A Wearable Exoskeletal Rehabilitation Robot for Interactive Therapy - Biomechatronics in Medicine and Healthcare - page 26

Tab l e 2 . 4 Comparison of movement accuracy and smoothness at early (therapy
session 1) and late (therapy session 9) in therapy

Movement Accuracy (%) a Movement Smoothness b Target

Early in Therapy Late in Therapy Early in Therapy Late in Therapy

C1 7 5 − 2.39 − 1.10

C2 4 0 − 2.11 − 1.11

C3 18 27 − 3.15 − 1.13

C4 19 12 − 2.87 − 1.13

R1 8 3 − 3.2 − 2.95

R2 13 5 − 2.92 − 1.74

R3 27 20 − 2.75 − 4.47

R4 32 14 − 2.92 − 3.60

a The best value of accuracy is 0. The values in the table are rounded off to the nearest integer.
b The best value of smoothness is

− 1.

(2) Speed Metric (Rohrer et al. (2002))
(3) Peaks Metric (Rohrer et al. (2002))
(4) Mean Arrest Period Ratio Metric (Rohrer et al. (2002))
(5) Spectral Metric (Balasubramanian et al. (2009))

A summary of the changes that occurred in one of the subject's movement
kinematics towards the end of the study relative to the start of the therapy
is shown in Fig. 2.6 ; this figure shows the qualitative change in the subject's
endpoint trajectory while performing a voluntary reaching movement to target R2
( Fig. 2.5(b) ) . The corresponding quantitative changes in the movement accuracy
and movement smoothness for the 8 targets used in the therapy are summarized
in Table 2.4 . It is evident from Fig. 2.6 and Table 2.4 t hat the reaching ability of the
subject to the different targets had improved the subject was able to reach further
away from the body, and the reaching movement was much smoother than at the
start of the therapy.

2.8 CONCLUSION

A wearable exoskeleton robot for upper-extremity stroke rehabilitation, its associ-
ated controller and therapy modes, and a virtual reality based biofeedback system
were described. The system described above is currently being used in a clinical
study to test the device's capability for therapeutic training in a clinical setting.
Our current work is focused towards developing adaptive robotic therapy
modes that modulate the robotic assistance according to the subject's motor ability.
The robotic therapy modes developed for RUPERT are still very basic in nature,
and do not account for the differences in motor abilities between different subjects.
One of the important factors for driving motor learning is active voluntary partic-