can be evaluated in real time. Accordingly, since it is not necessary to have a separate device for obtaining the collision pressure, collision force, movement speed, etc. acting on the test robot, safety evaluation can be performed at low cost.

In addition, according to the present invention, since the CAE information of the received robot is converted into big data and learned on the artificial neural network, the trained artificial neural network can be installed on the robot, so that a program implementing the algorithm can be installed on the robot with a small capacity. do.

In addition, according to the present invention, in a robot simulation that tracks the movement of the actual robot in real time, the risk of the corresponding part is displayed in color to enable safe work at the site where the robot is used, and collision damage is prevented when an unexpected accident occurs during operation. Can be minimized.

1 is a flowchart of a method for evaluating safety of a robot according to an embodiment of the present invention.

Figure 2 is a flow chart for the step of preparing the learned artificial neural network.

FIG. 3 is a diagram showing a process of selecting a collision factor and calculating the collision force according to the collision force to collect data, making the aggregated data into a data map or learning from an artificial neural network to predict the collision force of a robot in motion in real time. .

4 is a flow chart for explaining a detailed process of calculating the collision pressure and the collision force applied to the object by the test robot.

5 is a diagram schematically showing a state in which an actual robot and an object to be collided collide.

6 is a view showing a surface change of an object to be collided with an actual robot in FIG. 5.

7 is a diagram showing a three-dimensional shape of a test robot in FIG. 1.

8 is a view showing the magnitude of the contact pressure applied to the object to be collided according to the shape of each part of the test robot.

9 is a diagram showing values of collision pressure and collision force obtained through a 3D modeling program.

10 is a flow chart for explaining a step of converting the result of the safety evaluation of the robot into a numerical value, converting it into graphic information in a simulation program, and notifying it as visual information.

11 and 12 show the physical force of the collision acquired using information (joint angle, speed, weight information, etc.) of the robot in motion, safety evaluation in real time with the International Safety Standard (ISO), and the evaluation result numerically to provide a real-time graphic. Implementation drawing.

Hereinafter, a method for evaluating the safety of a robot according to a preferred embodiment will be described in detail with reference to the accompanying drawings. Here, the same symbols are used for the same configuration, and detailed descriptions of repetitive descriptions and known functions and configurations that may unnecessarily obscure the subject matter of the invention are omitted. Embodiments of the invention are provided to more completely explain the invention to those with average knowledge in the art. Accordingly, the shapes and sizes of elements in the drawings may be exaggerated for clearer explanation.

1 is a flow chart of a method for evaluating safety of a robot according to an embodiment of the present invention, and FIG. 2 is a flow chart for preparing a learned artificial neural network, and FIG. 3 is a flowchart of selecting a collision factor and calculating the collision physical force accordingly. It is a diagram showing a process of predicting in real time the collision physical force of a robot in motion by calculating and collecting data, making the collected data into a data map or learning from an artificial neural network.

1 to 3, a collision physics big data-based robot safety evaluation method (S100) capable of real-time robot collision risk monitoring using graphic information according to the present invention includes at least one of collision pressure and collision force. Preparing an artificial neural network in which data on physical forces and collision factors affecting it are learned (S110), and obtaining a 3D image or a 3D model of a test robot including shape information of the actual robot (S120) ), setting the movement time and movement path of the test robot by inputting profile information including movement time information and movement path information of the test robot (S130), and the collided object according to the collision factor of the test robot The step of extracting a collision physical force including at least one of a collision pressure and a collision force applied to the artificial neural network in real time (S140), and the magnitude of the extracted collision physical force falls within a preset maximum collision physical force It may include determining whether the safety of the test robot is evaluated (S150), and converting the result of the safety evaluation of the robot into graphic information in the simulation program and notifying the robot user as visual information (S160). .

