KR102513381B1 - Torque and posture-based robot teaching method using multimodal sensing - Google Patents

Abstract

Currently, most robots operate with motors, and there are robots that operate with other hydraulic and pneumatic powers, but in almost all cases, robots control the stiffness of robot motion with operating current, so that they can operate with appropriate stiffness in robot control. Since it is not easy to input teaching values or instruction values to control , it is difficult when a robot and a human work together as described above. In order to solve the above problems, the present invention is
One or more position sensors for detecting a change in position movement and/or tooling position for a worker's work; and one or more EMG sensors for measuring muscle strength used during the worker's work according to the type of the worker's work; and a position-based robot job information data storage step of position-based robot job information data for storing the EMG based on the position detected by the position sensor and the EMG measured by the EMG sensor. and time-based robot job information data storage step of time-based robot job information data for storing position and EMG based on time; and downloading robot work teaching information for transmitting the position information and EMG information to the same work robot as the worker. and selecting one of position-based robot job information data and time-based robot job information data during operation of the robot to operate the robot.
According to the above configuration of the present invention, even in precise work performed by a person, the robot provides the location information of the human worker's work tools and the magnitude of the force used at that time using the electromyogram sensor, so that the human work and More similar work can be done, and there is an effect of being able to implement human work more easily.

Description

Torque and posture-based robot teaching method using multimodal sensing{.}

The invention of this application relates to a position and a work teaching method for informing a work position of a robot. More specifically, it relates to a teaching method of a new concept of a robot that works together while performing the same or similar work in the same space, not the concept of an industrial robot in which a robot and a person work in the same workspace.

In the prior art prior to the invention of the present application, the axis angle θ1 of each control axis J of the arm is detected during direct teaching of the robot, and the standard point (K ), obtains the current position Pr, and generates a position command value Tcom having the target position Pcom as a position obtained by projecting the current position Pr onto a predetermined movement path (movement direction), and generating this position command value Tcom. ) to drive each control axis J based on the deviation between the command angle θ com corresponding to the position command value Tcom for each control axis J and the detected axis angle θ, or the position command value This is repeated until the deviation between (Tcom) and the detected axis angle (θ) and the current position (Pr) of the corresponding standard point (K) is less than or equal to a predetermined value. A method for teaching location information to a robot is disclosed.

Another prior art relates to a robot for direct teaching capable of collision detection, and more specifically, an F/T sensor is additionally mounted on an intermediate axis of a robot having at least 6 degrees of freedom, and the F/T sensor is provided at the front end of the robot. By using the F/T sensor mounted on the intermediate axis along with the T sensor, it satisfies the operator's desire to directly move the intermediate axis in teaching and at the same time enables convenient direct teaching, and the F/T sensor at the front of the robot If the value and the value of the F/T sensor of the intermediate axis do not correlate with each other, a robot technology for direct teaching capable of collision detection is disclosed that guarantees the safety of the operator by accepting it as an impact signal and stopping the robot.

Publication No. 10-2018-0059888 Publication of Patent Publication 10-2010-0096908

In the present invention, existing industrial robots or some automated robots can work while controlling some torque. However, it is not easy to control momentary torque according to the robot's movement speed and the type of work, and it is also difficult to transfer information about how to control torque to the robot.

As various artificial intelligence tools are developed with the development of the 4th industry, robots are living with us. Robots, which were previously used for welding, cutting and transport, are now gradually entering our lives, such as delivering coffee and delivering certificates.

By the way, since robots come into our lives and replace people's work, there is a very convenient point, but it has many risk factors. People's bodies aren't made of steel, and they don't always put strength on their bodies. Therefore, even if a person bumps into a person in a narrow space, there is little chance of injury in an accident. However, there are many cases where we get injured when we bump into an object while walking. This means that an object is always positioned with its own stiffness, and a person has low stiffness first, and if necessary, the stiffness can be slightly increased by applying force to the muscles. Therefore, when a human and a robot live together, the strong rigidity of the robot may injure a human, and it is necessary to adjust the rigidity to perform the same work as a human.

