Action control method and device

• US 10,642,346 B2
• Filed: 10/20/2017
• Issued: 05/05/2020
• Est. Priority Date: 10/31/2016
• Status: Active Grant

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Hayashida, Naoko
• Primary Examiner(s): Yoon, Sae Won
• Application Number:
  US15/788,890
• Publication Number:
  US 20180120928A1
• Time in Patent Office: 928 Days
• Field of Search: 345619
• CPC Class Codes: G06F 3/011 Arrangements for interaction with the human body, e.g. for user immersion in virtual reality blind teaching G09B21/00
• CPC Class Codes: G06T 13/40 of characters, e.g. humans, animals or virtual beings

Abstract

An action control method includes when a first state of a first user in communication with a first machine avatar is detected, specifying a first action of the first machine avatar for changing a state of the first user from the first state to a second state that is more desirable than the first state based on history information indicating changes in states of users in response to actions of machine avatars or different users in communication with the users, and causing the first machine avatar to perform the first action.

Claims

1. An action control method executed by a computer, the method comprising:

detecting, on the basis of data obtained by a sensor, a first state of a first user in communication with a first machine avatar;

extracting candidate data, from action history data, in which any user was in the first state;

selecting a time period from a start time to a detection time that the detecting detects the first state of the first user;

excluding one or more actions from the candidate data, the excluded one or more actions having taken place before the selected time period;

identifying an action from the candidate data in which the one or more actions are excluded, the identified action being an action resulting in the any user changing from the first state to a second state; and

causing the first machine avatar to take an action corresponding to the identified action.

2. The action control method according to claim 1, wherein the first state is non-verbal behavior.

3. The action control method according to claim 1, wherein the second state is a desirable non-verbal behavior for responding to the identified action.

4. The action control method according to claim 1, wherein the second state is a desirable social behavior for responding to the identified action.

5. The action control method according to claim 1, wherein

the first state is detected by monitoring a particular action in a given frequency band among actions performed by the first user.

6. The action control method according to claim 1, wherein

the first state is detected by monitoring a particular action not in a given frequency band among actions performed by the first user.

7. The action control method according to claim 1, wherein the one or more excluded actions are actions in which the any user is not in a desirable state for responding to the identified action.

8. The action control method according to claim 1, wherein the time period is a time period before the detection time.

9. The action control method according to claim 1, wherein the excluding includes calculating ratios for the candidate data relating to the any user changing from the first state to the second state, and the identifying being based on the calculated ratios.

10. The action control method according to claim 1, wherein the identifying includes synthesizing a candidate action and a switching action, the candidate action being specified based on the excluding.

11. An action control device comprising:

a memory; and

a processor coupled to the memory and the processor configured to;

detect, on the basis of data obtained by a sensor, a first state of a first user in communication with a first machine avatar;

extract candidate data, from action history data, in which any user was in the first state;

select a time period from a start time to a detection time that the detecting detects the first state of the first user;

exclude one or more actions from the candidate data, the excluded one or more actions having taken place before the selected time period;

identify an action from the candidate data in which the one or more actions are excluded, the identified action being an action resulting in the any user changing from the first state to a second state; and

cause the first machine avatar to take an action corresponding to the identified action.

12. The action control device according to claim 11, wherein the first state is non-verbal behavior.

13. The action control device according to claim 11, wherein the second state is a desirable non-verbal behavior for responding to the identified action.

14. The action control device according to claim 11, wherein the second state is a desirable social behavior for responding to the identified action.

15. The action control device according to claim 11, wherein

the first state is detected by monitoring a particular action in a given frequency band among actions performed by the first user.

16. The action control device according to claim 11, wherein

the first state is detected by monitoring a particular action not in a given frequency band among actions performed by the first user.

17. The action control device according to claim 11, wherein the one or more excluded actions include actions in which the any user is not in a desirable state for responding to the identified action.

18. A non-transitory computer-readable storage medium storing an action control program that causes a computer to execute a process comprising:

detecting, on the basis of data obtained by a sensor, a first state of a first user in communication with a first machine avatar;

extracting candidate data, from action history data, in which any user was in the first state;

selecting a time period from a start time to a detection time that the detecting detects the first state of the first user;

excluding one or more actions from the candidate data, the excluded one or more actions having taken place before the selected time period;

identifying an action from the candidate data in which the one or more actions are excluded, the identified action being an action resulting in the any user changing from the first state to a second state; and

causing the first machine avatar to take an action corresponding to the identified action.