In the step (S110) of preparing an artificial neural network in which data on the collision physical force including at least one of the collision pressure and the collision force and the collision factor affecting it is learned (S110), 1) conduct an actual collision experiment, or 2) Ansys By using a finite element simulation program such as 3) or by mathematically calculating the type and value of the collision factor corresponding to the input, collision physical force corresponding to the output can be obtained. Here, the collision factor may include at least one or more of the shape of the collision site, the effective mass, the moving speed, the collision direction, and the physical properties of the person, and the collision physical force may include at least one of collision pressure and collision force.

Extensive data on collision factors and collision forces obtained in this way can be stored in a database and learned through an artificial neural network. As an example, the artificial neural network may be in the form of a table in which a number of numerical values for a collision factor corresponding to an input and a collision force corresponding to an output are correlated.

In addition, the step (S110) of preparing the artificial neural network in which the collision physical force and data on the collision factors affecting the collision are learned (S110), consider the collision factor, and determine a combination of the collision factors (S111), and the It may include a step (S112) of acquiring a collision force according to a combination of collision factors, and a step (S113) of learning the acquired collision force data to the artificial neural network.

Specifically, by determining the combination of the collision factors among the collision factors such as the shape of the collision site, the effective mass, the moving speed, the collision direction, and the physical properties of the human body (for example, physical properties such as skin, bones, muscles, etc.) Can be selected.

In this way, collision forces may be obtained by combining collision factors and selecting the number of all cases according to the combination of collision factors. These collision physical forces can mean collision pressure, collision force, etc., and 1) actual collision tests are performed, 2) finite element simulation programs such as Ansys are used, or 3) collision pressure and collision are mathematically calculated. It is as described above that it is possible to obtain a collision physical force such as force.

Among the methods of obtaining the collision physical force from the collision factor, a method of obtaining the collision physical force through a mathematical calculation method will be described later.

In the step S120 of acquiring a 3D shape of the test robot, a 3D image or a 3D model of the test robot 10 including shape information of the actual robot R is obtained. Specifically, the test robot 10 may be formed of a 3D image formed by inputting shape information of the robot R into a simulation program or a 3D model formed through a 3D measurement sensor. That is, the test robot 10 is a 3D image formed by inputting shape information of the actual robot R into a simulation program such as a CAE (Computer Aided Engineering) program, or implemented through a 3D measurement sensor, and the actual robot R It can be made of a 3D model that is driven and controlled in the same way as.

The robot's CAE information includes the robot's shape, physical properties, and specifications for the robot's driving parts (motor performance and performance limits, joint angle limits, etc.). In addition, the CAE information of the robot may include collision factors such as the shape of the collision site, the effective mass, the moving speed, the collision direction, and the physical properties of the human body (eg, physical properties of skin, bones, muscles, etc.).

Although the type of the 3D modeled robot R is not limited, it may be a cooperative robot 10 that jointly processes tasks in a certain work space. Such a collaborative robot may be formed of a manipulator formed to have a mechanical hand at its tip to grasp and transport a specific object or perform a specific task. In addition, the test robot 10 may be formed to have a degree of freedom to move in at least one of the X-axis direction, Y-axis direction, Z-axis direction, pitch direction, yaw direction, and roll direction. have. That is, the test robot 10 may be formed of a manipulator having at least one degree of freedom and more than one degree of freedom.

Specifically, the test robot 10 may include at least two link units 12 connected through the joint 11 and an end-effector 13 connected to one of the link units 12. Here, the end effector refers to a part having a function that directly acts on a work object when the test robot 10 performs work, and may be, for example, a mechanical hand of a manipulator.

In the step of setting the movement time and movement path of the test robot (S130), the movement time and movement path of the test robot 10 are set by inputting profile information including movement time information and movement path information of the test robot 10 do. For example, when the test robot 10 is a 3D image, profile information including movement time information and movement path information is input to a simulation program to set the movement time and movement path of the test robot 10. In addition, when the test robot 10 is a 3D model, profile information including movement time information and movement path information is input to the control system of the test robot 10 to determine the movement time and movement path of the test robot 10. Set. Here, the method of controlling the driving of the test robot 10 through the simulation program and the control system of the robot is a known technique, and thus will be omitted.

