Machine learning device, industrial machine cell, manufacturing system, and machine learning method for learning task sharing among plurality of industrial machines
A machine learning device, which performs a task using a plurality of industrial machines and learns task sharing for the plurality of industrial machines, includes a state variable observation unit which observes state variables of the plurality of industrial machines; and a learning unit which learns task sharing for the plurality of industrial machines, on the basis of the state variables observed by the state variable observation unit.
1. A machine learning device for performing a task using a plurality of industrial machines and learning task sharing for the plurality of industrial machines, the device comprising:
a processor configured to
observe state variables of the plurality of industrial machines, and
learn task sharing for the plurality of industrial machines, on the basis of the observed state variables,
wherein when one of the plurality of industrial machines stops and the task is shared by other industrial machines of the plurality of industrial machines, the processor is configured to observe a change in a volume of production by the plurality of industrial machines and observe a task load on the stopped one of the plurality of industrial machines as zero, and
the plurality of industrial machines perform the task based on the learned task sharing,
wherein
the processor is configured to observe the state variables including
a task time from start to end of a series of tasks repeatedly performed by the plurality of industrial machines, and
a task load on each of the plurality of industrial machines in an interval from the start to the end of the series of task, the task load being a power consumption of said each of the plurality of industrial machines, and
the processor is configured to
set a negative reward when the observed task load exceeds a predetermined allowable range,
set a positive reward when the observed task time is shortened,
set no reward when the observed task time is not shortened, and
update an action value table that determines a value of work sharing for the plurality of industrial machines based on the reward.
2. The machine learning device according to claim 1 , wherein the processor is configured to decide and issue, as a command, a sharing detail of the task for the plurality of industrial machines by referring to the learned task sharing.
3. The machine learning device according to claim 2 , wherein
the machine learning device is connected to each of the plurality of industrial machines via a network,
the processor is configured to obtain the state variables of the plurality of industrial machines via the network, and
the processor is configured to send the sharing detail of the task to the plurality of industrial machines via the network.
4. The machine learning device according to claim 1 , wherein
the processor is configured to observe the state variables further including at least one of
a change in the volume of production by the plurality of industrial machines in an upstream process, or
a change in the volume of production by the plurality of industrial machines upon the stop of the one of the plurality of industrial machines for maintenance performed periodically.
5. The machine learning device according to claim 1 , wherein
when said one of the plurality of industrial machines stops and the task is shared by the other industrial machines of the plurality of industrial machines, the processor is configured to learn task sharing for
maintaining a volume of production by the other industrial machines of the plurality of industrial machines,
averaging a load on each of the other industrial machines of the plurality of industrial machines, and
maximizing a volume of the task performed by the other industrial machines of the plurality of industrial machines.
6. The machine learning device according to claim 1 , wherein
each of the plurality of industrial machines comprises a robot, and
the plurality of robots is configured to perform the task on the basis of the learned task sharing.
7. A machine learning device for performing a task using a plurality of industrial machines and learning task sharing for the plurality of industrial machines, the device comprising:
a processor configured to
observe state variables of the plurality of industrial machines, and
learn task sharing for the plurality of industrial machines, on the basis of the observed state variables,
wherein when one of the plurality of industrial machines stops and the task is shared by other industrial machines of the plurality of industrial machines, the processor is configured to observe a change in a volume of production by the plurality of industrial machines and observe a task load on the stopped one of the plurality of industrial machines as zero, and
the plurality of industrial machines perform the task based on the learned task sharing,
wherein
the processor is configured to observe the state variables including
a task time from start to end of a series of tasks repeatedly performed by the plurality of industrial machines, and
a task load on each of the plurality of industrial machines in an interval from the start to the end of the series of task, the task load being a power consumption of said each of the plurality of industrial machines, and
the processor is configured to
set a negative reward when the observed task load exceeds a predetermined allowable range,
set a positive reward when the observed task time is shortened,
set no reward when the observed task time is not shortened, and
update a value function for determining a value of task sharing for the plurality of industrial machines on the basis of the observed state variables and the reward.
8. The machine learning device according to claim 1 , wherein the machine learning device further comprises a neural network.
9. An industrial machine cell comprising the plurality of industrial machines; and the machine learning device according to claim 1 .
10. A manufacturing system comprising a plurality of industrial machine cells according to claim 9 , wherein
the machine learning devices are provided in correspondence with the industrial machine cells, and
the machine learning devices provided in correspondence with the industrial machine cells are configured to share or exchange data with each other via a communication medium.
11. The manufacturing system according to claim 10 , wherein the machine learning devices are located on a cloud server.
12. A machine learning method for performing a task using a plurality of industrial machines and learning task sharing for the plurality of industrial machines, the method comprising:
observing state variables of the plurality of industrial machines;
learning task sharing for the plurality of industrial machines, on the basis of the observed state variables; and
when one of the plurality of industrial machines stops and the task is shared by other industrial machines of the plurality of industrial machines,
observing a change in a volume of production by the plurality of industrial machines and observing a task load on the stopped one of the plurality of industrial machines as zero,
wherein
the plurality of industrial machines perform the task based on the learned task sharing,
said observing comprises observing the state variables including
a task time from start to end of a series of tasks repeatedly performed by the plurality of industrial machines, and
a task load on each of the plurality of industrial machines in an interval from the start to the end of the series of task, the task load being a power consumption of said each of the plurality of industrial machines, and
the method further comprises
setting a negative reward when the observed task load exceeds a predetermined allowable range,
setting a positive reward when the observed task time is shortened,
setting no reward when the observed task time is not shortened, and
updating an action value table that determines a value of work sharing for the plurality of industrial machines based on the reward.
13. The machine learning method according to claim 12 , wherein
said observing the state variables comprises observing the state variables including at least one of
an achievement level of a series of tasks performed by the plurality of industrial machines, or
a difference in task volume in each of the plurality of industrial machines.
14. The machine learning device according to claim 1 , wherein
the processor is configured to observe the state variables including at least one of
an achievement level of a series of tasks performed by the plurality of industrial machines, or
a difference in task volume in each of the plurality of industrial machines.