Operational control method, program, and recording media for robot device, and robot device

• US 6,697,711 B2
• Filed: 10/18/2002
• Issued: 02/24/2004
• Est. Priority Date: 02/21/2001
• Status: Expired due to Fees

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Yokono, Jun, Sabe, Kohtaro, Ohashi, Takeshi, Costa, Gabriel
• Primary Examiner(s): Cuchlinski, Jr., William A.
• Assistant Examiner(s): Marc, McDieunel, Marc, McDieunel
• Application Number:
  US10/258,110
• Publication Number:
  US 20030144764A1
• Time in Patent Office: 494 Days
• Field of Search: 700/245, 700/257, 700/260, 700/259, 700/262, 700/247, 700/249, 700/258, 318/568.1
• US Class Current: 700/245
• CPC Class Codes: G06N 3/008 based on physical entities controlled by simulated intelligence so as to replicate intelligent life forms, e.g. based on robots replicating pets or humans in their appearance or behaviour

Abstract

A robot apparatus (1) includes leg blocks (3A to 3D), head block (4), etc. as a moving part (16), a motion controller (102), learning unit (103), prediction unit (104) and a drive unit (105). When the moving part (106), any of the blocks, is operated from outside, the learning unit (103) learns a time-series signal generated due to the external operation. The motion controller (102) and drive unit (105) control together the moving part (106) based on a signal generated at the moving part (106) due to an external force applied to the robot apparatus (1) and a signal having already been learned by the learning unit (103) to make an action taught by the user. The prediction unit (105) predicts whether the moving part (106) makes the taught action according to the initial signal generated at the moving part (106) due to the applied external force. Thus, the robot apparatus (1) can learn an action taught by the user and determine an external force-caused signal to make the taught action.

Claims

1. A method of controlling the operation of a robot apparatus having a moving part, the method comprising steps of:

learning a time-series signal generated at the moving part correspondingly to an operation of the moving part; and

controlling the moving part according to the time-series signal learned in the learning step;

whereby learning said time-series signal is performed in response to an external force applied to said moving part during a learning operation.

2. The method as set forth in claim 1, wherein the signal indicates a position of the moving part.

3. The method as set forth in claim 2, wherein:

the robot apparatus includes means for detecting, when controlling the motion of the moving part, a position to which the moving part has moved; and

the signal is a detection signal from the position detecting means.

4. The method as set forth in claim 1, wherein in the motion controlling step, a motion is predicted based on an initial signal generated, after the learning, at the moving part due to the external force applied to the moving part, the predicted value is compared with the time-series signal learned in the learning step, and the moving part is controlled on the result of the comparison.

5. The method as set forth in claim 1, wherein in the learning step, the time-series signal is learned by a neural network in which a signal is inputted toward an input layer, hidden layer and output layer.

6. The method as set forth in claim 5, wherein the neural network is a recurrent neural network having a loop for a feedback from the output layer toward the input layer.

7. A program for controlling the operation of a robot apparatus, the program allowing a robot apparatus to execute steps of:

learning a time-series signal generated at the moving part correspondingly to an operation of the moving part; and

controlling the moving part according to the time-series signal learned in the learning step;

whereby learning said time-series signal is performed in response to an external force applied to said moving part during a learning operation.

8. A recording medium having recorded therein a program for controlling the operation of the robot apparatus having a moving part, the program allowing the robot apparatus to execute steps of:

learning a time-series signal generated at the moving part correspondingly to an operation of the moving part; and

controlling the moving part according to the time-series signal learned in the learning step;

whereby learning said time-series signal is performed in response to an external force applied to said moving Part during a learning operation.

9. A robot apparatus having a moving part, comprising:

means for learning a time-series signal generated at the moving part correspondingly to an operation of the moving part; and

motion controlling means for controlling the moving part according the time-series signal learned in the learning step;

whereby learning said time-series signal is performed in response to an external force applied to said moving part during a learning operation.

10. The apparatus as set forth in claim 9, wherein the signal indicates a position

of the moving part.

11. The apparatus as set forth in claim 10, wherein:

the robot apparatus includes means for detecting, when controlling the motion of the moving part, a position to which the moving part has moved; and

the signal is a detection signal from the position detecting means.

12. The apparatus as set forth in claim 9, further comprising means for predicting an action from an initial signal generated, after the learning, at the moving part due to the external force applied to the moving part;

the motion controlling means comparing the predicted value from the predicting means and the time-series signal learned by the learning means with each other and controlling the moving part on the basis of the result of comparison.

13. The apparatus as set forth in claim 9, wherein the learning means learns

the time-series signal by means of a neural network in which a signal is inputted toward an input layer, hidden layer and output layer.

14. The apparatus as set forth in claim 13, wherein the neural network is a recurrent neural network having a loop for a feedback from the output layer toward the input layer.