In the step (S140) of extracting a collision physical force including at least one of a collision pressure and a collision force applied to the object to be collided according to the collision factor of the test robot 10 from the artificial neural network prepared in advance (S140), the test robot 10 ), including the shape of the collision site, effective mass, moving speed, collision direction, and physical properties of a person (for example, physical properties of skin, bones, muscles, etc.) included in the CAE information received when acquiring the 3D shape of As data of the collision physical force, such as the corresponding collision pressure and collision force, are extracted from the artificial neural network prepared in advance, it is possible to extract the collision physical force in real time because it takes less time than the conventional one. Accordingly, the time for evaluating the safety of the robot is drastically shortened, the safety at the industrial site where the robot is used is greatly improved, and collision damage is minimized even in the event of an unexpected accident.

In the step of evaluating the safety of the test robot (S150), the magnitude of the collision pressure (P) and collision force (F _C ) corresponding to the collision factor of the test robot 10 extracted in real time from the artificial neural network is a preset maximum collision pressure. It is determined whether it falls within the magnitude of (P _MAX ) and the maximum impact force (F _MAX ).

If the magnitude of the collision pressure (P) and collision force (F _C ) applied to the object to be collided 20 in the step (S150) of evaluating the safety of the robot is preset maximum collision pressure (P _MAX ) and maximum collision force collision force If it falls within the size of (F _MAX ), the test robot 10 is determined to be safe. Conversely, if the magnitudes of the impact pressure P and the impact force F _C are greater than or equal to the maximum impact pressure P _MAX and the maximum impact force F _MAX , the test robot 10 is determined to be unsafe.

In the step (S150) of evaluating the safety of the robot, if the magnitude of the collision pressure (P) and the collision force (F _C ) is greater than or equal to the maximum collision pressure (P _MAX ) and the maximum collision force (F _MAX ), the object to be collided ( 20), the moving speed of the test robot 10 can be controlled so that the impact pressure (P) and the impact force (F _C ) applied to it are less than the magnitudes of the maximum impact pressure (P _MAX ) and the maximum impact force (F _MAX ). have. This is because reducing the moving speed of the test robot 10 reduces the force applied to the object to be collided 20. As the force applied to the object to be collided 20 decreases, the pressure also decreases, so if the moving speed of the test robot 10 is properly adjusted, it is possible to implement a robot R that does not injure the human body while achieving the maximum speed. You lose.

In the step of evaluating the safety of the robot (S150), the preset maximum impact pressure (P _MAX ) and maximum impact force (F _MAX ) are sized according to the International Organization for Standardization (ISO), more specifically TS 15066. I can. TS 15066 of the International Organization for Standardization (ISO) discloses the maximum allowable pressure and allowable force for each body part of a person, so the maximum impact pressure (P _MAX ) and maximum impact force (F _MAX ) If set to size, the stability of the robot R can be further improved.

A description of the step (S160) of converting the result of the safety evaluation of the robot into a numerical value, converting it into graphic information in the simulation program, and notifying the robot user as visual information will be described later.

4 is a flow chart for explaining a detailed process of calculating a collision pressure and a collision force applied to an object to be collided by a test robot, and FIG. 5 is a diagram schematically showing a state in which the actual robot and the object to be collided collide 6 is a diagram showing a change in the surface of an object colliding with an actual robot in FIG. 5, FIG. 7 is a diagram showing a three-dimensional shape of a test robot in FIG. 1, and FIG. 8 is a test It is a diagram showing the magnitude of the contact pressure applied to the object to be collided according to the shape of each part of the robot.

Referring to FIGS. 4 through 8, a method of obtaining data on collision physical force through a mathematical calculation method among methods of obtaining data on collision physical force according to a collision factor is as follows.

