[home] [Personal Program] [Help]

---

Session: Poster session I
Session starts: Thursday 24 January, 15:00

---

Bram Koopman (university of twente)
Edwin van Asseldonk (university of twente)
Herman van der Kooij (university of twente)

Abstract:
Introduction: During the last decade robotic gait training devices have been increasingly used as a clinical tool to provide neurological patients with task oriented, high intensity and repetitive training. These traditional robotic gait trainers are now expanding to mobile systems, that can be used outside the clinic and that can be used for several applications, such as rehabilitation, personal assistance or even human augmentation. Although a broad variety of control strategies exist for these devises, for most applications some sort of reference pattern, in order to determine the amount of assistance, is still required. These patterns are often based on pre-recorded trajectories from unimpaired volunteers. The major limitation of these patterns is that they are not publically available. Additionally, most patterns are recorded at a limited number of speeds, while the progress of the patients’ preferred walking speed can be as small as 0.1 km/h. In this study we will present a new method of constructing normative angular trajectories at different speeds. Methods: 15 elderly subjects walked on a treadmill at 7 different speeds, ranging from 0.5 to 5 km/h. Their angular trajectories (abduction/adduction of the hip and flexion/extension of hip and knee) were parameterized by defining different key events, which consisted of a selection of extreme values in position and velocity data. For each joint 6 key events were selected. Each key event was parameterized by an index, representing the percentage of the gait cycle at which the key event occurred, and its position, velocity and acceleration. Finally, the walking speed and body height dependency, of the parameters, were determined by regression models. To create the required subject- and speed dependent reference pattern, spines are fitted between the predicted key events. Results: For most of the key events, the index, position, velocity and acceleration were dependent on the walking speed. The body height contributed to the predictability of the regression models to a lesser extent. The results showed that the reconstructed reference patterns fitted the measured data well. The root mean square error (RMSE) between the reconstructed trajectories and the actual joint trajectories (averaged across subjects, joints and different walking speeds) was 2.6 degrees. As a reference; the RMSE between the right- and left actual joint trajectories was 2.1 degrees, indicating that the prediction error is close to the natural variation between left and right leg. Conclusion: In this study we derived and provided regression models that can be used to reconstruct patient-specific joint angle trajectories, based the subjects body height and walking speed. This will enable therapists, patients or users of robotic gait applications to easily change their gait speed, without the need to manually adjust the reference patterns.