Bayesian-centric autonomous robotic learning

• US 9,984,332 B2
• Filed: 07/27/2016
• Issued: 05/29/2018
• Est. Priority Date: 11/05/2013
• Status: Expired due to Fees

• Current Assignee: NPC Robotics, Inc.
• Original Assignee: NPC Robotics, Inc.
• Inventors: Garrod, Michael
• Primary Examiner(s): Holloway, Jason
• Application Number:
  US15/221,461
• Publication Number:
  US 20160332298A1
• Time in Patent Office: 671 Days
• CPC Class Codes: B25J 9/163 learning, adaptive, model based, rule based expert control
• CPC Class Codes: G05B 2219/39376 Hierarchical, learning, recognition and skill level and adaptation servo level
• CPC Class Codes: G05D 1/0088 characterized by the autonomous decision making process, e.g. artificial intelligence, predefined behaviours using knowledge based models G06N5/00
• CPC Class Codes: G05D 1/0274 using mapping information stored in a memory device navigation using map-matching G01C21/30
• CPC Class Codes: G06N 3/008 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 Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: Y10S 901/01 Mobile robot
• CPC Class Codes: Y10S 901/02 Arm motion controller

Abstract

Various apparatus and methods include autonomous robot operations to perturb a current Bayesian equation and determining whether the perturbed Bayesian equation yields an improved probability of success of achieving a goal relative to the current Bayesian equation. In an illustrative example, the perturbation may modulate a coefficient of a parameter in the Bayesian equation. In some examples, the perturbation may include assessment of whether adding or removing a parameter may improve the probability of success of achieving the goal. The parameters of the Bayesian equation may include, for example, current state information, alone or in combination with sensor input values and/or historical information, for example. In some implementations, the robot may advantageously autonomously optimize its operations by perturbing a current Bayesian equation associated with, for example, a current goal, sub-goal, task, or probability of success criteria.

Claims

1. An autonomous robotic system for learning, the system comprising:

a mobile platform having a central body;

a displacement module operably connected to the central body and configured to displace the mobile platform;

a payload module releasably connected to the central body and configured to perform a mission specific operation;

an input module configured to receive a user input command;

a controller disposed within the central body and operably connected to the displacement module and the input module;

a data store operably coupled to the controller, wherein the data store comprises a program of instructions that, when executed by the controller, cause the controller to perform operations to adaptively optimize a control system of the mobile platform, the operations comprising;

receive, via the input module, a predetermined target goal;

store the predetermined goal in the data store;

retrieve the predetermined target goal from the data store;

retrieve a set of parameters associated with the predetermined target goal;

retrieve a set of coefficients associated with the retrieved set of parameters;

determine a current success probability of achieving the predetermined target goal based on a Bayesian equation formed by the retrieved set of parameters and the retrieved set of coefficients;

receive a perturbation signal;

modify a selected one of the retrieved coefficients or a selected one of the retrieved parameters in response to the received perturbation signal;

determine a perturbed success probability based on the Bayesian equation using the selected one of the retrieved coefficients or the selected one of the retrieved parameters as modified by the received perturbation signal; and

,

if the perturbed success probability exceeds the current success probability, then store the selected one of the retrieved coefficients or the selected one of the retrieved parameters as modified by the received perturbation signal and in association with the predetermined target goal.

2. The system of claim 1, wherein the perturbation signal comprises one of the retrieved set of parameters.

3. The system of claim 1, wherein the perturbation signal comprises one of the retrieved set of coefficients.

4. The system of claim 3, wherein the one of the retrieved set of coefficients equals zero.

5. The system of claim 1, wherein the predetermined target goal comprises one or more sub-goals.

6. The system of claim 1, wherein the predetermined target goal is formed from a composite goal structure.

7. The system of claim 1, the input module comprises a network interface in operative communication with a user.

8. The system of claim 1, further comprising a sensor input operably connected to the controller and configured to receive sensor information about environmental surroundings of the central body.

9. The system of claim 8, wherein the operation to determine a perturbed success probability is further based on the received sensor information.

10. The method of claim 1, the operations further comprising:

generate an action command based on the perturbed success probability, wherein in response to the generated action command, the displacement module and the payload module perform operations towards achieving the predetermined target goal.

11. An autonomous robotic system for learning, the system comprising:

a mobile platform having a central body;

a displacement module operably connected to the central body and configured to displace the mobile platform;

a payload module releasably connected to the central body and configured to perform a mission specific operation;

a controller disposed within the central body and operably connected to the displacement module;

a data store operably coupled to the controller, wherein the data store comprises a program of instructions that, when executed by the controller, cause the controller to perform operations to adaptively optimize a control system of the mobile platform, the operations comprising;

retrieve a predetermined target goal from the data store;

retrieve a set of parameters associated with the predetermined target goal;

retrieve a set of coefficients associated with the retrieved set of parameters;

determine a current success probability of achieving the predetermined target goal based on a Bayesian equation formed by the retrieved set of parameters and the retrieved set of coefficients;

receive a perturbation signal;

modify a selected one of the retrieved coefficients or a selected one of the retrieved parameters in response to the received perturbation signal;

determine a perturbed success probability based on the Bayesian equation using the selected one of the retrieved coefficients or the selected one of the retrieved parameters as modified by the received perturbation signal; and

,

if the perturbed success probability exceeds the current success probability, then store the selected one of the retrieved coefficients or the selected one of the retrieved parameters as modified by the received perturbation signal and in association with the predetermined target goal.

12. The system of claim 11, further comprising an input module for receiving a user input command and operably connected to the controller, wherein the predetermined target goal is retrieved from the received user input command.

13. The system of claim 11, wherein the perturbation signal comprises one of the retrieved set of parameters.

14. The system of claim 11, wherein the perturbation signal comprises one of the retrieved set of coefficients.

15. The system of claim 14, wherein the one of the retrieved set of coefficients equals zero.

16. An autonomous robotic system for learning, the system comprising:

a mobile platform having a central body;

means for displacing the central body;

a payload module releasably connected to the central body and configured to perform a mission specific operation;

a controller disposed within the central body and operably connected to the displacement module;

a data store operably coupled to the controller, wherein the data store comprises a program of instructions that, when executed by the controller, cause the controller to perform operations to adaptively optimize a control system of the mobile platform, the operations comprising;

retrieve a predetermined target goal from the data store;

retrieve a set of parameters associated with the predetermined target goal;

retrieve a set of coefficients associated with the retrieved set of parameters;

determine a current success probability of achieving the predetermined target goal based on a Bayesian equation formed by the retrieved set of parameters and the retrieved set of coefficients;

receive a perturbation signal;

modify a selected one of the retrieved coefficients or a selected one of the retrieved parameters in response to the received perturbation signal;

determine a perturbed success probability based on the Bayesian equation using the selected one of the retrieved coefficients or the selected one of the retrieved parameters as modified by the received perturbation signal; and

,

if the perturbed success probability exceeds the current success probability, then store the selected one of the retrieved coefficients or the selected one of the retrieved parameters as modified by the received perturbation signal and in association with the predetermined target goal.

17. The system of claim 16, further comprising an input module for receiving a user input command and operably connected to the controller, wherein the predetermined target goal is retrieved from the received user input command.

18. The system of claim 16, wherein the perturbation signal comprises one of the retrieved set of parameters.

19. The system of claim 16, wherein the perturbation signal comprises one of the retrieved set of coefficients.

20. The system of claim 19, wherein the one of the retrieved set of coefficients equals zero.