The impact pressure (P) and impact force (force, F _C ) applied to the object to be collided 20 are determined by considering the shape, effective mass, movement speed, and direction of the test robot 10 at risk of injury. Can be calculated.

Specifically, the collision force (F _C ) applied to the object to be collided 20 can be implemented through Equation 1 below. Here, the object to be collided 20 may be a human, and the effective mass (M _i ) for the collision portion of the test robot is calculated by kinematic theory, and the effective mass (M _h ) for the collision portion of the object to be collided is to the user. It can be input by and predetermined. The displacement of the impact area of the test robot (y _i ) and the displacement of the impact area of the object to be collided (y _h ) can be obtained through the CAE system.

[Equation 1]

M _i : Effective mass for the collision area of the test robot

M _h : Effective mass for the collision part of the object to be collided

F _C : Collision force

y _i : Displacement of the collision area of the test robot

y _h : Displacement of the collision area of the object to be collided

In addition, the collision pressure P applied to the object to be collided 20 can be implemented through Equation 2 below. Here, the skin elasticity (K) of the object to be collided and the skin thickness (h) of the object to be collided may be input by the user through the CAE system and stored in advance.

[Equation 2]

δ: amount of skin deformation of the object to be collided

α: Collision angle between the test robot and the object to be collided

F _c : impact force p: impact pressure

K: skin elasticity of the object to be collided h: skin thickness of the object to be collided

x, y: collision plane coordinate system

In the step (S112) of obtaining data on the collision physical force according to the collision factor in a mathematically calculated manner, the collision pressure (P) and the collision force (F _C ) applied to the collided object 20 are calculated for each predetermined time unit. can do. This is to improve the speed of the calculation by reducing the amount of data for calculating the collision pressure (P) and the collision force (F _C ), and to prevent a load from being applied. Here, the time determined in a predetermined unit may vary according to the shape of the test robot 10. That is, as the shape of the test robot 10 becomes more complex, the unit time may be shorter.

The step (S112) of obtaining data on the collision physical force according to the collision factor in a mathematically calculated manner may further include a step (S1121) of setting an injury-causing risk area of the test robot.

In the step (S1121) of setting an injury-causing risk area of the test robot, the contact pressure applied to the object to be impacted 20 is calculated according to the area of each area of the test robot 10, and the test robot is based on the calculated contact pressure value. (10) At least one injury-causing risk area is set.

For example, when the test robot 10 has a cylindrical shape, each part of the test robot 10 for setting an injury-causing risk part is a circumferential surface, an upper surface, a lower surface, an upper edge, and a lower surface. It can be a corner. Then, the contact pressure applied to the object to be collided 20 is calculated according to the area of each portion. Here, the method of calculating the contact pressure can be calculated through a relational expression of P=F/A (P: pressure, F: force, A: area).

When the contact pressure for each part of the test robot 10 is calculated through this process, the part having the largest value among the contact pressures or all parts in which the contact pressure exceeds a preset value may be selected as the risk of injury. .

On the other hand, one or two or more areas of risk causing injury of the test robot 10 may be selected by the user. Specifically, an injury-prone risk site of the test robot 10 may be one or two or more selected from the link unit 12 and the end effector 13. As described above, when the risk-rising part of the test robot 10 is previously set by the user, the step of setting the risk-producing part of the test robot (S1221) may be omitted.

When collision data such as collision pressure (P) and collision force (F _C ) are acquired by the above-described method, these data are collected and the aggregated data is made into a data map, or the collision physics of the robot in motion is learned by learning from an artificial neural network. Can be predicted in real time.

On the other hand, the step of obtaining data on the collision physical force according to the collision factor in a mathematically calculated manner (S112) includes calculating the collision pressure and collision force applied to the object to be collided according to the change of the posture of the test robot (S1222). It may contain more.

