Framework for flexible cognitive perception and action selection

• US 8,990,139 B1
• Filed: 07/23/2012
• Issued: 03/24/2015
• Est. Priority Date: 07/23/2012
• Status: Active Grant

• Current Assignee: HRL Laboratories LLC (The Boeing Co.)
• Original Assignee: HRL Laboratories LLC (The Boeing Co.)
• Inventors: Howard, Mike, Bhattacharyya, Rajan
• Primary Examiner(s): CHANG, LI WU
• Application Number:
  US13/556,014
• Time in Patent Office: 974 Days
• Field of Search: None
• US Class Current: 706/46
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06Q 10/101 Collaborative creation, e.g. joint development of products or services

Abstract

Described is a system for flexible cognitive perception and selection. A pre-processing recognition module filters and tags input data from an environment resulting in a tagged percept. The tagged percept is stored and associated with a knowledge frame by a memory module based on shared descriptors, resulting in an activated knowledge frame. A utility rating is then supplied to each activated knowledge frame based on a set of reward values by an evaluation module. The activated knowledge frames are sorted, compared, and evaluated for a goodness of fit between the utility ratings of the activated knowledge frames and the input data by a hypothesis module. A best hypothesis for a current situation in the environment is determined based on a current highest rated activated knowledge frame.

Claims

1. A system for flexible cognitive perception and action selection, the system comprising:

one or more processors and a non-volatile memory having instructions therein such that when the instructions are executed by the one or more processors, the one or more processors perform operations of;

filtering and tagging of input data from an external environment by a pre-processing recognition module, resulting in at least one tagged percept;

storing the at least one tagged percept and associating the tagged percept with at least one knowledge frame based on shared descriptors between the at least one tagged percept and the knowledge frame by a frame memory module, resulting in an activated knowledge frame;

supplying a utility rating to each activated knowledge frame based on a set of reward values by an evaluation module;

sorting and comparing the activated knowledge frames, and evaluating the activated knowledge frames for a goodness of fit between the utility ratings of the activated knowledge frames and the input data by a hypothesis module;

determining, with the hypothesis module, a best hypothesis for a current situation in the external environment based on a current highest rated activated knowledge frame; and

wherein the pre-processing recognition module performs operations of;

extracting, and identifying a set of salient features in the input data;

filtering the set of salient features for relevancy based on similarity to the current highest rated activated knowledge frame; and

semantically tagging percepts.

2. The system for flexible cognitive perception and action selection as set forth in claim 1, wherein a knowledge frame in the memory module is an associative structure comprised of a set of bits that represents salient features of an experience, wherein each hit is associated with a hierarchy of semantic descriptors and related bits.

3. The system for flexible cognitive perception and action selection as set forth in claim 2, wherein the hypothesis module comprises a set of registers each containing at least one activated knowledge frame from the memory module, wherein each register has an associated utility rating.

4. The system for flexible cognitive perception and action selection as set forth in claim 3, wherein the hypothesis module further performs operations of:

sorting the set of registers based on each register'"'"'s associated utility rating; and

re-evaluating the current highest rated activated knowledge frame when the set of registers is updated.

5. The system for flexible cognitive perception and action selection as set forth in claim 4, wherein the evaluation module further performs operations of:

receiving a set of feedback from the external environment;

determining the utility rating for each activated knowledge frame based on a reward value that is based on how successful a hypothesis is for accomplishing a goal of the system.

6. A computer-implemented method for flexible cognitive perception and action selection, comprising an act of:

causing a data processor to execute instructions stored on a non-volatile memory such that upon execution the data processor performs operation of;

filtering and tagging of input data from an external environment by a pre-processing recognition module, resulting in at least one tagged percept;

storing the at least one tagged percept and associating the tagged percept with at least one knowledge frame based on shared descriptors between the at least one tagged percept and the knowledge frame by a frame memory module, resulting in an activated knowledge frame;

supplying a utility rating to each activated knowledge frame based on a set of reward values by an evaluation module;

sorting and comparing the activated knowledge frames, and evaluating the activated knowledge frames for a goodness of fit between the utility ratings of the activated knowledge frames and the input data by a hypothesis module;

determining, with the hypothesis module, a best hypothesis for a current situation in the external environment based on a current highest rated activated knowledge frame; and

extracting and identifying a set of salient features in the input data;

filtering the set of salient features for relevancy based on similarity to the current highest rated activated knowledge frame; and

semantically tagging percepts.

7. The method for flexible cognitive perception and action selection as set forth in claim 6, wherein a knowledge frame in the memory module is an associative structure comprised of a set of bits that represents salient features of an experience, wherein each bit is associated with a hierarchy of semantic descriptors and related bits.

8. The method for flexible cognitive perception and action selection as set forth in claim 7, wherein the hypothesis module comprises a set of registers each containing at least one activated knowledge frame from the memory module, wherein each register has an associated utility rating.

9. The method for flexible cognitive perception and action selection as set forth in claim 8, further comprising acts of:

receiving a set of feedback from the external environment;

determining the utility rating for each activated, knowledge frame based on a reward value that is based on how successful a hypothesis is for accomplishing a goal of the system.

10. A computer program product for flexible cognitive perception and action selection, the computer program product comprising computer-readable instruction means stored on a non-transitory computer-readable medium that are executable by a computer having a processor for causing the processor to perform operations of:

filtering and tagging of input data from an external environment by a pre-processing recognition module, resulting in at least on tagged percept;

storing the at least one tagged percept and associating the tagged percept with at least one knowledge frame based on shared descriptors between the at least one tagged percept and the knowledge frame by a frame memory module, resulting in an activated knowledge frame;

supplying a utility rating to each activated knowledge frame based on a set of reward values by an evaluation module;

sorting and comparing the activated knowledge frames, and evaluating the activated knowledge frames for a goodness of fit between the utility ratings of the activated knowledge frames and the input data by a hypothesis module;

determining, with the hypothesis module, a best hypothesis for a current situation in the external environment based on a current highest rated activated knowledge frame; and

extracting and identifying a set of salient, features in the input data;

filtering the set of salient features for relevancy based on similarity to the current highest rated activated knowledge frame; and

semantically tagging percepts.

11. The computer program product for flexible cognitive perception and action selection as set forth in claim 10, wherein a knowledge frame in the memory module is an associative structure comprised of a set of bits that represents salient features of an experience, wherein each bit is associated with a hierarchy of semantic descriptors and related bits.

12. The computer program product for flexible cognitive perception and action selection as set forth in claim 11, wherein the hypothesis module comprises a set of registers each containing at least one activated knowledge frame from the memory module, wherein each register has an associated utility rating.

13. The computer program product for flexible cognitive perception and action selection as set forth in claim 12, further comprising instruction means for causing the processor to perform operations of:

receiving a set of feedback from the external environment;

determining, the utility rating for each activated knowledge frame based on a reward value that is based on how successful a hypothesis is for accomplishing, a goal of the system.