Landmark-based location belief tracking for voice-controlled navigation system

• US 9,127,950 B2
• Filed: 03/13/2013
• Issued: 09/08/2015
• Est. Priority Date: 05/03/2012
• Status: Active Grant

• Current Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Raux, Antoine, Gupta, Rakesh, Ramachandran, Deepak, Ma, Yi
• Primary Examiner(s): Yang, Qian
• Application Number:
  US13/801,441
• Publication Number:
  US 20130297321A1
• Time in Patent Office: 909 Days
• US Class Current: 1/1
• CPC Class Codes: G01C 21/26 specially adapted for navigation in a road network
• CPC Class Codes: G01C 21/3608 using speech input, e.g. using speech recognition
• CPC Class Codes: G01C 21/3644 Landmark guidance, e.g. using POIs or conspicuous other objects

Abstract

An utterance is received from a user specifying a location attribute and a landmark. A set of candidate locations is identified based on the specified location attribute, and a confidence score can be determined for each candidate location. A set of landmarks is identified based on the specified landmark, and confidence scores can be determined for the landmarks. An associated kernel model is generated for each landmark. Each kernel model is centered at the location of the associated landmark on a map, and the amplitude of the kernel model can be based on landmark attributes, landmark confidence scores, characteristics of the user, and the like. The candidate locations are ranked based on the amplitudes of overlapping kernel models at the candidate locations, and can also be ranked based on confidence scores associated with the candidate locations. A candidate location is selected and presented to the user based on the candidate location ranking.

Claims

1. A computer-based method of determining a location, comprising:

receiving a signal representing an utterance from the user, the utterance specifying a location attribute and a landmark;

identifying a set of candidate locations based on the specified location attribute;

identifying a set of landmarks based on the specified landmark;

generating an associated kernel model for each landmark in the set of landmarks, each kernel model comprising a three-dimensional model centered on a map at the location of a landmark associated with the kernel model;

ranking the candidate locations based on kernel model amplitudes at each candidate location, wherein at least one candidate location is associated with multiple kernel models, and wherein the at least one candidate location is ranked based on a sum of the amplitudes of the multiple kernel models at the at least one candidate location; and

selecting a location to provide to the user based on the ranked candidate locations.

2. The method of claim 1, wherein the signal representing the utterance is converted to text comprising location attribute text representing the specified location attribute and landmark text representing the specified landmark.

3. The method of claim 1, wherein identifying a set of candidate locations comprises performing semantic belief tracking to identify candidate locations representative of the specified location attribute.

4. The method of claim 3, wherein semantic belief tracking is performed over a plurality of dialog turns.

5. The method of claim 1, further comprising determining an associated confidence score for each candidate location representing the likelihood that the candidate location represents the specified location attribute.

6. The method of claim 1, further comprising determining an associated confidence score for each landmark representing the likelihood that the landmark represents the specified landmark.

7. The method of claim 1, wherein at least one kernel model has a circular base, wherein the amplitude of the kernel model is greatest at the location on the map of the associated landmark, and wherein the amplitude of the kernel model decays with an increase in distance from the location of the associated landmark according to a Gaussian curve.

8. The method of claim 1, wherein the shape of at least one kernel model base is based on a type of the landmark associated with the kernel model.

9. The method of claim 1, wherein the shape of at least one kernel model base is based on a shape of the aerial boundary of the landmark associated with the kernel model on the map.

10. The method of claim 1, wherein the amplitude of at least one kernel model is based on a confidence score of the landmark associated with the kernel model.

11. The method of claim 1, wherein the amplitude of at least one kernel model is based on one or more attributes of the landmark associated with the kernel model.

12. The method of claim 11, wherein the amplitude of the kernel model is based on a landmark type of the associated landmark.

13. The method of claim 12, wherein the amplitude of the kernel model is based on a prevalence of the landmark type.

14. The method of claim 11, wherein the amplitude of the kernel model is based on reviews associated with the associated landmark.

15. The method of claim 1, wherein the received signal representing the utterance further specifies a direction, and wherein the amplitude of at least kernel model is based on the specified direction.

16. The method of claim 1, wherein the signal representing the utterance is received by a search engine at a search engine user interface, and further comprising:

providing the ranked candidate locations to the user as search results via the search engine user interface.

17. The method of claim 1, further comprising:

multiplying the sum of the kernel model amplitudes by a confidence score of the at least one candidate location to form a modified kernel model amplitude sum; and

ranking the at least one candidate location based on the modified kernel model amplitude sum.

18. The method of claim 1, wherein selecting a candidate location comprises selecting the highest ranked candidate location.

19. A system for determining a location for recommendation to a user, comprising:

a non-transitory computer readable storage medium comprising executable computer modules configured to;

receive a signal representing an utterance from the user, the utterance specifying a location attribute and a landmark;

identify a set of candidate locations based on the specified location attribute;

identify a set of landmarks based on the specified landmark;

generate an associated kernel model for each landmark in the set of landmarks, each kernel model comprising a three-dimensional model centered on a map at the location of a landmark associated with the kernel model;

rank the candidate locations based on kernel model amplitudes at each candidate location, wherein at least one candidate location is associated with multiple kernel models, and wherein the at least one candidate location is ranked based on a sum of the amplitudes of the multiple kernel models at the at least one candidate location; and

select a location to provide to the user based on the ranked candidate locations; and

a processor configured to execute the computer modules.

20. The system of claim 19, wherein the executable computer modules are further configured to:

convert the signal representing an utterance to text comprising location attribute text representing the specified location attribute and landmark text representing the specified landmark.

21. The system of claim 19, wherein identifying a set of candidate locations comprises performing semantic belief tracking to identify candidate locations representative of the specified location attribute.

22. The system of claim 19, wherein the executable computer modules are further configured to:

determine at least one of;

an associated confidence score for each candidate location representing the likelihood that the candidate location represents the specified location attribute, and an associated confidence score for each landmark representing the likelihood that the landmark represents the specified landmark.

23. The system of claim 19, wherein at least one kernel model has a circular base, wherein the amplitude of the kernel model is greatest at the location on the map of the associated landmark, and wherein the amplitude of the kernel model decays with an increase in distance from the location of the associated landmark according to a Gaussian curve.

24. The system of claim 19, wherein the shape of at least one kernel model base is based on at least one of:

a type of the landmark associated with the kernel model, and a shape of the aerial boundary of the landmark associated with the kernel model on the map.

25. The system of claim 19, wherein the amplitude of at least one kernel model is based on at least one of:

a confidence score of the landmark associated with the kernel model, one or more attributes of the landmark associated with the kernel model,

a landmark type of the associated landmark, and reviews associated with the associated landmark.

26. The system of claim 19, wherein the received signal representing the utterance further specifies a direction, and wherein the amplitude of at least kernel model is based on the specified direction.

27. The system of claim 19, further comprising:

multiplying the sum of the kernel model amplitudes by a confidence score of the at least one candidate location to form a modified kernel model amplitude sum; and

ranking the at least one candidate location based on the modified kernel model amplitude sum.

28. The system of claim 19, wherein selecting a candidate location comprises selecting the highest ranked candidate location.