Biomechanics and energetics of walking in powered ankle exoskeletons using myoelectric control versus mechanically intrinsic control.

BACKGROUND: Controllers for assistive robotic devices can be divided into two main categories: controllers using neural signals and controllers using mechanically intrinsic signals. Both approaches are prevalent in research devices, but a direct comparison between the two could provide insight into their relative advantages and disadvantages. We studied subjects walking with robotic ankle exoskeletons using two different control modes: dynamic gain proportional myoelectric control based on soleus muscle activity (neural signal), and timing-based mechanically intrinsic control based on gait events (mechanically intrinsic signal). We hypothesized that subjects would have different measures of metabolic work rate between the two controllers as we predicted subjects would use each controller in a unique manner due to one being dependent on muscle recruitment and the other not. METHODS: The two controllers had the same average actuation signal as we used the control signals from walking with the myoelectric controller to shape the mechanically intrinsic control signal. The difference being the myoelectric controller allowed step-to-step variation in the actuation signals controlled by the user's soleus muscle recruitment while the timing-based controller had the same actuation signal with each step regardless of muscle recruitment. RESULTS: We observed no statistically significant difference in metabolic work rate between the two controllers. Subjects walked with 11% less soleus activity during mid and late stance and significantly less peak soleus recruitment when using the timing-based controller than when using the myoelectric controller. While walking with the myoelectric controller, subjects walked with significantly higher average positive and negative total ankle power compared to walking with the timing-based controller. CONCLUSIONS: We interpret the reduced ankle power and muscle activity with the timing-based controller relative to the myoelectric controller to result from greater slacking effects. Subjects were able to be less engaged on a muscle level when using a controller driven by mechanically intrinsic signals than when using a controller driven by neural signals, but this had no affect on their metabolic work rate. These results suggest that the type of controller (neural vs. mechanical) is likely to affect how individuals use robotic exoskeletons for therapeutic rehabilitation or human performance augmentation.

Motor modules during adaptation to walking in a powered ankle exoskeleton.

BACKGROUND: Modules of muscle recruitment can be extracted from electromyography (EMG) during motions, such as walking, running, and swimming, to identify key features of muscle coordination. These features may provide insight into gait adaptation as a result of powered assistance. The aim of this study was to investigate the changes (module size, module timing and weighting patterns) of surface EMG data during assisted and unassisted walking in an powered, myoelectric, ankle-foot orthosis (ankle exoskeleton). METHODS: Eight healthy subjects wore bilateral ankle exoskeletons and walked at 1.2 m/s on a treadmill. In three training sessions, subjects walked for 40 min in two conditions: unpowered (10 min) and powered (30 min). During each session, we extracted modules of muscle recruitment via nonnegative matrix factorization (NNMF) from the surface EMG signals of ten muscles in the lower limb. We evaluated reconstruction quality for each muscle individually using R2 and normalized root mean squared error (NRMSE). We hypothesized that the number of modules needed to reconstruct muscle data would be the same between conditions and that there would be greater similarity in module timings than weightings. RESULTS: Across subjects, we found that six modules were sufficient to reconstruct the muscle data for both conditions, suggesting that the number of modules was preserved. The similarity of module timings and weightings between conditions was greater then random chance, indicating that muscle coordination was also preserved. Motor adaptation during walking in the exoskeleton was dominated by changes in the module timings rather than module weightings. The segment number and the session number were significant fixed effects in a linear mixed-effect model for the increase in R2 with time. CONCLUSIONS: Our results show that subjects walking in a exoskeleton preserved the number of modules and the coordination of muscles within the modules across conditions. Training (motor adaptation within the session and motor skill consolidation across sessions) led to improved consistency of the muscle patterns. Subjects adapted primarily by changing the timing of their muscle patterns rather than the weightings of muscles in the modules. The results of this study give new insight into strategies for muscle recruitment during adaptation to a powered ankle exoskeleton.

Comparing neural control and mechanically intrinsic control of powered ankle exoskeletons.