Currently, most robots operate with motors, and there are robots that operate with other hydraulic and pneumatic pressures. It is difficult to input a value or an instruction value, so it is difficult when a robot and a person work together as described above.

The invention of the present application is intended to solve the above problems using the following problem solving means.

one or more position sensors for detecting a position movement for a worker's work and/or a change in position and posture of an end effector; and

one or more EMG sensors for measuring muscle force used during the worker's work according to the type of the worker's work; and

Position-based robot job information data storage step of position-based robot job information data for storing the EMG based on the position detected by the position sensor and the EMG measured by the EMG sensor and/or

Time-based robot job information data storage step of time-based robot job information data for storing position and electromyogram based on time; and

Downloading robot work teaching information for transmitting the position information and EMG information to a work robot performing the same work as the worker; and

It provides a teaching method of a robot using an EMG sensor, characterized in that the robot is operated by selecting one of position-based robot work information data and time-based robot work information data during operation of the robot.

In addition, the one or more position sensors may be any one or more of an acceleration sensor, a position sensor, a lidar sensor, an image sensor, a wearable type position sensor, and a skeletal robot controller, and measure a change in position of at least the tip of the work tool It provides a teaching method of a robot using an EMG sensor, characterized in that it can do.

In addition, when the robot is an articulated robot or a pararobot, any one of the one or more position sensors corresponds one to one with the position of the work tool attached to the robot or the end effector of the robot. It provides a teaching method of the robot used.

In addition, when the robot is an articulated robot or a pararobot, one of the one or more EMG sensors corresponds one-to-one with a work tool attached to the robot or an end-effector driving unit of the robot. A teaching method for robots is provided.

In addition, when the robot is an articulated robot or a para robot, the physical configuration such as joint structure and link length of the robot is different from that of the operator, so that the result value of the position sensor for measuring the operator's movement corresponds to each joint and link of the robot. If this cannot be done, only the position of the work tool attached to the robot or the end effector of the robot is exactly matched, and the rest of the position sensors are used as reference values.

In addition, when the robot is an articulated robot or a pararobot, the number of one or more EMG sensors and the number of joint drive units and/or rotation drive units of the robot do not match, so information from all EMG sensors corresponds to the drive unit of the robot. If this cannot be done, the work tool attached to the robot or the end effector of the robot and the corresponding drive unit correspond to the EMG sensor, and the non-corresponding EMG sensor value is maintained only as a reference value. Provides teaching methods.

In addition, the information of the one or more EMG sensors corresponds to the driving unit of the robot as much as possible, and the position control information by inverse kinematics of the robot calculated by the position sensor information is used to drive the driving unit of the robot to determine the position. In the control, the current of the driving unit corresponding to the EMG sensor is controlled based on the teaching value of the EMG sensor, and when the current control value based on the EMG sensor is a small value that cannot drive the robot normally, the robot A method for teaching a robot using an EMG sensor, characterized in that it uses a set minimum current value that can operate normally.

In addition, the determination of a small value in which the current control value based on the EMG sensor cannot drive the robot normally means a current value that can handle the payload of the robot including the work tool attached to the robot in the corresponding drive unit. It provides a teaching method of a robot using an EMG sensor, characterized in that.

In addition, when the payload can be changed by replacing the work tool, a load cell capable of measuring the payload is further provided in the end effector of the robot. Provided is a method for teaching a robot using an electromyographic sensor. .

In addition, a method for teaching a robot using an EMG sensor is provided, wherein location information and EMG values of a tool used by a worker are simultaneously measured and used as task teaching information for the robot.

In addition, the measured position information is used as a reference value for path creation of the robot end effector, and the EMG value is used as a reference value for force generated in the robot end effector. to provide.

In addition, a robot teaching method using an EMG sensor is provided, wherein the speed of an end effector of the robot is controlled in proportion to the EMG value.

In addition, a method for teaching a robot using an EMG sensor is provided, wherein the moving speed of the robot can be calculated using the following equation from the location information and the EMG value.

