Limits...
Selective control of gait subtasks in robotic gait training: foot clearance support in stroke survivors with a powered exoskeleton.

Koopman B, van Asseldonk EH, van der Kooij H - J Neuroeng Rehabil (2013)

Bottom Line: The provided support did not result in reliance on the support for both groups.This enables the therapist to focus the support on the subtasks that are impaired, and leave the other subtasks up to the patient, encouraging him to participate more actively in the training.Additionally, the speed-dependent reference patterns provide the therapist with the tools to easily adapt the treadmill speed to the capabilities and progress of the patient.

View Article: PubMed Central - HTML - PubMed

Affiliation: Institute for Biomedical Technology and Technical Medicine MIRA, Department of Biomechanical Engineering, University of Twente, Enschede, The Netherlands. b.koopman@utwente.nl

ABSTRACT

Background: Robot-aided gait training is an emerging clinical tool for gait rehabilitation of neurological patients. This paper deals with a novel method of offering gait assistance, using an impedance controlled exoskeleton (LOPES). The provided assistance is based on a recent finding that, in the control of walking, different modules can be discerned that are associated with different subtasks. In this study, a Virtual Model Controller (VMC) for supporting one of these subtasks, namely the foot clearance, is presented and evaluated.

Methods: The developed VMC provides virtual support at the ankle, to increase foot clearance. Therefore, we first developed a new method to derive reference trajectories of the ankle position. These trajectories consist of splines between key events, which are dependent on walking speed and body height. Subsequently, the VMC was evaluated in twelve healthy subjects and six chronic stroke survivors. The impedance levels, of the support, were altered between trials to investigate whether the controller allowed gradual and selective support. Additionally, an adaptive algorithm was tested, that automatically shaped the amount of support to the subjects' needs. Catch trials were introduced to determine whether the subjects tended to rely on the support. We also assessed the additional value of providing visual feedback.

Results: With the VMC, the step height could be selectively and gradually influenced. The adaptive algorithm clearly shaped the support level to the specific needs of every stroke survivor. The provided support did not result in reliance on the support for both groups. All healthy subjects and most patients were able to utilize the visual feedback to increase their active participation.

Conclusion: The presented approach can provide selective control on one of the essential subtasks of walking. This module is the first in a set of modules to control all subtasks. This enables the therapist to focus the support on the subtasks that are impaired, and leave the other subtasks up to the patient, encouraging him to participate more actively in the training. Additionally, the speed-dependent reference patterns provide the therapist with the tools to easily adapt the treadmill speed to the capabilities and progress of the patient.

Show MeSH

Related in: MedlinePlus

Validation of the reconstructed ankle-height patterns. A: RMSE between the left ankle-height pattern and the reconstructed spline (black line), and the RMSE between the left ankle-height pattern and the right ankle-height pattern (gray). Both measures were averaged across subjects for each walking speed. The error bars indicate the standard deviation. B: Correlation between the left ankle-height pattern and the reconstructed spline (black line), and the correlation between the left ankle-height pattern and the right ankle-height pattern (gray line).Graph C and D show similar figures for the validation of the velocity profile.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3585791&req=5

Figure 6: Validation of the reconstructed ankle-height patterns. A: RMSE between the left ankle-height pattern and the reconstructed spline (black line), and the RMSE between the left ankle-height pattern and the right ankle-height pattern (gray). Both measures were averaged across subjects for each walking speed. The error bars indicate the standard deviation. B: Correlation between the left ankle-height pattern and the reconstructed spline (black line), and the correlation between the left ankle-height pattern and the right ankle-height pattern (gray line).Graph C and D show similar figures for the validation of the velocity profile.

Mentions: From the predicted key events, a reference ankle-height pattern was reconstructed for every subject and walking speed. We validated these patterns by comparing them with the measured patterns of the left leg (NB the regression equations were fitted on data of the right leg). The reconstructed patterns fitted the measured data well (see Figure 6). The RMSE, averaged across subjects, was around 1 cm for all walking speeds and the average correlation coefficient was larger than 0.95, for the low speeds, and showed even larger values for higher walking speeds. Since the error in predicting the key events is reflected in the reconstructed patterns, these RMSE values were close to the average RMSE in the prediction of the position of the key events (see Table 3, average position RMSE = 0.79 cm). The large correlation coefficients are in line with the small RMSE in the prediction of the timing of the key events. Also, the reconstructed velocity profiles matched the measured velocity profiles well (see Figure 6), though the correlations were a bit lower, especially for the lower velocities.