The step of calculating the collision pressure and the collision force applied to the object to be collided according to the change of the posture of the test robot (S1222) is by adjusting the angle of the joint 11 of the test robot 10, the link unit 12 and the end effector 13 ) To change the posture, and calculate the collision pressure (P) and the collision force (F _C ) applied to the object to be collided 20 by the area at risk of causing injury according to the change of the posture.

The reason for calculating the collision pressure (P) and collision force (F _C ) applied to the object to be collided 20 while changing the posture of the test robot 10 is the change in the posture of the link unit 12 and the end effector 13 This is because the distance and the contact area with the object to be collided 20 are changed according to. In addition, when the distance and the contact portion with the object to be collided 20 are different, the magnitude of the collision pressure P and the collision force F _C applied to the object to be collided 20 is also changed.

Therefore, if the magnitude of the collision pressure (P) and the collision force (F _C ) applied to the collided object 20 according to the change of posture is obtained, even if the robot R moves along the same movement speed and movement path, the collided object It becomes possible to implement a posture in which minimal impact is applied to (20).

8 is a diagram showing the magnitude of a contact pressure applied to an object to be collided according to the shape of each part of the test robot. In addition, FIG. 9 is a diagram showing values of collision pressure and collision force acquired through a 3D modeling program.

10 is a flow chart for explaining the step of converting the safety evaluation result of the robot into a numerical value, converting it into graphic information in the simulation program, and notifying it as visual information, and FIGS. 11 and 12 are information on the robot in motion (joint angle, speed, weight It is a diagram that implements real-time graphicization by evaluating safety in real time with international safety standards (ISO), and quantifying the evaluation results in real-time using the collision force acquired using information, etc.).

10 to 12, the step (S160) of converting the result of the safety evaluation of the robot into a numerical value and converting it into graphic information in the simulation program and notifying the robot user as visual information (S160) is an injury that may be dangerous when a collision occurs within the robot. Selecting an inducing region (S161), and calculating collision information including at least one of force, pressure, collision energy, and deformation amount that will occur when the injury-causing region and a human collide with the robot during movement (S162). ), converting the calculated risk of collision information into color (S163), and inputting the color of the injury-causing area drawing as the converted color value on a robot simulation that tracks the movement of the actual robot in real time. It may include step S164.

Particularly, in the step (S164) of inputting the color of the injury-causing area drawing as the converted color value in a robot simulation that tracks the movement of the actual robot in real time, the color of the area where collision information is not calculated in the drawing of the robot is The color value of the part where the collision information is calculated is interpolated and input. Accordingly, colors can be continuously connected to the surface of the robot in the simulation without being cut off.

The color interpolation method is basically a method of calculating in proportion to the distance from the color value calculation point, and the specific formula is as follows.

[Correction 02.03.2020 pursuant to Rule 91]

[Correction 02.03.2020 pursuant to Rule 91]
Here, A is a calculated point, V _a is a calculated value of A, X is an uncalculated point, V _x is an interpolated value, B is a calculated point, and V _b is a calculated value of B. In addition, dis_A is the distance between X and A, and dis_B is the distance between X and B.

Claims (11)

1. Preparing an artificial neural network from which data on collision physical force including at least one of collision pressure and collision force and collision factors affecting the collision force are learned;

   Obtaining a 3D image or a 3D model of the test robot including shape information of the actual robot;

   Setting a movement time and a movement path of the test robot by inputting profile information including movement time information and movement path information of the test robot;

   Extracting in real time a collision physical force including at least one of a collision pressure and a collision force applied to the object to be collided according to the collision factor of the test robot from the artificial neural network prepared in advance;

   Evaluating the safety of the test robot by determining whether the magnitude of the extracted collision physical force falls within a preset maximum collision physical force; And

   Converting the result of the safety evaluation of the robot into a numerical value and converting it into graphic information in the simulation program, and notifying the robot user as visual information;

