Control device and learning device
A control device that outputs a command for a robot includes a machine learning device that learns a command for the robot. The machine learning device includes a state observation unit that observes a state of the robot and a state of a person present in a peripheral area of the robot, as state variables representing a current state of an environment, a determination data acquisition unit that acquires determination data representing an interference state between the robot and the person, and a learning unit that learns the state of the robot, the state of the person present in the peripheral area of the robot, and the command for the robot obtained by associating the state of the robot and the state of the person present in the peripheral area of the robot by using the state variables and the determination data.
1. A control device that outputs a command for a robot, the control device comprising:
a processor configured to:
observe a state of the robot including a posture and a moving speed of a manipulator of the robot and a state of a person present in a peripheral area of the robot including an entering direction and a traffic line of the person, as state variables representing a current state of an environment,
acquire determination data representing an interference state between the robot and the person, and learn the state of the robot, learn the state of the person present in the peripheral area of the robot, and learn the command for the robot by associating the state of the robot and the state of the person present in the peripheral area of the robot, by using the state variables and the determination data.
2. The control device according to claim 1 , wherein the determination data includes at least any one of whether or not the robot and the person have collide with each other, a relative distance between the robot and the person, a magnitude of the collision force, and throughput.
3. The control device according to claim 1 , wherein the processor is further configured to:
obtain a reward related to an interference state between the robot and the person, and
update a function representing a value of the command for the robot with respect to the state of the robot and the state of the person present in the peripheral area of the robot, by using the reward.
4. The control device according to claim 1 , wherein the processor is further configured to calculate the state variables and the determination data in a multilayer structure calculation.
5. The control device according to claim 1 , wherein the processor is further configured to output a command value representing the command for the robot, based on a learning learned by the processor.
6. The control device according to claim 1 , wherein the processor is further configured to learn the command for the robot by using the state variables and the determination data, the state variable and the determination data being obtained from a plurality of robots.
7. The control device according to claim 1 , wherein the processor exists in a cloud server.
8. A learning device that learns a command for a robot, the learning device comprising:
a processor configured to:
observe a state of the robot including a posture and a moving speed of a manipulator of the robot and a state of a person present in a peripheral area of the robot including an entering direction and a traffic line of the person, as state variables representing a current state of an environment;
acquire determination data representing an interference state between the robot and the person; and
learn the state of the robot, learn the state of the person present in the peripheral area of the robot, and learn the command for the robot by associating the state of the robot and the state of the person present in the peripheral area of the robot, by using the state variables and the determination data.