식(1)

Equation (1)

식(2)

Equation (2)

는 로봇 암의 움직이는 속도,

는 근전도 센서의 힘값, T는 상기 로봇암의 위치 편차를 제어를 통하여 위치 편차를 0으로 만드는데 필요한 시간X is the position deviation,

is the moving speed of the robot arm,

is the force value of the EMG sensor, T is the time required to make the position deviation of the robot arm zero through control

According to the present invention, according to the configuration as described above, even in the precise work performed by a person, the robot provides the location information of the human operator's work tools and the magnitude of the force used at that time using the electromyography sensor, thereby helping the human work and More similar work can be done, and there is an effect of being able to implement human work more easily.

1 is a view of setting a working position of a robot using a teaching pad prior to the present invention.
Figure 2 is a view of inputting the working position of the robot without a teaching pad by directly moving the joints of the robot prior to the present invention.
Figure 3 shows a person working with a drill in metal.
Figure 4 is a diagram of the robot working with a drill in metal.
5 is a view of a person working with a drill in acrylic.
Figure 6 is a view of the robot working with a drill in acrylic.
7 is a configuration diagram for attaching an EMG sensor to a person's arm and storing information on the position of a working end to be used for teaching a robot.
8 is a configuration diagram for capturing the position of a human operator's work tool with a camera or the like.

1 is a diagram of setting the working position of a robot using a teaching pad prior to the present invention. When the first industrial robot was released, a plurality of welding positions, standby positions, etc. It is a drawing to control the robot by input.

Figure 2 is a diagram in which the robot's working position is input without a teaching pad by directly moving the joints of the robot prior to the present invention. Once the position input by the teaching pad is input, it can be continuously used. It was uncomfortable. 2 is an improvement of this point, and in a teaching mode in which the robot can move by external force, the operator inputs the position while moving the position of the robot point by point, so that the user can give work instructions even if the user does not understand the robot coordinate system. There are advantages to being able to

FIG. 3 is a diagram of a person working with a drill on metal, and FIG. 4 is a diagram showing a robot working on metal using a drill. In these tasks, there is little difference between what humans do and what robots do.

5 is a diagram showing a person working with a drill on acrylic, and FIG. 6 is a diagram showing a robot working using a drill on acrylic. If the force is not well controlled during work, the acrylic may break as shown in FIG. If the blade of the drill is sharp and friction heat is not generated, robots can work without any problems, but in other cases, unlike humans, the work may not work well. It is possible to install a camera outside to monitor the working condition moment by moment and measure and control the degree of warpage of the acrylic, but this also has an uneconomical aspect.

In order to solve this problem, FIG. 7 is a configuration diagram for attaching an EMG sensor to a person's arm and storing information on the position of the working end to be used for teaching the robot. The tip of the drill bit may be measured using a displacement sensor, an acceleration sensor, or a wearable input unit. That is, during the initial drilling operation, even if the rotational speed of the drill is high, the forwarding speed toward the acrylic side can be controlled by controlling the forwarding speed of the robot arm according to the working position, thereby controlling the working speed according to the position of the drill.

Incidentally, in the case of using materials with low strength such as acrylic in drilling work, it is necessary to increase the rotation of the drill bit in the beginning and drill at a slow speed to work slowly, and as the work progresses, the drill bit advances. You can speed up your work to get it done. On the other hand, if the robot is to drive a nail into a tree, it must strike with a hammer at maximum speed at the point where the nail head is located. When the nail is driven to a certain depth, the position of the head of the nail is calculated again and the robot is controlled to hit the hammer at the maximum speed at the position of the head of the nail.

When carving a tree with a knife, apply strong force at first, but at the end, release the force and lift it slightly to remove the piece of wood. As mentioned above, when using tools, including when people work directly with their hands and feet, there is a difference in how they are used depending on the type of tool and the object for which the tool is used. In other words, there is a difference in the method of giving power.

With the development of technology that was not thought of in the past, the era where robots can take the place of the above tasks is coming. The invention of the present application is intended to control not only the position control of the robot, but also the position and work speed of the robot together as one means of a method for recognizing human motion of a robot instead of human work.