There are an infinite number of ways to control an assistive robotic device; however, there is little consensus on which ways are better than others and why. We designed this study to compare the control of powered ankle exoskeletons using neural measurements to drive control versus that using mechanically intrinsic measurements. The controller driven by neural measurements was a dynamic gain proportional myoelectric controller using user's soleus muscle activity for an actuation signal. The controller driven by mechanically intrinsic measurements was a timing-based controller using detected heel-strikes of the user to appropriately time actuation. We designed these two controllers in such a way that both had the same average actuation signal and tested them with 8 healthy subjects. Results show no significant difference in metabolic work rate between the two controllers. Both controllers resulted in reductions in metabolic work rate of 19% below walking in the devices unpowered. We found that subjects using the timing-based mechanically intrinsic controller exhibited less positive and negative total ankle power than when using the dynamic gain proportional myoelectric controller. This finding was coupled with a reduction of 11.8% in soleus muscle activity. We believe these finding can have large implications for applications in rehabilitation and lend insight to when one controller is more appropriate to use than another.

Confidence in the curve: Establishing instantaneous cost mapping techniques using bilateral ankle exoskeletons.

Lower extremity robotic prostheses and exoskeletons can require tuning a large number of control parameters on a subject-specific basis to reduce users' metabolic power during locomotion. We refer to the functional relationship between control parameter configurations and users' metabolic power as the metabolic cost landscape. Standard practice for estimating a metabolic cost landscape, and thus identifying optimal parameter configurations, is to vary control parameters while measuring steady-state metabolic power during walking. This approach is time consuming, tedious, and inefficient. We have developed an instantaneous cost mapping analysis that allows for an estimate of the metabolic cost landscape without the explicit need for steady-state measurements. Here we present novel methods to quantify the confidence in an estimated metabolic cost landscape, allowing for an objective subject-specific comparison of protocols regardless of which metabolic analysis is used. We validated these techniques by estimating metabolic cost landscapes for healthy subjects walking with bilateral robotic ankle exoskeletons using a standard practice protocol and two innovative protocols that use an instantaneous cost mapping analysis. All cost landscapes were a function of the devices' actuation timing. Results showed that for this device a protocol using an instantaneous cost mapping analysis could accurately identify optimal parameter configurations in 20 min, where the standard practice protocol required 42 min. Additionally, using an instantaneous cost mapping analysis with the standard practice's parameter exploration significantly improved fit confidence. These methods could greatly improve real-time optimization of robotic assistive devices or studies focused on biomechanical manipulations of locomotion. NEW & NOTEWORTHY: We are presenting novel subject-specific metabolic cost landscape confidence analyses. These confidence analyses can greatly improve experimental design, intersubject analysis, and the comparison of landscape mapping protocols. We validated these methods by mapping subject-specific metabolic cost landscapes using bilateral ankle exoskeletons and are presenting the first full study using instantaneous cost mapping techniques to optimally tune an assistive robotic device.

Learning to walk with an adaptive gain proportional myoelectric controller for a robotic ankle exoskeleton.

BACKGROUND: Robotic ankle exoskeletons can provide assistance to users and reduce metabolic power during walking. Our research group has investigated the use of proportional myoelectric control for controlling robotic ankle exoskeletons. Previously, these controllers have relied on a constant gain to map user's muscle activity to actuation control signals. A constant gain may act as a constraint on the user, so we designed a controller that dynamically adapts the gain to the user's myoelectric amplitude. We hypothesized that an adaptive gain proportional myoelectric controller would reduce metabolic energy expenditure compared to walking with the ankle exoskeleton unpowered because users could choose their preferred control gain. METHODS: We tested eight healthy subjects walking with the adaptive gain proportional myoelectric controller with bilateral ankle exoskeletons. The adaptive gain was updated each stride such that on average the user's peak muscle activity was mapped to maximal power output of the exoskeleton. All subjects participated in three identical training sessions where they walked on a treadmill for 50 minutes (30 minutes of which the exoskeleton was powered) at 1.2 ms(-1). We calculated and analyzed metabolic energy consumption, muscle recruitment, inverse kinematics, inverse dynamics, and exoskeleton mechanics. RESULTS: Using our controller, subjects achieved a metabolic reduction similar to that seen in previous work in about a third of the training time. The resulting controller gain was lower than that seen in previous work (ß=1.50±0.14 versus a constant ß=2). The adapted gain allowed users more total ankle joint power than that of unassisted walking, increasing ankle power in exchange for a decrease in hip power. CONCLUSIONS: Our findings indicate that humans prefer to walk with greater ankle mechanical power output than their unassisted gait when provided with an ankle exoskeleton using an adaptive controller. This suggests that robotic assistance from an exoskeleton can allow humans to adopt gait patterns different from their normal choices for locomotion. In our specific experiment, subjects increased ankle power and decreased hip power to walk with a reduction in metabolic cost. Future exoskeleton devices that rely on proportional myolectric control are likely to demonstrate improved performance by including an adaptive gain.