Adaptive method with intercessory feedback control for an intelligent agent

• US 6,754,643 B1
• Filed: 06/14/2001
• Issued: 06/22/2004
• Est. Priority Date: 10/03/2000
• Status: Active Grant

• Current Assignee: National Technology & Engineering Solutions Of Sandia LLC (Honeywell International Inc.)
• Original Assignee: Sandia Corporation (Martin Marietta Corporation)
• Inventors: Goldsmith, Steven Y.
• Primary Examiner(s): Davis, George B.
• Application Number:
  US09/882,397
• Time in Patent Office: 1,104 Days
• Field of Search: 706/14
• US Class Current: 706/14
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation

Abstract

An adaptive architecture method with feedback control for an intelligent agent provides for adaptively integrating reflexive and deliberative responses to a stimulus according to a goal. An adaptive architecture method with feedback control for multiple intelligent agents provides for coordinating and adaptively integrating reflexive and deliberative responses to a stimulus according to a goal. Re-programming of the adaptive architecture is through a nexus which coordinates reflexive and deliberator components.

Claims

1. An adaptive method for controlling an intelligent agent to respond to a stimulus according to a goal, the method implemented as an adaptive architecture running on at least one processor, comprising:

a) generating a reflex action to the stimulus by the intelligent agent;

b) generating a planned action to the stimulus by the intelligent agent, according to the goal; and

c) coordinating the reflex action generation and the planned action generation by a nexus to re-program the adaptive architecture, wherein the nexus comprises an intercessory feedback control.

2. The adaptive method of claim 1, wherein coordinating the reflex action generation and the planned action generation by a nexus to re-program the adaptive architecture, comprises inter-operatively updating the reflex action coordinated by the intercessory feedback control.

3. The adaptive method of claim 1, wherein coordinating the reflex action generation and the planned action generation by a nexus to re-program the adaptive architecture, comprises:

a) integrating through a percept the reflex action generation and an association to a schema for generating the planned action; and

b) integrating through the schema the planned action generation and an elaboration to the percept.

4. The adaptive method of claim 1, wherein:

a) generating the reflex action comprises;

i) elaborating the stimulus;

ii) forming a percept of the stimulus; and

b) generating the planned action comprises associating the percept to a schema;

c) wherein the coordination integrates the elaborated stimulus with the generated planned action.

5. The adaptive method of claim 4, wherein the step of coordinating comprises feeding the percept back through the nexus to the step of generating the planned action.

6. The adaptive method of claim 1, wherein the intercessory feedback control comprises a sensation policy, wherein:

a) the step of generating the reflex action comprises;

i) filtering the stimulus, according to the sensation policy; and

ii) organizing the stimulus; and

b) the step of coordinating comprises triggering the reflex action to the classified stimulus.

7. The adaptive method of claim 6, wherein the step of coordinating further comprises updating the sensation policy through the intercessory feedback control.

8. The adaptive method of claim 6, wherein the intercessory feedback control further comprises an elaboration policy, wherein the step of generating the reflex action further comprises:

a) forming a percept of the organized stimulus; and

b) elaborating the percept, according to the elaboration policy.

9. The adaptive method of claim 8, wherein the step of coordinating further comprises updating the elaboration policy through the intercessory feedback control.

10. The adaptive method of claim 8, wherein the intercessory feedback control further comprises an association policy, wherein the step of generating the planned action comprises associating the percept to a schema, according to the association policy.

11. The adaptive method of claim 10, wherein the step of coordinating further comprises updating the association policy through the intercessory feedback control.

12. The adaptive method of claim 10, wherein the intercessory feedback control further comprises a deliberation policy, wherein the step of coordinating further comprises executing a plan, according to the deliberation policy.

13. The adaptive method of claim 12, wherein the step of coordinating further comprises updating the deliberation policy through the intercessory feedback control.

14. The adaptive method of claim 10, wherein the intercessory feedback control further comprises a feedback policy, wherein the step of coordinating further comprises setting a new upstream policy, according to the feedback policy.

15. The adaptive method of claim 1, wherein the intercessory feedback control comprises an elaboration policy, wherein the step of generating the reflex action comprises:

a) forming a percept of the stimulus; and

b) elaborating the percept, according to the elaboration policy.

16. The adaptive method of claim 15, wherein the step of coordinating comprises updating the elaboration policy through the intercessory feedback control.

17. The adaptive method of claim 1, wherein the intercessory feedback control comprises an association policy, wherein;

a) the step of generating the reflex action comprises forming a percept of the stimulus; and

b) the step of generating the planned action comprises associating the percept to a schema, according to the association policy.

18. The adaptive method of claim 17, wherein the step of coordinating comprises updating the association policy through the intercessory feedback control.

19. The adaptive method of claim 1, wherein the intercessory feedback control comprises a deliberation policy, wherein the step of generating the planned action comprises executing a plan, according to the deliberation policy.

20. The adaptive method of claim 19, wherein the step of coordinating comprises updating the deliberation policy through the intercessory feedback control.

21. The adaptive method of claim 1, wherein the intercessory feedback control comprises a feedback policy, wherein the step of coordinating further comprises setting a new upstream policy, according to the feedback policy.

22. The adaptive method of claim 1, wherein the nexus further comprises a learning-reflection intercessory feedback control, wherein the method further comprises:

a) assessing a performance metric of the intelligent agent; and

b) adaptively re-programming the adaptive architecture to modify the performance metric, through the learning-reflection intercessory feedback control.

