Method for dialogue between a machine, such as a humanoid robot, and a human interlocutor utilizing a plurality of dialog variables and a computer program product and humanoid robot for implementing such a method

Patent metadata

Field	Value
Publication number	US-10127226-B2
Application number	US-201415025540-A
Country	US
Kind code	B2
Filing date	Sep 29, 2014
Priority date	Oct 1, 2013
Publication date	Nov 13, 2018
Grant date	Nov 13, 2018

Abstract

A method for performing a dialog between a machine, preferably a humanoid robot, and at least one human speaker, comprises the following steps, implemented by a computer: a) identifying the human speaker; b) extracting from a database a speaker profile comprising a plurality of dialog variables, at least one value being assigned to at least one of the dialog variables; c) receiving and analyzing at least one sentence originating from the speaker; and d) formulating and emitting at least one response sentence as a function at least of the sentence received and interpreted in step c) and of one dialog variable of the speaker profile.

First claim

The invention claimed is:

1. A method for performing a dialog between a machine and at least one human speaker, comprising the following steps, implemented by said machine:
a) identifying said human speaker with a processor;
b) extracting from a database a speaker profile comprising a plurality of dialog variables, at least one value being assigned to at least one of said plurality of dialog variables;
c) receiving with a sound acquisition device at least one sentence originating from said speaker, analyzing said sentence or each said sentence originating from said speaker with the processor to extract therefrom at least one value to be assigned to at least one dialog variable of said speaker profile and store said value or each said value in said speaker profile in the database; and
d) formulating with the processor and emitting at least one response sentence through a sound emission device as a function of at least of said sentence received and interpreted in step c) and of one said dialog variable of said speaker profile;
wherein said sentence or at least one said sentence received and analyzed in step c) is a sentence spoken by said speaker spontaneously or following a non-interrogative sentence emitted by the sound emission device of said machine; and
wherein the analysis of said sentence or at least one sentence originating from said speaker and the formulation of said or at least one response sentence are performed by means of a plurality of sentence patterns represented by respective syntax trees.

2. The method for performing a dialog as claimed in claim 1 , wherein at least some of said sentence patterns contain at least one pointer to a dialog variable of said speaker profile.

3. The method for performing a dialog as claimed claim 1 , wherein at least some of said sentence patterns contain at least one pointer to at least one set of terms that are interchangeable in a sentence, called concept.

4. The method for performing a dialog as claimed in claim 1 , wherein said step d) is implemented by means of a plurality of rules associating at least one sentence pattern originating from said speaker and at least one response sentence pattern.

5. The method for performing a dialog as claimed in claim 4 , wherein, in said step d), at least one of said plurality of rules associates, with a same sentence pattern originating from said speaker, a plurality of sentence patterns that serve as response sentences; and wherein, out of said response sentence patterns, those which contain a pointer to a dialog variable to which no value is assigned in said speaker profile are discarded.

6. The method for performing a dialog as claimed in claim 4 , wherein:
at least one said rule is marked by a tag identifying a conversation topic;
in said step d), the rule or rules marked by at least one tag identifying a determined conversation topic are activated or deactivated as a function of at least one dialog parameter.

7. The method for performing a dialog as claimed in claim 1 , wherein said speaker profile also comprises a dialog history and, in said step d), said response or at least one said response is also formulated as a function of said dialog history.

8. The method for performing a dialog as claimed in claim 1 , wherein, in said step d), said response or at least one said response is also formulated as a function of at least one contextual parameter acquired or determined by said machine.

9. The method for performing a dialog as claimed in claim 1 , wherein, in said step a), said human speaker is identified by an image acquisition device, by facial recognition or by reading a graphic code presented to said device.

10. The method for performing a dialog as claimed in claim 1 , wherein, in said step b), the machine accesses a remote server via a communication network and downloads said speaker profile from said database which is stored on said remote server.

11. The method for performing a dialog as claimed in claim 10 , wherein said machine, having accessed said remote server via said communication network, loads into said database a speaker profile updated by the storage of at least one value assigned to at least one dialog variable extracted, in said step c), from at least one sentence originating from said speaker.

12. The method for performing a dialog as claimed in claim 1 , wherein said machine comprises the processor implemented as a processor embedded in a humanoid robot.

13. A humanoid robot wherein the processor comprises an embedded processor programmed to implement a method as claimed in claim 12 .

14. The humanoid robot as claimed in claim 13 , further comprising:
an image acquisition device cooperating with said embedded processor to identify a human speaker;
the sound acquisition device cooperating with said embedded processor to receive at least one sentence spoken by said human speaker; and
the sound emission device cooperating with said embedded processor to emit one said response sentence.

15. A non-transitory computer program product for implementing a method as claimed in claim 1 , wherein said non-transitory computer program is executed on the processor.

16. A method for performing a dialog between a machine and at least one human speaker, comprising the following steps, implemented by said machine:
a) identifying said human speaker with a processor;
b) extracting from a database a speaker profile comprising a plurality of dialog variables, at least one value being assigned to at least one of said plurality of dialog variables;
c) receiving with a sound acquisition device at least one sentence originating from said speaker, analyzing said sentence or each said sentence originating from said speaker with the processor to extract therefrom at least one value to be assigned to at least one dialog variable of said speaker profile and store said value or each said value in said speaker profile in the database; and
d) formulating with the processor and emitting at least one response sentence through a sound emission device as a function of at least of said sentence received and interpreted in step c) and of one said dialog variable of said speaker profile;
wherein said sentence or at least one said sentence received and analyzed in step c) is a sentence spoken by said speaker spontaneously or following a non-interrogative sentence emitted by the sound emission device of said machine;
wherein the analysis of said sentence or at least one sentence originating from said speaker and the formulation of said or at least one response sentence are performed by means of a plurality of sentence patterns represented by respective syntax trees;
wherein at least some of said sentence patterns contain at least one pointer to a dialog variable of said speaker profile;
wherein at least some of said sentence patterns contain at least one pointer to at least one set of terms that are interchangeable in a sentence, called concept; and
wherein said step d) is implemented by means of a plurality of rules associating at least one sentence pattern originating from said speaker and at least one response sentence pattern.

Assignees

• Softbank Robotics Europe

Inventors

• Patris Magali
• Houssin David
• Monceaux Jérôme

Classifications

• G06F40/211 Primary

  Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars · CPC title

• G06F40/35 Primary

  Discourse or dialogue representation · CPC title

• G06K7/1404

  Methods for optical code recognition · CPC title

• G06N3/008 Primary

  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 · CPC title

• G10L13/02

  Methods for producing synthetic speech; Speech synthesisers · CPC title