Selective control of gait subtasks in robotic gait training: foot clearance support in stroke survivors with a powered exoskeleton.

Koopman B, van Asseldonk EH, van der Kooij H - J Neuroeng Rehabil (2013)

Validation of the reconstructed ankle-height patterns. A: RMSE between the left ankle-height pattern and the reconstructed spline (black line), and the RMSE between the left ankle-height pattern and the right ankle-height pattern (gray). Both measures were averaged across subjects for each walking speed. The error bars indicate the standard deviation. B: Correlation between the left ankle-height pattern and the reconstructed spline (black line), and the correlation between the left ankle-height pattern and the right ankle-height pattern (gray line).Graph C and D show similar figures for the validation of the velocity profile.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3585791&req=5

Figure 6: Validation of the reconstructed ankle-height patterns. A: RMSE between the left ankle-height pattern and the reconstructed spline (black line), and the RMSE between the left ankle-height pattern and the right ankle-height pattern (gray). Both measures were averaged across subjects for each walking speed. The error bars indicate the standard deviation. B: Correlation between the left ankle-height pattern and the reconstructed spline (black line), and the correlation between the left ankle-height pattern and the right ankle-height pattern (gray line).Graph C and D show similar figures for the validation of the velocity profile.
Mentions: From the predicted key events, a reference ankle-height pattern was reconstructed for every subject and walking speed. We validated these patterns by comparing them with the measured patterns of the left leg (NB the regression equations were fitted on data of the right leg). The reconstructed patterns fitted the measured data well (see Figure 6). The RMSE, averaged across subjects, was around 1 cm for all walking speeds and the average correlation coefficient was larger than 0.95, for the low speeds, and showed even larger values for higher walking speeds. Since the error in predicting the key events is reflected in the reconstructed patterns, these RMSE values were close to the average RMSE in the prediction of the position of the key events (see Table 3, average position RMSE = 0.79 cm). The large correlation coefficients are in line with the small RMSE in the prediction of the timing of the key events. Also, the reconstructed velocity profiles matched the measured velocity profiles well (see Figure 6), though the correlations were a bit lower, especially for the lower velocities.

Bottom Line: The provided support did not result in reliance on the support for both groups.This enables the therapist to focus the support on the subtasks that are impaired, and leave the other subtasks up to the patient, encouraging him to participate more actively in the training.Additionally, the speed-dependent reference patterns provide the therapist with the tools to easily adapt the treadmill speed to the capabilities and progress of the patient.

View Article: PubMed Central - HTML - PubMed

Affiliation: Institute for Biomedical Technology and Technical Medicine MIRA, Department of Biomechanical Engineering, University of Twente, Enschede, The Netherlands. b.koopman@utwente.nl

ABSTRACT

Background: Robot-aided gait training is an emerging clinical tool for gait rehabilitation of neurological patients. This paper deals with a novel method of offering gait assistance, using an impedance controlled exoskeleton (LOPES). The provided assistance is based on a recent finding that, in the control of walking, different modules can be discerned that are associated with different subtasks. In this study, a Virtual Model Controller (VMC) for supporting one of these subtasks, namely the foot clearance, is presented and evaluated.

Methods: The developed VMC provides virtual support at the ankle, to increase foot clearance. Therefore, we first developed a new method to derive reference trajectories of the ankle position. These trajectories consist of splines between key events, which are dependent on walking speed and body height. Subsequently, the VMC was evaluated in twelve healthy subjects and six chronic stroke survivors. The impedance levels, of the support, were altered between trials to investigate whether the controller allowed gradual and selective support. Additionally, an adaptive algorithm was tested, that automatically shaped the amount of support to the subjects' needs. Catch trials were introduced to determine whether the subjects tended to rely on the support. We also assessed the additional value of providing visual feedback.

Results: With the VMC, the step height could be selectively and gradually influenced. The adaptive algorithm clearly shaped the support level to the specific needs of every stroke survivor. The provided support did not result in reliance on the support for both groups. All healthy subjects and most patients were able to utilize the visual feedback to increase their active participation.

Conclusion: The presented approach can provide selective control on one of the essential subtasks of walking. This module is the first in a set of modules to control all subtasks. This enables the therapist to focus the support on the subtasks that are impaired, and leave the other subtasks up to the patient, encouraging him to participate more actively in the training. Additionally, the speed-dependent reference patterns provide the therapist with the tools to easily adapt the treadmill speed to the capabilities and progress of the patient.

Show MeSH
Related in: MedlinePlus