Back to results
US 12,005,582 B2Granted
Controller and control system
AssigneeFanuc Corporation
InventorTakashi Satou
Life of this patent
Estimating the term…
Abstract
The controller acquires a force applied to a manipulator of a robot, to generate, based on the acquired data, force state data containing information related to the force applied to the manipulator and control command adjustment data indicating an adjustment behavior of a control command related to the manipulator as state data, thereby executing, based on the generated state data, a process of machine learning related to the adjustment behavior of the control command related to the manipulator.
Claims10
A controller configured to control an industrial robot having a function of detecting at least a force applied to a manipulator, the controller comprising:
- at least one memory;
- and
- at least one processor configured to:
- obtain at least the force applied in a feed direction of the manipulator and a moment about an axis of the feed direction of the manipulator, and a control command of the industrial robot as acquisition data, wherein the feed direction is a fitting direction of a part grasped by the manipulator to a predetermined position of another part;
- generate, based on the acquisition data, data including information indicating the force applied to the manipulator in only the feed direction and information indicating the control command of the industrial robot;
- and
- generate, based on the generated data, a neural network configured to output a value to determine an adjustment action of the control command for a change in at least one of a force, speed or torque applied in only the feed direction of the manipulator,
- wherein a change is made of a command value for at least one of the force, speed or torque of the control command related to the feed direction of the manipulator when the determined value exceeds a predetermined threshold during the control of the industrial robot.
5 more elements
A controller configured to control an industrial robot having a function of detecting at least a force applied to a manipulator, the controller comprising:
- at least one memory;
- and
- at least one processor configured to:
- obtain at least the force applied in a feed direction of the manipulator and a moment about an axis of the feed direction of the manipulator, and a control command of the industrial robot as acquisition data, wherein the feed direction is a fitting direction of a part grasped by the manipulator to a predetermined position of another part;
- generate, based on the acquisition data, data including information indicating the force applied to the manipulator in only the feed direction and information indicating the control command of the industrial robot;
- and
- generate, based on the generated data, a learning model of a neural network obtained by applying reinforcement learning to determine a value for an adjustment behavior of the control command for a change in at least one of a force, speed or torque related to the manipulator with respect to a state of the force applied in only the feed direction of the manipulator,
- wherein a change is made of a command value for at least one of the force, speed or torque of the control command related to the feed direction of the manipulator when the determined value exceeds a predetermined threshold during the control of the industrial robot.
5 more elements
A controller configured to control an industrial robot having a function of detecting at least a force applied to a manipulator, the controller comprising:
- at least one memory;
- and
- at least one processor configured to:
- obtain at least the force applied in a feed direction of the manipulator and a moment about an axis of the feed direction of the manipulator, and a control command of the industrial robot as acquisition data, wherein the feed direction is a fitting direction of a part grasped by the manipulator to a predetermined position of another part;
- generate, based on the acquisition data, data including information indicating the force applied to the manipulator in only the feed direction and information indicating the control command of the industrial robot;
- and
- store in the at least one memory a learning model obtained by applying reinforcement learning to determine a value for an adjustment behavior of the control command for a change in at least one of a force, speed or torque related to the manipulator with respect to a state of the force applied in only the feed direction of the manipulator by considering a work time of the manipulator based on the control command;
- and
- estimate, based on the generated data, the adjustment behavior of the control command related to the manipulator using the learning model stored in the at least one memory,
- wherein a change is made of a command value for at least one of the force, speed or torque of the control command related to the feed direction of the manipulator when the determined value exceeds a predetermined threshold during the control of the industrial robot.
7 more elements
A controller configured to control an industrial robot having a function of detecting at least a force applied to a manipulator, the controller comprising:
- at least one memory;
- and
- at least one processor configured to:
- obtain at least the force applied in a feed direction of the manipulator and a moment about an axis of the feed direction of the manipulator, and a control command of the industrial robot as acquisition data, wherein the feed direction is a fitting direction of a part grasped by the manipulator to a predetermined position of another part;
- generate, based on the acquisition data, data including information indicating the force applied to the manipulator in only the feed direction and information indicating the control command of the industrial robot;
- and
- obtain a value of an adjustment action of the control command for a change in at least one of a force, speed or torque applied in only the feed direction of the manipulator based on an output from a neural network when the generated data is input to a pre-generated neural network,
- wherein a change is made of a command value for at least one of the force, speed or torque of the control command related to the feed direction of the manipulator when the obtained value of the adjustment action exceeds a predetermined threshold during the control of the industrial robot.
5 more elements
Dependencies are inferred from the claim text.