Human walking has remarkable robustness and energy efficiency. In my thesis I sought to unravel the biomechanical mechanisms underlying walking via modeling approaches, by investigating what gait strategies are robust/energy efficient in walking models and whether these gait strategies represent human walking. Motivated by the increasing fall risk in the aging population, I also sought to accurately predict gait robustness or fall risk using stability measures.
In Chapter 2, I devised and validated several phase-dependent stability measures. The validation was realized by evaluating the correlations between phase-dependent stability measures and the largest allowable step perturbation that can be handled by two different walkers with varied configuration parameters. I found that the correlations differ between models and also between state space forms. The only exception was the divergence at foot strike, which displayed reasonably good correlations without state space dependency. Overall, it seems that even in simple walking models, phase-dependent stability measures and gait robustness are not one-to-one related.
In Chapter 3, I applied push and pull perturbations to the simple point-feet walker model at different phases of the single stance phase and investigated whether phase-dependent stability measures are related to phase-dependent gait robustness. I did not find strong association between them. Combining this with the previous assessment of gait robustness, I must conclude that phase-dependent stability measures fail to accurately predicting gait robustness.
In Chapter 4, I investigated whether humans employ a push-off control strategy to stabilize gait that has been suggested based on both modeling and experimental studies. According to Wang and Srinivasan, foot placement deviates in the same direction as the center-of-mass states in the preceding swing phase. When assuming this covariance to serve gait stabilization, the residual variances in foot placement from a linear regression model should be seen as foot placement ‘errors’. I analysed existing steady-state walking data of thirty participants and investigated whether these anterior-posterior foot placement errors were cor-related to push-off kinetic time series of the subsequent double stance phase. I found the trailing leg’s anterior-posterior ground reaction forces to be correlated to the foot placement errors at normal and slow speeds, with a largest peak of the mean correlations of up to 0.45, followed by the peak of the mean correlations for the trailing leg’s ankle moment of up to 0.39. These findings suggest that humans use a push-off control strategy to correct foot placement errors for gait stability in steady-state walking.
In Chapter 5, I investigated what gait strategies are energy efficient in a simple walking model. I investigated the effect of hip flexion and retraction actuation on energy efficiency, by finding the optimal stable periodic gaits in terms of the estimated metabolic cost of transport (MCOT). It turned out that the addition of hip flexion and retraction actuation could only reduce the estimated MCOT at medium speeds by up to 6% compared to ankle actuation only. At both low and high speeds the addition of hip actuation did not reduce the MCOT. These results suggest that in human walking hip flexion actuation does not improve energy efficiency but may serve other purposes such as improving gait robustness.
In Chapter 6, I reflected on the quantification of gait robustness and discussed the potential of using discrete stability measures like the divergence of foot strike for predicting gait robustness. I also provided an outlook on the opportunities and challenges of a mechanistic understanding of human walking. Despite these solid results my work should be considered only a first step towards prediction of gait robustness and energy efficiency, and a humble one towards a mechanistic understanding of human walking.
