Predictive robotic controller apparatus and methods

• US 9,950,426 B2
• Filed: 04/18/2016
• Issued: 04/24/2018
• Est. Priority Date: 06/14/2013
• Status: Active Grant

• Current Assignee: Brain Corporation
• Original Assignee: Brain Corporation
• Inventors: Laurent, Patryk, Passot, Jean-Baptiste, Sinyavskiy, Oleg, Ponulak, Filip, Gabardos, Borja Ibarz, Izhikevich, Eugene
• Primary Examiner(s): Tran, Khoi H
• Assistant Examiner(s): Peche, Jorge O
• Application Number:
  US15/132,003
• Publication Number:
  US 20160303738A1
• Time in Patent Office: 736 Days
• Field of Search: 700245, 700250, 700255, 700257, 700264, 901 1, 901 2, 901 4
• CPC Class Codes: B25J 9/163 learning, adaptive, model based, rule based expert control
• CPC Class Codes: G05B 2219/39271 Ann artificial neural network, ffw-nn, feedforward neural network
• CPC Class Codes: G05B 2219/39289 Adaptive ann controller
• 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 3/049 Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs

Abstract

Robotic devices may be trained by a user guiding the robot along target action trajectory using an input signal. A robotic device may comprise an adaptive controller configured to generate control signal based on one or more of the user guidance, sensory input, performance measure, and/or other information. Training may comprise a plurality of trials, wherein for a given context the user and the robot'"'"'s controller may collaborate to develop an association between the context and the target action. Upon developing the association, the adaptive controller may be capable of generating the control signal and/or an action indication prior and/or in lieu of user input. The predictive control functionality attained by the controller may enable autonomous operation of robotic devices obviating a need for continuing user guidance.

Claims

1. A computerized method of training a robotic apparatus comprising:

performing a physical training action in a context determined from at least one or more inputs from one or more sensors, the physical training action performed based on at least a training instruction provided by a user;

performing a first physical action in the context, the first physical action comprising a movement of the robotic apparatus along a first trajectory, the first physical action being determined by at least a user instruction and a first controller instruction indicative at least in part of a first predicted physical action determined from at least the context and the physical training action;

receiving a corrective input signal based on the performed first physical action; and

performing a second physical action in the context based on the corrective input signal, the second physical action comprising a movement of the robotic apparatus along a second trajectory, the second physical action further being determined by at least a second controller instruction generated from at least the first physical action, the physical training action, and a second predicted physical action determined from at least the context, wherein a deviation between performance measures of the second physical action and the physical training action is less than a deviation between performance measures of the first physical action and the physical training action.

2. The method of claim 1, further comprising computing a learning parameter that characterizes a neuron process, the computing of the learning parameter being based on at least the deviation between performance measures of the first physical action and the physical training action, and the second controller instruction is further generated from the learning parameter.

3. The method of claim 2, wherein the computing of the learning parameter comprises using a supervised learning process configured based on at least the context and a combination of the first action and the training action.

4. The method of claim 1, wherein performing each of the physical training action, first physical action, first predicted physical action, second physical action, and second predicted physical action comprises navigating a trajectory.

5. The method of claim 1, wherein performing the physical training action, first physical action, first predicted physical action, second physical action, and second predicted physical action each comprise manipulating a motorized operational element of the robotic apparatus.

6. The method of claim 1, wherein determining the context comprises generating an object representation, and the performing of the physical training action comprises at least one of performing an object approach maneuver and performing an object avoidance maneuver.

7. The method of claim 1, further comprising calculating the discrepancy between the first physical action and the physical training action from a proximity measure between the first action and the training action.

8. A trainable robotic apparatus comprising:

a sensor configured to generate a sensory input;

an interface configured to receive one or more user instructions;

a motorized operational element configured to perform physical actions; and

a processor configured to;

determine a context from the sensory input,

provide a training control signal based on at least a training instruction received through the interface, the training control signal causing the motorized operation element to perform a physical training action in the context,

provide a first control signal based on at least a user instruction received through the interface and a first predicted physical action determined from at least the context and the physical training action, the first control signal causing the motorized operation element to perform a first physical action in the context, the first physical action comprising at least a movement of the robotic apparatus along a first trajectory via the motorized operation element,

compute a learning parameter that characterizes a neuron process, the computation of the learning parameter being based on at least a deviation between performance measures of the first physical action and the physical training action, and

provide a second control signal based on at least the learning parameter and a second predicted physical action based on at least the context, the second control signal causing the motorized operation element to perform a second physical action in the context, the second physical action comprising at least a movement of the robotic apparatus along a second trajectory via the motorized operation element, wherein a deviation between performance measures of the second physical action and the physical training action is less than a deviation between performance measures of the first physical action and the physical training action.

9. The trainable robotic apparatus of claim 8, wherein the sensor includes a camera.

10. The trainable robotic apparatus of claim 8, wherein the physical training action, first physical action, first predicted physical action, second physical action, and second predicted physical action each comprise navigation of respective trajectories.

11. The trainable robotic apparatus of claim 8, wherein the context comprises an object representation and the training action comprises at least one of an object approach maneuver or an object avoidance maneuver.

12. The trainable robotic apparatus of claim 8, wherein the deviation between performance measures of the first physical action and the physical training action is calculated from a proximity measure between the first action and the training action.

13. The trainable robotic apparatus of claim 8, wherein computing the learning parameter comprises using a supervised learning process configured based on at least the context and a combination of the first action and the training action.

14. A robotic apparatus comprising:

one or more sensors;

a processor apparatus; and

a non-transitory computer-readable storage medium having a computer program stored thereon, the computer program comprising a plurality of instructions configured to, when executed by the processor apparatus, cause the robotic apparatus to;

perform a physical training action in an environmental context determined from at least one or more inputs from the one or more sensors, the physical training action performed based on at least a training instruction provided by a user;

perform a first physical action in the context, the first physical action comprising a movement of the robotic apparatus along a first trajectory, the first physical action being determined by at least a user instruction and a first controller instruction indicative at least in part of a first predicted physical action determined from at least the context and the physical training action;

compute a learning parameter based on at least a deviation between performance measures of the first physical action and the physical training action; and

perform a second physical action in the context, the second physical action comprising a movement of the robotic apparatus along a second trajectory, the second physical action being determined by at least a second controller instruction generated from at least the learning parameter and a second predicted physical action determined from at least the context;

wherein a deviation between performance measures of the second physical action and the physical training action is less than the deviation between the performance measures of the first physical action and the physical training action.

15. The robotic apparatus of claim 14, wherein the one or more sensors comprise at least a camera.

16. The robotic apparatus of claim 14, wherein each of the physical training action, the first predicted physical action, and the second predicted physical action comprises a movement of the robotic apparatus along a respective trajectory.

17. The robotic apparatus of claim 16, wherein:

the robotic apparatus further comprises at least one motorized operational element;

the plurality of instructions are further configured to, when execute by the processor apparatus, cause the robotic apparatus to manipulate the at least one motorized operational element to perform the physical training action, the first predicted physical action, and the second predicted physical action.

18. The robotic apparatus of claim 14, wherein the context comprises an object representation, and the physical training action comprises at least one of an object approach maneuver and an object avoidance maneuver.

19. The robotic apparatus of claim 14, wherein the deviation between the performance measures of the first physical action and the physical training action is calculated from a proximity measure between the first physical action and the physical training action.

20. The robotic apparatus of claim 14, wherein the computation of the learning parameter comprises a usage of a supervised learning process configured based on at least the context and a combination of the first physical action and the physical training action.