On the other hand, there are similarities to operating a robot or controlling the operation of a robot using an existing wearable controller, but there are differences when viewed in more detail.

That is, a feature of the present invention is that force control is performed even in a static operation.

Taking position control as an example, if the robot has a position error of 10 mm, in order to eliminate the position error, move the robot by multiplying the speed (V) by the movement time (T) as in the formula of dX = V*T is to set to 0.

However, here, the speed V is changed along a trapezoidal speed profile so that the robot can move stably for the maximum speed or section movement built into the robot.

Therefore, there is no means to change the speed V according to external operating conditions. It can be changed by parameter change or program, but there is no standard input value.

The end effector of the robot tries to move by removing the position error of 10 mm, but suppose there is a concrete wall in front of it. It is a common occurrence in our daily life. When trying to open a door, it is common for the distance to have a 10mm error. I try to pick up a cup, but there are many cups with a difference of more than 10 mm in size. In people's daily lives, it is impossible to set everything accurately. There is an error, and it is an object of the present invention to provide a control means that allows the robot to operate despite the error.

Again, going back to the motion of the robot, it is impossible to overcome dX with V*T. At this time, you can think about how a person would have responded. At first, you will try to push it as hard as you can, but if you do not push it with your maximum force, even if there is a command to move 10mm further forward, you will just relax and stay in contact.

The invention of the present application is intended to use the EMG value of a worker input from the outside for such control. It is intended to use the EMG value of the operator as a teaching value for controlling the V. By doing so, it will be possible to avoid a situation in which the robot operates to overcome a position error of 10 mm infinitely in the above situation and the robot may break down. In a situation where there is a position error of 10 mm and the robot cannot remove the position error by moving, it provides a means for not performing an operation to remove the position error.

If the robot has a positional error of 10 mm and there is a Styrofoam in front of it, the robot will pass through the Styrofoam and move forward by 10mm.

However, if V is reduced to a very small value in dX = V*T, the robot may not be able to penetrate the styrofoam and remain stationary.

The method for controlling the acrylic drilling operation is to lower the current of the forward driving part of the robot so that it can be pushed backward by the acrylic, since the initial drilling operation does not yet drill a hole in the acrylic at a position where the movement is small, and gradually by the drill It has the feature of being able to work by increasing the current of the driving part in the forward direction when the acrylic is pierced.

delete

8 is a configuration diagram of capturing the position of a human operator's work tool with a camera or the like. It goes without saying that image processing technology has been developed with the recent development of AI, and it is possible to precisely position the tip of a worker's tool using this image processing technology.

That is, the present invention simultaneously measures the positional information and EMG values of the tools used by the operator and uses them as task teaching information of the robot, thereby teaching the robot to work without setting which tool to use for which object. is to provide the means.

The speed control of the robot can all be different depending on the specifications of the servomotor constituting the robot, so the EMG value is multiplied by a proportional constant so that the torque of the servomotor used by the robot and the force according to the reduction ratio are proportional to the EMG value to be used for robot control. It can be used by multiplying by a proportional constant so that it can be used, or by adjusting the amplification ratio of the EMG value.

The problem-solving means exhibiting the above operational effects are as follows.

The invention of the present application is intended to solve the above problems using the following problem solving means.

One or more position sensors for detecting a change in position movement and/or tooling position for a worker's work; and

one or more EMG sensors for measuring muscle force used during the worker's work according to the type of the worker's work; and

Position-based robot job information data storage step of position-based robot job information data for storing the EMG based on the position detected by the position sensor and the EMG measured by the EMG sensor; and

Time-based robot job information data storage step of time-based robot job information data for storing position and electromyogram based on time; and

Downloading robot work teaching information for transmitting the position information and EMG information to the same working robot as the worker; and

It provides a teaching method of a robot using an EMG sensor, characterized in that the robot is operated by selecting one of position-based robot work information data and time-based robot work information data during operation of the robot.

