Robot device controller, robot device arrangement and method for controlling a robot device

Patent metadata

Field	Value
Publication number	US-10913152-B2
Application number	US-202016891917-A
Country	US
Kind code	B2
Filing date	Jun 3, 2020
Priority date	Jun 7, 2019
Publication date	Feb 9, 2021
Grant date	Feb 9, 2021

Abstract

A robot device controller including a memory configured to store a statistical model trained to implement a behaviour of the robot device, one or more processors configured to determine a nominal trajectory represented by the statistical model, determine an expected force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory, determine a measured force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory and adapt the statistical model based on a reduction of the difference between the measured force and the expected force.

First claim

What is claimed is:

1. A robot device controller, comprising:
a memory configured to store a statistical model trained to implement a behaviour of a robot device; and
one or more processors configured to:
determine a nominal trajectory represented by the statistical model;
determine an expected force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory;
determine a measured force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory; and
adapt the statistical model based on a reduction of the difference between the measured force and the expected force.

2. The robot device controller according to claim 1 , wherein the one or more processors are configured to adapt the statistical model by adapting one or more parameters of the statistical model including at least one parameter of one or more probability distributions.

3. The robot device controller according to claim 2 , wherein the one or more processors are configured to adapt the one or more parameters based on an optimization.

4. The robot device controller according to claim 3 , wherein the optimization is based on an objective function based on a difference between an actual force and an experienced force.

5. The robot device controller according to claim 3 , wherein the optimization is based on a Bayesian optimization or a Covariance Matrix Adaptation Evolution Strategy.

6. The robot device controller as recited in claim 3 , wherein the optimization is based on reducing a difference between the expected force and the measured force.

7. The robot device controller according to claim 1 , wherein the one or more processors are configured to adapt the statistical model by adapting one or more parameters of the statistical model including at least one of a mean and a standard deviation of one or more Gaussian distributions.

8. The robot device controller according to claim 1 , wherein the statistical model has a plurality of hidden states, each of the states being associated with an observation probability distribution.

9. The robot device controller according to claim 8 , wherein the one or more processors are configured to detect a trajectory adaptation event based on a difference between the measured force and the expected force, to determine a state of the statistical model representing a part of the nominal trajectory in which the trajectory adaptation event has occurred and to perform an optimization over parameters of the observation probability distribution.

10. The robot device controller according to claim 9 , wherein the one or more processors are configured to further perform the optimization over parameters of a duration probability distribution associated with the determined state.

11. The robot device controller according to claim 9 , wherein the one or more processors are configured to omit from the optimization parameters of probability distributions associated with other states than the determined state.

12. The robot device controller according to claim 9 , wherein the one or more processors are configured to limit the optimization to parameters of probability distributions associated with the determined state.

13. The robot device controller as recited in claim 9 , wherein the optimization is based on reducing a difference between the expected force and the measured force.

14. The robot device controller according to claim 1 , wherein the one or more processors are configured to train the statistical model based on one or more demonstrations of at least one trajectory.

15. The robot device controller according to claim 1 , wherein the statistical model is a Gaussian mixture model, or a hidden Markov model, or a hidden semi-Markov model.

16. The robot device controller as recited in claim 1 , wherein the statistical model is a Gaussian mixture model, or a hidden Markov model, or a hidden semi-Markov model, and wherein the nominal trajectory is a planned trajectory for the robot device.

17. The robot device controller as recited in claim 16 , wherein the statistical model encodes the nominal trajectory using a set of states.

18. A robot device arrangement, comprising:
a robot device; and
a robot device controller including:
a memory configured to store a statistical model trained to implement a behaviour of a robot device; and
one or more processors configured to:
determine a nominal trajectory represented by the statistical model;
determine an expected force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory;
determine a measured force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory; and
adapt the statistical model based on a reduction of the difference between the measured force and the expected force.

19. The robot device arrangement as recited in claim 18 , wherein the statistical model is a Gaussian mixture model, or a hidden Markov model, or a hidden semi-Markov model, and wherein the nominal trajectory is a planned trajectory for the robot device.

20. The robot device arrangement as recited in claim 19 , wherein the statistical model encodes the nominal trajectory using a set of states.

21. A method for controlling a robot device, comprising the following steps:
storing a statistical model trained to implement a behaviour of a robot device;
determining a nominal trajectory represented by the statistical model;
determining an expected force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory;
determining a measured force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory; and
adapting the statistical model based on a reduction of the difference between the measured force and the expected force.

22. The method as recited in claim 21 , wherein the statistical model is a Gaussian mixture model, or a hidden Markov model, or a hidden semi-Markov model, and wherein the nominal trajectory is a planned trajectory for the robot device.

23. The method as recited in claim 22 , wherein the statistical model encodes the nominal trajectory using a set of states.

Assignees

• Bosch Gmbh Robert

Inventors

• Rozo Leonel

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• B25J9/1689

  Teleoperation · CPC title

• B25J15/08

  having finger members (B25J15/02, B25J15/04 take precedence) · CPC title

• B25J9/1664 Primary

  characterised by motion, path, trajectory planning · CPC title

• B25J9/163 Primary

  learning, adaptive, model based, rule based expert control · CPC title