Alternative human-in-the-loop exoskeleton assistance strategies: heuristic-based exoskeleton control for co-adaptive locomotor assistance

Researchers recently demonstrated that exoskeleton assistance strategies aimed at keeping the human “in-the-loop” can achieve significant improvements in locomotor performance. Given these findings, we sought to develop an alternative human-in-the-loop strategy rooted in the idea of co-adaptation. We formulated an algorithm, based on heuristics about effective interactions between the user and the device, to drive the evolution of an exoskeleton torque pattern. When implemented on bilateral ankle exoskeletons, the algorithm generated patterns of assistance that continuously responded to the changing coordination patterns of each naïve exoskeleton user and significantly reduced whole-body metabolic rate. Similar co-adaptive strategies will likely enable the discovery of effective assistance patterns for new devices and populations.