Learning device, controller, and control system

1. A controller which machines a workpiece by controlling a pressing force of a tool mounted on the tip of an arm of a robot, the controller comprising:

a processor configured to:

acquire a present state of the robot as a state variable;

acquire, as a label, a detected value of a force sensor attached to the arm and configured to detect necessary data for control of the pressing force; and

generate a learning model indicative of the correlation between the state variable acquired in a no-load state and the label acquired under the state variable acquired in the no-load state and to estimate the detected value of the force sensor corresponding to the present state of the robot, based on the learning model,

wherein

the processor is further configured to control the pressing force by using the detected value of the force sensor acquired in the present state of the robot acquired in a loaded state and the estimated detected value of the force sensor based on the present state of the robot acquired in the loaded state,

the tool is configured to machine the workpiece by vibration, and

the present state of the robot includes at least a vibrational frequency of the tool.

2. The controller according to claim 1 , wherein the detected value of the force sensor is a force acting on the force sensor and/or a vibrational frequency acting on the force sensor.

3. The controller according to claim 1 , wherein the present state of the robot is at least one of factors including the operating direction, speed, acceleration, jerk, and posture of the robot and the rotational and vibrational frequencies of a drive unit of the tool.

4. The controller according to claim 1 , wherein the robot performs deburring and/or polishing of the workpiece.

5. A control system which comprises a robot configured to machine a workpiece by means of a pressing force of a tool mounted on the tip of an arm and a controller configured to control the operation of the robot,

the controller comprising a processor configured to:

acquire a present state of the robot as a state variable;

acquire, as a label, a detected value of a force sensor attached to the arm and detect necessary data for control of the pressing force; and

generate a learning model indicative of the correlation between the state variable acquired in a no-load state and the label acquired under the state variable acquired in the no-load state and to estimate the detected value of the force sensor corresponding to the present state of the robot, based on the learning model,

wherein

the processor is further configured to control the pressing force by using the detected value of the force sensor acquired in the present state of the robot acquired in a loaded state and the estimated detected value of the force sensor based on the present state of the robot acquired in the loaded state,

the tool is configured to machine the workpiece by vibration, and

the present state of the robot includes at least a vibrational frequency of the tool.

6. A learning device comprising:

a processor configured to:

acquire a present state of a robot as a state variable;

acquire, as a label, a detected value of a force sensor attached to an arm of the robot and detect necessary data for control of a pressing force of a tool mounted on the tip of the arm; and

generate a learning model indicative of the correlation between the state variable acquired in a no-load state and the label acquired under the state variable acquired in the no-load state and to estimate the detected value of the force sensor corresponding to the present state of the robot, based on the learning model, wherein

the tool is configured to machine the workpiece by vibration, and

the present state of the robot includes at least a vibrational frequency of the tool.

7. The controller according to claim 1 , further comprising

a preprocessor configured to

preprocess for at least one of (1) data obtained from an operation program of the robot, (2) data indicative of the present states of the robot or a drive unit of the robot, or (3) data obtained from the force sensor, and

output the preprocessed data to the processor.

8. The controller according to claim 7 , wherein the preprocessor is further configured to adjust the number of sampling data by combining at least two of (1) thinning of moving averages and data, (2) reduction of the number of data by partial extraction, (3) intermediate point interpolation, or (4) increasing of data number by fixed value addition.

9. The control system according to claim 5 , wherein the controller further comprises

a preprocessor configured to

preprocess for at least one of (1) data obtained from an operation program of the robot, (2) data indicative of the present states of the robot or a drive unit of the robot, or (3) data obtained from the force sensor, and

output the preprocessed data to the processor.

10. The control system according to claim 9 , wherein the preprocessor is further configured to adjust the number of sampling data by combining at least two of (1) thinning of moving averages and data, (2) reduction of the number of data by partial extraction, (3) intermediate point interpolation, or (4) increasing of data number by fixed value addition.

11. The learning device according to claim 6 , further comprising

a preprocessor configured to

preprocess for at least one of (1) data obtained from an operation program of the robot, (2) data indicative of the present states of the robot or a drive unit of the robot, or (3) data obtained from the force sensor, and

output the preprocessed data to the processor.

12. The learning device according to claim 11 , wherein the preprocessor is further configured to adjust the number of sampling data by combining at least two of (1) thinning of moving averages and data, (2) reduction of the number of data by partial extraction, (3) intermediate point interpolation, or (4) increasing of data number by fixed value addition.