Fig 1.
Reduction in metabolic rate from each device, compared to its peak positive power.
The vertical axis shows the percent reduction in gross whole-body metabolic rate, averaged over the gait cycle. The horizontal axis shows the peak instantaneous positive power of the device over the gait cycle, normalized by subject mass. The peak positive device power is the maximum, over time, of the sum of the positive power output of both actuators (left and right legs). Each crosshair provides the mean and standard deviation across 7 subjects for a single device. The hip flexion, knee flexion, and hip abduction devices provided greater metabolic savings than the ankle plantarflexion device (Tukey post-hoc test, p < 0.05).
Table 1.
Device performance and power.
Fig 2.
Device torque profiles compared to net joint moments.
The device actuator torque (blue), net joint moment generated by muscles in the presence of assistance (red), and net joint moment generated by muscles when not being assisted (black) are shown for each device. The moments are normalized by subject mass. Curves are averages over 7 subjects; shaded regions indicate ±1 standard deviation. The “muscles, no assistance” moment is the tracked net joint moment for both the no assistance and the assisted simulations; the “muscles, assisted” and “device, assisted” moments approximately sum to the “muscles, no assistance” moment. In many cases, the optimal device torque differed substantially from the net joint moment generated by muscles without assistance.
Fig 3.
Device power profiles compared to net joint powers.
The device actuator mechanical power (blue), net joint power generated by muscles in the presence of assistance (red), and net joint power generated by muscles when not being assisted (black) are shown for each device. All powers are normalized by subject mass. Curves are averages over 7 subjects; shaded regions indicate ±1 standard deviation. The “muscles, no assistance” power is the tracked net joint power for both the no assistance and assisted simulations; the “muscles, assisted” and “device, assisted” powers approximately sum to the “muscles, no assistance” power. Most devices exhibited substantially more positive work than negative work.
Fig 4.
Ankle plantarflexion device: Joint moments from key muscles.
The joint moments from the soleus (green), medial gastrocnemius (brown), and rectus femoris (purple) muscles about the (a) knee flexion/extension and (b) ankle dorsiflexion/plantarflexion degrees of freedom are shown without (solid) and with (dashed) the ankle plantarflexion device. For comparison, we also show the net moments from all muscles (gray) and from the device (dashed blue). The moments are normalized by subject mass and averaged across the 7 subjects. The ankle plantarflexion device largely bore the role of the soleus, while the gastrocnemius remained active to provide a knee flexion moment.
Fig 5.
Knee flexion device: Joint moments from key muscles.
The moments from the soleus (green), medial gastrocnemius (brown), rectus femoris (purple), and iliacus (pink) muscles about the (a) hip flexion/extension, (b) knee flexion/extension, and (c) ankle dorsiflexion/plantarflexion degrees of freedom. Muscle moments are shown without (solid) and with (dashed) the knee flexion device. For comparison, we also show the net moments from all muscles (gray) and from the device (dashed blue). The moments are normalized by subject mass and averaged across the 7 subjects. The device replaced much of the knee moment ordinarily generated by the gastrocnemius, which resulted in a decrease in gastrocnemius activity but an increase in the demand on the soleus to generate the plantarflexion moment. The rectus femoris countered the device to replace the iliacus and psoas (not shown) in providing a hip flexion moment.
Fig 6.
Hip flexion device: Joint moments from key muscles.
The moments from the soleus (green), medial gastrocnemius (brown), rectus femoris (purple), and iliacus (pink) muscles about the (a) hip external/internal rotation, (b) hip flexion/extension, (c) knee flexion/extension, and (d) ankle dorsiflexion/plantarflexion degrees of freedom are shown without (solid) and with (dashed) the hip flexion device. For comparison, we also show the net moments from all muscles (gray) and from the device (dashed blue). The moments are normalized by subject mass and averaged across the 7 subjects. The device mostly replaced the iliacus—which normally counters the net hip internal rotation moment—and psoas (not shown). The device also partially replaced the rectus femoris, allowing for a decrease in co-contraction at the knee. The soleus replaced some of the ankle moment that had been provided by the gastrocnemius.
Fig 7.
Hip abduction device: Joint moments from key muscles.
The moments from the soleus (green), medial gastrocnemius (brown), rectus femoris (purple), iliacus (pink), adductor longus (orange), and gluteus medius (anterior portion; light green) muscles about the (a) hip external/internal rotation, (b) hip adduction/abduction, (c) hip flexion/extension, (d) knee flexion/extension, and (e) ankle dorsiflexion/plantarflexion degrees of freedom are shown without (solid) and with (dashed) the hip flexion device. For comparison, we also show the net moments from all muscles (gray) and from the device (dashed blue). The moments are normalized by subject mass and averaged across the 7 subjects. Without assistance, a large fraction of the hip abduction moment was generated by the gluteus medius. The device replaced the gluteus medius in late stance, but the gluteus medius remained active in early stance to provide a hip internal rotation moment. In late stance, the adductor longus countered the device to provide hip flexion more economically than could the iliacus or psoas (not shown).