Exoskeletons have the ability to improve locomotor performance for a wide range of individuals:​ they can improve the economy of normal walking, aid in load carriage for soldiers, assist individuals with walking disabilities, and serve as gait rehabilitation tools. Although it may seem obvious how exoskeletons should be developed to provide a benefit to the user, the complexity of the human neuromuscular system makes developing effective exoskeleton assistance strategies a challenge. Rather than using intuition to guide our attempts at the design and control of ankle exoskeletons, we need to garner a deeper understanding of how ankle exoskeletons affect locomotor coordination and utilize such findings to facilitate effective interaction between the device and the human.

This thesis details an iterative approach towards the development of ankle exoskeleton assistance strategies. We first performed a controlled experiment to observe the human response to specific assistance techniques. We then sought to explain the reasons for the observed responses by estimating muscle-tendon mechanics and energetics at the assisted joint using simulations of a musculoskeletal model. Through experimentation and simulation we found that individuals change and adapt their coordination patterns when walking with ankle exoskeletons, often in unexpected ways. Based on these findings, we developed and tested a novel ankle exoskeleton assistance strategy that adjusts exoskeleton behavior online in response to measured changes in the user. Such individualized, adaptive control approaches seem promising for discovering effective exoskeleton assistance strategies. Eventually we want to apply such strategies to populations with gait disabilities, but only once we have a better understanding of the mechanisms driving gait impairments. To that end, we designed and are conducting an experiment to investigate the relationship between features of post-stroke gait and energy economy. We expect our experimental findings to aid in the development of more accurate predictive models of human locomotion and to motivate new methods for developing assistive and rehabilitative techniques using robotic exoskeletons.