Ambulation prediction controller for lower limb assistive device

Patent metadata

Field	Value
Publication number	US-9443203-B2
Application number	US-201313925668-A
Country	US
Kind code	B2
Filing date	Jun 24, 2013
Priority date	Jun 24, 2013
Publication date	Sep 13, 2016
Grant date	Sep 13, 2016

Abstract

Embodiments of the systems and methods described herein relate to a control unit comprising a memory having stored thereon a feature database that includes feature data, a portion of which is labeled with a transition from an ambulation mode and a memory having stored thereon a pattern recognition controller that is trained using the labeled feature data, wherein the pattern recognition controller is configured to predict the ambulation mode of an assistive device and the control unit is configured to communicate with sensors coupled to the assistive device.

First claim

The invention claimed is:

1. A control unit comprising:
a. a memory having stored thereon a feature database that includes feature data, a portion of which is labeled with a transition from an ambulation mode; and
b. a memory having stored thereon a pattern recognition controller that is trained using the labeled feature data;
wherein the pattern recognition controller is configured to predict the ambulation mode of an assistive device and the control unit is configured to naturally transition the assistive device from operating in a prior stride to operating in the predicted ambulation mode in a next stride, wherein the prior stride is the stride immediately prior to the next stride.

2. The control unit of claim 1 :
a. wherein the feature database is further configured to be updated with features derived from data collected from said sensors during level walking of the assistive device; and
b. wherein the pattern recognition controller is further configured to be retrained on an updated feature database.

3. The control unit of claim 1 :
a. wherein the pattern recognition controller comprises a classifier for each controlled ambulation mode.

4. The control unit of claim 1 :
a. wherein the pattern recognition controller is configured to predict the next ambulation mode of the assistive device using features derived from the prior stride.

5. The control unit of claim 4 :
a. wherein the pattern recognition controller is further configured to predict the ambulation mode of the assistive device using at least one of an LDA, a dynamic Bayesian network, a hidden Markov model, or a Kalman filter.

6. The control unit of claim 1 :
a. wherein the feature database is configured to be updated with feature data derived from data collected from said sensors during the prior stride and labeled with a transition from an ambulation mode.

7. The control unit of claim 1 :
a. wherein the control unit is configured to predict the ambulation mode of the assistive device using data collected from at least one electrode sensor coupled to the assistive device.

8. The control unit of claim 1 , wherein the ambulation mode of the prior stride is a level walking mode and the ambulation mode of the next stride is a stair descent mode.

9. The control unit of claim 1 , wherein the ambulation mode of the prior stride is a level walking mode and the ambulation mode of the next stride is a stair ascent mode.

10. The control unit of claim 1 , wherein the ambulation mode of the prior stride is a level walking mode and the ambulation mode of the next stride is a ramp descent mode.

11. The control unit of claim 1 , wherein the ambulation mode of the prior stride is a level walking mode and the ambulation mode of the next stride is a ramp ascent mode.

12. The control unit of claim 1 , wherein the ambulation mode of the prior stride is a stair ascent mode and the ambulation mode of the next stride is a level walking mode.

13. The control unit of claim 1 , wherein the ambulation mode of the prior stride is a stair descent mode and the ambulation mode of the next stride is a level walking mode.

14. The control unit of claim 1 , wherein the ambulation mode of the prior stride is a ramp ascent mode and the ambulation mode of the next stride is a level walking mode.

15. The control unit of claim 1 , wherein the ambulation mode of the prior stride is a ramp descent mode and the ambulation mode of the next stride is a level walking mode.

16. The control unit of claim 1 , wherein the ambulation mode of the prior stride is the same as the ambulation mode of the next stride.

17. The control unit of claim 1 , wherein the pattern recognition controller is configured to predict the ambulation mode of the next stride with an error rate equal to or less than three percent.

18. The control unit of claim 1 , wherein the pattern recognition controller is configured to predict the ambulation mode of the next stride with an error rate equal to or less than two and a half percent.

19. The control unit of claim 1 , wherein the pattern recognition controller is configured to predict the ambulation mode of the next stride with an error rate equal to or less than two and a two tenths of a percent.

20. The control unit of claim 1 , wherein the feature data comprises the number of zero crossings in a signal collected from an electromyographic sensor.

21. The control unit of claim 1 , wherein the feature data comprises the number of slope sign changes in a signal collected from an electromyographic sensor.

22. The control unit of claim 1 , wherein the feature data comprises the waveform length of a signal collected from an electromyographic sensor.

23. The control unit of claim 1 , wherein the assistive device is an artificial knee assistive device.

24. The control unit of claim 1 , wherein the assistive device is an artificial knee prosthesis.

25. An assistive device comprising:
a. at least one powered joint;
b. a set of sensors coupled to the at least one powered joint;
c. a control unit, comprising a memory having thereon a pattern recognition controller, a feature database, and feature data contained in the feature database;
d. wherein the pattern recognition controller is configured to be trained on a portion of feature data that is labeled with a transition from an ambulation mode;
wherein the ambulation mode of the assistive device is configured to be predicted by the pattern recognition controller and wherein the assistive device is configured to provide a natural transition from operating in a prior stride to operating in the predicted ambulation mode in the next stride.

26. The assistive device of claim 25 :
a. wherein the feature database is configured to be updated with features derived from data collected from said sensors during level walking of the assistive device; and
b. wherein the pattern recognition controller is configured to be retrained on an updated feature database.

27. The assistive device of claim 25 :
a. wherein the pattern recognition controller comprises a classifier for each controlled ambulation mode.

28. The assistive device of claim 25 :
a. wherein the pattern recognition controller is configured to predict the next ambulation mode of the assistive device using features derived from the prior stride.

29. The assistive device of claim 25 :
a. wherein the feature database is configured to be updated with feature data derived from data collected from said sensors during the prior stride and labeled with a transition from an ambulation mode.

30. The assistive device of claim 25 :
b. wherein the pattern recognition controller comprises a classifier for each controlled ambulation mode;
c. wherein the pattern recognition controller is configured to predict the next ambulation mode of the assistive device using features derived from the prior stride;
d. wherein the feature database is configured to be updated with feature data derived from data collected from said sensors during the prior stride and labeled with a transition from an ambulation mode;
e. wherein the feature database is configured to be updated with features derived from data collected from said sensors during level walking of the assistive device;
f wherein the pattern recognition controller is configured to be retrained on an updated feature database; and

Assignees

• Chicago Rehabilitation Inst

Inventors

• Young Aaron
• Hargrove Levi

Classifications

• A61F2002/704

  computer-controlled, e.g. robotic control · CPC title

• A61F2/60

  Artificial legs or feet or parts thereof · CPC title

• G06N99/005 Primary

  Physics · mapped topic

• A61F2/70

  electrical · CPC title

• G06F17/30424

  Physics · mapped topic