23. An adaptive method for controlling a plurality of software intelligent agents to respond to a stimulus according to a goal, the method implemented as an adaptive architecture running on one or more processors, the method for each intelligent agent comprising:

a) generating a reflex action to the stimulus by the intelligent agent;

b) generating a planned action to the stimulus by the intelligent agent, according to the goal; and

c) coordinating the reflex action generation and the planned action generation by a nexus to re-program the adaptive architecture, wherein the nexus comprises an intercessory feedback control.

24. The adaptive method of claim 23, wherein coordinating the reflex action generation and the planned action generation by a nexus to reprogram the adaptive architecture, comprises inter-operatively updating the reflex action coordinated by the intercessory feedback control.

25. The adaptive method of claim 23, wherein coordinating the reflex action generation and the planned action generation by a nexus to reprogram the adaptive architecture, comprises:

a) integrating through a percept the reflex action generation and an association to a schema for generating the planned action; and

b) integrating through the schema the planned action generation and an elaboration to the percept.

26. The adaptive method of claim 23, wherein:

a) generating the reflex action comprises;

i) elaborating the stimulus;

ii) forming a percept of the stimulus; and

b) generating the planned action comprises associating the percept to a schema;

c) wherein the coordination integrates the elaborated stimulus with the generated planned action.

27. The adaptive method of claim 26, wherein the step of coordinating comprises feeding the percept back through the nexus to the step of generating the planned action.

28. The adaptive method of claim 23, wherein the intercessory feedback control comprises a sensation policy, wherein:

a) the step of generating the reflex action comprises;

i) filtering the stimulus, according to the sensation policy; and

ii) organizing the stimulus; and

b) the step of coordinating comprises triggering the reflex action to the classified stimulus.

29. The adaptive method of claim 28, wherein the step of coordinating further comprises updating the sensation policy through the intercessory feedback control.

30. The adaptive method of claim 28, wherein the intercessory feedback control further comprises an elaboration policy, wherein the step of generating the reflex action further comprises:

a) forming a percept of the organized stimulus; and

b) elaborating the percept, according to the elaboration policy.

31. The adaptive method of claim 30, wherein the step of coordinating further comprises updating the elaboration policy through the intercessory feedback control.

32. The adaptive method of claim 30, wherein the intercessory feedback control further comprises an association policy, wherein the step of generating the planned action comprises associating the percept to a schema, according to the association policy.

33. The adaptive method of claim 32, wherein the step of coordinating further comprises updating the association policy through the intercessory feedback control.

34. The adaptive method of claim 32, wherein the intercessory feedback control further comprises a deliberation policy, wherein the step of coordinating further comprises executing a plan, according to the deliberation policy.

35. The adaptive method of claim 34, wherein the step of coordinating further comprises updating the deliberation policy through the intercessory feedback control.

36. The adaptive method of claim 32, wherein the intercessory feedback control further comprises a feedback policy, wherein the step of coordinating further comprises setting a new upstream policy, according to the feedback policy.

37. The adaptive method of claim 23, wherein the intercessory feedback control comprises an elaboration policy, wherein the step of generating the reflex action comprises:

a) forming a percept of the stimulus; and

b) elaborating the percept, according to the elaboration policy.

38. The adaptive method of claim 37, wherein the step of coordinating comprises updating the elaboration policy through the intercessory feedback control.

39. The adaptive method of claim 23, wherein the intercessory feedback control comprises an association policy, wherein:

a) the step of generating the reflex action comprises forming a percept of the stimulus; and

b) the step of generating the planned action comprises associating the percept to a schema, according to the association policy.

40. The adaptive method of claim 39, wherein the step of coordinating comprises updating the association policy through the intercessory feedback control.

41. The adaptive method of claim 23, wherein the intercessory feedback control comprises a deliberation policy, wherein the step of generating the planned action comprises executing a plan, according to the deliberation policy.

42. The adaptive method of claim 41, wherein the step of coordinating comprises updating the deliberation policy through the intercessory feedback control.

43. The adaptive method of claim 23, wherein the intercessory feedback control comprises a feedback policy, wherein the step of coordinating further comprises setting a new upstream policy, according to the feedback policy.

44. The adaptive method of claim 23, wherein the nexus further comprises a learning-reflection intercessory feedback control, wherein the method further comprises:

a) assessing a performance metric of the intelligent agent; and

b) adaptively re-programming the adaptive architecture to modify the performance metric, through the learning-reflection intercessory feedback control.

45. The adaptive method of claim 23, further comprising:

a) sharing an agent state and an agent plan with the other intelligent agents; and

b) securely collaborating with the other intelligent agents to satisfy the goal.

46. An adaptive method for controlling a plurality of software intelligent agents to respond to a stimulus according to a goal, the method implemented as an adaptive architecture on one or more computing devices, the method for each intelligent agent comprising:

a) generating a reflex action to the stimulus by the intelligent agent;

b) generating a planned action to the stimulus by the intelligent agent, according to the goal; and

c) coordinating the reflex action generation and the planned action generation by a nexus to re-program the adaptive architecture to inter-operatively update the reflex actions, wherein the nexus comprises an intercessory feedback control.

47. The adaptive method of claim 46, wherein each of the computing devices comprises at least one of a RAM, a ROM, a disk, an ASIC, a PROM, and a re-programmable memory.