In addition, the one or more position sensors may be any one or more of an acceleration sensor, a position sensor, a lidar sensor, an image sensor, a wearable type position sensor, and a skeletal robot controller, and measure a change in position of at least the tip of the work tool It provides a teaching method of a robot using an EMG sensor, characterized in that it can do.

In addition, when the robot is an articulated robot or a pararobot, one of the one or more position sensors is a robot using an electromyographic sensor, characterized in that one to one correspondence with the position of the work tool or fingertip attached to the robot It provides a teaching method of

In addition, when the robot is an articulated robot or a pararobot, one of the one or more EMG sensors is in one-to-one correspondence with a work tool attached to the robot or a driving unit where a finger is located. A teaching method for robots is provided.

In addition, when the robot is an articulated robot or a pararobot, there is a difference in physical length such as length and rotation radius between the one or more position sensors and the joints of the robot, so that all positions of the one or more position sensors are determined by the robot. If it is not possible to correspond to the joints of the robot, a work tool or a finger attached to the robot accurately matches the position, and the rest of the position sensors provide a robot teaching method using an EMG sensor, characterized in that using reference values.

In addition, when the robot is an articulated robot or a pararobot, the number of the one or more EMG sensors and the number of joint drive units and/or rotation drive units of the robot do not match, so information from all EMG sensors is transferred to the drive unit of the robot. If it is not possible to do so, a robot teaching method using an EMG sensor characterized in that the EMG sensor is matched from a driving unit that corresponds to a work tool or finger attached to the robot, and the value of the EMG sensor that does not correspond is maintained only as a reference value. provides

In addition, the information of the one or more EMG sensors corresponds to the driving unit of the robot as much as possible, and the position control information by inverse kinematics of the robot calculated by the position sensor information is used to drive the driving unit of the robot to determine the position. In the control, the current of the drive unit corresponding to the EMG sensor is controlled based on the teaching value of the EMG sensor, and when the current control value based on the EMG sensor is a small value that cannot drive the robot normally, the robot A method for teaching a robot using an EMG sensor is provided, characterized in that the set minimum current value that can operate normally is used.

In addition, the determination of a small value in which the current control value based on the EMG sensor cannot drive the robot normally means a current value that can handle the payload of the robot including the work tool attached to the robot in the corresponding drive unit. It provides a teaching method of a robot using an electromyography sensor, characterized in that.

In addition, when the payload can be changed by replacing the work tool, a load cell capable of measuring the payload is further provided in the end effector of the robot. Provided is a method for teaching a robot using an electromyographic sensor. .

Claims (12)

one or more position sensors for detecting a position movement for a worker's work and/or a change in position and posture of an end effector; and
one or more EMG sensors for measuring muscle force used during the worker's work according to the type of the worker's work; and
Position-based robot job information data storage step of position-based robot job information data for storing the EMG based on the position detected by the position sensor and the EMG measured by the EMG sensor and/or
Time-based robot job information data storage step of time-based robot job information data for storing position and electromyogram based on time; and
Downloading robot work teaching information for transmitting the position information and EMG information to a work robot performing the same work as the worker; and
When the robot is working, select one of the position-based robot job information data and the time-based robot job information data to operate the robot,
The information of the one or more EMG sensors corresponds to the driving unit of the robot, and in controlling the position using the position control information by inverse kinematics of the robot calculated by the position sensor information to drive the driving unit of the robot , The current of the drive unit corresponding to the EMG sensor is controlled based on the teaching value of the EMG sensor, and if the current control value based on the EMG sensor is a small value that cannot normally drive the robot, the robot may operate normally. It uses the minimum current value that can be set,
Determination of a small value at which the current control value based on the EMG sensor cannot drive the robot normally means a current value that can handle the payload of the robot including the work tool attached to the robot in the driving unit,
A method for teaching a robot using an electromyographic sensor, characterized in that a load cell capable of measuring the payload is further provided in an end effector of the robot when the payload can be changed by replacing the work tool. delete delete delete delete delete delete delete delete delete delete delete

Images