Machine learning device that learns shocks to teaching device, shock prevention system of teaching device, and machine learning method

Patent metadata

Field	Value
Publication number	US-2018197112-A1
Application number	US-201815865920-A
Country	US
Kind code	A1
Filing date	Jan 9, 2018
Priority date	Jan 10, 2017
Publication date	Jul 12, 2018
Grant date	—

Abstract

A machine learning device, which learns shocks to a teaching device, includes a state observation unit which observes data based on an inclination of the teaching device or a present position of the teaching device; a label obtaining unit which obtains a label based on a shock received by the teaching device; and a learning unit which generates a learning model based on an output of the state observation unit and an output of the label obtaining unit.

First claim

What is claimed is:

1 . A machine learning device, which learns shocks to a teaching device, comprising:
a state observation unit which observes data based on an inclination of the teaching device or a present position of the teaching device; a label obtaining unit which obtains a label based on a shock received by the teaching device; and a learning unit which generates a learning model based on an output of the state observation unit and an output of the label obtaining unit.

2 . The machine learning device according to claim 1 , wherein
the state observation unit further observes data based on at least one of speed at which buttons of the teaching device are pressed, content of an operation involving the teaching device, length of time of the operation, a time zone of the operation, and an identity of an operator of the teaching device.

3 . The machine learning device according to claim 1 , wherein
the label obtaining unit obtains a label based on whether or not the shock received by the teaching device is greater than a threshold value.

4 . The machine learning device according to claim 1 , wherein the learning unit comprises:
an error calculation unit which calculates an error based on outputs of the state observation unit and the label obtaining unit; and a learning model update unit which updates a learning model for determining an error in a shock to the teaching device, based on outputs of the state observation unit and the error calculation unit.

5 . The machine learning device according to claim 1 , wherein
the machine learning device comprises a neural network.

6 . The machine learning device according to claim 1 , wherein
the machine learning device is connectable with at least one other machine learning device and exchanges or shares with the at least one other machine learning device a learning model generated by the learning unit of the machine learning device.

7 . A shock prevention system of a teaching device, comprising:
the machine learning device according to claim 1 ; the teaching device, the machine learning device learning shocks to the teaching device; and an output utilization unit which utilizes an output of the machine learning device, wherein
the output utilization unit operates in such a manner as to prevent a shock to the teaching device.

8 . The shock prevention system of the teaching device according to claim 7 , wherein
the output utilization unit outputs an alarm or advises caution before the teaching device receives a shock, based on an output of the machine learning device, to prevent the shock to the teaching device.

9 . The shock prevention system of the teaching device according to claim 8 , wherein
the output utilization unit uses at least one of an alarm using light or sound, voice, indication on a display, and vibration before the teaching device receives a strong shock greater than a certain threshold value to prevent the strong shock to the teaching device.

10 . A machine learning method of learning shocks to a teaching device, comprising:
observing state data based on an inclination of the teaching device or a present position of the teaching device; obtaining a label based on a shock received by the teaching device; and generating a learning model based on the observed state data and the label.

11 . The machine learning method according to claim 10 , wherein
the state data further comprises data based on at least one of speed at which buttons of the teaching device are pressed, content of an operation involving the teaching device, length of time of the operation, a time zone of the operation, and an identity of an operator of the teaching device.

12 . The machine learning method according to claim 10 , wherein
the label is based on whether or not the shock received by the teaching device is greater than a threshold value.

13 . The machine learning method according to claim 10 , wherein the generating the learning model comprises:
calculating an error based on the observed state data and the label; and updating a learning model for determining an error in a shock to the teaching device, based on the state data and the calculated error.

Assignees

• Fanuc Corp

Inventors

• Sugiyama Yuusuke

Classifications

• G05B2219/36162

  Pendant control box · CPC title

• G05B2219/39443

  Portable, adapted to handpalm, with joystick, function keys, display · CPC title

• G05B2219/33098

  Several nc machines, dnc, cnc · CPC title

• G05B19/409

  characterised by using manual data input [MDI] or by using control panel, e.g. controlling functions with the panel; characterised by control panel details or by setting parameters (G05B19/408, G05B19/4093 take precedence) · CPC title

• B25J9/163

  learning, adaptive, model based, rule based expert control · CPC title