   A robot safety evaluation method based on collision physics big data capable of monitoring a real-time robot collision risk using graphic information including a.
2. The method of claim 1,

   In the step of preparing the artificial neural network in which data on the collision physical force including at least one of the collision pressure and the collision force and the collision factor affecting the collision is learned,

   A method of performing a collision experiment between an actual robot and an object to be collided, performing a finite element simulation, or mathematically calculating, and includes at least one of the shape of the collision site, the effective mass, the moving speed, the collision direction, and the physical properties of the human body. A method for evaluating robot safety based on collision physical force big data capable of monitoring a collision risk of a robot in real time using graphic information, characterized in that acquiring data on a collision factor and collision physical force according to the collision factor.
3. The method of claim 2,

   In the step of preparing an artificial neural network in which data on collision physics and collision factors affecting it are learned,

   Considering a collision factor and determining a combination of the collision factors;

   Obtaining a collision physical force according to a combination of the collision factors; And

   Learning the acquired collision force data to the artificial neural network;

   A robot safety evaluation method based on collision physics big data capable of monitoring a real-time robot collision risk using graphic information including a.
4. The method of claim 1,

   The step of evaluating the safety of the test robot by determining whether the magnitude of the extracted collision physical force falls within a preset magnitude of the maximum collision physical force,

   A robot safety evaluation method based on collision physics big data that enables real-time robot collision risk monitoring using graphic information to evaluate the safety of a test robot by comparing the extracted collision force with the International Robot Collision Safety Standard (ISO).
5. The method of claim 1,

   The test robot,

   A robot safety evaluation method based on collision physical force big data that enables real-time robot collision risk monitoring using a three-dimensional image formed by inputting the shape information of the robot into the simulation program or a three-dimensional model formed through a three-dimensional measurement sensor.
6. The method of claim 1,

   The simulation program is a computer aided engineering (CAE) program, a robot safety evaluation method based on collision physics big data that enables real-time robot collision risk monitoring using graphic information.
7. The method of claim 2,

   Among the methods of obtaining data on the collision force according to the collision factor, the mathematical calculation method is:

   Real-time robot collision using graphic information including the step of changing the posture of the test robot and calculating a collision pressure and a collision force applied to the object to be collided by an injury-causing risk site of the test robot according to the change of the posture A robot safety evaluation method based on collision physics big data that can monitor risk.
8. The method of claim 7,

   The mathematical calculation method,

   The step of calculating a contact pressure applied to the object to be collided according to the area of each part of the test robot, and setting at least one injury-causing risk site for the test robot based on the calculated contact pressure value. A robot safety evaluation method based on collision physics big data that enables real-time robot collision risk monitoring using graphic information.
9. The method of claim 2,

   The mathematical calculation method,

   A robot safety evaluation method based on collision physics big data that enables real-time robot collision risk monitoring using graphic information that calculates collision pressure and collision force applied to the collided object for each predetermined time unit.
10. The method of claim 1,

    The step of quantifying the safety evaluation result of the robot and converting it into graphic information in the simulation program and notifying the robot user as visual information is:

    Selecting an injury-causing site that may be dangerous during a collision in the robot;

    Calculating collision information including at least one of force, pressure, collision energy, and deformation amount that will occur when a human collides with the injury-causing site during movement of the robot;

    Converting the calculated risk level of the collision information into color; And

    Inputting a color of the injury-causing area drawing as the converted color value on a robot simulation that tracks the movement of an actual robot in real time;

    A robot safety evaluation method based on collision physics big data capable of monitoring a real-time robot collision risk using graphic information including a.
11. The method of claim 10,

    In the step of displaying the color of the drawing of the corresponding part as the converted color on a robot simulation that tracks the movement of an actual robot in real time,

    The robot safety evaluation method based on collision physics big data that enables real-time robot collision risk monitoring using graphic information inputted by interpolating the color value of the part where collision information is calculated for the color of the part in the drawing of the robot for which collision information is not calculated.

Images